ThesisPDF Available

Exploring and evaluating the cross-cultural applicability of compassion-based approaches

Authors:

Abstract and Figures

The concept and benefits of practicing compassion have been recognised and discussed in the contemplative traditions for thousands of years. However, it is within the last two to three decades, that research and psychotherapy have shown an increased interest in integrating compassion for addressing mental health difficulties and increased well-being. Although heavily influenced by Buddhist philosophy and Eastern traditions, compassion related studies and interventions are mostly developed and applied in the Western communities. In fact, compassion-based studies are particularly scarce in the Asian context. Therefore, whilst briefly outlining the theories and existing compassion-based interventions, this thesis explored the cross-cultural applicability of compassion-based interventions in the Asian communities. A rigorous qualitative investigation discussed that compassion is a culturally embraced concept in Sri Lanka, a Buddhist influenced, collectivistic Asian community, and discussed the challenges Sri Lankan participants (n = 10) experience when practicing compassion. Participants discussed that showing compassion to others was easier than showing compassion to themselves, whilst religion, society, and upbringing influenced these experiences. To understand whether these compassionate experiences are similar across cultures, a cross-sectional quantitative study was conducted among Sri Lankan (n = 149) and UK (n = 300) participants. This study indicated that some similarities (e.g., compassion to and from others, depression, anxiety) and some differences (e.g., self-compassion and self-reassurance, fears of compassion and external shame were higher in the Sri Lankan group, and social safeness was higher in the UK group) existed in the levels of compassion, and facilitators and inhibitors of compassion across the two samples. Therefore, it was important to note that the impact of compassion-based interventions might have cross-cultural differences. To test this, a longitudinal Compassionate Mind Training was implemented among Sri Lankan (n = 21) and UK participants (n = 73), which produced promising results towards increasing compassion for the self and others, along with significant reductions in distress and improvements in well-being in participants across both countries. Thus, this thesis suggests that although research is limited in exploring the cross-cultural applicability of compassion, compassion-based interventions can be used effectively in the Asian communities.
Content may be subject to copyright.
University of Southampton Research Repository
Copyright © and Moral Rights for this thesis and, where applicable, any accompanying data are
retained by the author and/or other copyright owners. A copy can be downloaded for personal
non-commercial research or study, without prior permission or charge. This thesis and the
accompanying data cannot be reproduced or quoted extensively from without first obtaining
permission in writing from the copyright holder/s. The content of the thesis and accompanying
research data (where applicable) must not be changed in any way or sold commercially in any
format or medium without the formal permission of the copyright holder/s.
When referring to this thesis and any accompanying data, full bibliographic details must be given,
e.g.
Thesis: Author (Year of Submission) "Full thesis title", University of Southampton, name of the
University Faculty or School or Department, PhD Thesis, pagination.
Data: Author (Year) Title. URI [dataset]
University of Southampton
Faculty of Environmental and Life Sciences
School of Psychology
Exploring and Evaluating the Cross-Cultural Applicability of Compassion-Based
Approaches
by
Lasara Kariyawasam
MSc in Foundations of Clinical Psychology, BSc (Hons) Psychology
ORCID ID 0000-0002-2207-2182
Thesis for the degree of Doctor of Philosophy
November 2022
Abstract
The concept and benefits of practicing compassion have been recognised and discussed in the
contemplative traditions for thousands of years. However, it is within the last two to three
decades, that research and psychotherapy have shown an increased interest in integrating
compassion for addressing mental health difficulties and increased well-being. Although heavily
influenced by Buddhist philosophy and Eastern traditions, compassion related studies and
interventions are mostly developed and applied in the Western communities. In fact, compassion-
based studies are particularly scarce in the Asian context. Therefore, whilst briefly outlining the
theories and existing compassion-based interventions, this thesis explored the cross-cultural
applicability of compassion-based interventions in the Asian communities. A rigorous qualitative
investigation discussed that compassion is a culturally embraced concept in Sri Lanka, a Buddhist
influenced, collectivistic Asian community, and discussed the challenges Sri Lankan participants (n
= 10) experience when practicing compassion. Participants discussed that showing compassion to
others was easier than showing compassion to themselves, whilst religion, society, and upbringing
influenced these experiences. To understand whether these compassionate experiences are
similar across cultures, a cross-sectional quantitative study was conducted among Sri Lankan (n =
149) and UK (n = 300) participants. This study indicated that some similarities (e.g., compassion to
and from others, depression, anxiety) and some differences (e.g., self-compassion and self-
reassurance, fears of compassion and external shame were higher in the Sri Lankan group, and
social safeness was higher in the UK group) existed in the levels of compassion, and facilitators
and inhibitors of compassion across the two samples. Therefore, it was important to note that the
impact of compassion-based interventions might have cross-cultural differences. To test this, a
longitudinal Compassionate Mind Training was implemented among Sri Lankan (n = 21) and UK
participants (n = 73), which produced promising results towards increasing compassion for the
self and others, along with significant reductions in distress and improvements in well-being in
participants across both countries. Thus, this thesis suggests that although research is limited in
exploring the cross-cultural applicability of compassion, compassion-based interventions can be
used effectively in the Asian communities.
Table of Contents
i
Table of Contents
Table of Contents ........................................................................................................... i
Table of Tables............................................................................................................. ix
Table of Figures ........................................................................................................... xi
Research Thesis: Declaration of Authorship ................................................................ xiii
Acknowledgements .................................................................................................... xv
Definitions and Abbreviations ................................................................................... xvii
Chapter 1 An Introduction to Compassion and its Cross-Cultural Applicability, and the
Rationale, Aims of the Thesis, and Chapter Summaries ............................... 1
1.1 Definitions and History of Compassion ...................................................................... 1
1.2 Theoretical Models of Compassion ............................................................................ 4
1.2.1 Neff’s Theory of Compassion ............................................................................. 4
1.2.1.1 Criticisms of Neff’s Model ....................................................................... 5
1.2.2 Gilbert’s Model of Compassion .......................................................................... 6
1.2.2.1 Social Mentality Theory (SMT) ................................................................ 7
1.2.2.2 The Tripartite Model of Affective Regulation ......................................... 7
1.2.2.3 The Three Flows of Compassion.............................................................. 8
1.2.2.4 Facilitators and Inhibitors of Compassion ............................................... 9
1.2.2.5 Criticisms of Gilbert’s Model ................................................................. 11
1.2.3 Compassion-Based Interventions..................................................................... 11
1.2.3.1 Mindful Self-Compassion (MSC) ............................................................ 13
1.2.3.2 Compassion Focused Therapy (CFT)...................................................... 13
1.2.3.3 Compassion Cultivation Training (CCT) ................................................. 15
1.2.3.4 Cognitively Based Compassion Training (CBCT) .................................... 15
1.2.3.5 Compassion Meditations (CM) and Loving-Kindness Meditations (LKM)
............................................................................................................... 15
1.2.3.6 Cultivating Emotional Balance (CEB) ..................................................... 16
1.2.3.7 Similarities and Differences of Compassion-Based Interventions ........ 16
1.2.4 Cross-Cultural Applicability of Compassion ..................................................... 17
1.2.4.1 Cultural Differences of Compassion ...................................................... 18
1.3 Rationale for the Thesis............................................................................................ 20
Table of Contents
ii
1.3.1 Thesis Aims ....................................................................................................... 20
1.4 Summary of Chapters ............................................................................................... 22
Chapter 2 Methodological Approaches used in the Thesis ......................................... 27
2.1 Using a Mixed Methods Approach ........................................................................... 27
2.1.1 Paper 1: Meta-Analysis..................................................................................... 28
2.1.1.1 Analysis .................................................................................................. 30
2.1.2 Paper 2: Qualitative Study................................................................................ 30
2.1.2.1 Reflexivity .............................................................................................. 31
2.1.2.2 Interpretative Phenomenological Analysis ........................................... 32
2.1.2.3 Semi-Structured Interviewing ............................................................... 34
2.1.2.4 Quality in Qualitative Research ............................................................. 35
2.1.3 Paper 3: Quantitative Study ............................................................................. 36
2.1.3.1 Analysis .................................................................................................. 36
2.1.4 Paper 4: Experimental Study ............................................................................ 37
2.1.4.1 Randomised Controlled Trials (RCT) ...................................................... 38
2.1.4.2 Analysis .................................................................................................. 38
2.1.5 Considering Context ......................................................................................... 39
2.1.5.1 Questionnaire Use ................................................................................. 39
2.1.5.2 Internet Use ........................................................................................... 40
Chapter 3 Compassion-Based Interventions in Asian Communities: A Meta-Analysis of
Randomised Controlled Trials ................................................................... 43
3.1 Background............................................................................................................... 44
3.1.1 Models and Measures of Compassion ............................................................. 44
3.1.2 Compassion-Based Interventions..................................................................... 46
3.1.3 Rationale for the Meta-Analysis ....................................................................... 47
3.2 Method ..................................................................................................................... 48
3.2.1 Protocol and Registration................................................................................. 48
3.2.2 Eligibility Criteria .............................................................................................. 48
3.2.3 Search Strategy................................................................................................. 49
3.2.4 Data Extraction ................................................................................................. 50
Table of Contents
iii
3.2.5 Analysis Strategy .............................................................................................. 50
3.2.6 Risk of Bias within Studies ................................................................................ 50
3.3 Results ...................................................................................................................... 51
3.3.1 Systematic Search Results ................................................................................ 51
3.3.2 Quantitative Results ......................................................................................... 55
3.3.2.1 Intervention and Participant Characteristics ........................................ 55
3.3.3 Compassion Outcomes ..................................................................................... 62
3.3.3.1 Compassion-Based Interventions Compared to Waitlist Control Groups
............................................................................................................... 62
3.3.3.2 Compassion-Based Interventions Compared to Active Control Groups
............................................................................................................... 63
3.3.4 Risk of Bias within Studies ................................................................................ 63
3.3.5 Risk of Bias across Studies ................................................................................ 64
3.4 Discussion ................................................................................................................. 64
3.4.1 Strengths and Limitations ................................................................................ 65
3.4.2 Clinical Implications .......................................................................................... 68
3.4.3 Conclusion ........................................................................................................ 69
Chapter 4 Views and Experiences of Compassion in Sri Lankan Students: An
Exploratory Qualitative Study ................................................................... 71
4.1 Background............................................................................................................... 72
4.1.1 Theory and practice of compassion ................................................................. 73
4.1.2 Cultural influence ............................................................................................. 73
4.1.3 Rationale for the present study ....................................................................... 75
4.2 Method ..................................................................................................................... 75
4.2.1 Design and participants .................................................................................... 75
4.2.2 Interview structure ........................................................................................... 76
4.2.3 Procedure ......................................................................................................... 77
4.2.4 Data analysis ..................................................................................................... 78
4.3 Results ...................................................................................................................... 79
4.3.1 Superordinate theme: what compassion means to me ................................... 80
Table of Contents
iv
4.3.1.1 Subordinate theme: sympathetic consideration towards suffering ..... 80
4.3.1.2 Subordinate Theme: self and others: It is not the same ....................... 82
4.3.2 Superordinate Theme: What I make of it ........................................................ 82
4.3.2.1 Subordinate theme: positive vibes and genuine motivation ................ 83
4.3.2.2 Subordinate theme: Obligations and exhaustion ................................. 84
4.3.2.3 Subordinate theme: disclosure: nobody should feel bad about my life
............................................................................................................... 85
4.3.3 Superordinate theme: compassion through facilitators and inhibitors .......... 85
4.3.3.1 Subordinate theme: ‘god is good’: religion in shaping compassion ..... 86
4.3.3.2 Subordinate theme: being there for one another ................................ 86
4.3.3.3 Subordinate theme: compassion is conditional .................................... 87
4.3.3.4 Subordinate theme: society as an inhibitor .......................................... 88
4.4 Discussion ................................................................................................................. 90
4.4.1 What compassion means to me ....................................................................... 90
4.4.2 What I make of it? ............................................................................................ 91
4.4.3 Compassion through facilitators and inhibitors ............................................... 94
4.4.4 Strengths and limitations ................................................................................. 98
4.4.5 Implications for future research ...................................................................... 99
4.4.6 Conclusion ...................................................................................................... 100
Chapter 5 A Cross-Cultural Exploration of Compassion, and Facilitators and Inhibitors
of Compassion in Sri Lankan and UK People ............................................ 101
5.1 Background............................................................................................................. 102
5.1.1 Theoretical Perspective of Compassion ......................................................... 102
5.1.2 Inhibitors and Facilitators of Compassion ...................................................... 103
5.1.3 Compassion across Cultures........................................................................... 104
5.1.4 Rationale for the Present Study ..................................................................... 106
5.2 Method ................................................................................................................... 107
5.2.1 Design and Participants .................................................................................. 107
5.2.2 Measures ........................................................................................................ 107
5.2.3 Procedure ....................................................................................................... 108
5.2.4 Data Analysis Plan .......................................................................................... 108
Table of Contents
v
5.3 Results .................................................................................................................... 109
5.3.1 Participants ..................................................................................................... 109
5.3.2 Aim 1: Testing compassion, and inhibitors and facilitators of compassion
between Sri Lankan and UK participants ....................................................... 109
5.3.3 Aim 2: Predictors of the three flows of compassion in the UK and Sri Lankan
participants..................................................................................................... 113
5.3.3.1 Predictors of Self-Compassion in Sri Lankan and UK participants ...... 113
5.3.3.2 Predictors of Compassion to Others in Sri Lankan and UK participants
............................................................................................................. 115
5.3.3.3 Predictors of Compassion from Others in Sri Lankan and UK
participants .......................................................................................... 117
5.4 Discussion ............................................................................................................... 119
5.4.1 Self-Compassion ............................................................................................. 120
5.4.2 Compassion to Others .................................................................................... 122
5.4.3 Compassion from Others ............................................................................... 124
5.4.4 Strengths and Limitations .............................................................................. 126
5.4.5 Clinical Implications and Recommendations for Future Research ................ 127
5.4.6 Conclusion ...................................................................................................... 128
Chapter 6 Exploring the Cross-Cultural Applicability of a Brief Compassionate Mind
Training in Sri Lankan and UK People ...................................................... 129
6.1 Background............................................................................................................. 130
6.2 Method ................................................................................................................... 134
6.2.1 Design and Participants .................................................................................. 134
6.2.2 Measures ........................................................................................................ 136
6.2.2.1 Demographic Questions ...................................................................... 136
6.2.2.2 Compassionate Engagement and Actions Scales (CEAS) .................... 136
6.2.2.3 Fears of Compassion Scales (FOCS) ..................................................... 136
6.2.2.4 Forms of Self-Criticising/Attacking and Self-Reassuring Scale (FSCRS)
............................................................................................................. 136
6.2.2.5 Others as Shamer Scale (OAS) ............................................................. 137
6.2.2.6 Social Safeness and Pleasure Scale (SSPS) .......................................... 137
6.2.2.7 Generalised Anxiety Disorder-7 Scale (GAD-7) ................................... 137
Table of Contents
vi
6.2.2.8 Patient Health Questionnaire (PHQ-9) ................................................ 137
6.2.2.9 The Warwick-Edinburgh Mental Well-Being Scale (WEMWBS) .......... 137
6.2.2.10 CMT Engagement Feedback Questions............................................... 138
6.2.3 CMT: Design and Tasks ................................................................................... 138
6.2.4 Procedure ....................................................................................................... 140
6.2.5 Statistical Analysis .......................................................................................... 141
6.3 Results .................................................................................................................... 142
6.3.1 Participants ..................................................................................................... 142
6.3.2 Differences Between Countries and Groups at Baseline (T1) ........................ 142
6.3.3 Efficacy of the CMT: Sri Lankan Participants .................................................. 147
6.3.4 Efficacy of the CMT: UK Participants .............................................................. 153
6.3.5 Maintenance of Efficacy of the CMT at Follow-Up: Sri Lankan Participants . 159
6.3.6 Maintenance of Efficacy of the CMT at Follow-Up: UK Participants ............. 163
6.3.7 Feedback on the CMT Engagement ............................................................... 167
6.4 Discussion ............................................................................................................... 169
6.4.1 Efficacy of the CMT on Compassion ............................................................... 170
6.4.1.1 CMT on Compassion in Sri Lankan Participants .................................. 170
6.4.1.2 CMT on Compassion in UK Participants .............................................. 170
6.4.2 Efficacy of the CMT on Facilitators and Inhibitors of Compassion ................ 171
6.4.2.1 Differences in the Facilitators and Inhibitors of Compassion in Sri
Lankan Participants ............................................................................. 171
6.4.2.2 Differences in the Facilitators and Inhibitors of Compassion in UK
Participants .......................................................................................... 171
6.4.3 CMT Engagement Feedback Questions ......................................................... 176
6.4.4 Strengths and Limitations .............................................................................. 177
6.4.5 Clinical Implications ........................................................................................ 178
6.4.6 Conclusion ...................................................................................................... 179
Chapter 7 General Discussion ................................................................................. 181
7.1 Paper Findings ........................................................................................................ 181
Table of Contents
vii
7.1.1 Paper 1: A meta-analysis to explore whether compassion-based interventions
lead to increased levels of compassion in people living in Asian communities.
