ArticlePDF Available

The Longitudinal Relationship Between Youth Intergroup Contact and Social Cohesion Outcomes in Two Divided Societies

October 2024
European Journal of Social Psychology

DOI:10.1002/ejsp.3121

License
CC BY 4.0

Authors:

Shelley McKeown

University of Oxford

Christoph Daniel Schaefer

University of Oxford

Pier-Luc Dupont

Swansea University

Show all 8 authorsHide

Intergroup contact has long been established as a prejudice‐reduction tool in divided societies, with contact being particularly effective during adolescence. A large proportion of evidence, however, draws on cross‐sectional surveys or analytical approaches that do not distinguish between‐ and within‐person effects. In the present research, we address this by exploring the potential of intergroup contact longitudinally on social cohesion–related outcomes amongst youth (aged 14–19) in Belfast (Study 1, N = 231) and Bradford (Study 2, N = 159). Measures included intergroup contact, outgroup attitudes, intergroup anxiety, outgroup empathy and outgroup prosocial behaviour across three time points. Using random‐intercept cross‐lagged panel models, results demonstrate between‐person associations of contact with our outcomes, but limited within‐person changes. Our findings demonstrate the potential and limitations of intergroup contact for social cohesion–related outcomes for youth growing up in divided societies, pointing to the need for developmental‐focused future research.

Content uploaded by Sumedh Rao

Content may be subject to copyright.

European Journal of Social Psychology

RESEARCH ARTICLE

The Longitudinal Relationship Between Youth Intergroup

Contact and Social Cohesion Outcomes in Two Divided

Societies

Shelley McKeown1,2Christoph Daniel Schaefer1,2Shazza Ali1Pier-Luc Dupont3David Manley1

Sumedh Rao1,2Laura K. Taylor4Rose Meleady5

1School of Education, University of Bristol, Bristol, UK 2Department of Experimental Psychology, University of Oxford, Oxford, UK 3Department of Politics,

Philosophy and International Relations, Swansea University, Swansea, UK 4School of Psychology, University College Dublin, Dublin, UK 5School of

Psychology, University of East Anglia, Norwich, UK

Correspondence: Shelley McKeown Jones (shelley.mckeownjones@psy.ox.ac.uk)

Received: 15 January 2024 Accepted: 18 September 2024

Keywords: intergroup contact | prejudice | prosocial behaviour | social cohesion | youth

ABSTRACT

Intergroup contact has long been established as a prejudice-reduction tool in divided societies, with contact being particularly

effective during adolescence. A large proportion of evidence, however, draws on cross-sectional surveys or analytical approaches

that do not distinguish between- and within-person effects. In the present research, we address this by exploring the potential of

intergroup contact longitudinally on social cohesion–related outcomes amongst youth (aged 14–19) in Belfast (Study 1, N=231) and

Bradford (Study 2, N=159). Measures included intergroup contact, outgroup attitudes, intergroup anxiety, outgroup empathy and

outgroup prosocial behaviour across three time points. Using random-intercept cross-lagged panel models, results demonstrate

between-person associations of contact with our outcomes, but limited within-person changes. Our findings demonstrate the

potential and limitations of intergroup contact for social cohesion–related outcomes for youth growing up in divided societies,

pointing to the need for developmental-focused future research.

1 Introduction

High levels of ethnic, religious and racial prejudice are a prevail-

ing feature in many societies across the globe, especially those

with a long history of political violence and ethnic tensions. One

approach to reducing such prejudice is to implement strategies

that promote positive and meaningful interactions between the

groups in conflict, known as intergroup contact (Allport 1954).

Substantial evidence shows that intergroup contact entails pos-

itive outcomes in different forms and amongst different groups

(e.g., Paluck, Green, and Green 2019; Pettigrew and Tropp 2006).

Perhaps unsurprisingly, therefore, intergroup contact principles

are often applied to promote social cohesion–related outcomes

in divided societies, including through education and other

community-focused interventions (Al Ramiah and Hewstone

2013; McKeown and Cairns 2012).

A large proportion of the evidence for the benefits of intergroup

contact, however, comes either from cross-sectional studies

that cannot establish causality or from longitudinal studies that

do not clearly separate between-person variance (differences

between individuals over time) from within-person variance

(changes in individual values over time). Recent analyses

using statistical techniques have found limited evidence of

within-person reduction of prejudice following within-person

increases in intergroup contact (Friehs et al. 2024; Hodson and

Meleady 2024). Commentators have posited several reasons

as to why this might be the case. One suggestion is that there

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly

cited.

European Journal of Social Psychology, 2024; 0:1–16

https://doi.org/10.1002/ejsp.3121

1of16

could be ‘third variable’ explanations such that those who

are more likely to hold negative attitudes due to individual

difference factors are less likely to engage in contact (Hodson

and Meleady 2024). Another is the idea that the impact of contact

experiences on prejudice diminishes as they become more

common for an individual over time (Friehs et al. 2024;Wölfer

et al. 2016). There may, however, be a related developmental

explanation whereby contact is most influential in reducing

prejudice in early-to-mid childhood, with a continued reduced

effect through adolescence and into adulthood due to the

stabilisation of intergroup attitudes. Evidence for this assertion is

supported by Merrilees et al. (2023) who found that the strength

of the intergroup contact quality–prejudice relationship drops

through adolescence, and by Wölfer et al. (2016) who found that

intergroup contact as measured via friendship was a stronger

predictor of adolescent compared to adult intergroup attitudes.

Few studies, however, have differentiated between-person and

within-person effects of contact on prejudice amongst youth. An

exception is Friehs et al. (2024) who examined contact effects

on attitudes over time with a sample of school children in

England (Mage =12.11 years) and found evidence for between-

person but not within-person effects of contact on prejudice.

It remains unclear, however, whether the lack of observed

effects applies to mid-late adolescence, and further, to outcomes

beyond prejudice reduction. We know, for example, that contact

has proven to be associated with lower levels of intergroup

anxiety (Pettigrew and Tropp 2008), higher levels of empathy

(Pettigrew and Tropp 2008) and to encourage prosocial actions

targeted towards outgroup members (Koschate et al. 2012;

McKeown and Taylor 2018), but whether this can be observed

when tested for within-person changes remains untested

to date.

The present research, therefore, aims to examine the between-

and within-person potential of the quantity and quality of

intergroup contact experiences in adolescence on outcomes

that relate to social cohesion, including outgroup attitudes,

outgroup empathy, intergroup anxiety and outgroup prosocial

behaviour. We assess these relationships amongst youth (aged 14–

19) growing up in Belfast and Bradford through a three-time-point

survey study. Youth in both settings are experiencing the effects

of past and present ethno-religious tensions, which highlights

the importance of further exploring the potential of intergroup

contact to promote social cohesion agendas.

1.1 Intergroup Contact Theory

For decades, intergroup contact has been considered one of

psychology’s most effective strategies for improving intergroup

relations. Inspired by early observations of the benefits of

desegregation on racial attitudes, Allport’s (1954) ‘contact hypoth-

esis’ holds that encouraging interactions across group lines is

key to reducing hostilities and improving intergroup relations.

According to the original formulation of the contact hypothesis,

contact works best under four conditions: cooperation, common

goals, equal status and social/institutional support. Reflecting

its intuitive appeal and applied potential, this hypothesis has

become one of the most extensively tested ideas in psychology.

Multiple meta-analytic integrations attest to the robust, positive

impact of intergroup contact on prejudice (e.g., Davies et al. 2011;

Lemmer and Wagner 2015; Pettigrew and Tropp 2006). Across

different implementations, participant populations and bases for

group membership, more contact is generally associated with less

prejudice, even when not all Allport’s contact conditions are met.

This is the case for both intergroup contact quantity (how often

interactions occur) and intergroup contact quality (how positive

interactions are).

The contact hypothesis has now evolved into a sophisticated

theoretical framework, more complex and complete than All-

port’s (1954) original formulation, specifying how, when and

why contact is associated with reduced prejudice (Hodson and

Hewstone 2013; Pettigrew and Tropp 2011). Intergroup contact has

been found, for example, to play an important role in reducing

intergroup anxiety, promoting empathy (Pettigrew and Tropp

2008) and, in turn, reducing prejudice. These emotive factors

can also have knock-on effects on indicators of social cohesion.

Reducing intergroup anxiety through imagined contact can, for

example, lead to more contact motivation and a lower tendency to

avoid outgroup members (Turner, West, and Christie 2013). There

is also evidence that empathy can lead to more positive attitudes

and, in turn, to actions that have benefits for outgroup members

and society at large (see the empathy-attitudes-action model;

Batson et al. 1997;2002). A study conducted amongst children

and young people in Northern Ireland, for example, found that

empathy can impact outgroup attitudes, and, in turn, not only

individual-helping actions, but also actions that are indicative of

structural and cultural change (Taylor and McKeown 2021)that

are arguably linked to social cohesion. Evidence can also be found

for the direct effects of contact on prosocial actions, including

those intended to benefit outgroup members (Koschate et al. 2012;

McKeown and Taylor 2018). Outcomes such as these are arguably

crucial for building more cohesive societies. Taken together,

there is significant evidence that contact can reduce prejudice

and can have wider outcomes beyond prejudice reduction that

have potential to impact individuals, groups and society at

large. There are, however, several methodological critiques of

contact research that must be acknowledged, which we discuss

now.

1.2 Methodological Critiques of Contact

Research

A significant challenge to the apparent effects of intergroup

contact arises through the cross-sectional nature of many of the

studies on which the evidence base is constructed, as they cannot

establish causality. Around 71% of the studies included in the

seminal meta-analysis of Pettigrew and Tropp (2006) were cross-

sectional (Christ and Wagner 2013), and only 5% of the studies

were experimental, varying widely in methodology and outgroups

(Paluck, Green, and Green 2019). Whilst experimental evidence

is an ideal way to test the causal effects of contact and has

shown reductions in prejudice, experiments are less frequently

conducted, with most studies being short term and involving

limited and infrequent contact (Paluck, Green, and Green 2019).

These observations point to a need to consider alternative designs

and approaches if we are to better understand contact effects over

time—whether they exist and whether they persist. Longitudinal

research holds considerable promise to measure such naturally

occurring psychological processes as they unfold.