181
7.1.1.1 Additional Strengths and Limitations .................................................. 181
7.1.2 Paper 2: A qualitative investigation to explore the views and lived experiences
of compassion in Sri Lankan students. ........................................................... 183
7.1.2.1 Additional Strengths and Limitations .................................................. 186
7.1.3 Paper 3: An exploration of cross-cultural differences in compassion, and the
facilitators and inhibitors of compassion between Sri Lankan and UK
participants..................................................................................................... 187
7.1.3.1 Additional Strengths and Limitations .................................................. 189
7.1.4 Paper 4: A brief Compassionate-Mind Training (CMT) to increase compassion
in a cross-cultural group of Sri Lankan and UK people. ................................. 190
7.1.4.1 Additional Strengths and Limitations .................................................. 191
7.1.4.2 Recommendations for Intervention Development ............................. 194
7.2 Theoretical Implications ......................................................................................... 196
7.3 Research and Clinical Implications ......................................................................... 199
7.4 Conclusion .............................................................................................................. 204
Appendix A Assessment of Risk of Bias within Studies (Paper 1) .................................. 207
Appendix B Participant Information Sheet (Paper 2)…………………………………………….. 205
Appendix C Participant Consent Form (Paper 2) ........................................................ 219
Appendix D Debriefing Form (Paper 2) ...................................................................... 221
Appendix E Interview Guide (Paper 2) ....................................................................... 223
Appendix F IPA Coding Draft (Paper 2) ...................................................................... 227
Appendix G Information Sheet (Paper 3) ................................................................... 263
Appendix H Consent Form (Paper 3) .......................................................................... 269
Appendix I Debriefing Sheet (Paper 3) ...................................................................... 271
Appendix J Information Sheet and Consent Form (Paper 4)....................................... 273
Appendix K Debriefing Sheet (Paper 4) ...................................................................... 281
Appendix L Questionnaires used in Papers 3 and 4 .................................................... 283
L.1 Demographic Questionnaire .................................................................................. 283
Table of Contents
viii
L.2 The compassionate Engagement and Action Scales .............................................. 283
L.3 Fears of Compassion Scale ..................................................................................... 291
L.4 The Forms of Self-Criticising/Attacking & Self-Reassuring Scale (FSCRS) .............. 295
L.5 Other as Shamer Scale (OAS) ................................................................................. 297
L.6 Social Safeness and Pleasure Scale ........................................................................ 299
L.7 GAD-7: Anxiety ....................................................................................................... 300
L.8 PHQ-9: Depression ................................................................................................. 301
L.9 The Warwick-Edinburgh Mental Well-being Scale (WEMWBS) ............................. 303
Appendix M Study Advert and Email distributed for Participants (Paper 4) .............. 305
Appendix N CMT Participation Statements on the Qualtrics Website for each stage
(Paper 4) ................................................................................................ 307
Appendix O: Consort checklist of information to include when reporting a randomised
trial…………………………………………………………………………………………………………………………..313
List of References .................................................................................................... 3177
Table of Tables
ix
Table of Tables
Table 3.1. Inclusion and Exclusion Criteria for the Review. ........................................................ 49
Table 3.2. Intervention Characteristics........................................................................................ 57
Table 3.3. Post intervention effects on compassion and self-compassion. ................................ 62
Table 4.1. Interview Questions. ................................................................................................... 77
Table 4.2. Superordinate Themes and Subordinate Themes. ..................................................... 80
Table 5.1. Means, Standard Deviations and ANCOVA Results. ................................................. 110
Table 5.2. Regression Results for Predictors of Self-Compassion in Sri Lankan Participants. .. 113
Table 5.3. Regression Results for Predictors of Self-Compassion in UK Participants ............... 114
Table 5.4. Regression Results for Predictors of Compassion to Others in Sri Lankan Participants.
....................................................................................................................... 116
Table 5.5. Regression Results for Predictors of Compassion to Others in UK Participants. ..... 117
Table 5.6. Regression Results for Predictors of Compassion from Others in Sri Lankan Participants.
....................................................................................................................... 118
Table 5.7. Regression Results for Predictors of Compassion from Others in UK Participants. . 119
Table 6.1. Two-Week CMT as Informed by Matos et al.'s (2017a) Study Manual. ................... 139
Table 6.2. Timeline across the Two Groups............................................................................... 140
Table 6.3. Demographic Information of Sri Lankan and UK Participants. ................................. 142
Table 6.4. Sample Characteristics at Baseline-1. ....................................................................... 144
Table 6.5. Pre-Post Intention to Treat Analyses of the Sri Lankan Sample. .............................. 148
Table 6.6. Pre-Post Per Protocol Analyses of the Sri Lankan Sample. ....................................... 150
Table 6.7. Pre-Post Intention to Treat Analyses of the UK Sample. .......................................... 154
Table 6.8. Pre-Post Per Protocol Analyses of the UK Sample. ................................................... 156
Table 6.9. Changes across Time in the Sri Lankan Sample. ....................................................... 160
Table 6.10. Changes across Time in the UK Sample. ................................................................. 164
Table 6.11. CMT Engagement Feedback Questions and the Most Common Answers. ............ 168
Table of Figures
xi
Table of Figures
Figure 1. PRISMA Flow Diagram of Study Selection .................................................................... 53
Figure 2. The effect of compassion-based interventions with wait-list control groups on self-
compassion. ..................................................................................................... 63
Figure 3. The effect of compassion-based interventions with active control groups on self-
compassion. ..................................................................................................... 63
Figure 4. Risk of bias graph across studies. ................................................................................. 64
Figure 5. Participant flow chart for baseline, post-intervention and the two-week follow-up...133
Research Thesis: Declaration of Authorship
xiii
Research Thesis: Declaration of Authorship
I, Lasara Kariyawasam declare that this thesis and the work presented in it are my own and has
been generated by me as the result of my own original research.
Exploring and Evaluating the Cross-Cultural Applicability of Compassion-Based Approaches
I confirm that:
1. This work was done wholly or mainly while in candidature for a research degree at this
University;
2. Where any part of this thesis has previously been submitted for a degree or any other
qualification at this University or any other institution, this has been clearly stated;
3. Where I have consulted the published work of others, this is always clearly attributed;
4. Where I have quoted from the work of others, the source is always given. With the exception
of such quotations, this thesis is entirely my own work;
5. I have acknowledged all main sources of help;
6. Where the thesis is based on work done by myself jointly with others, I have made clear
exactly what was done by others and what I have contributed myself;
7. Parts of this work have been published as:
Kariyawasam, L., Ononaiye, M., Irons, C., Stopa, L., & Kirby, S.E. (2021). Views and experiences of
compassion in Sri Lankan students: An exploratory qualitative study. PLoS ONE, 16(11):
e0260475. https://doi.org/10.1371/journal.pone.0260475
Kariyawasam, L., Ononaiye, M., Irons, C. & Kirby, S.E. (2022). A cross-cultural exploration of
compassion, and facilitators and inhibitors of compassion in UK and Sri Lankan people.
Global Mental Health, 112. https://doi.org/10.1017/gmh.2022.10
Signature: Date: 24/11/2022
Acknowledgements
xv
Acknowledgements
First and foremost, I am extremely grateful to my supervisors Dr. Margo Ononaiye, Dr. Sarah
Kirby, and Dr. Chris Irons for their invaluable support, advise, and expertise during my PhD. I thank
Professor Lusia Stopa for co-supervising my first PhD paper. I would like to personally thank Dr.
Margo Ononaiye and Dr. Sarah Kirby for going above and beyond their position as my supervisors
for ensuring my well-being particularly during the COVID-19 period. This PhD journey has been
nothing short of a joyful ride for me because of the support and kindness I received from the two
of them.
I would also like to thank and acknowledge Dr. Jin Zhang, Dr. Samuele Cortese, Dr. Daniel Schoth,
and Ms. Vicky Fenerty for their technical support during my study. I would also like to thank all
the Sri Lankan and UK participants for their time and participation without whom this thesis
would not have been a success. I would like to take this opportunity to thank all my wonderful
friends, PhD colleagues, and flatmates who made my time in the UK a truly enjoyable one for me.
Finally, I would like to express my gratitude to my wonderful parents, brother, uncle and aunt,
and my partner for their tremendous love and support.
Definitions and Abbreviations
xvii
Definitions and Abbreviations
SCS: Self-Compassion Scale
SMT: Social Mentality Theory
CEAS: Compassionate Engagement and Action Scales
MSC: Mindful Self-Compassion
CFT: Compassion Focused Therapy
CMT: Compassionate Mind Training
CCT: Compassion Cultivation Training
CBCT: The Cognitively Based Compassion Training
CM: Compassion Meditations
LKM: Loving-Kindness Meditations
CEB: Cultivating Emotional Balance
RCT: Randomised Controlled Trial
IPA: Interpretative Phenomenological Analysis
C-MT: Compassion Mindfulness Therapy
Individualistic culture: A social context in which people have an independent concept of self,
pursue their goals independently from their ingroups, their social behaviors are driven by
personal attitude, values, and belief, and individuals make their own choice of social relationships
based on individual rationalisation (Bhawuk, 2017).
Collectivistic culture: A social context in which people’s concept of self is and goals are
interdependent and compatible with ingroups, and their social behaviours are norm driven and
guided by their social exchange with other people (Bhawuk, 2017).
Chapter 1: Introduction
1
Chapter 1 An Introduction to Compassion and its Cross-
Cultural Applicability, and the Rationale, Aims of the
Thesis, and Chapter Summaries
1.1 Definitions and History of Compassion
Compassion has been defined in a range of different ways and the Oxford English
dictionary recognises the word compassion as stemming from the Latin word compati, which
means “to suffer with” (Simpson & Weiner, 1989. p. 1340). Literature identifies with this notion as
most definitions view compassion as an antidote to suffering (e.g., Lampert, 2005; Richard, 2015).
For instance, Lazarus (1991, p.289) defined compassion as “being moved by another’s suffering
and wanting to help”, while Goetz and colleagues (2010, p.351) viewed it as “a feeling that arises
in witnessing another’s suffering and that motivates a subsequent desire to help”. The Dalai Lama
(1995) viewed compassion as an openness to the suffering of others with a commitment to
relieve it. In addition, two of the most popular scholars of compassion research, Kristin Neff
(2003a) and Paul Gilbert (2009a) provided their own definitions for self-compassion and
compassion. Neff (2003a, p.87) viewed self-compassion as compassion turned inward and defined
it as “being touched by and open to one’s own suffering, not avoiding or disconnecting from it,
generating the desire to alleviate one’s suffering and to heal oneself with kindness”. Gilbert and
Choden (2013, p.94) defined compassion as “a sensitivity to suffering in self and others with a
commitment to try to alleviate and prevent it”. Thus, definitions of compassion are varied (Strauss
et al., 2016), with some researchers suggesting it as an emotion (Goetz et al., 2010), a motivation
(Gilbert, 2014), or a multidimensional construct (Jazaieri et al., 2013). Drawing together on these
various definitions from Buddhist philosophy and Western psychological perspectives, Strauss et
al. (2016, p.25) identified five components of compassion, “recognition of suffering;
understanding its universality; feeling sympathy, empathy, or concern for those who are
suffering...; tolerating the distress associated with the witnessing of suffering; and motivation to
act or acting to alleviate the suffering”.
Chapter 1: Introduction
2
As with the various definitions, various possible underpinnings of compassion have also
been identified over the years. Some of the most agreed upon underpinnings of compassion are
evolutionary theory (Gilbert, 2000), attachment systems (Bowlby, 1969, 1982), biological
approaches (Goetz et al., 2010), and Buddhist philosophy (Dalai Lama, 1995).
Roots of compassion from an evolutionary perspective can be traced back to Darwin
(1871), who mentioned that only the societies with highest number of sympathetic members
would flourish best and produce most offspring. Evolutionary theory emphasises that kindness
and compassion emerged for human survival, and empathy is built from within (de Waal, 2009).
This theory also discusses that natural selection favours altruistic acts as they help one group
outweigh another even at a cost to an individual (Wilson, 2015; Wilson et al., 2009). Therefore,
the evolutionary approach to compassion recognises that the ability to feel compassion originated
from the evolutionary advantages of caring for others particularly offspring, kin (family and
relations), and in-group associates (Gilbert, 2015).
The attachment approaches (Bowlby, 1969, 1982) applied to compassion signify that
compassion is likely to have originated from early relationships with primary caregivers (Gilbert,
2009a; Gilbert & Procter, 2006, Neff, 2011; Neff & McGeehee, 2010). According to this approach,
care giving motivations develop from parent-child relationships and extends to non-kin meaning
beyond one’s family and relations (Gilbert, 2015; Neff, 2011). The attachment theory identifies
three main attachment styles: secure, avoidant, and anxious attachment style (Bowlby, 1969,
1982). A secure attachment style is formed through validation of the child’s emotions, which later
plays a role in adult affect regulation (Siegel, 2012). Individuals with a secure attachment style are
more likely to be motivated to care giving rather than individuals who are avoidant or anxious.
These individuals are also more likely to develop compassionate and empathic concerns towards
others and are receptive to compassion given from others (Bowlby, 1969; Mikulincer & Shaver,
2001). Absence of a secure attachment often makes children vulnerable to developing
psychopathology as adults, due to the inability of regulating their emotions (Gross & Munoz,
1995; Mikulincer & Shaver, 2012).
Chapter 1: Introduction
3
In contrast, insecure attachment with primary caregivers leads to avoidant or anxious
attachment styles in children (Mikulincer et al., 2003). This is when parents neglect, or do not
respond to the emotional needs of the children (Bowlby, 1969). Anxious attachment is displayed
in hyper-activating strategies, and avoidant attachment is displayed in deactivating strategies of
emotion. In other words, children with an anxious attachment style act out by being overly
emotional in an attempt to receive parental attention, whilst children with an avoidant
attachment style dissociate from the feelings of abandonment when their needs are not met by
the caregiver (Mikulincer et al., 2003). This leads to those with an avoidant attachment style to be
less empathetic, less compassionate, and less willing to help those in need, and those with an
anxious attachment style to feel personally distressed when having to help others in distress (Erez
et al., 2008; Mikulincer et al., 2005).
The biological approaches to compassion suggest that the human nervous system enables
the recognition of distress and the ability to empathise (Preston, 2013; Preston & de Waal, 2002).
Several cross-sectional (e.g., Lutz et al., 2008; Kim et al., 2009) and longitudinal (Klimecki et al.,
2012) studies suggested that compassion activates certain brain regions such as the insula, ventral
striatum, and medial orbitofrontal cortex, which are typically activated in the reception of reward,
love, and affiliation. In support, a compassion-training study found increased brain activations in
the medial orbitofrontal cortex, pregenual anterior cingulate cortex, and ventral striatum, all of
which have previously been identified to correlate with affiliation and positive affect (Kringelbach
& Berridge, 2009; Strathearn et al., 2009). Several compassion cultivating studies have also
observed various physiological functions associated with compassion. For instance, when
presented with compassion eliciting images, a greater amygdala activation has been found in
people who practice compassion (Desbordes et al., 2012). Furthermore, compassion training has
shown to increase the engagement of the dorsolateral prefrontal cortex regions that down-
regulate distress (Weng et al., 2013). Additionally, practicing compassion has shown to strengthen
the activity of certain brain functions such as the brain pathways between the Nigra and Orbital
Frontal cortex, a specific network in the brain recognised for triggering compassion (Singer &
Lamm, 2009), suggesting that compassion has a direct physiological basis.
Chapter 1: Introduction
4
Compassion is a concept embraced across all the world’s most practiced religions (e.g.,
Christianity, Judaism, Islamism) as a core component of their faith and is predominantly discussed
as a fundamental tenet of Buddhist philosophy (Germer & Siegel, 2012). Buddhism defines
compassion as a caring response to the suffering of the self and others with a deep commitment
to alleviate that suffering (Dalai Lama, 1995; Kuan, 2008). It is believed that the concept of self-
compassion originated from Buddhism (Neff, 2003a; Zeng et al., 2016). Buddhist teachings inform
compassion by the wisdom to understand the inner causes of suffering (Germer & Siegel, 2012)
and emphasise that practicing compassion has the capacity to awaken all the positive states of
mind to relieve such suffering (Makransky, 2012). Buddhist influenced compassion practices such
as mindfulness meditation and loving kindness meditation have helped people with depression
and anxiety (Herman, 2014). Studies have also suggested that people who follow Buddhism are
more self-compassionate than people who do not practice Buddhism (Kariyawasam et al., 2021;
Neff, 2003a).
Although the concept, benefits and endorsements of practicing compassion have been
discussed in the contemplative traditions for centuries, it is within the last two to three decades
or so that Western psychology and psychotherapy have identified compassion for its influencing
effects for reducing psychopathology and promoting prosocial behaviour and well-being (Gilbert,
2020). Therefore, several theorists have attempted to explore compassion and provide a
theoretical framework of how compassion may be a fundamental factor in dealing with human
suffering and in facilitating well-being (Irons, 2014).
1.2 Theoretical Models of Compassion
Kristin Neff and Paul Gilbert are pioneers in the field of compassion research and have
introduced the two most widely discussed theories of compassion. Neff’s work is centred on self-
compassion (2003a) whilst Gilbert’s (2014a) model focuses on compassion for the self and others.
1.2.1 Neff’s Theory of Compassion
Neff (2003a, b) was one of the first researchers to define and measure self-compassion,
forming the basis of extensive research in this area by several researchers over the years (Germer
Chapter 1: Introduction
5
& Neff, 2013). Being self-compassionate according to Neff’s approach (2003a), involves taking the
stance of a compassionate other towards oneself. Drawing on various Buddhist teachings, Neff
put forward a theory specifically focusing on self-compassion and conceptualised it as consisting
of three components with opposing negative counterparts: mindfulness/over-identification,
common humanity/isolation, and self-kindness/self-judgement. Mindfulness is described as the
non-judgemental, systematic observation of thoughts and feelings as they arise without denying
or suppressing them. Over-identification, on the other hand, is proposed as being caught up and
swept away by the negative reactivity caused by distressing thoughts and feelings. Neff
emphasised the importance of acknowledging that pain is a shared-human experience, known as
common humanity, whereas isolation is the perception that one is alone in their suffering. Self-
kindness is treating oneself gently and warmly, and self-judgement is treating oneself from a cold
and critical perspective when faced with failure and suffering. The interplay between mindfulness,
common humanity, and self-kindness facilitates one to create a self-compassionate frame of mind
(Germer & Neff, 2013). Considering these positive and negative components of self-compassion
theory, Neff (2003b) developed the Self-Compassion Scale (SCS). The SCS measures all six positive
and negative components of self-compassion, which are totalled to produce an overall score for
self-compassion. Whilst Neff’s approach is centred around self-compassion, she emphasised that
self-compassion enhances compassion and concern for the self and others.