2of16 European Journal of Social Psychology,2024

10990992, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.3121 by Sumedh Rao - <Shibboleth>-member@open.ac.uk , Wiley Online Library on [14/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

Longitudinal research on intergroup contact is growing. A

challenge, however, is that the majority of studies that have used

longitudinal designs to explore contact effects have relied on tra-

ditional methods of analysis such as the cross-lagged panel model

(CLPM) (e.g., Binder, Zagefka, and Brown 2009; Dhont, Van Hiel,

and Hewstone 2014;Swartetal.2011), which whilst valuable

are unable to clearly separate between-person variance from

within-person variance. This has implications for the conclusions

that can be drawn because, whilst we would expect to find that

in general people who engage in more contact report lower levels

of prejudice (between-person effects), we should also expect to

see individual variation such that someone who reports having

higher levels of contact also reports lower levels of prejudice

later (within-person effects). Here, the proportion of variance

that is stable across time can be understood as between-person

variance, whilst the proportion of variance that reflects individual

variability can be understood as within-person variance. While

between-person differences in the independent variable are not

necessarily problematic for causal inferences, between-person

differences reflected in the dependent variable can result in

inaccurate conclusions when using analytical methods such as

CLPMs. This is due to the influence of unobserved third variables

on the independent and the dependent variables. In such cases,

significant cross-lagged paths are assumed to represent evidence

that contact induces change that subsequently reduces prejudice,

but they may instead reflect stable differences (between people)

of third variables. Alternative statistical approaches are therefore

essential to differentiate within-effects from between-effects

more unequivocally, to move closer to a causal analysis in contact

research and to better understand the nature and effects of

intergroup contact on prejudice reduction and beyond.

1.3 Examining Within-Person Changes and

Between-Person Differences

Recent analyses using a new statistical technique capable

of separating within-person variance from between-person

variance—the ‘random intercept cross-lagged panel model’

(RI-CLPM)—have yielded significant between-person negative

associations between contact and prejudice, but no evidence

of within-person change of prejudice following within-person

change of contact (Friehs et al. 2024; Hodson and Meleady 2024).

These findings are incongruent with mainstream thinking about

contact as a process capable of producing changes within people

over time in the form of lower prejudice. Indeed, between-

person associations without within-person changes would

be consistent with third-variable explanations, according to

which the relation between contact and prejudice results from

differences in personality (i.e., the type of people that are involved

in contact also report lower prejudice) and not from a process

in which contact changes attitudes. Therefore, it is necessary

to ask why no associations of within-person changes are being

observed.

It is argued here that one reason evidence has not been found

that within-person change in intergroup contact predicts within-

person change in prejudice in these analyses is because we need

to consider if and how intergroup effects vary over time and

for different age groups. Evidence shows, for example, using

a five-wave study amongst youth in Northern Ireland that the

strength of the relationship between intergroup contact quality

and prejudice grows through adolescence, but then drops from

age 16 onwards (Merrilees et al. 2023). There is also evidence

that the relationship between contact, measured as friendship

through social networks, and prejudice reduction is stronger in

adolescence (aged 13–19) than in adulthood (aged 20–26, Wölfer

et al. 2016). It could be that these effects are due to the stabilisation

of attitudes during adolescence (Wölfer et al. 2016)andas

such, that contact may exert most of its within-person influence

in early-to-mid childhood, but then declines through adoles-

cence and into adulthood (leaving between-person differences

to drive more of the effects). That is, contact may exert mean-

ingful within-person effects mainly at a specific developmental

stage.

To our knowledge, few studies have considered whether there is

evidence for between- and within-person effects of contact during

adolescence as a critical period of development. An exception

is Friehs et al. (2024), who analysed social network data from a

five-wave social network sample of 11–12 year olds in England

to explore this and found no evidence of within-person change

in the contact–prejudice relationship. The authors note that this

finding may be due to a reduction of contact effects on prejudice

over time, i.e., the first series of intergroup interactions could be

particularly potent in reducing prejudice, but this effect could

reduce and stabilise with later repeated interactions. A connected

possibility, however, is that there are age-related effects. In

other words, it could be that between–within person dynamics

play out differently across stages of adolescence as indicated by

previous research exploring the contact–prejudice relationship

over time (Merrilees et al. 2023; Wölfer et al. 2016). It is the

nature of contact experiences in adolescence that we focus on

here.

2The Present Research

The present research pushes forward current knowledge by

exploring the potential impact of contact quantity and quality on

outcomes relevant to social cohesion for young people growing

up in socially divided societies. Previous research adopting an

RI-CLPM approach has explored the effects of contact measured

by quantity or friendship in the domain of harmony (i.e., liking;

Friehs et al. 2024; Hodson and Meleady 2024), and the domain

of equality (i.e., collective action, solidarity; Górska and Tausch

2023;Senguptaetal.2023), using data mostly from adult samples.

Here, we expand on these contributions in two ways by consider-

ing the longitudinal effects of both intergroup contact quantity

and intergroup contact quality: (1) amongst youth growing up in

divided contexts and (2) on outcomes related to social cohesion

that include outgroup attitudes, intergroup anxiety (measured as

comfort), outgroup empathy and outgroup prosocial behaviour.

Whilst we do not test developmental processes specifically, we

focus on exploring between- and within-person contact effects

during adolescence, as this may be one of the critical stages in

which to explore the nature of contact and its consequences, as

well as a time in which to intervene in promoting contact. If, for

example, we find that within-person effects of contact exist at a

specific developmental stage, then this suggests that contact is a

relevant intervention, but one that needs to be introduced earlier

than has been the focus in the field.

3of16

RI:

Attitudes

Attitudes T1 Attitudes T3Attitudes T2

Attitudes

T2 (W)

Attitudes

T3 (W)

RI:

Contact

Quantity

Attitudes

T1 (W)

RI:

Contact

Quality

Contact

Quality T1

Contact

Quality T3

Contact

Quality T2

Contact

Quality

T2 (W)

Contact

Quality

T3 (W)

Contact

Quality

T1 (W)

Contact

Quantity

T2 (W)

Contact

Quantity

T3 (W)

Contact

Quantity

T1 (W)

Contact

Quantity T1

Contact

Quantity T3

Contact

Quantity T2

FIGURE 1 Representation of our conceptual model used in our RI-CLPM analyses. The figure depicts the model with the outcome variable

‘outgroup attitudes’ as an example. In our statistical analyses, one model is specified for each of the outcome variables, which were: outgroup attitudes,

outgroup empathy, intergroup anxiety and outgroup prosocial behaviour. The latent variables (i.e., random intercepts and centred variables) are derived

from the measured variables. The random intercepts are marked with ‘RI’. The centred variables (marked with ‘(W)’) reflect fluctuations around

individual average values (within-person variance). Of central interest in the RI-CLPM are the cross-lagged paths between the centred variables. (Not

shown in the figure are the errors of the centred variables at T2 and T3; neither shown are the permitted correlations between the random intercepts,

between the centred variables at T1 and between the errors of the centred variables within T2 and within T3.)

A simplified version of our conceptual and RI-CLPM model is

presented in Figure 1. Here, we include only one outcome variable

as an example, to ease readability.

To achieve our aims, we draw on three-time-point survey data

collected as part of a larger study on youth intergroup contact in

two settings: Belfast, Northern Ireland (Study 1), and Bradford,

England (Study 2). This multicontext analysis enables us to

examine contact effects in one historical and ongoing conflict

between ethno-religious groups and in another context of more

contemporary division between ethnic groups. Whilst previous

RI-CLPM studies have been conducted in societies with ethnic

tensions such as England (Friehs et al. 2024), Germany (Friehs

et al. 2024) and New Zealand (Sengupta et al. 2023), to our

knowledge, few have explored within-person effects of contact

in what might be best described as conflict or post-conflict

settings where contact tends to be less frequent and perhaps even

less positive due to the nature of segregated communities and

inequalities that continue to permeate society. We explain more

about each of these contexts below.

2.1 Research Context

Youth growing up in Belfast and Bradford are experiencing the

effects of past and present ethnic tensions. The Northern Ireland

context can be understood as a setting marked by protracted

ethno-religious conflict between those who wish Northern Ire-

land to remain part of the UK (Protestant) and those who wish

for reunification with Ireland (Catholic). Despite the 1998 Peace

Agreement, violence remains (Taylor et al. 2016) and community

4of16 European Journal of Social Psychology,2024

relations are a government priority. For example, the Together:

Building a United Community Strategy (The Executive Office

[Northern Ireland] 2013) is a key focus of the Belfast City

Council’s Good Relations Programme.

Whilst Bradford does not have the same historical legacy of

conflict as Belfast, it is an ethnically, religiously and culturally

diverse context and thus a pertinent case of intergroup dynamics

to explore the effects of youth intergroup contact. According to

the 2021 census, most individuals living in Bradford belong to

the White/White British (56.7%) or Asian/Asian British (32.1%)

communities. Bradford is a site of integration focus in the

Government’s Integrated Communities Strategy Green Paper

(HM Government 2018) due to historical and continued ethnic

tensions. For example, in 2001, the Bradford riots, which occurred

due to increased tensions between the Asian and White British

communities following an antifascist rally in the city, resulted

in significant damage to property and to individuals, leading to

a large number of arrests (Waddington 2010). Due to instances

like this, promoting social cohesion is a priority in the Bradford

context, and policies such as ‘getting along’ (promoting greater

interaction) are a key pillar in the Bradford for Everyone Strategy

(Stronger Communities Partnership 2019). Given the socially

divided nature of both Belfast and Bradford, understanding the

potential effects of intergroup contact for prejudice reduction and

beyond is vital to inform policy and practice in both settings. In

the present research, we focus specifically on contact with and

attitudes towards the dominant ethno-religious groups in each

of the two contexts: Protestant and Catholic youth growing up

in Belfast (Study 1) and White/White British and Asian/Asian

British youth growing up in Bradford (Study 2).

3Study 1

Study 1 explores the within- and between-person effects of contact

quantity and quality on outgroup attitudes, outgroup empathy,

intergroup anxiety and outgroup prosocial behaviour amongst

youth in Belfast, Northern Ireland.

3.1 Sample

A total of 488 participants, recruited from three participating

secondary schools in Belfast, filled in the survey at Time 1, 419

participants at Time 2 and 460 at Time 3. Of those participants,

we were able to match data for 282 across all the three time

points. As we were interested specifically in Catholic–Protestant

relations, we removed 51 participants who could not be assigned

to either the Catholic or Protestant community and/or who were

not born in Northern Ireland, Ireland or the wider UK context.

This resulted in a final sample of 231 participants (Mage,Time1 =14.7

years, SDage,Time1 =0.71; 36% female, 58% male, 6% other gender or

not stated; 34% Catholic, 66% Protestants). Of the total sample,

30% reported receiving free school meals (an indicator of low

socio-economic status), 2% described the financial situation of

their family as not very well off or as not at all very well off,

55% as average and 35% as well off or very well off (7% missing

values).

3.2 Recruitment

Due to our interest in exploring interactions between Catholic

and Protestant youth, we initially aimed to recruit ethno-

religiously mixed schools in Belfast to take part in the research.