1.2.1.1 Criticisms of Neff’s Model
Although this approach has informed research across the world (e.g., Anuwatgasem et al.,
2020, Finlay-Jones et al., 2018; Mak et al., 2018), several criticisms have been made on Neff’s
(2003a, b) conceptualisation of self-compassion. One of the criticisms, is that although Neff’s
conceptualisation of self-compassion originated from Buddhism, several studies concluded that
her model is theoretically different from the Buddhist concept of self-compassion (Peng & Shen,
2012; Zeng et al., 2016). For instance, Peng and Shen (2012) argued that the concept of common
humanity discussed in Neff’s (2003a) theory, contradicts with the idea of common humanity in
the Buddhist philosophy. They emphasised that while Neff viewed common humanity as
acknowledging failure as a common weakness of humanity, Neff’s theory encourages one to
Chapter 1: Introduction
6
compare themselves against others. This contradicts from the Buddhist’s view of common
humanity, which emphasises the “oneness” of self and others by viewing oneself as being part of
the rest of humankind (Peng & Shen, 2012). Furthermore, they emphasised that the self-kindness
component of Neff’s (2003a) self-compassion theory, only focuses on serving the happiness of
oneself, which also contradicts with Buddhist compassion, that emphasises on developing
compassion and loving-kindness for all beings.
Many of the other criticisms also focus on Neff’s (2003b) SCS scale. Although SCS is the
most used measure of compassion (Kurebayashi, 2021), Lopez et al. (2015) suggested that from a
theoretical standpoint, it is more appropriate to separate the positive and negative items of the
SCS, as it seems to measure two different processes: self-compassion and self-criticism, rather
than one construct of self-compassion. This is in line with Gilbert et al.’s (2011) proposal that self-
compassion is distinct from self-criticism and they should not be measured together. Thus, Neff’s
(2003a, b) conceptualisation of self-compassion as a bipolar construct ranging from high self-
compassion (as indicated by the positive items of the scale) to high self-criticism (as indicated by
the negative items), has been questioned and critiqued by studies that evidenced self-compassion
and self-criticism as two independent processes (Gilbert et al. 2011; Lopez et al., 2015; Williams et
al., 2014).
1.2.2 Gilbert’s Model of Compassion
Gilbert’s (2014, 2019) theory is rooted in an evolutionary informed, biopsychosocial
approach, which recognises compassion as an “evolved strategy” to support basic survival needs.
According to this view, compassion is a multi-faceted process originally evolved to be shared
within one’s family and relations or to be reciprocated in relationships. Gilbert emphasised that
for an individual to develop compassion, they need to show a degree of motivation, willingness,
courage, and distress tolerance. This would include the identification of suffering (also referred to
as engagement with suffering), and also taking ownership to acquire the skills and ability needed
to relieve and prevent such suffering (referred to as action towards compassion). Thus, this theory
posits that compassion involves not just a wish to be helpful, but also a reasoning process and
judgment of what is best to do in a moment of distress (Gilbert, 2019). To enable this to happen,
Chapter 1: Introduction
7
it is proposed that one should display six essential competencies related to “motivation” to care
for well-being, “sensitivity” to suffering, “sympathy”, “distress tolerance”, “empathy”, and being
accepting and “non-judgmental” (Gilbert et al., 2017). In reality, however, Gilbert emphasised
that people can be good at some competencies of compassion, and not so good at others.
Therefore, he stressed that practicing compassion helps one to develop a compassionate identity,
known as a compassionate self, which in turn would train the mind to develop all six compassion
competencies, and expand one’s compassion beyond their family and existing relationships. The
compassionate self can be further developed through several practices such as meditations,
thought monitoring, and behaviour modification (Gilbert, 2019). Gilbert’s (2020) approach to
compassion is multidimensional and consists of several elements such as the Social Mentality
Theory (SMT), the tripartite model of affective regulation, the three flows of compassion, and
facilitators and inhibitors of compassion. These will be discussed in detail in the below sections.
1.2.2.1 Social Mentality Theory (SMT)
To understand how to develop a compassionate self, Gilbert (1989, 2000) developed the
Social Mentality Theory (SMT) and defined social mentalities as “internal systems that generate
patterns of cognition, affect and behaviour … that allow for the enactment of social roles”
(Gilbert, 2000, p.120). In other words, this theory suggests that social mentalities are internal
systems developed to create cognitions, affects and behaviours that are necessary for acts of
survival such as care-seeking and reproduction. According to SMT, people relate to themselves
through systems that were initially developed for relating to others, and as a result, social
mentalities activate in relationships with others as well as relationships within the self (Gilbert,
2000, 2005a).
1.2.2.2 The Tripartite Model of Affective Regulation
In addition to the SMT perspective, Gilbert (2015) suggested that there are three
interacting affective regulatory systems called the threat, drive, and soothing system, which are
triggered by different social mentalities to induce the basic human survival responses (Depue &
Moronne-Strupinsky, 2005). Perceived threats (e.g., perception of a predator) activate the threat
system, which triggers motivating feelings (e.g., anxiety, fear) that in turn creates defensive
Chapter 1: Introduction
8
strategies to protective behaviours (e.g., fight, flight, freeze). The drive system, on the other hand,
alerts individuals to potential opportunities and drives them towards accomplishment, such as the
evolutionary motivating force that guides animals to find food and humans to pursuit for social
rank. Whilst these systems are evolved to protect and motivate within an evolutionary context,
the model proposed that they could negatively affect a person’s well-being in this modern world
(Gilbert, 2009a). This is because the persistent activation of the threat and drive systems (e.g.,
fear of failure, high anxiety, work stress), can be exhausting and feed more adverse thoughts and
feelings (e.g., fear, rumination), leading to psychopathology (e.g., depression and/or anxiety). The
soothing system’s role then, is to calm the threat state, creating safety, and downregulate the
impact of the over aroused threat system. The soothing system can also generate compassionate
motivations by activating the care-seeking and caregiving social mentalities (Gilbert, 2005;
Hermanto & Zuroff, 2016). Gilbert proposed that the activation of the soothing system can be
impaired due to a variety of factors such as early attachment issues, adverse childhood
experiences, and an enduring activation of the threat system. This explains the biological,
evolutionary, and attachment basis of Gilbert’s (2000, 2009a) theory, which posits that practicing
compassion during distressing (threat activating) events allows for the activation of the soothing
system, which would in turn reduce the likelihood of psychological distress and illness (Allen &
Leary, 2010).
1.2.2.3 The Three Flows of Compassion
According to Gilbert (2000, 2019), compassion is experienced across three directional
flows, through the activation of a combination of care-seeking and caregiving social mentalities.
These three flows of compassion include, self-compassion, compassion towards others, and
compassion from others (Gilbert et al., 2017). Self-compassion refers to the compassion people
give to themselves, compassion to others refers to the compassion people give to others, and
compassion from others refers to one’s experiences of compassion from people around them
(whether one perceives others to be supportive towards them). Gilbert’s (2019) notion puts
forward that compassion can be understood as a reciprocal process between the provider of
compassion and the recipient of it. Whilst one would need practice to acclimatise themselves to
Chapter 1: Introduction
9
be open to receiving compassion from themselves or from others, the provider of compassion will
also focus on the impact of their caring on others (Gilbert, 2019). Thus, Gilbert and colleagues
(2017) discussed that compassion is not only felt towards the self, but also for others in the form
of offering compassion to others and receiving others’ compassion. To measure the
aforementioned three flows, and to avoid the issues discussed in the measurement of opposing
constructs within one measure as in Neff’s SCS (2003b), they developed a self-report measure,
named the Compassionate Engagement and Action Scales (CEAS). The CEAS measure one’s
“engagement with suffering and “action towards compassion across self-compassion,
compassion to others, and compassion from others (Gilbert et al., 2017).
1.2.2.4 Facilitators and Inhibitors of Compassion
Gilbert and Mascaro (2017) argued that the motivation to be compassionate, like any
other behaviour that requires motivation, has facilitators and inhibitors. They proposed that one
of the biggest facilitators of compassion is self-reassurance, which is the ability to show oneself
kindness, care, and support during times of distress and failure (Gilbert et al., 2004). Whilst
predicting increased well-being (Barnard & Curry, 2011), self-reassurance has negatively
correlated with the manifestation of psychopathology including depression and anxiety
(Hermanto & Zuroff, 2015). Self-reassurance is underlined by the care-seeking and care giving
social mentality (Gilbert, 2000, 2005), meaning self-reassurance is largely determined on a
person’s social relationships and attachments with others. Therefore, social safeness and
pleasure, which is the perception that the social world is safe and warm, has also shown strong
correlations with increased compassion (Alavi et al., 2017; Gilbert, 2005, 2015). The perception
that one’s immediate surrounding is safe calms the threat and drive systems and keeps the
soothing system activated (Gilbert, 2005, 2015). Given that self-reassurance and social safeness
predict increased well-being and higher self-compassion, Giblet et al. (2009) identified these
factors as facilitators of compassion.
In contrast, Gilbert and Mascaro (2017) suggested that inhibitors of compassion are
related to fears, blocks, and resistances (Gilbert et al., 2011, 2014), meaning that one’s
experiences of compassion can be inhibited by these factors. They posited that such fears relate
Chapter 1: Introduction
10
to negative and self-doubting thoughts of compassion, such as compassion is a weakness, or that
one would be judged as self-indulgent or manipulative. These also relate to the assumption that
one’s compassionate efforts would be rejected, shamed or unhelpful. It is noted that some people
could also be fearful that offering compassion would be too upsetting or cause them to
overwhelm (Gilbert & Mascaro, 2017; Vitaliano et al., 2013). Compassion could be blocked due to
lack of understanding or not knowing how to be compassionate. Other environmental
contingencies such as staff shortage or hectic health settings could also inhibit (or block) health
professionals from being as compassionate as they would like to be (Brown et al., 2014; Gilbert &
Mascaro, 2017). Resistances are related to simply not wanting to be compassionate, which can
also inhibit compassion (Gilbert, 2019). Sometimes, people resist showing compassion to others,
either when they are fearful of others, or hold grudges from previous experiences, which
negatively impact on their desire to be compassionate (Dalenberg & Paulson, 2009; Furnham et
al., 2013). Self-advantage such as not wanting to share their resources with others can also be
considered as a resistant factor for individuals to show compassion (Keltner, 2016).
Self-criticism and shame are known as two of the most pervasive features of
psychopathology (Gilbert & Irons, 2005; Gilbert et al., 2012). Self-criticism, which involves
constant negative self-labelling and harsh judgment of the self (Kannan & Levitt, 2013; Shahar,
2015), has been strongly associated with increased depression (Zuroff et al., 2005). Additionally,
Gilbert et al. (2014) emphasised that social comparison and the sense of inferiority one feels in
relation to a desired social context, is linked to self-inadequacy and criticism. This leads to
external shame (Matos et al., 2015) and internal shame (Gilbert, 2007), which are a self-
perception of scrutiny and negative judgement of self by significant others, and one’s own
negative judgements of oneself, respectively. Both external shame and internal shame have
shown strong correlations with a range of psychopathological traits including anxiety (Tangney et
al., 1992; Matos et al., 2015) and depression (Alexander et al., 1999; Gilbert & Irons, 2004; Matos
& Pinto-Gouveia, 2010). In addition to fears, blocks, and resistances, these factors such as self-
criticism, psychopathology, and shame, often overlap and interplay with one-another, inhibiting
one’s experiences of compassion (Gilbert et al., 2011).
Chapter 1: Introduction
11
1.2.2.5 Criticisms of Gilbert’s Model
Although Gilbert’s (2009, 2014) model provides a solid framework for understanding the
emergence of mental health problems and well-being (Gilbert & Proctor, 2006; Kelly et al., 2012),
several criticisms exist. For example, it has been suggested that the model’s heavy reliance on
early attachments with primary caregivers in the development of the soothing system, ignores the
valuable influence of affiliative experiences with significant others beyond the primary caregivers
(Allen & Land, 1999; Duarte & Pinto-Gouveia, 2017; Farr, 2019). In support, studies have found
that, in addition to the primary caregivers, strong relationships with other figures such as peers,
teachers, and even strangers can have an impact on increased well-being, self-compassion, and a
sense of belonging (Allen & Land, 1999; Cunha et al., 2017; Ferreira et al., 2020; Matos & Pinto-
Gouveia, 2014). These results also corroborate the research that has showed that adolescence is
an influential phase of the development of the soothing system and compassion (Farr, 2021;
Matos & Pinto-Gouveia, 2014). Additionally, Noh and Cho (2020) identified that Gilbert’s
approach does not fully account for the habitual aspect of self-criticism, which they argue is an
unconscious response triggered by self-critical schemas (Cho et al., 2019), that can lead to
psychopathology (Verplanken et al., 2007). Although Gilbert (2014) discussed the impact of self-
criticism on compassion and vice versa, the argument is that the formation of self-criticism is not
sufficiently detailed in the model (Noh & Cho, 2020).
1.2.3 Compassion-Based Interventions
Based on the various definitions (e.g., Lazarus, 1991; Strauss et al. 2016) and theoretical
approaches (e.g., Gilbert, 2010a; Neff, 2003a) to compassion, a multitude of clinical
psychotherapeutic and general population interventions have been developed to promote
compassion for the self and to/from others (Kirby, 2016). Many of these have indicated various
benefits of compassion for mental well-being, physiological health, and genetic expressions
(Fredrickson et al., 2013). Some of these include improved clinical outcomes (Epstein et al., 2005;
Sanghavi, 2006), higher life satisfaction (Yamagata et al., 2011), quality of life (Van Dam et al.,
2011), social, family, and maternal support (Neff & McGehee, 2010), mindfulness (Fredrickson et
al., 2008), and improved mental and personal well-being (Feldman & Kuyken, 2011; Neff et al.,
Chapter 1: Introduction
12
2007: Neely et al., 2009). Practicing self-compassion has shown to reduce interpersonal problems
and psychological distress (Mak et al., 2018; Schanche et al., 2011), personal pathology and
psychiatric symptoms including stress (e.g., Lutz et al., 2008), depression (Leary et al., 2007;
Shapira & Mongrain, 2010), and anxiety (e.g., Van Dam et al., 2011). Several longitudinal studies
and laboratory experiments have also discussed the impact of compassion on direct physiological
health improvements, such as reduced sympathetic nervous system activation, increased heart
rate variability, and reduced inflammatory responses following exposure to a lab stressor (Arch et
al., 2014; Breines et al., 2014, 2015; Crocker & Canevello, 2012; Rockliff et al., 2008), and indirect
health benefits caused from healthy behaviour such as healthy eating (Adams & Leary, 2007;
Schoenefeld & Webb, 2013), exercise (Magnus et al., 2010), medical adherence (Brion et al.,
2014), dietary adherence (Dowd & Jung, 2017), and terminating smoking (Kelly et al., 2010).
Considering the promising findings from compassion-based intervention research on a
range of presentations, an increased interest in compassion-based interventions that specifically
focus on compassion cultivation (e.g., Gilbert, 2014; Neff & Germer, 2013) has developed over the
past decade. For example, Kirby (2016) conducted a review on the effectiveness of compassion-
based interventions with a primary focus on compassion-cultivation and found that there were at
least six empirically supported interventions. These are Mindful Self-Compassion (MSC: Neff &
Germer, 2013), Compassion Focused Therapy (CFT: Gilbert, 2014), Cognitively Based Compassion
Training (CBCT: Pace et al., 2009), Compassion and Loving Kindness Meditations (LKM: Hofmann
et al., 2011) and Compassion Meditations (CM: Wallmark et al., 2013), Cultivating Emotional
Balance (Kemeny et al., 2012), and Compassion Cultivation Training (Jazaieri et al., 2013), which
will be discussed in detail in the following sections. A meta-analysis of these interventions found
evidence for the trans-diagnostic applicability of these interventions, with significant
improvements in compassion for the self (d = .70) and others (d =.55), well-being (d = .51), and
significant reductions in mental health problems such as depression (d = .64) and anxiety (d = .49)
(Kirby et al., 2017).
Chapter 1: Introduction
13
1.2.3.1 Mindful Self-Compassion (MSC)
Neff and Germer (2013) developed the MSC programme to cultivate self-compassion. It
has been effective among both clinical (Neff & Germer, 2012) and non-clinical (Neff & Germer,
2012; Yela et al., 2019) populations. This programme incorporates several practices such as
psychoeducation, mindfulness, loving-kindness, and self-compassion practices. Significant
improvements in self-compassion, mindfulness, and life satisfaction and significant reductions in
depression, anxiety, and stress have been found among people who were enrolled in the MSC
programme (Neff & Germer, 2013). Gilbert (2020) recognised the MSC as one of the most well-
developed, evidence-based approaches to address self-criticism. However, Kirby (2016)
emphasised that the programme lacks evidence-base underpinnings and evaluations in clinical
samples. Furthermore, the efficacy of the MSC is often assessed using the SCS (Neff, 2003b),
which has been criticised for its conceptualisation and structure as previously discussed.
1.2.3.2 Compassion Focused Therapy (CFT)
Gilbert (2014) developed CFT based on the previously discussed Social Mentality Theory
(Gilbert, 1989, 2000). CFT was intended to help people to be motivated to engage with suffering,
and to act on alleviating and preventing that suffering. CFT provides psychoeducation specifically
related to the three emotion-regulation systems by emphasising how the soothing motivational
system helps cultivate compassion to reduce the impact of self-criticism and shame caused by the
activation of threat and drive systems (Gilbert, 2014; Kirby, 2016). Clinicians and researchers have
used the CFT approach to help people with a range of mental health complications, such as anger
issues (Kolts, 2012), anxiety disorders (Tirch, 2012), and eating disorders (Goss & Allan, 2014).
Some researchers have also integrated CFT with other well-known approaches such as
Acceptance and Commitment Therapy (ACT: Hayes & Lillis, 2014), and Cognitive Behavioural
Therapy (CBT: Beck, 1970) and created new psychotherapeutic approaches such as the
Compassion Focused ACT (CFACT: Tirch et al., 2014) and Group CBT with Compassion Training
(Asano et al., 2017) respectively. A Japanese study designed a new group programme named the
Enhancing Self-Compassion Programme (ESP: Arimitsu, 2016) using CFT, which significantly
increased self-compassion in the participants. To date, CFT remains the most evaluated
Chapter 1: Introduction
14
intervention with a systematic review (Leaviss & Uttley, 2015) and meta-analysis (Kirby et al.,
2017) discussing the effects of CFT as a successful intervention for increased well-being and
reduced distress.
1.2.3.2.1 Compassionate Mind Training (CMT)
Compassionate Mind Training (Gilbert, 2000, 2009, 2010; Gilbert & Procter, 2006) is
another compassion-based approach, which is an integral part of CFT. CMT was originally
developed as a group-based therapy, to help people with high levels of shame and self-criticism.
This training incorporates several practices designed to develop physical and mental
competencies to promote self-grounding, a sense of compassionate self and sensitivity to the
suffering of others, and to help people regulate different emotions to face distress and life
difficulties (Crocker & Canevello, 2012; Gilbert, 2009b; Matos et al., 2017a). Some of these
practices include mindfulness training, soothing rhythm breathing, and compassion-based
imagery etc. (Beamont et al., 2021; Gilbert, 2009b, 2010). A substantial portion of CMT, when
compared to CFT, prioritises on psychoeducation and is known to be more suitable than CFT for
group-based treatment than specific individual cases (Matos et al., 2017a). It is also a hybrid of
several therapies, with an evolutionary basis similar to CFT. Interestingly, an initial CMT study
found that participants were reluctant to engage in a compassion-based intervention due to
beliefs that compassion is a weakness or a self-indulgent concept, and these thoughts were
challenged and completely changed upon receiving the CMT (Gilbet & Procter, 2006). This implies
that the perception that compassion is not something one should cultivate, may reduce people
from participating in compassion-based interventions, although the actual participation might
change these attitudes and bring positive outcomes.
Engaging in a two-week CMT increased participants’ self-compassion as well as
compassion for others. In addition, the two-week CMT intervention increased facilitators of
compassion such as safeness and contentment, and reduced inhibitors of compassion such as all
three forms of fears of compassion (fear of self-compassion, and fear of compassion to and from
others), self-criticism, and external shame (Matos et al., 2017a). CMT has also shown to
Chapter 1: Introduction
15
significantly reduce depression, anxiety, feelings of inferiority, and submissive behaviour (Gilbert
& Procter, 2006).
1.2.3.3 Compassion Cultivation Training (CCT)
Underpinned on Tibetan Buddhist contemplative practices and Western psychology
(Jazaieri et al., 2013), CCT aims to promote feelings of open-heartedness and connection with
others (Kirby, 2016). CCT is conducted in group settings in the form of group discussions, and
includes practices such as guided meditation, mindfulness practices and other interactive
exercises to induce self-compassion and compassion towards others (Kirby, 2016). Although CCT
has reported to increase compassion and mindfulness (Jazaieri et al., 2013, 2014), and reduce
attention to unpleasant topics (Jazaieri et al., 2016), CCT is still at an early developmental stage
with limited number of studies conducted (Kirby, 2016).
1.2.3.4 Cognitively Based Compassion Training (CBCT)
This programme was originated to help university students develop emotional resilience
(Ozawa-de Silva & Negi, 2013). Underpinned on Tibetan Buddhist traditions, mindfulness, and
cognitive-restructuring strategies, CBCT consists of eight stages, which include homework
exercises and meditation practices (Kirby, 2016). CBCT has been evaluated in multiple randomised
controlled trials and tested with adolescents in addition to adults in the general public. Some
benefits of using CBCT have been reported, although no independent evaluations of the efficacy
of CBCT have been published (Kirby, 2016).
1.2.3.5 Compassion Meditations (CM) and Loving-Kindness Meditations (LKM)
CM and LKM are combined in many Buddhist practices (Hoffmann et al., 2011). CM is a
Buddhist practice where the meditator wishes for others to be free from suffering. LKM involves
developing caring feelings starting with oneself, and expanding towards loved ones,
acquaintances, strangers and as far as towards those one may have had experienced difficulties
with, and towards all living beings with no distinction (Galante et al., 2014). Kirby (2016)
emphasised that CM and LKM are used in all compassion-based interventions as means to help
calm the mind, enhance compassion for the self and others, and to improve mental health.
Chapter 1: Introduction
16
1.2.3.6 Cultivating Emotional Balance (CEB)
This is another compassion-based intervention which is underpinned by Western
psychology, traditional Eastern attention focus, and contemplative practices (Ekman & Ekman,
2013). Through a variety of practices such as mindfulness and LKM, the CEB training encourages
individuals to recognise the suffering in the self and others and to tolerate distress through
compassion cultivation. Practicing CEB has significantly reduced depression, anxiety, rumination,
and negative affect, and increased positive affect and mindfulness (Kemeny et al., 2012).
However, Kirby (2016) emphasised that the studies exploring the efficacy of CEB have been
limited to female schoolteachers, implying the need to be further explored in different
populations and settings.
1.2.3.7 Similarities and Differences of Compassion-Based Interventions
Several similarities and differences have been observed between various compassion-
based interventions, indicating the multidimensional nature of compassion (Kirby, 2016). In
consideration of the similarities, all the interventions have been influenced by the Tibetan
Buddhism and incorporated some form of mindfulness practice. CFT and MSC programmes
focused less on mindfulness whilst the CCT, CBCT, and CEB programmes spent most of the
interventions focusing on mindfulness-based training. Importantly, all interventions included a
portion of psychoeducation, providing a rationale for engaging in the compassion-based training.
All interventions also entailed activities and tasks that participants actively practiced using specific
compassion strategies. These practices were similar in most interventions and contained
techniques such as breathing exercises, facial and body expressions, building compassionate inner
voices, compassionate letter writing, and imagery tasks aimed at producing calm and soothing
sensations by activating the parasympathetic system (soothing system). A homework component
was also incorporated in all interventions. Interestingly, these interventions also demonstrated
the ability to be delivered in group settings (Kirby, 2016).
In consideration of the differences, CFT was notably different from other compassion-
based interventions, as it is a form of psychotherapy, whereas the other interventions are simply
programmes developed to increase compassion. CFT can be tailored to meet the needs of the
Chapter 1: Introduction
17
individual, whilst the other interventions are delivered by following the prescribed session
content for each session (Kirby, 2016). All interventions, except MSC, focus on compassion as a
broader experience that spreads across the self and towards others, whilst MSC only focuses on
self-compassion as measured by the SCS (Neff & Germer, 2013). The MSC should be used with
caution as incorporating SCS (Neff, 2003b) to measure self-compassion, and Neff’s (2003a)
conceptualisation of self-compassion have been criticised (Kirby, 2016) as discussed earlier in the
introduction section.
Due to the scarcity of rigorous methodologies and research in clinical populations, Kirby
(2016) acknowledged that CFT is the most appropriate form of intervention especially for clinical
populations. Furthermore, CFT is the only model to address fears, blocks, and resistances to
compassion as well as affiliative feelings and behaviours that would promote compassion (Gilbert,
2014; Kirby, 2016). The consideration of inhibitors (fears, blocks, and resistances) and facilitators
(affiliative feelings and behaviours) of compassion in CFT further strengthens the use of this
intervention, over and above the other methods.
However, it is important to note that there are only a limited number of RCT studies (e.g.,
Arimitsu, 2016; Tung, 2020) conducted so far, to support the use of these compassion-based
interventions (Kirby, 2016; Matos et al., 2017a, 2022a, b). Furthermore, literature exploring
compassion-based interventions also seem to be at an infancy stage (Kirby, 2016), with the need
for more rigorous trials to explore the efficacy of compassion-based interventions across clinical
and non-clinical samples from a range of diverse backgrounds.
1.2.4 Cross-Cultural Applicability of Compassion
Many religions have discussed the perceived benefits of compassion in one way or
another (Vivinio et al., 2009). In fact, it is one of the key virtues in the Buddhist philosophy
(Keown, 2005). Self-compassion is believed to have originated in the Buddhist philosophy (Neff,
2003a), and compassion is also deeply embedded in the culture of Asian communities, especially
where Buddhism is the predominant religion (Shih et al., 2013). Due to these reasons, it has been
proposed that people in Asian communities maybe more self-compassionate than people in
Western communities (Markus & Kitayama, 1991). In fact, Neff (2003a) compared Buddhist
Chapter 1: Introduction
18
participants with a group of university students and found that the Buddhist participants were
more self-compassionate. Whilst this appears to be the only study conducted explicitly exploring
the interplay of Buddhism using the SCS, it has been criticised for the small sample size (n = 172
Buddhists) and for not examining the psychometric properties of the scale (Zeng et al., 2016). In
contrast, some studies (e.g., Kitayama & Markus, 2000; Kitayama et al., 1997; Neff et al., 2008)
have found that people from Asian Buddhist communities are more self-critical and less self-
compassionate when compared to people from Western backgrounds. This disparity signifies the
importance of exploring cross-cultural differences of compassion.
1.2.4.1 Cultural Differences of Compassion
Some cultures, such as the Japanese and other Confucian cultures (e.g., Chinese culture)
appreciate compassion for others, whilst condemning self-compassion (Arimitsu, 2016). These
cultures encourage self-criticism and believe that self-critical thoughts lead to self-improving
efforts by allowing one to achieve hierarchy and role mastery through developing
interconnectedness and interdependence belongingness (Heine et al., 2001). Furthermore,
collectivistic environments are known to be rich in emotional interdependence and communal
relationships, causing individuals from these backgrounds to become self-critical and feel more
pressure from their relationships with others during shortcomings and failures (Kitayama &
Uchida, 2003). This potentially explains the aforementioned low self-compassion found in
participants from Asian Buddhist communities, which could perhaps be due to their increased
self-criticism. This was also evident in a cross-cultural study by Neff et al. (2008), which indicated
that levels of self-compassion were lowest in Taiwan (an Asian collectivistic Confucian culture)
and highest in Thailand (an Asian collectivistic Buddhist culture), with the United States falling in
between (a Western individualistic culture). This implies that whilst Buddhism may be an indicator
of increased self-compassion, the Confucian and other cultural dynamics might inhibit this. A
study between Japan and the USA also found that self-compassion was strongly related to
increased positive affect in the USA compared to Japan, indicating that self-compassion promotes
more positive affect in Western individualistic cultures than in Asian collectivistic cultures
(Arimitsu, 2016). These findings suggest that self-compassion and compassion for others may be
Chapter 1: Introduction
19
moderated by cultural differences such as Confucian self-critical attitudes and other culture-
specific factors as well as religion (Arimitsu, 2016; Neff et al., 2008).
1.2.4.1.1 Sri Lanka
Sri Lanka is a collectivistic Asian country, where Buddhism is the predominant religion,
and is practised by 69% of the population (De Zoysa 2013; Marecek 1998). Despite this, Buddhist
practices (e.g., mindfulness and compassion) in Sri Lanka are mostly restricted to monastic
settings, with not many studies incorporating Buddhist philosophy into psychological research (De
Zoysa 2011, 2013). Mental health professionals are also reluctant to accept Buddhist influenced
psychotherapy in Sri Lanka, perhaps due to the lack of knowledge and social acceptance that
Buddhist practices, such as meditation, should be practiced in religious settings rather than in the
psychotherapeutic context (De Zoysa, 2013).
Sri Lanka has been faced with several catastrophes over the years, including a 30-year
long civil war that ended in 2009, which caused over 100,000 causalities and one million refugees
(Neumann & Fahmy, 2012), and the 2004 tsunami, which resulted in 35,000 deaths (Brun & Lund,
2008). Presently, in addition to the impact of the global COVID-19 pandemic (World Bank, 2021),
Sri Lanka is going through its worst economic and political crisis in history (Al-Jazeera, 2022). It has
also been shown that Sri Lankans experience higher rates of grief, domestic violence, learned
helplessness, alcohol abuse, self-harming, and attempted suicides (WHO, 2012), in addition to
increased levels of shame, self-criticism, depression, anxiety, and post-traumatic stress disorder
particularly in the war affected areas (Gunaratnam et al., 2003). A cross-cultural study between
UK and Sri Lankan university students who were not living in the war affected areas, also found
that Sri Lankan participants’ self-harming behaviour was significantly higher than that of the UK
participants (Kariyawasam et al., 2019). Therefore, it seems fair to suggest that a Buddhist
influenced country, such as Sri Lanka, may benefit from a compassion-based intervention, given
that all compassion-based interventions are influenced by the Buddhist philosophy (Kirby, 2016),
whilst approaches such as CFT and CMT focus on reducing shame, self-criticism, and
psychopathology (Gilbert, 2014) that are highly prevalent in Sri Lanka.
Chapter 1: Introduction
20
1.3 Rationale for the Thesis
In comparison to cross-cultural studies on self-compassion, few studies have explored
cultural differences on compassion towards and from others (Arimitsu, 2016). People in
collectivistic cultures (e.g., Japan, Sri Lanka), where people’s lifestyle is influenced by high levels of
social interdependence (Neff et al., 2008) are presumed to be more compassionate towards
others. People in such cultures tend to enjoy social relationships when there are compassionate
exchanges, given that social support and connections are considered as the ultimate goal of life
for people in these cultures (Arimitsu, 2016; Hitokoto & Uchida, 2015). This was evident in a study
that found Japanese participants’ compassion towards others was associated with
interdependent happiness when compared to American participants (Arimitsu, 2018). This is
further clarified in Gilbert’s (1989) SMT, as the theory discusses how compassion is developed
through social roles such as care seeking and care giving. In addition, SMT predicts that social rank
mentality, which creates a competitive, individualistic, and materialistic atmosphere with others
in society, therefore, will decrease compassion for others. This begs the question, whether people
in Western individualistic cultures who strive for personal success are more self-compassionate, in
contrast to Eastern collectivistic cultures where social acceptance is acquired for sharing
compassion towards others.
1.3.1 Thesis Aims
As this is an area of interest that is yet to be explored, the overarching aim of this thesis
was to explore the cross-cultural applicability of compassion-based approaches in an Asian
community, namely Sri Lanka. This was explored across seven chapters including four
independent research papers as outlined below.
1) Chapter 1: This chapter was written to introduce the concept of compassion and to
summarise the existing compassion-based theories and interventions. Objectives included:
a. To provide an overview of compassion in relation to history and definitions of
compassion.
b. To discuss the theoretical models and interventions based on compassion.
Chapter 1: Introduction
21
c. To discuss the experience of compassion in relation to cross-cultural differences.
d. To emphasise on the scarcity of cross-cultural studies on compassion and to lay the
groundwork for testing compassion-based studies in Sri Lanka and other Asian
communities.
2) Chapter 2: This chapter aimed to summarise the different methodological approaches used in
this thesis. Objectives included:
a. To discuss the reasons for the data collection and analyses methods used in the four
independent papers.
b. To discuss the importance of conducting mixed-methods studies.
c. To discuss the use of online studies and questionnaire designs.
3). Chapter 3 (Paper 1): The first paper aimed to explore compassion-based interventions that
have been tested in Asian communities. Objectives included:
a. To explore whether compassion-based interventions have been previously tested among
Asian communities.
b. To test whether if any, the existing compassion-based interventions have been effective
to be used in Asian communities.
4. Chapter 4 (Paper 2): The second paper aimed to understand the experience of compassion in
Sri Lankan university students. Objectives included:
a. To gain an in-depth idea of the views and lived experiences of compassion in Sri Lankan
students.
b. To understand the perceived facilitators and inhibitors of compassion in Sri Lankan
students.
5. Chapter 5 (Paper 3): The third thesis paper aimed to explore compassion and facilitators and
inhibitors of compassion in a cross-cultural group of Sri Lankan and UK people. Objectives
included:
a. To compare the levels of compassion between Sri Lankan and UK people.
b. To investigate the predictors of compassion in relation to the facilitators of compassion
and inhibitors of compassion for Sri Lankan and UK people separately.
Chapter 1: Introduction
22
c. To discuss (if) any cross-cultural differences in the levels of compassion and facilitators
and inhibitors of compassion, and the predictors of compassion in Sri Lankan and UK
people.
6. Chapter 6 (Paper 4): The fourth and final paper aimed to implement a Compassionate Mind
Training intervention to Sri Lankan and UK people. Objectives included:
a. To explore whether CMT will increase the three flows of compassion in the intervention
group when compared to the wait-list control group, regardless of the cultural
background.
b. To explore whether CMT will improve the facilitators and inhibitors of compassion, and to
explore whether these changes will be similar or different between Sri Lankan and UK
people.
c. To explore whether the improvements of CMT (if any) will be maintained at a two-week
follow-up.
7. Chapter 7: This chapter provides a summary of the findings of this programme of research. In
addition, strengths and limitations of each paper, and overall theoretical, research, and clinical
implications are discussed in more detail. The objectives of this chapter included:
a. To discuss findings that were not sufficiently explored in the individual papers.
b. To discuss the overall theoretical implications
c. To discuss the overall research and clinical implications
1.4 Summary of Chapters
The introduction to this thesis (Chapter 1) provided relevant background on compassion
from its history and definitions to theoretical underpinnings and models, and the existing
compassion-based interventions. The introduction chapter also discussed the cross-cultural
applicability of compassion and emphasised on the scarcity of compassion especially in the Asian
communities.
Chapter 2 discusses the methodological approaches used in the four independent
research papers of this thesis. This includes the rationale behind choosing the research methods
Chapter 1: Introduction
23
that were utilised to address the paper objectives. This chapter also outlines different elements of
qualitative research, in relation to paper 2. In addition to the methods used in this thesis, and
discussing the rationale for each paper, this chapter considers context in relation to internet and
questionnaire use.
Chapter 3 (Paper 1) is a systematic review and meta-analysis of existing compassion-
based interventions conducted among Asian communities. The aim was to understand the
prevalence and efficacy of such interventions regarding whether compassion-based interventions
lead to increased levels of compassion in people living in Asian communities. As most
compassion-based interventions are formed and tested in Western cultures, this paper also
investigated the cross-cultural applicability of these interventions in the Asian context. To test
these aims, a meta-analysis was conducted with randomised controlled trials (RCTs). The rigorous
methodology resulted in only eight compassion-based interventions conducted in a few Asian
countries. Results indicated that although limited in number, there is promise for implementing
compassion-based interventions to increase compassion and well-being in Asian communities.
This study has been submitted for publication and is under review on the peer reviewed journal:
Psychology and Psychotherapy: Theory, Research and Practice.