Schools were selected using publicly available demographic

school data on the Northern Ireland Department of Education

website. All secondary-level schools in Belfast where the compo-

sition of the Catholic or Protestant community was no more than

60% of the pupil population (n=7) were contacted in July 2021,

and two agreed to take part in the research: one grammar school

(1438 enrolled, 44% Protestant) and one integrated school (628

enrolled, 44% Protestant). A second school recruitment round

was carried out in September 2021 with a further 11 schools

with no more than 85% of youth from either the Catholic or

Protestant community contacted to take part. This resulted in one

additional school agreeing to participate (controlled boys’ school,

1093 enrolled, 85% Protestant).

Of the participants in the final, matched, sample used for the

data analysis (n=231), 52% were from the grammar school (n=

120), of which 33% reported having a Catholic background and

67% reported having a Protestant background. In the integrated

school (n=99), which accounted for 32% of the final sample, 48%

of participants reported having a Catholic background and 52%

reported having a Protestant background. In the controlled school

(n=38, 16% of the final sample), 100% of participants reported

having a Protestant background.

3.3 Measures

In addition to a series of demographic questions (i.e., age, gender,

community background and free school meal as a proxy for socio-

economic status), youth participants completed the following

survey measures:

3.3.1 Contact Quantity

Adapted from Tam et al. (2009), contact quantity was measured

by asking participants to rate on a four-item scale ranging from

0(none)to3(alot) how much contact they have with each

of the Protestant and Catholic communities ‘at school’, ‘in your

neighbourhood’, ‘across all social situations’ and ‘online/on social

media’. Responses to this question for the outgroup commu-

nity only were considered (e.g., if Protestant, responses about

Catholics were selected). A higher score implied higher frequency

of outgroup contact (Time 1: α=0.76; Time 2: α=0.74;

Time 3: α=0.75). To assess measurement invariance of contact

quantity across the three measurement waves for this sample, we

specified confirmatory factor analysis models. All four contact

quantity items were used as indicators for one latent variable

per time point while allowing the residual variances of the same

items to correlate across waves. The configural models yielded

acceptable fits (CFI =0.971; TLI =0.951; RMSEA =0.062; SRMR

=0.043). The constraints testing for metric invariance did not

lead to a significant deterioration of the fit (p=0.359). The

additional constraints testing for scalar invariance neither led to a

significant deterioration of the fit compared to the previous model

(p=0.072).

5of16

3.3.2 Contact Quality

To measure contact quality, youth responded to a series of bipolar

adjectives regarding how they find it when they meet people from

the Protestant and Catholic communities in general, adapted

fromTametal.(2009). This included: do you find the contact

‘pleasant or unpleasant?’ (0 =very unpleasant; 5=very pleasant),

‘competitive (trying to “win” or “beat” each other) or cooperative

(where you work together)?’ (0 =very competitive;5=very

cooperative), ‘casual (e.g., conversations that don’t really matter)

or meaningful (e.g., having deep or personal conversations)?’

(0 =very causal;5=very meaningful). An additional item was

developed to measure online contact quality: ‘When you interact

with people online from the following communities, do you

find the contact pleasant or unpleasant’ (0 =very unpleasant;5

=very pleasant). To examine outgroup contact specifically, we

used only the items that refer to the outgroup (e.g., for Catholic

participants we chose their responses about interactions with

Protestants).

To assess measurement invariance of outgroup contact quality

across the three measurement waves, we specified confirmatory

factor analysis models, using the four contact quality items as

indicators for a latent variable per time point and allowing the

residual variances of the same items to correlate across time

points. Since the configural models did not lead to an acceptable

fit, we omitted the third item above from the scale, which had the

lowest factor loadings (Time 1: α=0.72; Time 2: α=0.75; Time 3: α

=0.77). The resulting three-item measure resulted in good fits in

the configural models (CFI =0.995; TLI =0.988; RMSEA =0.034;

SRMR =0.033). The constraints testing for metric invariance did

not lead to a significant deterioration of the fit (p=0.566). The

additional constraints testing for scalar invariance neither led to a

significant deterioration of the fit compared to the previous model

(p=0.076).

3.3.3 Outgroup Attitudes

To assess outgroup attitudes, youth completed a feeling ther-

mometer (Cairns et al. 2006) asking them to rate how favourable

they feel towards the outgroup (Protestant or Catholic) commu-

nity (0 =unfavourable, 100 =favourable). Before entering this

variable into the statistical models, it was divided by a constant,

in this case by 20. This procedure is recommended by Muthén

and Muthén (1998–2017: 524) who specify observed variables with

variances greater than 10 should be divided by a constant.

3.3.4 Outgroup Empathy

To measure outgroup empathy, two items were adapted from

Hughes et al. (2013). Youth were asked to rate the following

statements: ‘I care about the problems faced by the [respective

outgroup]’ and ‘I find it easy to see things from the point of view

ofthe[respectiveoutgroup]’.Thescalesrangedfrom0(strongly

disagree)to5(strongly agree). The Spearman–Brown coefficient

was ρ=0.69 for Time 1, ρ=0.80 for Time 2 and ρ=0.82 for

Time 3 (Eisinga, Grotenhuis, and Pelzer 2013, demonstrating that

the Spearman–Brown coefficient is the most suitable reliability

measure for two-item scales).

3.3.5 Intergroup Anxiety

Youth were asked to respond to a single item statement on their

levels of comfort when interacting with members of the outgroup

as an indicator of intergroup anxiety, specifically: ‘How comfort-

able do you feel talking to people from the following communities

(more than just saying hello)?’. The scale ranged from 0 (very

uncomfortable)to5(very comfortable) and participants reported

feelings towards interacting with both the Catholic and Protestant

communities. Here, we focus on responses for outgroup contact

only. Responses to this item were reverse coded, so that a high

score indicates high levels of intergroup anxiety.

3.3.6 Outgroup Prosocial Behaviour

Youth were first asked to indicate how often they engaged in

concrete acts of prosocial behaviour (such as helping, cooperation

and concern) (adapted from Taylor et al. 2014). They were then

asked to use this list to report how often they implemented these

acts towards the Protestant or Catholic communities on a scale

ranging from 0 (never)to5(often). It is responses to this single

item that we used to measure outgroup prosocial behaviour in the

present research. Again, we focus only on responses to the item

referring to the outgroup community specifically.

3.4 Procedure

Prior to data collection in the three participating schools, ethical

approval was obtained from the University of Bristol and infor-

mation sheets were sent out to all potential participant’s parents

providing them with the opportunity to opt out their young

person as well as to participate in the research themselves. Survey

data were collected in each of the three Belfast secondary schools

during November 2021–January 2022 (Time 1), March 2022 (Time

2) and May 2022 (Time 3). Surveys took place during school hours

and in the presence of classroom teachers and researchers. In two

of the schools, surveys were completed in computer rooms via

Qualtrics whilst in the third school, they were completed using

paper and pen versions. Each classroom teacher was provided

with a survey information pack prior to data collection and

following this, all pupils were informed about the purpose of the

research, asked to read the information sheet provided and if they

wished to take part, to confirm written consent. The classroom

teachers and researchers were available to answer any questions

during the 30–40-minute data-collection period. All participants

who completed a survey were given a £10 Amazon voucher as

remuneration. Each school was also given a £500 payment at the

end of the project.

3.5 Data Analysis Plan

In our statistical analyses, we examined the effects of both contact

quality and contact quantity on our outcome variables using RI-

CLPMs in Mplus (version 8.5, Muthén and Muthén 1998–2017).

The RI-CLPMs were specified as proposed by Hamaker (2015)as

well as Mulder and Hamaker (2021), separating between-person

variances from within-person variances of both the independent

and the dependent variables. This was implemented by specifying

6of16 European Journal of Social Psychology,2024

random intercepts as latent variables for all independent and

dependent variables. There is one random intercept as a latent

variable for each independent variable (contact quantity and

contact quality) and each dependent variable (outgroup attitudes,

outgroup empathy, intergroup anxiety and outgroup prosocial

behaviour), while each random intercept latent variable has

the respective measured variable of all waves as an indicator.

For example, there is one latent variable reflecting the random

intercept of contact quality which has the measured contact

quality variables of Time 1, Time 2 and Time 3 as its three

indicators. The factor loadings are set as 1. These latent variables

are allowed to correlate, representing correlation of between-

person differences (similar to cross-sectional correlations). Group

membership, Catholic or Protestant, was used as a predictor for

the random intercepts.

Additionally, each measured variable is an indicator for another

latent variable which captures the deviation of the measured

variable from the personal average value (of each individual) at

each time point. These time-specific latent variables thus reflect

within-person variations of the respective construct. Structural

relations were specified between the within components anal-

ogously to the traditional cross-lagged panel models, specifying

auto-regressive and cross-lagged effects. The difference to the

traditional cross-lagged panel model is that within RI CLPMs,

within-person deviations from the respective individual average

are used when looking at cross-lagged and autoregressive effects.

For example, within-person deviations of attitudes at Time 2 are

regressed on within-person deviations of attitudes at Time 1 and

on within-person deviations of contact quality at Time 1.

We specified separate models for each outcome variable to

prevent estimation problems resulting from a high number

of parameters. While there is no universal direct relationship

between the number of parameters and the required sample size,

models with more parameters tend to require larger sample sizes

(Jackson 2003; Kline 2016). This approach led to four models, one

for each of the outcome variables: outgroup attitudes, outgroup

empathy, intergroup anxiety and outgroup prosocial behaviour.

Cross-lagged effects of independent variables on the respective

dependent variable were specified. In addition, cross-lagged

effects of the dependent variable on the independent variables

were specified. Autoregressive effects were also specified for all

variables. Full information maximum likelihood estimation was

used to deal with missing values in our dataset.

3.6 Results

Descriptive statistics and bivariate correlations between all

variables at each time point are presented in Table 1.Eachof

the four RI-CLPM models, one associated with each outcome

variable, yielded a good fit to the data (see Table 2for fit indices)

using the criteria specified by Hu and Bentler (1999)whereby

cut-off values close to 0.95 of the CFI and the TLI, values close

to 0.08 of the SRMR and values close to 0.06 of the RMSEA are

indications of a good fit.

At the between-person level, we found that contact quantity was

positively correlated with outgroup attitudes (B=0.30, SE =0.05,

p<0.001), outgroup empathy (B=0.30, SE =0.05, p<0.001) and

outgroup prosocial behaviour (B=0.22, SE =0.03, p<0.001),

and negatively correlated with intergroup anxiety (B=−0.25,

SE =0.04, p<0.001). Similarly, we found that contact quality

was positively correlated with outgroup attitudes (B=0.38,

SE =0.09, p<0.001), outgroup empathy (B=0.38, SE =0.07,

p<0.001) and outgroup prosocial behaviour (B=0.25, SE =0.05,

p<0.001), and negatively correlated with intergroup anxiety (B=

−0.37, SE =0.07, p<0.001). These between-person correlations

indicate the degree to which the participants’ average levels of

contact (across time points) are associated with participants’

average levels of the respective dependent variable (across time

points). The correlations are thus similar to traditional cross-

sectional correlations and the findings align with what would

be expected based on the contact literature, that is, the positive

effects of high quality and frequent contact on outcomes being

associated with stronger social cohesion.