Chapter 4 (Paper 2) is a qualitative study conducted in Sri Lanka. As there were no
compassion-based studies that had been conducted (prior to this thesis) in Sri Lanka, a qualitative
study among ten Sri Lankan undergraduate students was conducted. The aim was to gain an in-
depth understanding of whether participants were familiar with the concept of compassion and if
so, to explore their views and experiences of practicing or engaging in compassion. This study also
sought to gain an understanding of whether Sri Lankan people would be open to receiving a
compassion-based intervention. Study findings showed that Sri Lankan participants were well
aware of the concept of compassion and its impact on increasing their well-being and reducing
distress. However, most participants struggled to experience compassion due to several inhibitors
such as social shame, self-criticism, and depression whilst they discussed that religion and certain
cultural values facilitated them to experience compassion towards and with others. This study has
been published in PLOS ONE, a peer reviewed journal:
Chapter 1: Introduction
24
Kariyawasam, L., Ononaiye, M., Irons, C., Stopa, L., & Kirby, S.E. (2021). Views and experiences of
compassion in Sri Lankan students: An exploratory qualitative study. PLoS ONE, 16(11):
e0260475. https://doi.org/10.1371/journal.pone.0260475
Chapter 5 (Paper 3) is a quantitative cross-sectional study that compared all three flows of
compassion (self-compassion, compassion to and from others) and explored facilitators and
inhibitors of compassion affecting these three flows. This study aimed to explore whether there
were any cross-cultural differences in the three flows of compassion and compared a sample of
300 participants from the UK, an individualistic, Western country (Gardner et al., 1999), with a
sample of 149 participants from Sri Lanka, a collectivistic, Asian country (Freeman, 1997). Results
indicated that there were differences in the levels of self-compassion, which was significantly
higher in the Sri Lankan group, although no significant differences were found in compassion to or
from others between the two countries. In addition, there were some similarities (anxiety did not
significantly differ) and some differences (self-reassurance, fears of compassion and external
shame were significantly higher in the Sri Lankan group, social safeness was significantly higher in
the UK group) across the two countries, when exploring the facilitators and inhibitors of
compassion. This study also explored predictors of each flow of compassion (in relation to the
facilitators and inhibitors of compassion, and psychopathology) and found that there were some
cross-cultural similarities and differences in these predictors (for instance, self-reassurance
predicted self-compassion in both countries whilst the lack of fear of self-compassion and
following Buddhism also predicted self-compassion only in the Sri Lankan group). Findings of this
study suggested that society and culture seem to significantly influence one’s experiences of the
three flows of compassion, highlighting that these specific factors should be carefully considered
when implementing Western compassion-based approaches in non-Western contexts such as Sri
Lanka. This study has been published in Global Mental Health, a peer reviewed journal:
Kariyawasam, L., Ononaiye, M., Irons, C. & Kirby, S.E. (2022). A cross-cultural exploration of
compassion, and facilitators and inhibitors of compassion in UK and Sri Lankan people.
Global Mental Health, 112. https://doi.org/10.1017/gmh.2022.10
Chapter 1: Introduction
25
Chapter 6 (Paper 4) implemented a two-week, online, Compassionate Mind Training
(CMT) to explore whether engaging in the CMT will result in any improvements across the three
flows of compassion, and if there will be any changes in the inhibitors and facilitators of
compassion post CMT, between a cross-cultural group of Sri Lankan and UK people. This
longitudinal study also investigated if CMT would result in any lasting effects, at a two-week
follow-up test. In total, 21 Sri Lankan and 73 UK participants, who were randomly allocated to the
CMT group, completed the training. The results suggested that CMT was effective in increasing all
three flows of compassion (self-compassion, compassion to others, and compassion from others)
in both Sri Lankan and UK participants, highlighting the efficacy and cross-cultural applicability of
the CMT. In addition, several improvements were observed in the facilitators and inhibitors of
compassion with significant increases of positive affect and significant decreases of negative
affect in the two countries. There were some cross-cultural similarities (e.g., fear of compassion
from others, fear of self-compassion) and some differences (fear of compassion to others,
reassured-self, social safeness and pleasure, anxiety, depression, and well-being) in the
improvements observed across participants from the two countries, upon practicing the CMT. All
improvements were sustained at a two-week follow up with further improvements in some
variables (anxiety, and depression reduced in the Sri Lankan group, and social safeness increased
in the UK group). Thus, results of this study indicated that a two-week CMT showed promise in
consideration of cross-cultural efficacy (Sri Lanka) in not only increasing compassion, but also
increasing well-being and reducing distress. This paper has been submitted to the “Mindfulness”
journal and is presently being peer reviewed.
Finally, a general discussion is presented in Chapter 7, which begins with a summary of
findings of the four papers. Drawing from the findings, this discussion provides recommendations
for prospective studies and compassion-based intervention development and discusses the
clinical implications from conducting compassion-based interventions in cross-cultural settings.
This chapter also discusses the findings in relation to theoretical perspectives, the strengths and
weaknesses of the present thesis, and highlights areas for future research.
Chapter 2: Methodological Approaches
27
Chapter 2 Methodological Approaches used in
the Thesis
The methods for all four papers discussed in this thesis, were carefully selected within an
overarching mixed methods approach. The following sections will discuss the rationale for the
data collection methods and the analyses chosen for each of the four papers of this thesis.
2.1 Using a Mixed Methods Approach
This thesis used a mixed methods approach by conducting both qualitative and
quantitative studies. Altogether, this thesis included a systematic review and meta-analysis, a
qualitative study, a quantitative study following a cross-sectional survey design, and a longitudinal
intervention study following a Randomised Controlled Trial (RCT) design. Well-designed mixed-
method studies have the ability to combine qualitative and quantitative methods to mitigate the
limitations that arise in qualitative and quantitative studies individually (Young, 2016). Mixed
methods studies serve three purposes: triangulation, facilitation, and complementarity (Young,
2016). Triangulation refers to the combination of quantitative and qualitative methods to
corroborate the findings of each other. This means results obtained from one approach is
validated if data obtained from another method produces convergent results. Triangulation was
expected to be observed in this thesis as the qualitative analysis in Paper 2 and the cross-sectional
quantitative design in Paper 3 aimed to discover a shared set of variables, which are common
inhibitors and facilitators of compassion (Gilbert, 2014). Facilitation refers to the use of one
research approach to make another research approach more effective. For instance, qualitative
information gained through interviews (Paper 2) may facilitate the formation of a hypothesis that
will be tested in a quantitative study (as tested in Paper 3 and Paper 4). Complementarity is when
two different research approaches (qualitative and quantitative) are selected to understand
different aspects of an investigation to gain a clear picture of a broader issue. This is reflected
across the studies as the understanding gained from Paper 2 especially related to Sri Lankan
society being an inhibitor of compassionate experiences, helped the understanding of the cross-
Chapter 2: Methodological Approaches
28
cultural differences observed in Paper 3 and Paper 4. In addition, Sri Lankan participants’ lived
experiences and the societal challenges of compassion (Paper 2) broadened the understanding of
the cross-cultural differences between Sri Lankan and UK people.
It is important to note that this programme of research was conducted in English across
both Sri Lanka and UK people. This includes the interviews, questionnaires used, and the final
intervention study. The majority of the Sri Lankan participants were university students who were
studying in English. However, not all Sri Lankans are fluent in English and therefore, additional
research should be conducted in the participants’ native languages before generalising these
findings. Such research should however confirm congruency between the terms and their true
meaning in the language to which studies are translated (Kalfoss, 2019). For instance, when
translating the term “compassion”, researchers should thoroughly study whether the translated
term conveys the true meaning of the construct to fully understand participants’ experiences
from their perspectives.
2.1.1 Paper 1: Meta-Analysis
For the first paper of this thesis, a systematic review and meta-analysis was conducted to
explore existing compassion-based interventions in Asian communities and to understand
whether compassion-based interventions lead to increased levels of compassion in people living
in Asian communities. A meta-analysis is a scientific and objective method of combining and
examining different results (Ahn & Kang, 2018). The use of high-quality RCTs is recommended in
systematic reviews and meta-analyses to obtain more reliable results and are considered as the
pinnacle of evidence-based research (Akhter et al., 2019; Uetani et al., 2009). Therefore, this
meta-analysis limited the search to RCTs. Previous meta-analyses and systematic reviews of
compassion interventions have been limited to papers published in English (Austin et al., 2020;
Ferrari et al., 2019; Kirby et al., 2017), conducted in adult populations (Kirby et al., 2017) or
people with long term physical conditions (Austin et al., 2020), and/or only focused on self-
compassion of the three flows of compassion (Austin et al., 2020; Ferrari et al., 2019). The present
Chapter 2: Methodological Approaches
29
meta-analysis was not restricted by the publication date, the language the papers were published
in, or the age of the participants and focused on compassion-based interventions in Asian
communities aimed at increasing one or all three flows of compassion (self-compassion,
compassion to others, compassion from others).
A meta-analytic approach was deemed suitable, as there is no restriction on the similarity
of studies based on the interventions, participants or exposures when conducting meta-analyses
(Borenstein, 2009; Borenstein et al., 2009). Borenstein (2009) emphasised that a certain amount
of diversity among studies is inevitable and in fact desirable in a meta-analysis. In addition,
generalisability and usefulness of meta-analyses are increased noticeably when there is
heterogeneity in populations, settings, and other variables across studies (Gotzsche, 2000).
Borenstein further highlighted that a good meta-analysis should anticipate this diversity and
interpret findings with caution. This is because, for a meta-analysis to be meaningful, researchers
should pay careful attention to the diversity of studies filtered and create specific eligibility
criteria to obtain comprehensive and methodologically rigorous studies (Akhter & Khan, 2019;
Borenstein et al., 2009). The present meta-analysis adhered to these guidelines and narrowed the
search to RCTs conducted in at least one Asian country, aiming to generate compassion (to and/or
from others) or self-compassion, and included at least one self-report measure of compassion.
As the meta-analysis within this thesis aimed to explore existing compassion-based
interventions and to investigate their effectiveness among Asian communities, a meta-analytic
approach over just a systematic review seemed more suitable. This is because meta-analyses
employ statistical methods to synthesise results across multiple studies to uncover the true
effects buried within data by examining and comparing findings of multiple studies (Akhter &
Khan, 2019; Wright et al., 2007). This is a notable benefit of meta-analyses as pooling studies
increases statistical power, which would otherwise be unattainable in individual studies (Akhter et
al., 2019). Adhering to various guidelines for presenting meta-analyses, the present meta-analysis
followed criteria outlined by the Preferred Reporting Items for Systematic Reviews and Meta-
analysis (PRISMA), a research quality improvement body (Page et al., 2020).
Chapter 2: Methodological Approaches
30
2.1.1.1 Analysis
The pooling and analysis of data in this meta-analysis was conducted using the version 5.4
of RevMan statistical software package (The Cochrane Collaboration, 2020). The effect sizes were
interpreted using the Cohen’s (1992) guidelines of small (d = .2), medium (d = .5), and large (d
= .8) effects. Each trial of the meta-analysis was assigned a weighting based on its size and the
precision of study findings (effect size) using a random effects model (Rodseth & Marais, 2016). A
random effects model appeared suitable for this meta-analysis as it assumes that the observed
estimates of treatment effect can vary across studies due to differences in the treatment effects
and due to variations in the samples (chance). Therefore, heterogeneity in treatment effects is
expected in meta-analyses as differences in study variables such as different populations,
different types of interventions, and the length of interventions can all contribute to these
variabilities (Riley et al., 2011). Considering the expected heterogeneity in studies (due to the
scarcity of research in Asian context), a random effects model was selected.
2.1.2 Paper 2: Qualitative Study
This paper used an Interpretative Phenomenological Analysis (IPA) to understand Sri
Lankan students’ lived experiences of offering and receiving compassion. This involved exploring
their thoughts, attitudes, and perceptions of compassion, and lived experiences with a specific
focus on the influence of culture, religion, and societal upbringing. Participants’ perceived
inhibitors and facilitators of compassion were also explored in line with the Gilbert’s (2014)
model. Conducting qualitative studies is highly important to fully comprehend the challenges that
mental health problems pose to experiencing compassion and to discover the potential inhibitors
that wider cultural discourses produce (Campion & Glover, 2017). Therefore, to capture a detailed
personal account from each participant, a qualitative interview approach was adhered.
A contextual constructionist perspective (Madill et al., 2000), which stands in between the
epistemological (knowledge) and realist (the idea that world exists independently to our
perception of it) perspectives was used to comprehend the information gathered. A contextual
constructionist perspective holds the idea that research findings are context bound and therefore,
Chapter 2: Methodological Approaches
31
dependent and variable on the framework in which the data is collected and analysed (Jaeger &
Rosnow, 1998; Madill et al., 2000). This approach allows the researcher to actively contribute to
the research process with knowledge discovery and construction (Jaeger & Rosnow, 1998).
Additionally, Pidgeon and Henwood (1997) declared that a contextual constructionist standpoint
can be affected by the researcher’s interpretation, participants’ personal understanding as well as
the cultural setting where both understandings are immersed.
2.1.2.1 Reflexivity
A contextual constructionist position seemed particularly useful when reflecting on the
reflexivity of this research. Reflexivity is an integral part of qualitative research which describes
the researcher’s role in the research (Palaganas et al., 2017). Reflexivity facilitates transparency by
allowing the researchers to openly consider their position, background, and motivations within
the research (Dowling, 2006). As researchers closely engage with the analysis process of
qualitative research, it is impossible to completely avoid personal bias. Therefore, by discussing
reflexivity, researchers are encouraged to introduce themselves to the readers, and clarify their
experiences, training, and personal characteristics (gender, occupation etc). This will improve the
credibility of findings by allowing the readers to determine how discussions of papers might have
been influenced by the researcher’s interpretations (Tong et al., 2007). The following section will
be written in first person to reflect on the reflexivity of this research.
As the primary researcher of this programme, I have a close relationship with the research
conducted. I am a Sri Lankan female, with experience in cross-cultural research settings,
particularly with Sri Lankan and UK people. In addition, I am also a Buddhist follower, with lived
experience on how Buddhism influences the Sri Lankan society. My passion for cross-cultural
research and understanding how Western methods and treatment delivery can be successfully
applied to non-Western settings adhering with cultural sensitivity, guided the interests of this
research. The reflexive position of me as the primary researcher may have also influenced the
interview flow and participants’ responses, as the participants were aware of my position as a Sri
Lankan researcher. This may have allowed participants (particularly female participants) to share
information more freely and in detail. My perspectives will also be integrated through the analysis
Chapter 2: Methodological Approaches
32
and interpretation of the study findings of Paper 2. This is, however, not a limitation of this study,
but a reflection of the interpretative nature of qualitative research as the “objectivity” strived for
in quantitative research is not the goal of qualitative research (Yardley, 2000). In fact,
understanding the socio-cultural context in which qualitative research is conducted (e.g., me
being a Sri Lankan and understanding its culture) and the researcher’s ability to draw on their
personal interpretations is considered to enrich the research process and its outcomes (Palaganas
et al., 2017; Yardley, 2000). Therefore, participants’ lived experiences discussed in this paper were
comprehended using a subjective standpoint within a relativist position (Willig, 2013).
2.1.2.2 Interpretative Phenomenological Analysis
An Interpretative Phenomenological Analysis (IPA: Smith et al., 2009) was used as a
methodological approach to analyse the interview data. This method was chosen due to its
specific focus on perception and experience, that allows the researcher to discover participants’
understanding and experiences of compassion in a rich, detailed manner. This allowed the
researchers to make logical interpretations of the discussed phenomenon (Smith et al., 2009;
Tindall, 2009). IPA is a reflexive, transparent approach which provides a thorough understanding
of individuals’ lived experiences with meaningful interpretations in reflection with their
relationships to the world and others (Bhaskar, 2008: Smith et al., 2009).
Three key areas; phenomenology, hermeneutics, and ideography comprise the basis of IPA
(Smith, 2011). Langdridge (2007, p. 11) described phenomenology as focusing on “people’s
perceptions of the world or the perception of the things in their appearing”. Husserl, the founder
of the phenomenological view explained that intentionality, a critical feature of consciousness,
allows us to direct our minds towards how people perceive matters as they present themselves to
consciousness (Langdridge, 2007; Larkin et al., 2011). Hermeneutics, the theory of interpretation,
enables the underpinning for interpretations across wider contexts (Langdridge, 2007; Smith et
al., 2009). This is where the link between phenomenology and hermeneutics is created as
people’s engagement with the world and their making sense of matters are often retrieved
through the interpretation in which our past experiences, assumptions and preconceptions are
drawn upon (Heidegger, 1988). Unlike other popular qualitative approaches where human
Chapter 2: Methodological Approaches
33
behaviour is generalised in their overarching claims, IPA is ideographic in nature (Smith, 2004).
This allows a specific focus on personal experiences and perspectives of individual cases rather
than grouping participants’ answers to make common claims (Smith, 2004; Smith et al., 2009).
IPA has been comprehensively used in psychological research, especially within the clinical
and social context (Smith, 2004). Additionally, Reid and colleagues (2005) emphasised the
valuable applicability of using IPA in research areas that significantly lack previous investigations.
Given that there was no compassion-related psychological research that has been conducted in a
Sri Lankan population, IPA was used as the most appropriate methodology to analyse the data
gathered from Paper 2. Additionally, the inductive nature of the analysis also meant that this
particular study did not necessarily need previous literature to build upon as the inductive
approach facilitated the possibility of unexpected and novice experiences to emerge (Eatough &
Smith, 2008).
As with any analysis approach, several strengths and limitations of IPA have been
recognised. The exploration of individuals’ subjective experiences is particularly beneficial for
studies that aim to understand people’s unique personal experiences and how they make sense of
it, within a given cultural context (Shaw, 2001). In addition, the inductive nature of the
questioning, may facilitate unexpected discoveries leading the IPA research in new directions. This
would allow the researchers to uncover areas that they were not previously aware of as beneficial
for their research (Eatough & Smith, 2008; Noon, 2018).
On the other hand, one of the most agreed upon limitations of IPA is the language barrier.
This emphasises that the interpretation of the analysis relies on the representational validity of
language (Willig, 2013). This can be especially problematic, when people with language
difficulties, or those of whom English is not their first language are interviewed (Noon, 2018).
Although Paper 2 was conducted among Sri Lankan students, whose first language was not
English, they were all fluent in English and were undertaking psychology undergraduate degrees
in English.