The statistical within-person cross-lagged effects on the depen-

dent variables in the four models are reported in Table 2.A

within-person variable reflects, for each time point and person,

the deviation from an individual’s average value. Between these

within-person variables, cross-lagged and autoregressive paths

were specified, analogously to the paths in traditional cross-

lagged panel models. These paths thus represent associations

between deviations from individual average values, which are

separate from between-person differences. Of special interest in

the present context are the cross-lagged paths which are linking

the individual’s deviation from the individual’s average contact

score with the individual’s deviation from its average attitude

score at the following wave (see Hamaker, Kuiper, and Grasman

2015). Significant effects were observed for contact quantity at

Time 1 on more positive attitudes at Time 2 and contact quality at

Time 2 on all four dependent variables at Time 3. These statistical

within-person cross-lagged effects reflect associations between

time-shifted deviations from individual average values. These

within-person effects thus indicate that individual deviations

(from the individual average values) in contact predicted changes

in the dependent variable at the following time point (in the form

of individual deviations of the dependent variable when control-

ling for the deviations of the previous time point). These findings

demonstrate some, but at the same time, limited evidence for

within-person change of contact on social cohesion outcomes.

3.7 Discussion

The aim of Study 1 was to explore the between- and within-person

variances of contact quantity and quality in relation to outgroup

attitudes, outgroup empathy, intergroup anxiety and outgroup

prosocial behaviour amongst youth in Belfast. Similar to previous

RI-CLPM studies (Friehs et al. 2024; Hodson and Meleady

2024), we demonstrate evidence for between-person differences

of contact quality and quantity in relation to all the measured

outcomes with more frequent and better-quality contact being

associated with more positive outgroup attitudes, higher levels of

outgroup empathy, lower levels of intergroup anxiety and more

outgroup prosocial behaviour coupled with limited evidence for

within-person effects. Specifically, we find that contact quantity

at Time 1 entailed more positive outgroup attitudes at Time

2 and that contact quality at Time 2 entailed more positive

outgroup attitudes, higher outgroup empathy, higher engagement

7of16

TABLE 1 Means, standard deviations and intercorrelations for the main variables of the Belfast sample.

Measure MSD 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

1. Contact Quantity T1 1.67 0.70 0.62 0.66 0.48 0.44 0.40 0.48 0.46 0.46 0.48 0.39 0.40 −0.34 −0.37 −0.41 0.55 0.40 0.43

2. Contact Quantity T2 1.69 0.66 0.55 0.38 0.50 0.28 0.42 0.50 0.43 0.39 0.49 0.31 −0.30 −0.42 −0.34 0.44 0.45 0.35

3. Contact Quantity T3 1.65 0.65 0.44 0.39 0.47 0.42 0.35 0.45 0.41 0.34 0.40 −0.27 −0.33 −0.43 0.43 0.43 0.46

4. Contact Quality T1 3.47 0.87 0.53 0.54 0.61 0.43 0.43 0.47 0.43 0.43 −0.45 −0.39 −0.45 0.48 0.30 0.36

5. Contact Quality T2 3.37 0.98 0.63 0.53 0.63 0.57 0.42 0.61 0.44 −0.26 −0.67 −0.48 0.54 0.47 0.45

6. Contact Quality T3 3.49 0.92 0.43 0.40 0.51 0.34 0.31 0.41 −0.30 −0.41 −0.63 0.43 0.32 0.44

7. Intergroup Attitudes T1 3.56 1.20 0.62 0.62 0.51 0.51 0.51 −0.39 −0.46 −0.45 0.55 0.38 0.44

8. Intergroup Attitudes T2 3.45 1.25 0.70 0.37 0.51 0.38 −0.30 −0.52 −0.49 0.47 0.47 0.37

9. Intergroup Attitudes T3 3.55 1.21 0.40 0.38 0.42 −0.30 −0.48 −0.51 0.40 0.37 0.44

10. Outgroup Empathy T1 2.90 1.17 0.58 0.61 −0.33 −0.37 −0.34 0.49 0.39 0.42

11. Outgroup Empathy T2 2.95 1.20 0.60 −0.25 −0.43 −0.29 0.44 0.38 0.29

12. Outgroup Empathy T3 3.03 1.18 −0.17 −0.40 −0.35 0.42 0.30 0.44

13. Intergroup Anxiety T1 1.03 1.08 0.32 0.40 −0.34 −0.30 −0.23

14. Intergroup Anxiety T2 1.18 1.21 0.42 −0.47 −0.40 −0.46

15. Intergroup Anxiety T3 1.15 1.07 −0.37 −0.38 −0.41

16.Outgr.Prosoc.Beh.T1 2.06 0.92 0.47 0.53

17.Outgr.Prosoc.Beh.T2 2.18 0.82 0.47

18.Outgr.Prosoc.Beh.T3 2.05 0.82

Note: For all variables, a higher score indicates a higher level of the construct in question. For the intercorrelations, ns varied between 191 and 229, owing to missing values. All correlation coefficients (in bold) are significant

(ps<0.05, two-tailed).

8of16 European Journal of Social Psychology,2024

TABLE 2 Statistical within-person cross-lagged effects of contact quantity and contact quality, with fit indicators of the respective model. There

was one model for each dependent model (i.e., four models in total).

Measures Model Fit Coefficients

DV IV CFI TLI RM SEA SRMR BSE p

Outgroup Attitudes T2 Contact Quantity T1 1.00 0.99 0.03 0.02 0.52 0.24 0.034

Contact Quality T1 0.14 0.20 0.471

Outgroup Empathy T2 Contact Quantity T1 1.00 1.00 0.01 0.02 0.22 0.23 0.336

Contact Quality T1 0.26 0.19 0.176

Intergroup Anxiety T2 Contact Quantity T1 1.00 1.00 0.00 0.02 −0.35 0.27 0.190

Contact Quality T1 −0.17 0.22 0.449

Outgroup Prosocial Behaviour T2 Contact Quantity T1 1.00 1.00 0.00 0.02 0.09 0.19 0.634

Contact Quality T1 −0.04 0.15 0.780

Outgroup Attitudes T3 Contact Quantity T2 −0.18 0.21 0.385

Contact Quality T2 0.36 0.12 0.004

Outgroup Empathy T3 Contact Quantity T2 −0.34 0.27 0.213

Contact Quality T2 0.36 0.17 0.038

Intergroup Anxiety T3 Contact Quantity T2 0.21 0.24 0.391

Contact Quality T2 −0.40 0.16 0.011

Outgroup Prosocial Behaviour T3 Contact Quantity T2 −0.24 0.19 0.217

Contact Quality T2 0.31 0.10 0.003

Note: For all variables, a higher score indicates a higher level of the construct in question. Outgroup attitudes had been divided by 20 to avoid large differences

between variances. (Autoregressive effects and cross-lagged effects of the respective dependent variable on the independent variables were also specified within

all models.) Full information maximum likelihood was used.

in prosocial outgroup behaviour and lower levels of intergroup

anxiety at Time 3. No effects of contact quantity were observed

on any of our outcomes from Time 2 to Time 3. Although these

findings suggest that there is some potential for within-person

change of contact, we are cautious about this due to the lack

of consistency across time and for our two contact measures.

It is worth noting here, however, that our findings may have

been impacted by the specific time lags in this study whereby

the Time 1–2 interval varied between 1.5 and 3.5 months (due

to school availability), while the Time 2–3 interval was constant

at 2 months and 1 week. This matters because the RI-CLPM

can only detect time-delayed associations between fluctuations

around personal average value, and so is sensitive towards the

choice of the examined time intervals (Hamaker, Kuiper, and

Grasman 2015). Nevertheless, our findings demonstrate the need

to continue to investigate the potential for intergroup contact

to evince within-person change on social cohesion outcomes

amongst adolescents.

4Study 2

Study 2 followed the design of Study 1 to replicate and explore

the within-person changes and between-person differences of

contact quantity and quality in relation to outgroup attitudes,

outgroup empathy, intergroup anxiety and outgroup prosocial

behaviour amongst youth growing up in a different divided

context: Bradford, England.

4.1 Sample

A total of 530 participants completed the survey at Time 1, 334

participants completed the survey at Time 2 and 221 completed

the survey at Time 3. Of those participants, we were able to match

data for 167 across the three time points. We focused specifically

on youth who identified as one of the two largest ethnic groups

in Bradford: White/White British and Asian/Asian British. Of

those participants, 102 specified their ethnicity as Pakistani, 18

as Bangladeshi, 7 as Indian, 1 as Punjabi and 2 as ‘other Asian’;

20 participants specified their ethnic group as White British

and 2 as White European; 9 participants did not specify their

ethnicity beyond White/White British or Asian/Asian British.

Eight participants who did not identify as either White/White

British or Asian/Asian British were removed, resulting in a final

sample of 159 participants (Mage,Time1 =16.5 years, SDage,Time1 =

0.75; 64% female, 31% male, 4% other or no stated gender; 84%

Asian, 16% White; 23% reported receiving free school meals, 72%

stated that they do not receive free school meals; 24% described

the financial situation of their family as not very well off or as

not at all very well off, 58% as average and 11% as well of or very

well off). Participants were recruited through local youth groups,

several grassroots organisations and a local college.1

4.2 Measures

Youth participants completed the same demographic and

survey measures as that of Study 1 with adaptations made

9of16

for the Bradford context. The main variables of interest were

analogous to the items which have been used in the Belfast study.

Specifically, participants were asked to rate contact quantity,

contact quality, outgroup attitudes, outgroup empathy and

intergroup anxiety in relation to three ethnic groups: White,

Black and Asian. In the analysis that follows, we focus on

intergroup relations between White/White British and Asian

Asian/British participants, as the two largest ethnic groups in

Bradford. Tests for the scale characteristics of the multi-item

measured constructs are as follows (while intergroup anxiety and

outgroup prosocial behaviour were single item measures and not

tested for reliability but were still included in the survey):

4.2.1 Contact Quantity

The same four items used in Study 1 in Belfast were used for the

Bradford sample in Study 2, with the target groups White/White

British and Asian/Asian British (Time 1: α=0.68; Time 2: α=

0.62; Time 3: α=0.72). The residual variances of the same items

were again allowed to correlate across waves. The configural mod-

els which were used as a basis to assess measurement invariance

across the three waves yielded acceptable fits (CFI =0.988; TLI =

0.980; RMSEA =0.032; SRMR =0.046). The constraints testing

for metric invariance did not lead to a significant deterioration

of the fit (p=0.108). The additional constraints testing for scalar

invariance neither led to a significant deterioration of the fit

compared to the previous model (p=0.639).

4.2.2 Contact Quality

The three items which had been used in Study 1 were also used for

Study 2 (with adapted target groups) to measure contact quality

(Time 1: α=0.70; Time 2: α=0.64; Time 3: α=0.72). Allowing the

residual variances of the same items to correlate across waves, the

three-item measure resulted in good fits in the configural models

used to assess measurement invariance across the three waves

(CFI =0.986; TLI =0.967; RMSEA =0.048; SRMR =0.034). The

constraints testing for metric invariance did not lead to a signifi-

cant deterioration of the fit (p=0.260). The additional constraints

testing for scalar invariance neither led to a significant deteriora-

tion of the fit compared to the previous model (p=0.322).

4.2.3 Outgroup Empathy

The same two items which had been used in Study 1 were used for

the Bradford sample in Study 2 (with adapted target groups). The

Spearman–Brown coefficient was ρ=0.70 for Time 1, ρ=0.59 for

Time 2, and ρ=0.51 for Time 3.

4.3 Procedure

Ethical approval was obtained from the University of Bristol

prior to data collection. To develop the Bradford sample, a 12-

month period of recruitment was undertaken, supported by local

organisations. Emails were sent to ethnically diverse secondary

schools, colleges, and grassroots organisations to invite them to

take part in the research. The recruitment search resulted in a

number of youth groups and grassroots organisations as well as a

college agreeing to take part. No school participation was secured,

and although several expressed interest, they were already part

of other projects and were unable to find the time to participate.

Survey data were collected at three time points between 5.10.22

and 30.11.22 (Time 1), 20.02.22–24.04.22 (Time 2) and 05.06.23–

26.08.23 (Time 3) amongst youth participants across a range of

settings. In all cases, youth were asked to complete the survey

online in Qualtrics either in the presence of the researcher or

a youth leader (for those under 16) or on their own (if over

16). Where relevant, the researchers provided a summary video

about the research (using the same information given via the

information sheet to participants) that was shown to participants

in advance, and the researchers held video calls with youth groups

and grassroots organisations before sharing the survey link with

those interested in taking part. All youth participants were

asked to give informed consent prior to completing the survey.

The survey took approximately 30–40 minutes to complete, and

participants were sent a £10 Amazon voucher. At later data

collection points, all youth were contacted directly and invited

to take part. For those under 16 and recruited through youth

groups, youth leaders were asked to facilitate discussions with

the participants in advance of sending the survey link. Following

completion of the project, supporting organisations were thanked

for their time and facilitation with a £500 payment.

4.4 Data Analysis Plan

Descriptive statistics and bivariate correlations between all vari-

ables at each time point are presented in Table 3. We specified

RI-CLPMs with random intercepts as latent variables, repre-

senting individual average values to ensure the modelling was

analogous to Study 1. These latent variables were allowed to

correlate, indicating between-person associations. Group mem-

bership, White/White British or Asian/Asian British, was used

as a predictor for the random intercepts. Additionally, for each

construct and for each time point, a latent variable was specified

that reflected within-person differences. For all of these within-

person latent variables, autoregressive and cross-lagged effects

were specified. Full information maximum likelihood estimation

was used to deal with missing values.

4.5 Results

The four RI-CLPM models yielded good fit to the data. For

between-person associations, we found that contact quantity was

positively correlated with outgroup attitudes (B=0.15 SE =0.04,

p=0.001), with outgroup empathy (B=0.13, SE =0.04, p=

0.001) and with outgroup prosocial behaviour (B=0.19, SE =

0.04, p<0.001), while it was negatively correlated with intergroup

anxiety (B=−0.10, SE =0.04, p=0.015). We also found that

contact quality was positively correlated with outgroup attitudes

(B=0.32 SE =0.08, p<0.001) and with outgroup empathy (B=

0.29, SE =0.06, p<0.001), while it was negatively correlated with

intergroup anxiety (B=−0.29, SE =0.07, p<0.001), but not with

outgroup prosocial behaviour (B=0.08, SE =0.06, p=0.210).

These between-person correlations indicate the degree to which

the individual average level of contact is associated with the

individual average level of the respective dependent variable. The

10 of 16 European Journal of Social Psychology,2024

TABLE 3 Means, standard deviations, and intercorrelations for the main variables of the Bradford sample.

Measure MSD 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

1. Contact Quantity T1 1.49 0.63 0.59 0.56 0.26 0.09 0.14 0.28 0.22 0.16 0.39 0.31 0.25 −0.23 −0.14 −0.18 0.38 0.26 0.34

2. Contact Quantity T2 1.55 0.58 0.59 0.18 0.07 0.17 0.14 0.20 0.11 0.27 0.30 0.22 −0.13 −0.21 −0.26 0.28 0.24 0.24

3. Contact Quantity T3 1.56 0.62 0.27 0.20 0.28 0.27 0.24 0.18 0.28 0.31 0.32 −0.14 −0.23 −0.29 0.42 0.30 0.40

4. Contact Quality T1 3.42 0.85 0.49 0.48 0.51 0.39 0.37 0.47 0.33 0.41 −0.58 −0.23 −0.38 0.29 0.30 0.18

5. Contact Quality T2 3.39 0.80 0.64 0.42 0.51 0.52 0.28 0.31 0.25 −0.33 −0.46 −0.44 0.20 0.30 0.26

6. Contact Quality T3 3.40 0.83 0.36 0.39 0.60 0.29 0.35 0.49 −0.26 −0.35 −0.59 0.19 0.30 0.30

7. Intergroup Attitudes T1 3.55 1.15 0.53 0.47 0.46 0.28 0.27 −0.38 −0.37 −0.29 0.33 0.23 0.28

8. Intergroup Attitudes T2 3.41 1.06 0.49 0.37 0.25 0.28 −0.23 −0.41 −0.32 0.27 0.40 0.25

9. Intergroup Attitudes T3 3.38 1.11 0.25 0.22 0.38 −0.20 −0.46 −0.45 0.22 0.23 0.41

10. Outgroup Empathy T1 3.15 1.11 0.53 0.56 −0.45 −0.14 −0.23 0.30 0.25 0.19

11. Outgroup Empathy T2 3.09 1.13 0.54 −0.28 −0.27 −0.20 0.36 0.20 0.23

12. Outgroup Empathy T3 3.14 1.01 −0.27 −0.23 −0.38 0.27 0.35 0.38

13. Intergroup Anxiety T1 1.37 1.05 0.19 0.24 −0.25 −0.28 −0.26

14. Intergroup Anxiety T2 1.45 1.14 0.39 −0.15 −0.29 −0.29

15. Intergroup Anxiety T3 1.31 1.02 −0.18 −0.20 −0.20

16.Outgr.Prosoc.Beh.T1 2.11 0.83 0.46 0.43

17. Outgr. Prosoc. Beh. T2 2.06 0.81 0.46

18.Outgr.Prosoc.Beh.T3 2.05 0.82

Note: For all variables, a higher score indicates a higher level of the construct in question. For the intercorrelations, ns varied between 153 and 159, owing to missing values. Correlation coefficients in bold are significant (ps

<0.05, two-tailed).

11 of 16

TABLE 4 Statistical within-person cross-lagged effects of contact quantity and contact quality, with fit indicator of the respective model. There

was one model for each dependent model (i.e., four models in total).

Measures Model Fit Coefficients

DV IV CFI TLI RM SEA SRMR BSEp

Outgroup Attitudes T2 Contact Quantity T1 0.99 0.97 0.05 0.03 −0.04 0.27 0.878

Contact Quality T1 0.03 0.19 0.860

Outgroup Empathy T2 Contact Quantity T1 1.00 1.00 0.00 0.03 0.32 0.26 0.222

Contact Quality T1 −0.05 0.17 0.777

Intergroup Anxiety T2 Contact Quantity T1 1.00 0.99 0.02 0.03 0.07 0.34 0.834

Contact Quality T1 0.25 0.27 0.358

Outgroup Prosocial Behaviour T2 Contact Quantity T1 1.00 1.00 0.01 0.03 −0.34 0.19 0.070

Contact Quality T1 0.30 0.13 0.025

Outgroup Attitudes T3 Contact Quantity T2 −0.40 0.34 0.241

Contact Quality T2 0.36 0.23 0.113

Outgroup Empathy T3 Contact Quantity T2 −0.30 0.40 0.449

Contact Quality T2 −0.40 0.27 0.136

Outgroup Anxiety T3 Contact Quantity T2 −0.29 0.31 0.344

Contact Quality T2 −0.19 0.21 0.360

Outgroup Prosocial Behaviour T3 Contact Quantity T2 −0.45 0.29 0.122

Contact Quality T2 0.15 0.19 0.426

Note: For all variables, a higher score indicates a higher level of the construct in question. Outgroup attitudes had been divided by 20 to avoid large differences

between variances. (Autoregressive effects and cross-lagged effects of the respective dependent variable on the independent variables were specified with the same

models.) Full information maximum likelihood was used.

statistical within-person cross-lagged effects on the dependent

variables in the four models are displayed in Table 4.There

were no significant within-person cross-lagged effects, except for

contact quality at Time 1 on outgroup prosocial behaviour at

Time 2. Thus, there was only one instance in which individual

deviations (from the individuals’ average values) in contact

statistically predicted deviations in the dependent variable at the

respective later time point. This contrasts with the Belfast sample,

as in the Belfast sample, contact quality at Time 2 was consistently

associated with all outcome variables at Time 3; this is considered

further in the general discussion.

4.6 Discussion

Study 2 aimed to provide a second test of the within- and between-

person effects of contact quantity and quality on outgroup

attitudes, outgroup empathy, intergroup anxiety and prosocial

behaviour amongst youth in a different context, Bradford. Find-

ings demonstrate evidence for between-person associations of

contact quality and quantity with most of the measured out-

comes, with more frequent and better-quality contact being

associated with more positive intergroup attitudes, higher levels

of outgroup empathy and lower levels of intergroup anxiety.

However, only contact quantity was associated with greater

outgroup prosocial behaviour. Unlike in Study 1, we did not

observe any effects of within-person changes of contact quality

and quantity at Time 1 on within-person changes of the out-

come measures at Time 2 or effects of within-person changes

of contact quality and quantity at Time 2 on within-person

changes of outcome measures at Time 3. This aligns with

previous research using the RI-CLPM which has failed to

find associations of within-person changes between contact

and attitudes and further related outcomes (e.g., Friehs et al.