Another criticism is that search for common themes could reduce the idiographic focus of
the analysis (Arroll, 2015). In response to this criticism, whilst acknowledging the challenge of
Chapter 2: Methodological Approaches
34
maintaining an idiographic focus, Noon (2017) stressed that emphasising each participant’s
unique idiosyncrasies within shared concepts was indeed a possibility. Whilst a smaller sample is
recommended to maintain the idiographic focus (Smith et al., 2009), using a smaller sample
inhibits the generalisability of findings to a larger population (Charlick et al., 2016). IPA
researchers, however, do not consider this as a great limitation, as they highlight that the
objective of IPA is not to uncover a phenomenon in every setting, but rather the perception of a
selected group within their setting (Noon, 2017). In fact, IPA is considered in relation to theory
(meaning researchers should identify connections between the IPA findings and literature) rather
than an empirical generalisability (Smith & Osborn, 2003).
2.1.2.3 Semi-Structured Interviewing
Following the IPA guidelines, to enable participants to engage with an in-depth interview
freely, and reflectively (Smith et al., 2009), a series of semi-structured questions were
incorporated to an interview guide. This ensured that topics discussed during the interviews were
consistent with the research questions that were informed by Gilbert's (2014) theoretical
approach. Participants were interviewed individually in person using a semi-structured interview
that enables the interviewer and participant to engage in a conversation while making
modifications to the interview guide to better understand the participant’s experiences (Smith &
Osborn, 2007). This also allowed the researcher to probe significant areas that arose through the
conversation (Smith & Osborn, 2007; Smith et al., 2009). Semi-structured interviewing is known as
the best method of collecting data for an IPA report (Smith & Osborn, 2007). It has been proposed
that IPA research would ideally incorporate a semi-structured interview with a maximum of 25
and a minimum of two participants (Smith & Osborn, 2007). Furthermore, Smith et al. (2009)
emphasised that a smaller sample of four to ten participants is advised for doctoral studies, to
maintain the idiographic commitment of the analysis. Therefore, ten participants were
purposively recruited for individual face-to-face interviews (Alase, 2017).
The semi-structured interviews in Paper 2 followed a carefully preconstructed interview
guide. The face-to-face interviews were audio recorded with participants’ consent and later
transcribed verbatim into written transcripts. Each transcript was read multiple times to allow the
Chapter 2: Methodological Approaches
35
researcher to immerse themselves in the original data. Whilst reading and re-reading the
transcripts, the primary researcher took notes in the margin of each transcript. These notes
helped the researcher to re-read chunks of transcript and analyse the notes, to develop initial
emergent themes. This process was repeated for each transcript. Emergent themes within
transcripts were studied to search for connections across themes, with the aim of integrating any
related themes. However, where there were no clear connections, the researcher kept an open
mind and did not force themes to be integrated, maintaining the individuality of each case.
Transcripts were further studied to explore any patterns across participants and to identify
emerging themes and noting idiosyncratic occurrences. Once these themes were identified, the
researcher conducted a rigorous analysis by utilising metaphors and importing other theories to
make sense of the participants’ interpretations (Charlick et al., 2016; Smith et al., 2009).
2.1.2.4 Quality in Qualitative Research
Yardley (2000) suggested four essential criteria for assessing validity and quality in
qualitative research. These are namely, sensitivity to context, commitment and rigour,
transparency and coherence, and impact and importance. Paper 2 considered and applied these
criteria where appropriate. Sensitivity to context refers to the theoretical and empirical data
within research and the context in which researchers’ and participants’ perspectives and
experiences are considered. This also emphasises the importance of awareness of socio-cultural
contexts in which the study is conducted. In consideration of this, previous literature was
explored, and the interview guide was developed with an awareness of the socio-cultural
background of the participants, and questions and probes were asked with cultural sensitivity.
The ability to ask interview questions with cultural sensitivity was further facilitated by the
primary researcher’s personal experiences and perspectives as a Sri Lankan. Commitment and
rigour refer to the in-depth immersion of the topic and development of skills and competencies to
conduct methodologically rigorous research. In line with this, the primary researcher studied the
existing literature on the topic as well as IPA manuals and learned interview skills prior to
conducting the study. Careful consideration was also given to the depth and breadth of the
analysis. Transparency and coherence refer to the clarity in the rationale, methods and
Chapter 2: Methodological Approaches
36
presentation of the study and the reflexivity of the researcher. Whilst clearly stating the study
objectives and detailing the methodological steps of this study, the researcher also maintained a
diary for each interview to reflect on the researcher’s experiences, challenges faced, and
suggestions for conducting prospective interviews. The diary also allowed the researcher to
reflect on their own emotions from the topics discussed in the interviews and to expect and
manage these emotions appropriately in future interviews. Finally, impact and importance refer
to the theoretical, socio-cultural, and practical impact of the study. This relates to the relevance of
the study to theoretical models, and the general usefulness of the research for the participant
population, the public, and other researchers. In line with this, the socio-cultural impact of
compassion was reflected throughout the interview discussions and elaborated in the clinical and
research implications section of Paper 2.
2.1.3 Paper 3: Quantitative Study
An online, questionnaire-based cross-sectional quantitative survey was used in this study
among Sri Lankan and UK people, to compare their levels of the three flows of compassion, and a
series of facilitators and inhibitors of compassion as identified by Gilbert (2010b, 2014).
Questionnaires are the most common form of conducting surveys (Young, 2016), and the most
widely used tools in the social sciences research (Fife-Schaw, 2006). Questionnaires are easy to
access, can be transformed into online versions, can be used to obtain large datasets with a
relative ease, and the data gathered through questionnaires can be analysed and generalised to a
larger population relatively easily (Demetriou et al., 2017; Young, 2016). The increased use of
well-designed questionnaire studies has been seen to obtain data relating to demographic and
background information, attitudes, opinions of participants, and for determining prospective
interventions (Young, 2016), all of which were aims of Paper 3. Therefore, a series of validated
questionnaires were delivered online.
2.1.3.1 Analysis
The primary aim of this study was to explore the cross-cultural similarities and differences
in the three flows of compassion (self-compassion, compassion to others, and compassion from
Chapter 2: Methodological Approaches
37
others), facilitators of compassion (self-reassurance, and social safeness), inhibitors of
compassion (fears of compassion, self-criticism, external shame), and psychopathology
(depression, and anxiety) between Sri Lankan and UK participants. To test this, analyses of
covariance (ANCOVA) tests were conducted whilst controlling for age and gender. This approach
of using ANCOVA allowed the researchers to compare the Sri Lankan and UK groups whilst
controlling for the effect of the demographic characteristics. The second aim was to identify the
predictors of the three flows of compassion (in relation to facilitators and inhibitors of
compassion, and psychopathology) and to test whether there will be any cross-cultural
differences in the predictors of the three flows of compassion between the Sri Lankan and UK
participants. The second aim was tested using six hierarchical multiple linear regressions (one for
each flow of compassion in each country). Religion, age, and gender were controlled in the first
block, depression and anxiety scores were controlled in the second block, and then the final block
contained all the controlled variables and all the scales measured in the study. This method of
hierarchical regressions was useful to test the true impact of each predicting variable without the
influence of other variables (such as age, gender, religion, depression, and anxiety) as they were
controlled in the first blocks. All the statistical analyses were performed using the SPSS version 28
and only complete datasets were downloaded and analysed. Chi square tests were performed to
explore differences between age, gender, and religion in the two countries. Exploratory analyses
to identify relevant mediators were carried out using PROCESS (Gray & Kinnear, 2012).
2.1.4 Paper 4: Experimental Study
This paper examined a brief online CMT intervention with pre-post and a two-week
follow-up design, among a cross-cultural group of Sri Lankan and UK participants. This study used
an RCT design where participants were randomly allocated to either the intervention group (CMT
group) or the Wait-List Control group (WLC group) on a 1:1 ratio. Participants in the control group
were also given access to the two-week CMT, after the waiting period. Randomisation of groups
was conducted by an independent researcher using a computer randomisation programme
(Qualtrics, 2022), so the main researcher was blinded to the condition allocated.
Chapter 2: Methodological Approaches
38
2.1.4.1 Randomised Controlled Trials (RCT)
RCTs are considered as the gold standard of evidence-based research due to their ability
to minimise bias (Bondemark & Ruf, 2015). Randomisation in RCTs also provides a rigorous tool to
examine causal relationships between an intervention and its corresponding outcome.
Researchers are advised to carefully select the target group, and the interventions using which
outcomes would be compared (Hariton et al., 2018). This is because randomisation allows both
known and unknown determinants to be evenly distributed into different groups (WLC and CMT)
minimising the assessment bias of differences in effects between the two groups (Bondemark &
Ruf, 2015). To further reduce bias, a process known as concealment (also known as blinding) is
suggested. This is where both the researcher and the participant are unaware of the group that
participants would be allocated to (Hariton et al., 2018). In line with this, participants from
universities and the general public were recruited from Sri Lanka and the UK, and a two-week
CMT design was compared with a wait-list group (participants in the control group had no tasks to
complete). In addition, the primary researcher as well as participants were unaware of which
group participants would be allocated to, at the time of the initial baseline data collection. The
participants were then randomised into their group and given the participatory nature of the
intervention could not be further blinded. However, the primary research continued to remain
blinded for the post and follow up assessments.
2.1.4.2 Analysis
Statistical analyses were conducted using SPSS version 28. Only complete datasets were
analysed. Variables that did not meet normality were either bootstrapped, or re-coded based on
the severity of the skewness. Differences in the variables across the two countries (Sri Lanka and
UK) and the two conditions (CMT and WLC) at the baseline were tested using chi square and
independent samples t-tests. The pre-test post-test efficacy of the CMT was investigated using a 2
× 2 mixed ANOVA design with the two conditions (CMT vs WLC) as the between-group factor, and
time (T1: pre-test and T2: post-test) as the within-group factor. Analyses were repeated for Sri
Lankan and UK samples separately. To investigate whether the intervention efficacy was
maintained at a two-week follow-up, a repeated measures ANOVA was carried out on the CMT
Chapter 2: Methodological Approaches
39
group in relation to the three time points (T1: before CMT, T2: immediately post CMT, and T3:
two-weeks post CMT). The Greenhouse-Geisser correction was used for F-test comparisons when
sphericity was not met. Analyses were repeated for Sri Lankan and UK samples separately.
Both Intention to Treat (ITT) and Per-Protocol (PP) analyses were conducted to observe
the intervention efficacy. The ITT takes all the participants originally randomised into account,
including those that dropped out halfway through the intervention. The ITT recommends the last
value of missing observations to be carried forward (Shah, 2011). Per-protocol analyses on the
other hand, only include data of the participants who adhered to the study protocol and
completed the intervention, in the primary analyses (Ranganathan et al., 2016). ITT analyses are
usually conducted as the exclusion of missing cases violates the principle of randomisation, results
in a decrease of sample size, and fails to estimate the number of participants who would benefit
from the prospective intervention. Therefore, the inclusion of both ITT and PP analyses are
recommended to facilitate a realistic interpretation of the effect of an intervention in RCTs
(Ranganathan et al., 2016; Schulz et al., 2010). ITT was used for the pre- and post-intervention
(T1-T2) analyses only (no further analyses were conducted at T3, as this would have over-inflated
the study outcomes), with PP analyses being conducted using data from all three time points (pre-
post and follow-up: T1-T3).
2.1.5 Considering Context
2.1.5.1 Questionnaire Use
All the interventions chosen for the meta-analysis in Paper 1, and the measures used for
Paper 3 and Paper 4 relied on self-report measures. A common limitation of self-report measures
is response bias (Ivtzan et al., 2017). A response bias is a phenomenon that takes place when
individuals complete self-report measures and provide biased estimates of self-assessed
behaviours (Rosenman, 2011). On such occasions, some participants might respond in a certain
pattern regardless of the question presented (for example only selecting “yes” responses or “no”
responses), which could affect the reliability and validity of questionnaires (Demetrious et al.,
2017). One of the most common reasons for response bias is social desirability bias, where
Chapter 2: Methodological Approaches
40
individuals complete self-report measures to appear to be displaying socially accepted behaviour,
or to “look good”, even when the surveys are completely anonymous. This is a serious problem in
interventions when a recalibration of bias is caused after completing an intervention (Howard,
1980). This is, when there is a response-shift bias, meaning that respondents’ answers may vary
across measurement time points although this might not be an actual representation of the
intervention efficacy, but of a respondents change in perception or internal calibration of the
construct being measured (Roesenman, 2011). As this thesis focused on Asian populations, with a
specific interest in Sri Lankan people in comparison to UK people in the experimental chapters,
this could be particularly problematic. This is because social desirability bias might be higher in
Asian people including Sri Lankan people due to the stigma of mental illness restricting Asian
people from reporting affective complaints (Wong & Mak, 2016). Another limitation of
questionnaire use is that the lack of clarity in questions may lead participants to interpret
questions differently (Demetrious et al., 2017; Young, 2016), which could potentially occur among
the participants in this thesis for whom English is not a first language.
To minimise the biases of questionnaire use, researchers have recommended the use of
multimethod studies (more than one method of assessment) to corroborate the findings of one
another. In addition, the use of carefully thought-through and clarified, well-structured
questionnaires are recommended (Demetrious et al., 2017). Internet-based studies have also
found lower rates of social desirability responding, suggesting that the confidentiality and
anonymity in online studies might reduce the pressure of participants to respond in socially
desirable ways (Nayak & Narayan, 2019). Therefore, considering these suggestions, to minimise
such biases, this thesis adopted a mixed methods approach using qualitative and quantitative
studies, used a series of validated measures, and converted the questionnaires to be delivered
online.
2.1.5.2 Internet Use
The meta-analysis conducted in Paper 1 discussed that both online and face-to-face
interventions were equally effective. Paper 3 and Paper 4 of this programme of research used
online survey designs. The online approach was especially useful as data collection for these
Chapter 2: Methodological Approaches
41
studies took place during the COVID-19 pandemic where face to face studies were not
possible/advised (Pfefferbaum & North, 2020), as well as during a period of political and
economic unrest in Sri Lanka (Al-Jazeera, 2022). Additionally, online interventions are cost
effective, self-administered, feasible, and accessible to a larger sample (Chi, 2013; Mak et al.,
2018). The use of online survey designs also enable data collection from participants who might
otherwise have hesitated to meet face-to-face (Wright, 2005). For instance, the questionnaires
used in Paper 3 and Paper 4 included depression, anxiety scales, and questions related to social
shame, fears of compassion, and self-criticism, which are related to sensitive information that
people would be reluctant to share openly (Halamova et al., 2020; WHO, 2001). Furthermore,
considering that disclosing mental illness is stigmatised in countries such as Sri Lanka
(Kariyawasam et al., 2021, Lauber & Rossler, 2007), the use of internet approaches seemed
particularly useful for the present thesis. Whilst internet-based studies facilitate the recruitment
of diverse samples across age, gender, geographic setting, and socioeconomic status, findings
from internet-based studies are found to be consistent with findings from traditional face-to-face
methods (Gosling et al., 2004). In addition, studies have discussed that people prefer mobile
phone-based interventions as opposed to other internet-based interventions (Berry et al., 2016).
Considering this, questionnaires in Paper 3, and the intervention and questionnaires in Paper 4
were made available via both mobile phones and other online formats (e.g., laptop, iPad).
Chapter 3: Meta-Analysis of Compassion-Based Interventions
Chapter 3 Compassion-Based Interventions in
Asian Communities: A Meta-Analysis of
Randomised Controlled Trials
Abstract
Compassion is known as a sensitivity to suffering and being motivated to relieve such suffering in
the self and others. Research has shown that practicing compassion increases well-being and
reduces depression, anxiety, and psychological distress among clinical and non-clinical
populations. Despite a rapid increase of compassion-based interventions within the past two
decades, the reviews are limited to predominantly Western cultures. Therefore, this systematic
review and meta-analysis aimed to evaluate the literature attempting to promote and increase
compassion in Asian communities. The effectiveness of eight Randomised Controlled Trials
conducted across 1012 participants from Thailand, Japan, China, and Hong Kong was explored
using a random effects model. Significant between-group differences in change scores were
reported on self-compassion with large effect sizes in interventions with wait-list control groups
(d = .86) and small effect sizes in interventions with active control groups (d = .19). The findings
suggest that although the existing compassion-based interventions are heterogeneous in nature
and limited in scope, there is promising evidence of improving self-compassion in Asian
communities, also supporting for their cross-cultural applicability. However, research within the
Asian context is limited and at an infancy stage, signifying the importance of conducting further
compassion-based interventions in clinical and non-clinical groups living in the Asian
communities.
Keywords: compassion, self-compassion, RCT, efficacy, Asian, cross-cultural
Chapter 3: Meta-Analysis of Compassion-Based Interventions
44
3.1 Background
The concept of compassion has been widely discussed in Buddhist philosophy and other
practiced religions (Germer & Siegel, 2012; Strauss et al., 2016). Compassion is commonly
understood as an openness to consciously turn towards suffering, rather than away from it
(Gilbert, 2014a). Whilst compassion-based meditations have formed a central part of some
spiritual traditions (e.g., Buddhism: Lama & Thupten, 1995), they have also been incorporated
into treatment approaches in psychotherapy (Gilbert, 2013; Neff, 2003a). Practicing compassion
has shown increased improvements in psychological and physiological well-being in clinical and
non-clinical populations (Germer, 2009; Gilbert, 2013; MacBeth & Gumley, 2012; Neff, 2003a).
Therefore, the existing literature provides evidence to support the notion that compassion
cultivation and practice may have a positive impact on a range of emotional, physical, and life
experiences whilst reducing psychopathology (Kirby, 2016).
3.1.1 Models and Measures of Compassion
Neff (2003a) introduced one of the earliest and most widely used approaches to self-
compassion (Germer & Neff, 2013), and viewed self-compassion as being moved by one’s own
suffering and turning towards the suffering to alleviate it with kindness and non-judgment. Drawn
from Buddhist philosophy, this model outlines three key components of self-compassion:
mindfulness, the acknowledgement and the non-judgmental acceptance of suffering, common
humanity, the recognition that suffering is common to all humankind, and self-kindness, showing
kindness to oneself during times of distress (Neff & Dahm, 2015). Neff emphasised that these
three key components of self-compassion are conceptually distinct, yet overlap with one another,
and is best understood when combined with their negative counterparts: overidentification, being
absorbed by one’s own negative thoughts and feelings, isolation, perception that one is isolated
from the rest of humanity, and self-judgment, negative and harsh judgments of oneself during
their shortcomings.