2024; Hodson and Meleady 2024). In the general discussion,

we bring together the findings of our two studies and con-

sider implications for research and practice in socially divided

contexts.

5General Discussion

The present research explored the potential between- and within-

person impact of intergroup contact on social cohesion–related

outcomes amongst youth in two socially divided contexts, draw-

ing on three-time-point survey data amongst youth in Belfast

(Study 1) and Bradford (Study 2). Findings from across our

two studies align with previous RI-CLPM studies which have

demonstrated the presence of between-subjects effects of contact

on outcomes such as prejudice coupled with an absence of within-

person changes (e.g., Friehs et al. 2024; Hodson and Meleady

2024). We now discuss these collective findings in relation to the

wider literature.

Our finding of consistent between-person differences across our

two studies bolsters previous evidence that those who hold

12 of 16 European Journal of Social Psychology,2024

outgroup friendships (Friehs et al. 2024) or have more frequent

intergroup contact (Hodson and Meleady 2024) also report more

positive outgroup attitudes, but pushes this forward by observing

that those who report frequent quantity and better quality contact

also report lower levels of intergroup anxiety (measured as

levels of comfort), higher levels of outgroup empathy and higher

likelihood of engaging in prosocial acts to support outgroup

members (for contact quality only in Study 1) as indicators related

to social cohesion. These between-person effects, however, were

observed in the partial absence of within-person changes. And

whilst we found that contact quantity at Time 1 entailed more

positive outgroup attitudes at Time 2 and that contact quality

at Time 2 entailed more positive outgroup attitudes, higher

outgroup empathy, higher engagement in prosocial outgroup

behaviour and lower levels of intergroup anxiety at Time 3 in

Study 1, we are unable to claim consistent within-person effects

over time or between our studies.

A possible explanation for the lack of within-person effects

observed in our studies, in line with the arguments put forward

by previous research (Friehs et al. 2024; Wölfer et al. 2016),

is that initial engagement in contact may be more powerful

than later engagement in contact in predicting our outcomes of

interest. As such, our participants may have already experienced

both frequent and good quality intergroup contact in the past

which means that we have not captured the potentially trans-

formative power of initial intergroup interactions. A further and

related explanation is that perhaps our participants were at an

age in which the strength of the relationship between contact

and prejudice had already declined. Our Study 2 participants

were older on average (Mage =16.5 years at Time 1) than our

Study 1 participants (Mage =14.7 years at Time 1) but due

to inconsistent findings across time points, we are unable to

determine whether the observation of some within-person effects

in Study 1 is due to developmental differences. Whilst Friehs

et al. (2024), who analysed social network data from a five-wave

social network sample of 11–12 year olds in England, found no

evidence of within-person change in the contact–prejudice rela-

tionship, questions still remain about potential developmental

pathways.

There may, however, be methodological reasons for our lack of

observed within-person effects. First, it is worth considering the

time lags that were used in the present research. Previous RI-

CLPM studies on contact effects have tended to have time lags

ranging from several months (Friehs et al. 2024: Study 1; Hodson

and Meleady 2024) to several years (Sengupta et al. 2023). Given

that longitudinal studies often have time lags that are too long to

detect within-person change (Dormann and Griffin 2015)andthat

one of the few studies that has detected significant within-person

contact effects on collective action (Górska and Tausch 2023)had

time lags of 2 weeks, our aim was to employ relatively short time

lags to increase the chances of observing within-person effects.

It could be that an even shorter time frame than in our studies

is required to find within-person effects. This is something that

should be explored in future research.

Second, it is possible that our measurement of intergroup con-

tact, using self-reported items of contact quality and contact

quantity, does not fully capture the nature of participants’

interaction experiences and consequently we were unable to

detect consistent within-person change. This is plausible given

substantial empirical evidence that when primed, intergroup

contact has been found to reduce prejudice amongst those in

an experimental condition compared to a control condition

(Paluck, Green, and Green 2019). Future research, therefore,

may wish to explore intergroup contact experiences in more

depth.

5.1 Implications for Theory and Practice

Taken together, our two studies yield important theoretical

implications for the research literature on intergroup contact

as well as practical implications for social cohesion agendas.

In relation to theory, we provide further evidence for a lack of

consistent within-person changes in the relationship not only

between intergroup contact and prejudice but also between

intergroup contact and a wider range of social cohesion–related

outcomes. Our observation of consistent between-person effects

in the absence of within-person effects in line with previous

studies (e.g., Hodson and Meleady 2024;Senguptaetal.2023)

highlights the need to theoretically determine the conditions

under which and for whom within-person changes of contact

might be observed and on which outcomes. This requires further

theoretical development, building on the work of Merrilees

et al. (2023) and Wölfer et al. (2016) on the temporal nature

of intergroup contact, including, for example, if and when we

might expect within-person changes to occur across the lifespan

in relation to both prior contact experiences and developmental

trajectories. Our findings also suggest that the field may need

to consider how intergroup contact experiences are measured

in survey research, to truly determine whether within-person

changes are evinced following contact experiences.

Our findings also have practical implications for promoting

social cohesion in divided societies. We show, for example, that

youth who report having good quality and frequent contact

with outgroup members also report high scores on a range of

positive outcomes that can build stronger and more cohesive

societies, such as more positive outgroup attitudes, higher levels

of outgroup empathy, lower levels of outgroup anxiety and

higher levels of self-reported engagement in outgroup prosocial

behaviours. Whilst we are unable to provide evidence for within-

person changes of contact on our outcomes across time points

and samples, this does not, in our view, demonstrate that contact

is not effective in promoting social cohesion, rather that we

need to further explore how and when we might expect within-

person changes to occur. Indeed, we know from experimental

research that contact reduces prejudice (Paluck, Green, and

Green 2019) and as such, it remains an important tool for pro-

moting more positive group relations in divided societies. What

we would suggest, however, is that contact should not be seen

as a panacea and other approaches towards social cohesion are

needed.

5.2 Limitations and Future Directions

Whilst our research has strengths in exploring contact effects

longitudinally amongst youth in two different contexts, there

are some important limitations. First, our sample sizes are

13 of 16

relatively small and not fully representative of the ethnic groups

present within each context, especially in Study 2. While previous

research with larger samples (e.g., Hodson and Meleady 2024;

Sengupta et al. 2023) has failed to observe within-person effects of

contact, we cannot rule out that our failure to observe consistent

results is due to our underpowered samples. Future work should,

therefore, seek to recruit a higher number of participants to

allow a more robust analysis of within-person effects across time

amongst adolescents and for different ethnic groups. The need

for larger sample sizes is underlined by the results of our post

hoc power analyses. In the Monte Carlo simulations based on

the Belfast data, a simulated sample size of 440 was required

to ensure that all replications could be computed without any

estimation problems (while entailing power values of over 80%

for the significant within-person effects), as reported above. For

two outcome variables in the Bradford sample, even a simulated

sample size of 10,000 was not sufficient to yield replications that

were consistently without estimation problems. This suggests

that our empirical sample sizes should have ideally been larger

to provide robust parameter estimates. Additionally, needing a

simulated sample size of 600 to reach power estimates of over 80%

in the Bradford sample suggests that the studies might have been

underpowered.

Second, whilst our sample included youth participants across

mid-late adolescence, we did not have a sample over a long

enough period of time to be able to examine developmental

trajectories. A crucial next step will be to conduct research

that enables an analysis of the effects of contact on outcomes

across different developmental ages to determine an under-

standing of whether, and if so, under which context conditions

there is a point in which within contact effects weaken with

age.

Third, several practical considerations meant that participants

did not always complete the survey in the same format (online

vs. on paper), not always across the same time intervals, and

not always in the same setting (some in school, some in college,

some in youth groups). A more controlled approach would

enable stronger conclusions to be made regarding the relation-

ship between intergroup contact and social cohesion–related

outcomes across the two research contexts.

Fourth, we relied on single-item measures of intergroup anxiety

and outgroup prosocial behaviour. This enabled us to have a

more comprehensive survey with our young people by reducing

item response burden, but there may be consequences for overall

validity. For example, our measure of intergroup anxiety may

be related more to feelings of comfort specifically, although

it is worth noting that comfort is one of the most measured

affective components of intergroup anxiety alongside others such

as feeling anxious, at ease, etc. (Stephan 2014). Future research,

however, should aim to measure outcomes using a wider range of

items.

Finally, whilst our research expands the range of outcomes of

intergroup contact that have been previously explored through

RI-CLPM studies, it would be advantageous to assess tertiary-

related outcomes. This is because within-person change might

be observed for outcomes that transcend beyond prejudice

reduction—such as cognitive flexibility or creativity.

6 Conclusion

Intergroup contact has long been established as a prejudice-

reduction tool in divided societies. A large proportion of the

evidence for the benefits of intergroup contact, however, comes

either from cross-sectional studies that cannot establish causality

or from longitudinal studies, many of which do not clearly

separate between-person variance from within-person variance.

The present research aimed to address this limitation by exploring

between-person associations and within-person changes regard-

ing the relationship of intergroup contact with social cohesion

outcomes amongst youth in two divided cities: Belfast and

Bradford. Drawing on three-time-point survey data, our findings

extend previous research demonstrating between-subject effects

of contact quality and quantity on outcomes including outgroup

attitudes, empathy, intergroup anxiety and prosocial behaviour,

but limited evidence of within-person effects on these outcomes.

By focusing on contact effects amongst youth in socially divided

contexts and by exploring a wide range of social cohesion–related

outcomes, our research offers a grounded assessment of the

potential and limitations of intergroup contact to promote social

cohesion. And, whilst failing to observe consistent within-person

effects of contact in our research, we recognise the importance

of continuing to recognise the potential of intergroup contact as

a prejudice-reduction tool, as evidenced in experimental studies

in the field, but point to the need to better understand potential

within-person effects through robust longitudinal studies that

consider the development of intergroup contact experiences over

time and across the lifespan.

Author Contributions

McKeown, Taylor and Manley acquired funding for this research and

led the research design. Data were collected by McKeown and Ali, input

by Ali, analysed by Schaefer and supported by Meleady and Taylor.

All authors contributed to the conceptualisation and write up of the

manuscript.

Acknowledgements

This work was supported by funding obtained from the Economic and

Social Research Council [ES/T014709/1]. Our deep appreciation to the

organisations, administrators, teachers, youth leaders and young people

for supporting and participating in this project.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of the study will be made available on

the UK Data Archive following the completion of the research project.

Endnotes

1The original aim was to recruit participants aged 14–16 through schools

as per Study 1, but the team was unable to secure access to local schools

and so proceeded with alternative recruitment approaches, supported by

local organisations and the project advisory group, and included a higher

upper age boundary to enable a larger sample to be collected.

14 of 16 European Journal of Social Psychology,2024

References

Allport, G. W. 1954. The Nature of Prejudice. Cambridge, MA: Addison

Wes ley.

Al Ramiah, A., and M. Hew stone. 2013. “Intergroup Contact as a Tool

for Reducing, Resolving, and Preventing Intergroup Conflict: Evidence,

Limitations, and Potential.” American Psychologist 68, no. 7: 527–542.

https://doi.org/10.1037/a0032603.

Batson, C. D., M. P. Polycarpou, E. Harmon-Jones, et al. 1997. “Empathy

and Attitudes: Can Feeling for a Member of a Stigmatized Group Improve

Feelings Toward the Group?.” Journal of Personality and Social Psychology

72, no. 1: 105–118. https://doi.org/10.1037/0022-3514.72.1.105.

Batson, C. D., J.Chang, R. Orr, and J. Rowland. 2002. “Empathy,Attitudes,

and Action: Can Feeling for a Member of a Stigmatized Group Motivate

one to Help the Group?.” Personality and Social Psychology Bulletin 28, no.

12: 1656–1666.

Binder, J., H. Zagefka, R. Brown, et al. 2009. “Does Contact Reduce

Prejudice or Does Prejudice Reduce Contact? A Longitudinal Test of

the Contact Hypothesis Among Majority and Minority Groups in Three

European Countries.” Journal of Personality and Social Psychology 96, no.

4: 843–856. https://doi.org/10.1037/a0013470.

Cairns, E., J. Kenworthy, A. Campbell, and M. Hewstone. 2006. “The

Role of In-Group Identification, Religious Group Membership and Inter-

group Conflict in Moderating In-Group and Out-Group Affect.” British

Journal of Social Psychology 45, no. 4: 701–716. https://doi.org/10.1348/

014466605X69850.

Christ, O., and U. Wagner. 2013. “Methodological Issues in the Study of

Contact: Towards a New Wave of Research.” In Advances in Intergroup

Contact, edited by G. Hodson and M. Hewstone, 233–261. London:

Psychology Press.

Davies, K., L. R. Tropp, A. Aron, T. F. Pettigrew, and S. C. Wright. 2011.

“Cross-Group Friendships and Intergroup Attitudes: A Meta-analytic

Review.” Personality and Social Psychology Review 15, no. 4: 332–351.

https://doi.org/10.1177/1088868311411103.

Dhont, K., A. Van Hiel, and M. Hewstone. 2014.“Changing the Ideological

Roots of Prejudice: Longitudinal Effects of Ethnic Intergroup Contact on

Social Dominance Orientation.” Group Processes & Intergroup Relations

17, no. 1: 27–44. https://doi.org/10.1177/1368430213497064.

Dormann, C., and M. A. Griffin. 2015. “Optimal Time Lags in Panel

Studies.” Psychological Methods 20, no. 4: 489–505. https://doi.org/10.

1037/met0000041.

Eisinga, R., M. Grotenhuis, and B. Pelzer. 2013. “The Reliability of a

Two-Item Scale: Pearson, Cronbach,or Spearman-Brown?.” International

Journal of Public Health 58: 637–642. https://doi.org/10.1007/s00038-012-

0416-3.

Friehs, M.-T., C. Bracegirdle, N. K. Reimer, et al. 2024. “The Between-

Person and Within-Person Effects of Intergroup Contact on Out-

group Attitudes: A Multi-Context Examination.” Social Psychologi-

cal and Personality Science 15, no. 2: 125–141. https://doi.org/10.1177/

19485506231153017.

Górska, P., and N. Tausch. 2023. “Dynamic, Yet Stable: Separating

Within- and Between-Person Components of Collective Action in Support

of a Disadvantaged Outgroup and Its Antecedents.” Social Psycholog-

ical and Personality Science 14, no. 7: 875–887. https://doi.org/10.1177/

19485506221133882.

Hamaker, E. L., R. M. Kuiper, and R. P. P. P. Grasman. 2015. “A Critique of

the Cross-Lagged Panel Model.” Psychological Methods 20, no. 1: 102–116.

https://doi.org/10.1037/a0038889.

HM Government. 2018. Integrated Communities Strategy Green Paper.

London: HM Government. https://assets.publishing.service.gov.uk/

media/5ac490f4e5274a0b1849f8c2/Integrated_Communities_Strategy.

pdf.

Hodson, G., and M. Hewstone. 2013. Advances in Intergroup Contact.

London: Psychology Press. https://doi.org/10.4324/9780203095461.

Hodson, G., and R. Meleady. 2024. “Replicating and Extending Sengupta,

et al. (2023): Contact Predicts No Within-Person Longitudinal Outgroup-

Bias Change.” American Psychologist 79, no. 3: 451–462. https://doi.org/

10.1037/amp0001210.

Hu, L., and P. M. Bentler. 1999. “Cutoff Criteria for Fit Indexes in Covari-

ance Structure Analysis: Conventional Criteria Versus New Alternatives.”

Structural Equation Modeling: A Multidisciplinary Journal 6, no. 1: 1–55.

https://doi.org/10.1080/10705519909540118.

Hughes, J., A. Campbell, S. Lolliot, M. Hewstone, and T. Gallagher. 2013.

“Inter-Group Contact at School and Social Attitudes: Evidence From

Northern Ireland.” Oxford Review of Education 39, no. 6: 761–779. https://

doi.org/10.1080/03054985.2013.857595.

Jackson, D. L. 2003. “Revisiting Sample Size and Number of Parameter

Estimates: Some Support for the N:q Hypothesis.” Structural Equation

Modeling: A Multidisciplinary Journal 10, no. 1: 128–141. https://doi.org/

10.1207/S15328007SEM1001_6.

Kline, R. B. 2016. Principles and Practice of Structural Equation Modeling,

4th ed. New York: Guilford Press.

Koschate, M., S. Oethinger, D. Kuchenbrandt, and R. Van Dick. 2012.

“Is an Outgroup Member in Need a Friend Indeed? Personal and

Task-Oriented Contact as Predictors of Intergroup Prosocial Behavior.”

European Journal of Social Psychology 42, no. 6: 717–728. https://doi.org/

10.1002/ejsp.1879.

Lemmer, G., and U. Wagner. 2015. “Can We Really Reduce Ethnic

Prejudice Outside the Lab? A Meta-analysis of Direct and Indirect Contact

Interventions.” European Journal of Social Psychology 45, no. 2: 152–168.

https://doi.org/10.1002/ejsp.2079.

McKeown, S., and E. Cairns. 2012. “Peacemaking Youth Programmes in

Northern Ireland.” Journal of Aggression, Conflict and Peace Research 4,

no. 2: 69–75. https://doi.org/10.1108/17596591211208274.

McKeown, S., and L. K. Taylor. 2018. “Perceived Peer and School Norm

Effects on YouthAntisocial and Prosocial Behaviours Through Intergroup

Contact in Northern Ireland.” British Journal of Social Psychology 57, no.

3: 652–665. https://doi.org/10.1111/bjso.12257.

Merrilees,C.E.,L.K.Taylor,M.Klotz,M.C.Goeke-Morey,P.Shirlow,and

E. M. Cummings. 2023. “Timing is Everything: Developmental Changes

in the Associations Between Intergroup Contact and Bias.” International

Journal of Behavioral Development 47, no. 3: 243–252. https://doi.org/10.

1177/01650254221146409.

Mulder, J. D., and E. L. Hamaker. 2021. “Three Extensions of the Random

Intercept Cross-Lagged Panel Model.” Structural Equation Modeling:

A Multidisciplinary Journal 28, no. 4: 638–648. https://doi.org/10.1080/

10705511.2020.1784738.

Muthén, L. K., and B. O. Muthén. 1998–2017. Mplus User’s Guide, 8th ed.

Los Angeles, CA: Muthén & Muthén.

Paluck, E. L., S. A. Green, and D. P. Green. 2019. “The Contact Hypothesis

Re-Evaluated.” Behavioural Public Policy 3, no. 2: 129–158. https://doi.org/

10.1017/bpp.2018.25.

Pettigrew, T. F., and L. R. Tropp. 2006. “A Meta-analytic Test of Intergroup

Contact Theory.” Journal of Personality and Social Psychology 90, no. 5:

751–783. https://doi.org/10.1037/0022-3514.90.5.751.

Pettigrew, T. F., and L. R. Tropp. 2008. “How Does Intergroup Contact

Reduce Prejudice? Meta-analytic Tests of Three Mediators.” European

Journal of Social Psychology 38, no. 6: 922–934. https://doi.org/10.1002/

ejsp.504.

Pettigrew, T. F., and L. R. Tropp. 2011. When Groups Meet: The Dynamics

of Intergroup Contact. New York: Psychology Press.

Sengupta, N. K., N. K. Reimer, C. G. Sibley, and F. K. Barlow. 2023.

“Does Intergroup Contact Foster Solidarity With the Disadvantaged? A

Longitudinal Analysis Across 7 Years.” American Psychologist 78, no. 6:

750–760. https://doi.org/10.1037/amp0001079.

Stephan, W. G. 2014. “Intergroup Anxiety: Theory, Research, and Prac-

tice.” Personality and Social Psychology Review 18, no. 3: 239–255. https://

doi.org/10.1177/1088868314530518.

15 of 16

Stronger Communities Partnership. 2019. Bradford for Everyone

Strategy: Building Stronger Communities Together: 2018–2023.Bradford

Council. https://bradfordforeveryone.co.uk/wp-content/uploads/2021/

11/Bradford-for-Everyone-Strategy-2018-2023.pdf.

Swart, H., M. Hewstone, O. Christ, and A. Voci. 2011. “Affective Mediators

of Intergroup Contact: A Three-Wave Longitudinal Study in South

Africa.” Journal of Personality and Social Psychology 101, no. 6: 1221–1238.

https://doi.org/10.1037/a0024450.

Tam, T., M. Hewstone, J. Kenworthy, and E. Cairns. 2009. “Intergroup

Trust in Northern Ireland.” Personality and Social Psychology Bulletin 35,

no. 1: 45–59. https://doi.org/10.1177/0146167208325004.

Taylor, L. K., and S. McKeown. 2021. “Adolescent Outgroup Helping, Col-

lective Action, and Political Activism in a Setting of Protracted Conflict.”

International Journal of Intercultural Relations 85: 37–46. https://doi.org/

10.1016/j.ijintrel.2021.09.001.

Taylor, L. K., C. E. Merrilees, M. C. Goeke-Morey, P. Shirlow, E.

Cairns, and E. M. Cummings. 2014. “Political Violence and Adolescent

Out-Group Attitudes and Prosocial Behaviors: Implications for Positive

Inter-Group Relations.” Social Development 23, no. 4: 840–859. https://doi.

org/10.1111/sode.12074.