Chapter 3: Meta-Analysis of Compassion-Based Interventions
To measure self-compassion using its key components and the negative counterparts,
Neff (2003b) developed the Self-Compassion Scale (SCS). However, several studies have
questioned the use of this scale as combining the positive and negative components of self-
compassion inaccurately represents the concept of self-compassion. For instance, Lopez et al.
(2015) suggested that the positive and negative components seem to measure two different
processes: self-compassion and self-criticism, rather than self-compassion as one construct.
Furthermore, Gilbert et al. (2011) proposed that self-compassion is distinct from self-criticism and
should not be measured together. More recently, Neff and colleagues (2021) developed two new
state measures of self-compassion, named the State Self-Compassion Scale-Long Form (SSCS-L) to
measure the six components of self-compassion, and the State Self-Compassion Scale-Short Form
(SSCS-S) to measure the global state of self-compassion. Although they did not address the
criticisms of the SCS, these new scales were developed to complete the void of a state measure to
assess causal inferences of self-compassion (Neff et al., 2021).
Gilbert, another pioneer in the compassion field developed the Social Mentality Theory
(SMT: 1989, 2014, 2017) and suggested that compassion emerges from the evolution of the
mammalian care-giving motivational system designed to regulate negative affect. These
motivational systems are referred to as social mentalities, which evolved to overcome challenges
essential for survival, such as care-seeking and caregiving (Gilbert, 2005, 2014). Gilbert (2014)
emphasised that one could feel compassion for the self and others, and therefore, compassion
can be experienced across three directional flows, namely self-compassion, compassion to others,
and compassion from others (Gilbert, 2009b).
To measure all three flows of compassion, as well as addressing the issues of the SCS,
Gilbert and colleagues (2017) developed a self-report measure called the Compassionate
Engagement and Action Scales (CEAS). The CEAS has been found to be a psychometrically robust
measure to measure the three flows of compassion in clinical and non-clinical populations
(Davalos-Batallas et al., 2020; Lindsey, 2017) cross-culturally (Asano et al., 2020).
Chapter 3: Meta-Analysis of Compassion-Based Interventions
46
3.1.2 Compassion-Based Interventions
To date, only two meta-analyses have investigated the efficacy of existing compassion-
based interventions for the use of public, with randomised controlled trials (Ferrari et al., 2019;
Kirby et al., 2017). Of these, only one review has provided a rigorous overview of the aims, design,
and evidence underpinning the existing compassion-based interventions (Kirby, 2016). Kirby
(2016) identified at least six empirically supported interventions designed with a specific focus on
developing a more compassionate stance. These are Compassion Focused Therapy (CFT: Gilbert,
2010b), Mindful Self-Compassion (MSC: Neff & Germer, 2013), Cognitively Based Compassion
Training (CBCT: Pace et al., 2009), Compassion and Loving Kindness Meditations (LKM: Hofmann
et al., 2011) and Compassion Meditations (CM: Wallmark et al., 2013), Cultivating Emotional
Balance (CEM: Kemeny et al., 2012), and Compassion Cultivation Training (CCT: Jazaieri et al.,
2013). Providing evidence for the trans-diagnostic applicability, Kirby et al.’s meta-analysis (2017)
found that compassion-based interventions improved self-reported compassion (d =.55), self-
compassion (d = .70), and well-being (d = .51), and decreased mental health indicators such as
depression (d = .64) and anxiety (d = .49). They also concluded that although there are multiple
similarities in these interventions, CFT, which was developed by Gilbert (1989, 2005) is notably
different due to its theoretical basis of evolutionary psychology, attachment theory, and SMT. In
addition, Ferrari et al. (2019) found that when compared to the control groups, self-compassion
interventions indicated significant improvements in self-compassion (g = 0.75) and several other
psychological outcomes including eating behaviour (g = 1.76), rumination (g = 1.37), stress (g =
0.67), depression (g = 0.66), mindfulness (g = 0.62), self-criticism (g = 0.56), and anxiety (g = 0.57).
Despite these promising results, the majority of the studies included in these reviews were based
in Western countries, and neither review (Ferrari et al., 2019; Kirby et al., 2017) assessed the
potential influence of culture on the efficacy of these compassion-based interventions.
Chapter 3: Meta-Analysis of Compassion-Based Interventions
3.1.3 Rationale for the Meta-Analysis
Although there is an increased interest in developing compassion-based interventions and
promising evidence for their use for a range of clinical presentations (Kirby et al., 2017), most
interventions have been limited to Western cultures (Fredrickson et al., 2008; Jazaieri et al., 2013;
Neff & Germer, 2013). However, cross-cultural investigations are important, as cultural
differences have been found in the experiences of compassion and their relationship with well-
being and psychopathology (Arimitsu et al., 2019).
It is often assumed that people in Asian collectivistic cultures (where one’s lifestyle and
decision making may be influenced by their society) would experience more compassion towards
and from others, as these cultures are rich in interpersonal connectedness, social conformity and
caring for one another (Arimitsu et al., 2019; Gardner et al., 1999; Markus & Kitayama, 1991;
Steindl et al., 2020). However, several cross-cultural studies have found that when compared to
Western cultures, people from Asian collectivistic cultures such as Singapore (Steindl et al., 2020)
and Sri Lanka (Kariyawasam et al., 2022) were less likely to experience compassion towards and
from others possibly due to their heightened fears of the society, and perceived external shame.
Kariyawasam et al. (2021, 2022) discussed that compassion and help-seeking behaviour is
considered as weak and shameful in such collectivistic societies. Increased fears and self-criticism
also seemed to inhibit self-compassion in Japanese people in a related cross-cultural study
(Kitayama & Markus, 2000). These findings signify the importance of conducting further studies in
Asian communities to understand the cross-cultural differences of compassion and imply that
efficacy of compassion-based interventions may vary due to these cultural differences.
Due to the cross-cultural differences and that, several compassion-based approaches
(Gilbert, 2010b) propose a cross-cultural applicability, the present study aimed to explore the
efficacy of the existing compassion-based interventions conducted in Asian communities.
Chapter 3: Meta-Analysis of Compassion-Based Interventions
48
Therefore, this meta-analysis aimed to answer the question: do compassion-based interventions
lead to increased levels of compassion for people living in Asian communities?
3.2 Method
3.2.1 Protocol and Registration
This meta-analysis adhered to the general principles of the Preferred Reporting Items for
Systematic Reviews and Meta-Analyses (PRISMA: Page et al., 2021). The protocol was
prospectively registered in PROSPERO, the international prospective register of systematic
reviews, under the registration number CRD42020201832.
3.2.2 Eligibility Criteria
The primary researcher and two voluntary research assistants carried out searches
independently. Studies that met the following criteria were included: (a) a Randomised Controlled
Trial (RCT) in which the primary focus was to purposively generate compassion or self-
compassion; (b) conducted in at least one Asian country; and (c) included at least one self-report
measure related to compassion or self-compassion. Both clinical and non-clinical populations of
all ages were included. No publication date, language, or study design restrictions were applied.
Eligibility criteria was based on the population, intervention, comparator, outcomes, and the
study type (Table 3.1). Studies conducted across countries in the Middle East and North Africa
(MENA) region were excluded as although some Middle Eastern countries are situated in the
Asian continent, they are considered as countries in the MENA region (separately from other
Asian countries) and share certain cultural and religious norms that are different from other Asian
cultures (Alkaiyat & Weiss, 2013; Kabasakal et al., 2012).
Chapter 3: Meta-Analysis of Compassion-Based Interventions
Table 3.1. Inclusion and Exclusion Criteria for the Review.
Inclusion Criteria
Exclusion Criteria
Population
Asian communities
Participants in non-Asian settings/ non-
Asian
Intervention
RCTs aimed to increase compassion
Non-RCTs/ aim is not compassion (e.g.,
mindfulness)
Comparator
Waitlist control, active control group
No comparator/control
Outcome
Measures compassion/self-
compassion
Does not measure compassion/ self-
compassion
Studies
Published/unpublished studies, all
languages
Literature reviews, opinion papers,
abstracts, policy reports
3.2.3 Search Strategy
The systematic literature search was conducted using Scopus, Medline, Web of Science,
AMED, APA PsycINFO, Ovid (EMBASE) and CINAHL databases. Cochrane Library, ProQuest for
Dissertations and Theses, and Open-Dissertations databases were systematically searched to
detect any relevant grey literature. The final search took place on the 10th of March 2022. The
following search terms were developed with a research librarian: TI (compassion*) AND AB
(random* control*) AND AB (trial OR interven* OR stud* OR program* OR therap* OR training)
AND TX (Asia* OR East* OR “Eastern culture*” OR Japan* OR Chin* OR Vietnam* OR Malaysia*
OR Singapore* OR “Hong Kong” OR Korea* OR India* OR Pakistan* OR Bangladesh* OR “Sri
Lanka*”). Although various other interventions have integrated compassion (e.g., Mindfulness-
Based Compassion Training: Lo, 2011) or produced increased compassion (e.g., Mindfulness-
Based Cognitive Therapy: Segal et al., 2002a, b), their primary focus is not compassion cultivation.
Therefore, such interventions were excluded from the search results and only the interventions
with a specific focus of compassion/self-compassion cultivation were included.
Chapter 3: Meta-Analysis of Compassion-Based Interventions
50
3.2.4 Data Extraction
The following data on study characteristics were extracted: Name of authors and year of
publication, country, intervention name, design, and underpinning theory/model, aim of the
study, target population, measures used, duration of the intervention, intervention tasks, and the
main findings of the study. For the meta-analyses, the means, standard deviations, and sample
sizes for each group at pre- and post-interventions were extracted.
3.2.5 Analysis Strategy
Version 5.4 of the RevMan software (The Cochrane Collaboration, 2020) was used for the
analyses. Cohen’s (1992) guidelines of small (0.2), medium (0.5), and large (0.8) effects were used
when interpreting effect sizes, represented by d. Computations were based on a weighted-
average of the effect sizes using a random-effects model, as the random effects model assumes
that study variations can be systematic and not only due to random error. A random-effects
model is also appropriate as true effects of interventions are likely to vary depending on the
sample characteristics and implementation of the intervention (Borenstein et al., 2009).
The efficacy of the interventions was compared in relation to the control groups of either
waitlist control (WLC) or active control (AC) with the effectiveness of the compassion
interventions analysed separately. WLC groups received no intervention, and the AC groups
received a different form of intervention than the intervention groups (Kirby et al., 2017). It was
assumed that studies with an AC would report smaller effect sizes than studies with a WLC group,
as the different interventions received by the AC groups would also influence the outcome
variables (Cuijpers et al., 2016, Kirby et al., 2017).
3.2.6 Risk of Bias within Studies
As the PRISMA statement suggests the inclusion of a risk of bias assessment (Page et al.,
2021), risk of bias within studies was assessed using the Cochrane risk of bias tool of the Revman
Chapter 3: Meta-Analysis of Compassion-Based Interventions
software (Higgins et al., 2011). Critical assessments were made separately for each study for the
following domains: sequence generation, allocation concealment, blinding of participants and
personnel, blinding of outcome assessment, incomplete outcome data, selective outcome
reporting, and for other biases. Studies that adequately described these domains were given a
judgment of ‘low risk’, studies that stated the domains were not addressed were given a ‘high
risk’, and studies that did not describe the process of these domains were given a judgment of an
‘unclear’ risk of bias (results for each study are reported in Appendix A).
3.3 Results
3.3.1 Systematic Search Results
The initial database search resulted in 266 records, which were transferred to EndNote, a
reference management software. Sixty-three duplicates were detected by EndNote, and 19
duplicates were detected manually. All 82 duplicates were removed. Titles and abstracts of the
remaining 184 papers were screened and 159 papers were excluded, as they were not related to
the search. After a full text screen of 25 papers, 16 were excluded based on the eligibility criteria.
Of the final 9 results, three papers reported on one study (Mak et al., 2018, Mak et al., 2019, Yip
et al., 2018) and therefore, only one of them was retained (Mak et al., 2018). Four
conference/meeting abstracts were obtained although they were excluded as the full papers of
these could not be acquired (after contacting and receiving no response from the authors of the
abstracts). Reference lists of the chosen studies and other resources were searched for any
potential studies and one suitable study was found from ‘ResearchGate’, a social networking
website for researchers and scientists. This resulted in eight studies with quantitative data that
were included in this meta-analysis. All studies were allocated a number from 1 to 8 (see Table
Chapter 3: Meta-Analysis of Compassion-Based Interventions
52
3.2) and are referred to by their assigned number (e.g., 2, 4) going forward. Figure 1 details the
search strategy.
Chapter 3: Meta-Analysis of Compassion-Based Interventions
Records identified from databases (n = 266)
Amed + CINHAL+ MEDLINE + Open
Dissertations +APA PsycInfo (n = 62) +Scopus
(n = 103)
Web of Science (n = 45)
Cochrane (n = 31)
Proquest (n = 20)
Ovid (EMBASE) (n = 5)
Records removed before screening:
Duplicate records removed (n = 82)
Records screened (n = 184)
Records excluded** (n = 159)
Conducted in the MENA region or not
conducted in Asia (n = 72)
Not compassion related (n = 15)
Intervention focus is not compassion (n =10)
Not an intervention (n = 45)
Protocol/prospective/ongoing (n =15)
Uncontrolled (n = 2)
Reports sought for retrieval (n = 25)
Reports assessed for eligibility (n = 25)
Reports excluded (n = 18)
Not an intervention or an RCT (n = 2)
Did not measure compassion (n = 5)
Conference/meeting abstract (n = 4)
Intervention focus is not compassion (n = 5)
Papers discuss the same study (n = 2)
Records identified from:
Websites (n = 1)
Citation searching (n = 0)
Identification of studies via databases and registers
Identification of studies via other methods
Identification
Screening
Studies included in review (n = 8)
Reports sought for
retrieval (n = 1)
Reports assessed for
eligibility (n = 1)
Figure 1. PRISMA Flow Diagram of Study Selection
Included
Chapter 3: Meta-Analysis of Compassion-Based Interventions
3.3.2 Quantitative Results
3.3.2.1 Intervention and Participant Characteristics
A total of six of the eight studies included at least one of the six compassion-based
interventions that Kirby (2016) outlined. Four studies were based on MSC (Neff & Germer, 2013) 1,
4, 6, 7, one was based on CFT (Gilbert, 2010b) 2, and another study incorporated both MSC and CMT
approaches 5. Although not outlined in Kirby’s review, the remaining two studies in the present
review were based on approaches by Neff (2009) and colleagues (2021) with one study
conducting a self-compassion writing exercise (Neff, 2009) 8 and the other looking at a new
approach named the Self-Compassionate Mind-state Induction (SCMI: Neff, 2021)3. Four studies
were delivered in person, 1, 2, 5, 7 and four were delivered online 3, 4, 6, 8, and used a group1, 2, 3, 5, 7 or
a self-delivered approach 4, 6, 8. All studies, except for one 3, used the SCS (Neff, 2003b) to measure
self-compassion. Some studies used the complete scale (with 26 items) of the SCS 1, 2, 4, 5, 7, and
other studies only used the 13 positive items of the scale 6, 8. One study used the State
Selfcompassion ScaleLong Form 3 (SSCS-L: Neff et al., 2021).
Intervention duration varied from one to eight weeks. One study did not specify the
duration of the intervention 3. The authors were contacted to ascertain the information regarding
the intervention duration, but no response was received. Four studies included a waitlist-control
group (WLC) 2, 4, 5, 7, and four studies included an active control group (AC) 1, 3, 6, 8. The AC groups
received a form of intervention different to the compassion-based interventions given to
intervention groups, whilst the WLC groups received no treatment/intervention. The type of
intervention received by the AC groups varied between a standard psychotherapy 1, neutral
writing condition 3, cognitive behavioural therapy 6, and a control writing condition 8 . The
majority of the studies also reported follow-up data, 2, 4, 5, 6, 7, 8 with follow-up periods ranging from
one to twelve months post-intervention (Table 2 gives a summary of the study characteristics). All
studies were conducted within a five-year period (2016-2021) in various Asian countries including
Chapter 3: Meta-Analysis of Compassion-Based Interventions
56
Thailand 1, Japan 2, China 3, 4, 5, and Hong Kong 6, 7, 8. The studies included university students 2, 3, 5, 7,
8, adults from the public 1, 6, and pregnant women 4. Some studies specifically recruited adults with
low self-compassion 2, and symptoms of anxiety and/or depression 1, 4 at the baseline level. One
paper did not discuss the gender of the participants 2, one paper consisted of only female
participants 4, one paper had more male participants than female participants 3, and the other
five out of eight papers had more female participants than male participants 1, 5, 6, 7, 8. All the
papers that considered gender differences in the analyses discussed that no significant impact of
gender was reported in the findings 3, 5, 6.
Chapter 3: Meta-Analysis of Compassion-Based Interventions
57
Table 3.2. Intervention Characteristics.
No
Author, Year
and Country
Intervention, Design,
Underpinning Model and
Duration
Aim and Target Population
Measures
Comparator
Tasks
Findings
1
Anuwatgasem
et al. (2020)
Thailand
Mindfulness and Self-
Compassion-based
therapy (MSC), RCT,
group, in person study
design, based on MSC, 7
weeks
To compare the effect of
MSC on group
psychotherapy on people
with a DSM-V diagnosis of
Major Depressive Disorder
(n=23 intervention group,
n=11 control group)
MADRS, SCS - Thai,
PSQI, HADS, Thai-
PSS-10, RSES,
WHOQOL
Pre-test and
post-test
against a
standard
intervention
(AC group)
Activities to promote
self-kindness,
common humanity,
mindfulness via
meditation and
compassionate body
scan etc.
Significant
decreases in
depression,
anxiety and
stress, self-
esteem, and
quality of life
2
Arimitsu
(2016)
Japan
Enhancing Self-
Compassion Programme
(ESP), RCT, in person
study design, based on
CFT, CMT, 7 weeks
To develop an ESP and test
the efficacy of the
programme in enhancing
self-compassion in low
compassionate Japanese
psychology university
students
(n=20 intervention, n =20
control group)
Acceptability
questionnaire, SCS -
Japanese, RSES, BDI-
2, STAI, DACS, MMS,
SCES
Pre-test post-
test and
3month
follow up
against a WLC
group
Loving-kindness
meditation,
mindfulness training,
compassionate mind
training using
imagery,
compassionate letter
writing, three-chair
work, homework
increases in
each subscale
of self-
compassion
except for
mindfulness,
reduced
negative
thoughts and
emotions
Chapter 3: Meta-Analysis of Compassion-Based Interventions
58
No
Author, Year
and Country
Intervention, Design,
Underpinning Model and
Duration
Aim and Target Population
Measures
Comparator
Tasks
Findings
3
Guan et al.
2021
China
Self-Compassionate
Mindstate Induction
(SCMI) RCT, online
group, based on SCMI by
Neff (2020), duration not
specified
To investigate online self-
compassion exercises’
effectiveness in alleviating
negative affect in university
students during the COVID-
19 pandemic (n=50
intervention, n=45 control)
Demographic
information, SSCS-L,
PANAS-negative
affect, STAI-S
Pre-test post-
test against a
neutral
control group
(AC group)
Writing
task containing a
series of writing
prompts that aimed
to induce
mindfulness,
common humanity,
and self-kindness
Increases in
self-
compassion
and decreases
in negative
affect when
compared to
the control
condition
4
Guo, Zhang,
Mu & Ye
(2020)
China
Mindful Self-Compassion
Programme (MBSP), RCT,
online study design,
based on MSC,
6 weeks
To explore MBSP’s effects in
preventing the
development of PPD in
women in 2nd or 3rd
trimester of pregnancy with
antenatal depressive or
anxiety symptoms
(n=144 intervention, n = 140
control group)
MAAS, EPDS, STAI 1
and 2, BDI 2, SCS-
Chinese, WBI of
WHO-5, Chinese
PSI, The Scales of
warmth and
negativity of the
CPBQ, IBQ-Short
Form
Pre-test
(2nd/3rd
trimester),
post-test (3
month post-
partum), one-
year post-
partum and
control group
Six sequential steps
involving different
types of exercises
with guided
instructions were
performed in a
stepwise way
(steps/tasks not
specified)
Reduced
anxiety,
improved
mindfulness,
self-
compassion,
and well-being
in the MBSP
group
5
Huang et al.
(2021)
China
Self-Compassion
Intervention, RCT, in
person, group study
design, based on MBCT,
CMT, and MSC, 4 weeks
To test the effects of the
self-compassion
intervention on future-
oriented coping and
psychological distress in
Chinese college students
SCS, The 16-item
Future-Oriented
Coping Inventory,
DASS
Pre-test post-
test and 1
month follow
up against a
WLC group
Psychoeducation,
observing body
sensations
under stress,
affectionate
breathing meditation,
Improvements
in self-
compassion,
future-oriented
coping,
depression,
stress
Chapter 3: Meta-Analysis of Compassion-Based Interventions
59
No
Author, Year
and Country
Intervention, Design,
Underpinning Model and
Duration
Aim and Target Population
Measures
Comparator
Tasks
Findings
(n=32 intervention group,
n=34 control group)
loving-kindness
meditation
6
Mak et al.
(2018)
Hong Kong
Self-Compassion
Programme (SCP)
3-arm randomised,
parallel, positive-
controlled, noninferiority
trial, online study, based
MSC, 4 weeks
To examine the efficacy of a
mobile app-based self-
compassion programme in
improving mental well-
being and reducing distress
among adults in general
population
(n=180 intervention group,
n=160 cognitive behavioural
group)
WHO’s 50item WBI,
The 6-item K6,
MAAS, SCS (13
items only),
Depressed Mood
and Anxiety
Subscales of the
ACS, 9-item
Discomfort with
Ambiguity sub scale
from the NCS, CSQ
Pre-test post-
test, 3 month
follow up
against a
cognitive
behavioural
programme:
CBP (AC
group)
Compassionate body
scan, affectionate
breathing, loving-
kindness meditation,
compassionate
walking, self-
compassion break,
self-compassion
journaling
Improved
mental well-
being and
reduced
psychological
distress.
Enhanced
mindfulness
awareness at
post
programme
7
Tung (2020)
Hong Kong
Mindful Self-Compassion
Programme, RCT, in
person, group study
design, based on MSC, 8
weeks
To increase self-compassion
and reduce stress in nursing
students in Hong Kong
(n=33 intervention group,
n=44 control group)
Chinese versions of
PSS, SCS, ProQOL-5,
FFMQ
Pre-test post-
test, 1 month
follow up
against a
waitlist group
Meditations
(affectionate
breathing,
compassionate body
scan, loving-
kindness), informal
Reduced stress,
improved self-
compassion
Chapter 3: Meta-Analysis of Compassion-Based Interventions
60
Note. SCS Self-Compassion Scale, HADS Hospital Anxiety Depression Scales; FFMQ Five Facet Mindfulness Questionnaire; CS Compassion for others
Scale; MADRS Montgomery-Åsberg Depression Rating Scale; PSQI Pittsburgh Sleep Quality Index; PSS Perceived Stress Scale; RSES Rosenberg Self-
Esteem Scale; WHOQOL World Health Organization Quality of Life; BDI Beck’s Depression Inventory; RRS – Ruminative Responses Scale; STAI - State-
Trait Anxiety Inventory; MAAS Mindfulness Attention Awareness Scale; WBI Well-Being Inventory; CHIPS Cohen-Hoberman Inventory of Physical
Symptoms; TMMS Trait Meta-Mood Scale; CESD Center for Epidemiological Studies Depression Scale; FOCS Fear of Compassion Scale; SMS State
Mindfulness Scale; DACS Depression Anxiety Cognition Scale; MMS Multiple Mood Scale; SCES Self-Conscious Emotions Scale; DASS Depression
Anxiety Stress Scale; APS R Almost Perfect Scale Revised; EPDS Edinburgh Postnatal Depression Scale; WHO World Health Organisation; PSI
Parenting Stress Index; CPBQ Comprehensive Parenting Behaviour Questionnaire; IBQ Infant Behaviour Questionnaire; K6 Kessler Psychological
No
Author, Year
and Country
Intervention, Design,
Underpinning Model and
Duration
Aim and Target Population
Measures
Comparator
Tasks
Findings
practices (soothing
touch, compassionate
walking, letter
writing, listening),
homework
8
Wong & Mak
(2016)
Hong Kong
Self-Compassion Writing,
Mixed research design
with a RCT, online study,
based on Neff (2009),
Leary et al. (2007), 1
week
To examine the efficacy of
self-compassion writing on
post-writing mood, physical
and psychological health in
Hong Kong Chinese
university students
(n=33 intervention group,
n=32 control group)
SCS (13 items only),
PANAS, The 10-item
CESD, The 33-item
CHIPS, The 30-item
TMMS
Baseline, 1
month and 3
month follow
up against a
control
writing group
(AC group)
Writing on an adverse
recent event and
experiences about
this event using
mindfulness, common
humanity and self-
kindness
Increased post-
writing
negative affect,
reduced
physical
symptoms in
intervention
group.
Chapter 3: Meta-Analysis of Compassion-Based Interventions
61
Distress Scale; FSCRC Fear of Self-Criticising/Attacking and Self-Reassuring Scale; ACS Affective Control Scale, NCS Need for Closure Scale; NAS Non-
Attachment Scale; CSQ The client satisfaction questionnaire; LCS Loving-Kindness Compassion Scale; ISS Internalised Shame Scale; SWLS Satisfaction
with Life Scale; ProQOL-5 Professional Quality of Life Scale; PANAS Positive And Negative Affect Scale; SHS - Subjective Happiness Scale; CAMS-R -
Cognitive and Affective Mindfulness Scale; SSCS L - State Selfcompassion ScaleLong Form.
Chapter 3: Meta-Analysis of Compassion-Based Interventions
62
3.3.3 Compassion Outcomes
Separate analyses were conducted for studies with WLC groups and studies with AC
groups respectively. Prior to that, effect sizes and heterogeneity statistics for self-compassion (SC)
were tested for the two categories (Table 3.3). A random effects model was conducted for all
analyses.
Table 3.3. Post intervention effects on self-compassion.
Category
Outcome
k
N
d
z
p for d
Q
p for Q
I2
Studies with a
WLC
SC
4
478
0.86
4.27
<0.0001
8.95
0.03
66%
Studies with
an AC
SC
4
534
0.19
2.06
0.04
3.15
0.25
5%
Note. SC = self-compassion; k = number of samples; N = number of participants contributing to
the outcome; d = standardised mean difference effect size; z = z-score; Q = test statistic for
heterogeneity; p = test for significance evaluated against .05; I2 = measure of degree of
heterogeneity.
3.3.3.1 Compassion-Based Interventions Compared to Waitlist Control Groups
A significant large effect size was found for self-compassion, d =.86, k = 4, 95% CI [0.46-
1.25], p < .0001 when comparing the intervention group with a WLC group (Figure 2). There was a
significant amount of heterogeneity in effect sizes for self-compassion, Q(3) = 8.95, p = .03, I2 =
66%. The high degree of statistical heterogeneity suggests that results should be interpreted with
caution (Kirby et al., 2017).
Chapter 3: Meta-Analysis of Compassion-Based Interventions
63
Figure 2. The effect of compassion-based interventions with wait-list control groups on self-
compassion.
3.3.3.2 Compassion-Based Interventions Compared to Active Control Groups
When looking at compassion intervention groups compared with AC groups, the results
indicated a significant small effect size for self-compassion, d =.19, k = 4, 95% CI [-.57-.40], p = .04.
Heterogeneity of variance in the effect sizes for self-compassion was not significant, Q(3) = 3.15, p
= .37, I2 = 5%. This means, that despite the differences in the intervention design, population and
other variables, the overall variability in studies with an AC group was relatively negligible (Weiss
et al., 2016). See Figure 3 for a visual representation of the effects.
Figure 3. The effect of compassion-based interventions with active control groups on self-
compassion.
3.3.4 Risk of Bias within Studies
The risk of bias evaluation is displayed in Figure 4. Overall, the summary figure of risk of
bias indicated a low risk of bias across studies (as indicated in the grey area). However, several
studies failed to report the method of randomisation 2, 3, 5, 8, performance bias 5, 7, and detection
Chapter 3: Meta-Analysis of Compassion-Based Interventions
64
bias 1, 2, 4, 6, 7. Whilst all studies discussed the attrition rates and possible reasons for participant
dropouts, they indicated a low risk of bias for reporting bias, selection bias, and other sources of
bias.
Figure 4. Risk of bias graph across studies.
3.3.5 Risk of Bias across Studies
Due to the limited number of studies included in the present meta-analysis, this study was
not sufficiently powered to assess publication bias across studies (Borenstein et al., 2021).
3.4 Discussion
This is the first meta-analysis to explore the efficacy and cross-cultural applicability of
compassion-based interventions in Asian populations. Incorporating eight RCT studies which
gathered data from 1012 participants from Thailand, Japan, China, and Hong Kong, this study
aimed to answer the question ‘can compassion-based interventions increase compassion in
people living in Asian communities?’.
In consideration of this, significant effect sizes were reported for increased levels of self-
compassion in the intervention groups when compared to both WLC groups and AC groups. As
predicted, effect sizes of studies including an AC group were lower, when compared to studies
with WLC groups (Cujipers et al., 2016; Kirby et al., 2017). This implies that the AC interventions
may have also increased self-compassion among participants in the AC groups to some extent
Chapter 3: Meta-Analysis of Compassion-Based Interventions
65
(Kirby et al., 2017). This raises the question whether the AC interventions also incorporated
compassion-enhancing tasks, or whether engaging in any intervention (possibly with a well-being
indicator) increases self-compassion in general. It is important to note, that although an overall
significant large effect size was observed for self-compassion in studies containing WLC groups,
they also indicated a significantly large variability for self-compassion, reducing the confidence in
the interventions used. This implies that despite the encouraging results from the compassion-
based interventions (with WLC groups), the variability across studies were considerably high
(Weiss et al., 2016). In consideration of the research question, this meta-analysis evidenced
promising findings that compassion-based interventions can increase self-compassion in
participants from Asian communities.
Kirby et al. (2017) concluded in their review, that there is an evident lack of clarity in
relation to the most appropriate measure of self-compassion. This suggestion is still apparent in
this review as almost all the studies discussed in this review used the SCS to measure self-
compassion, whilst some of them acknowledged the criticisms of the scale (e.g., Arimitsu, 2016;
Huang et al., 2021). As the CEAS (Gilbert et al., 2017) was developed to measure all three flows of
compassion (self, to/from others) whilst also addressing issues surrounding the SCS, it seems fair
to propose that the CEAS may be a more appropriate measure of compassion. Indeed, recent
research has used this measure in Asian countries such as Japan (Asano et al., 2020) and Sri Lanka
(Kariyawasam et al., 2021; Kariyawasam et al., 2022), and emphasised on the advantages of using
this measure (Asano et al., 2020). Thus, it would be useful to investigate the effectiveness of
compassion-based interventions using the CEAS in Asian communities to further understand the
interplay between these flows of compassion and well-being.
3.4.1 Strengths and Limitations
A strength of this meta-analysis is that it only included papers with a specific focus on
compassion cultivation and excluded interventions that prioritised other elements such as
Chapter 3: Meta-Analysis of Compassion-Based Interventions
66
mindfulness (e.g., Mindfulness Based Stress Reduction, Mindfulness Based Cognitive Therapy).
One intervention (Huang et al., 2021) however, was informed by both MSC (Neff & Gerrmer,
2013) and CMT (Gilbert, 2009b) approaches. This means that is difficult to differentiate which
approach produced which outcome (e.g., whether self-compassion increased due to the
mindfulness element, or the compassion element or both) when assessing the efficacy of
compassion interventions based on integrated approaches (Kirby et al., 2017).
Overall, this review highlights the lack of research exploring the effectiveness of
compassion-based interventions in Asian communities, as there were only four Asian countries to
have conducted RCTs of compassion-based interventions so far. This was surprising given that
many Asian countries such as Japan (Arimitsu, 2016) and Sri Lanka (Kariyawasam et al., 2021) are
familiar with the concept of compassion due to the significant Buddhist influence in these
cultures.
Whilst the rigorous inclusion criteria helped to choose studies with higher methodological
quality, this was also a limitation of this meta-analysis as that lead to the exclusion of several
studies that did not meet the eligibility criteria (e.g., Finlay-Jones et al., 2018, Noh & Cho, 2020;
Yeung et al., 2022). Although there were a few existing compassion-based interventions in Asian
communities which were able to promote compassion and self-compassion, these were not RCTs
(e.g., Finlay-Jones et al., 2018), or did not include measures to assess compassion (e.g., Lo et al.,
2015), and therefore, had to be excluded from this review and the analysis.
Another limitation this meta-analysis discovered was that all the studies discussed in this
meta-analysis only assessed self-compassion, disregarding the other two flows, namely
compassion to others, and compassion from others (Gilbert et al., 2017). Ferrari et al. (2019) also
narrowed the search to self-compassion-based interventions only, when conducting their meta-
analysis of compassion-based interventions. Gilbert et al. (2017) argued that compassion is not
only felt for the self, but also towards and from others, whilst studies have discussed how these
flows interact with one another (Rashid et al., 2021), and are linked with increased well-being
(Asano et al., 2020; Gilbert et al., 2017).
Chapter 3: Meta-Analysis of Compassion-Based Interventions
67
In addition, findings of the studies included in this review were largely limited to non-
clinical populations (e.g., Arimitsu, 2016; Guan et al., 2021; Huang et al., 2021), indicating the
need for further research to be conducted among both clinical and non-clinical populations in
Asian communities. Despite the common use of small underpowered sample sizes in majority of
evidence-based interventions (Kirby et al., 2017), this review noted that the papers generally
included a small sample size, which limits the generalisability of the findings (Huang et al., 2021).
The limited number of RCTs also meant that a funnel-plot was not suitable to assess the risk of
publication bias (Higgins et al., 2011) leaving the risk of publication bias undetermined.
Similar to Kirby et al. (2017), this meta-analysis did not include studies that assessed
compassion using heart rate variability and other bodily measures. Incorporation of these
measures would have increased the researchers’ understanding of the effectiveness of the
interventions (Luo et al., 2018; Tian et al., 2020) at a physiological level. Furthermore, the RCTs
discussed in this meta-analysis included a range of self-reported measures of depression, anxiety,
psychological distress, and well-being questionnaires, which may be particularly problematic due
to the stigma of mental illness in Asian cultures as people might respond with a social desirability
bias (Wong & Mak, 2016). Therefore, it seems fair to propose that future research should focus on
using physiological measures (Finlay-Jones et al., 2018) in addition to self-report measures to help
build a comprehensive understanding of the efficacy of compassion-based approaches.
Another limitation was that the secondary gain relating to well-being or distress was not
tested in this analysis, as the primary aim was to focus on understanding the impact of
compassion-based interventions on increasing compassion in Asian communities. However, in
addition to increased self-compassion, there is clear evidence to suggest that the compassion-
based interventions also increase well-being and reduce distress in Asian communities. In fact,
studies included in this meta-analysis reported significant increases in mindfulness (Mak et al.,
2018), coping (Huang et al., 2021), and quality of life (Anuwatgasem et al., 2020), and significant
decreases in depression, anxiety, and stress (Anuwatgasem et al., 2020; Guan et al., 2021; Guo et
al., 2020; Huang et al., 2021). Now that this review has found promising evidence for the increase
Chapter 3: Meta-Analysis of Compassion-Based Interventions
68
in self-compassion using these interventions, future research would benefit from exploring the
implications for well-being and reduced distress and psychopathology.
Additionally, this meta-analysis comprised studies that varied in multiple components
such as the intervention duration, content, and targeted population. Thus, prospective
interventions should investigate the contents and structure of compassion-based interventions, to
determine the most suitable intervention for their targeted populations.
3.4.2 Clinical Implications
This review found potentia