Taylor, L. K., C. E. Merrilees, M. C. Goeke-Morey, P. Shirlow, and E.

M. Cummings. 2016. “Trajectories of Adolescent Aggression and Family

Cohesion: The Potential to Perpetuate or Ameliorate Political Conflict.”

Journal of Clinical Child & Adolescent Psychology 45, no. 2: 114–128. https://

doi.org/10.1080/15374416.2014.945213.

The Executive Office. 2013. Together: Building a United Community

Strategy (TBUC). Belfast: The Execute Office. https://www.

executiveoffice-ni.gov.uk/sites/default/files/publications/ofmdfm_dev/

together-building-a- united-community-strategy.pdf.

Turner, R. N., K. West, and Z. Christie. 2013. “Out-Group Trust, Inter-

group Anxiety, and Out-Group Attitude as Mediators of the Effect of

Imagined Intergroup Contact on Intergroup Behavioral Tendencies.”

Journal of Applied Social Psychology 43, no. Suppl no. 2: E196–E205.

https://doi.org/10.1111/jasp.12019.

Waddington, D. P. 2010. “Applying the Flashpoints Model of Public

Disorder to the 2001 Bradford Riot.” The British Journal of Criminology

50, no. 2: 342–359. https://doi.org/10.1093/bjc/azp082.

Wölfer, R., K. Schmid, M. Hewstone, and M. Van Zalk. 2016. “Develop-

mental Dynamics of Intergroup Contact and Intergroup Attitudes: Long-

Term Effects in Adolescence and Early Adulthood.” Child Development

87, no. 5: 1466–1478. https://doi.org/10.1111/cdev.12598.

16 of 16 European Journal of Social Psychology,2024

ResearchGate has not been able to resolve any citations for this publication.

Replicating and Extending : Contact Predicts No Within-Person Longitudinal Outgroup-Bias Change

Article

Full-text available

Nov 2023

Intergroup contact has long been touted as a premier means to reduce prejudice and forge positive bonds with outgroups. Given its origins in psychological research, it is perhaps of little surprise that contact is expected to induce change within people over time. Yet using random-intercepts crossed-lagged modeling that parses within-person from between-person effects, Sengupta et al. (2023) recently found no evidence of within-person change, only unexplained between-person effects, regarding contact’s effects on outgroup solidarity in New Zealand. We conceptually replicated their study, focusing on modern racism and an affect thermometer as the outcomes, in a three-wave study of White British participants (NT1 = 946, NT2 = 667, NT3 = 591) and their attitudes toward foreigners. We replicated the general pattern described by Sengupta and colleagues, confirming between-person effects without within-person effects, suggestive of third-variable explanations. As a novel finding, we discover that differences in social dominance orientation (SDO) and right-wing authoritarianism (RWA) can account for the observed between-person effects. Problematically for contact theory, contact effects, at least those relying on self-reported accounts, increasingly appear to reflect differences between people (person factors) rather than being context-driven (situation factors)—such that those lower (vs. higher) in SDO and RWA are more favorable toward outgroups, rather than intergroup contact bringing about positive outcomes itself. Implications for theory development and intervention are discussed.

Does Intergroup Contact Foster Solidarity With the Disadvantaged? A Longitudinal Analysis Across 7 Years

Article

Full-text available

Jan 2023

Contact theory is a well-established paradigm for improving intergroup relations—positive contact between groups promotes social harmony by increasing intergroup warmth. A longstanding critique of this paradigm is that contact does not necessarily promote social equality. Recent research has blunted this critique by showing that contact correlates positively with political solidarity expressed by dominant groups toward subordinate groups, thus furthering the goal of equality. However, this research precludes causal inferences because it conflates within-person change (people with higher contact subsequently expressing higher solidarity) and between-person stability (people with chronically high contact simultaneously expressing chronically high solidarity, and vice versa). We addressed this problem in a highly powered, seven-wave study using two different measures of contact and three different measures of political solidarity (N = 22,646). Results showed no within-person change over a 1-year period (inconsistent with a causal effect), but significant between-person stability (consistent with third-variable explanations). This reinforces doubts about contact as a strategy for promoting equality.

Adolescent outgroup helping, collective action, and political activism in a setting of protracted conflict

Article

Full-text available

Sep 2021
INT J INTERCULT REL

This article examines the role of empathy for outgroup helping, collective action and political activism among youth in Northern Ireland, a setting of protracted conflict. Integrating the Empathy-Attitudes-Action model with the Developmental Peacebuilding Model, a two-wave study was conducted to assess youth’s behavioural intentions and actual behaviours toward refugees. Across two waves (N = 383, 52% male, 48% female; 14-16 years old), empathy at Time 1 predicted more positive attitudes toward ethnic minorities at Time 2, which in turn was positively related to four outcomes aiming to foster prosocial change for refugees: helping behaviour and realistic helping at the interpersonal level, collective action intentions at the structural level, and signing a petition aiming for cultural change. That is, outgroup attitudes mediated the link from empathy to three types of prosocial action toward refugees. The findings suggest that youth not only volunteer to help an individual outgroup member, but also support broader structural and cultural change that will benefit those they may never meet. Implications for recognising and supporting the constructive agency of youth toward disadvantaged groups in conflict settings are discussed.

Three Extensions of the Random Intercept Cross-Lagged Panel Model

Article

Full-text available

Aug 2020

The random intercept cross-lagged panel model (RI-CLPM) is rapidly gaining popularity in psychology and related fields as a structural equation modeling (SEM) approach to longitudinal data. It decomposes observed scores into within-unit dynamics and stable, between-unit differences. This paper discusses three extensions of the RI-CLPM that researchers may be interested in, but are unsure of how to accomplish: (a) including stable, person-level characteristics as predictors and/or outcomes; (b) specifying a multiple-group version; and (c) including multiple indicators. For each extension, we discuss which models need to be run in order to investigate underlying assumptions, and we demonstrate the various modeling options using a motivating example. We provide fully annotated code for lavaan (R-package) and Mplus on an accompanying website.

The contact hypothesis re-evaluated

Article

Full-text available

Jul 2018

This paper evaluates the state of contact hypothesis research from a policy perspective. Building on Pettigrew and Tropp's (2006) influential meta-analysis, we assemble all intergroup contact studies that feature random assignment and delayed outcome measures, of which there are 27 in total, nearly two-thirds of which were published following the original review. We find the evidence from this updated dataset to be consistent with Pettigrew and Tropp's (2006) conclusion that contact “typically reduces prejudice.” At the same time, our meta-analysis suggests that contact's effects vary, with interventions directed at ethnic or racial prejudice generating substantially weaker effects. Moreover, our inventory of relevant studies reveals important gaps, most notably the absence of studies addressing adults' racial or ethnic prejudices, an important limitation for both theory and policy. We also call attention to the lack of research that systematically investigates the scope conditions suggested by Allport (1954) under which contact is most influential. We conclude that these gaps in contact research must be addressed empirically before this hypothesis can reliably guide policy.

Perceived peer and school norm effects on youth antisocial and prosocial behaviours through intergroup contact in Northern Ireland

Article

Full-text available

Mar 2018

In adolescence, youth spend a high proportion of their time with their peers and in school; it is hardly surprising therefore, that perceptions of peer and school norms have a strong influence on their attitudes and behaviours. These norms, however, do not always influence youth in the same way. Building on past research, the present study examines the role of peer norms and school norms in influencing the quantity and quality of intergroup contact, as well as the impact of such contact on positive and negative intergroup behaviours. Youth (aged 14-16) living in Northern Ireland (N = 466, evenly split by religion and gender) were recruited through their school as part of a two-wave study and completed a series of survey measures including intergroup contact (quality and quantity), norms (peer and school), participation in sectarian antisocial behaviour and outgroup prosocial behaviour. Mediation analysis was conducted in Mplus. Controlling for wave 1 responses on contact and behavioural outcomes, findings demonstrate that more positive peer norms are associated with less participation in antisocial behaviour and more participation in prosocial outgroup behaviours through increased and better quality intergroup contact. Positive school norms were also associated with increased prosocial behaviour, but only though better quality contact. Findings demonstrate the relative importance of peer norms compared to school norms for this age group. The results have implications for school-based interventions that aim to improve intergroup relations, and highlight the importance of peer networks to promote more positive outgroup behaviours in divided societies such as Northern Ireland.

Developmental Dynamics of Intergroup Contact and Intergroup Attitudes: Long-Term Effects in Adolescence and Early Adulthood

Article

Full-text available

Sep 2016
CHILD DEV

Intergroup contact represents a powerful way to improve intergroup attitudes and to overcome prejudice and discrimination. However, long-term effects of intergroup contact that consider social network dynamics have rarely been studied at a young age. Study 1 validated an optimized social network approach to investigate intergroup contact (N = 6,457; Mage = 14.91 years). Study 2 explored the developmental trajectories of intergroup contact by applying this validated network approach in a cross-sequential design (four-cohort-four-wave; N = 3,815; 13-26 years). Accelerated growth curve models showed that contact predicts the development of attitudes in adolescence, whereas acquired attitudes buffer against decreasing contact in adulthood. Findings highlight the potential of social network analysis and the developmental importance of early intergroup contact experiences.

The Between-Person and Within-Person Effects of Intergroup Contact on Outgroup Attitudes: A Multi-Context Examination

Article

Feb 2023

The extensive literature on the contact hypothesis reports a positive association between intergroup contact and outgroup attitudes, yet it remains unknown whether this association reflects within-person (i.e., situational changes within individuals) or between-person (i.e., stable differences between individuals) effects. To investigate this question, we applied (random-intercept) cross-lagged panel models in two studies featuring different samples, measurements, and contexts. We found longitudinal contact–attitude associations in cross-lagged panel models, which cannot differentiate within-person and between-person effects. In random-intercept cross-lagged panel models, we identified between-person effects but not within-person effects. These results conflict with the contact hypothesis, which assumes that contact leads to intra-individual attitude change. We further investigated whether between-person effects represent spurious correlations caused by potential confounders (demographic characteristics, personality, and intergroup ideologies), but found that this was not the case. Our findings highlight the need to further investigate within-person effects and potential explanations of between-person differences in contact and attitudes.

Timing is everything: Developmental changes in the associations between intergroup contact and bias

Article

Jan 2023

Identifying developmental patterns in intergroup contact and its relation with bias is crucial for improving prevention strategies around intergroup relations. This study applied time-varying effects modeling (TVEM) to examine age-based changes in relations between contact and bias in a divided community that included 667 youth ( M age = 15.74, SD = 1.97) from Belfast, Northern Ireland, a conflict-affected setting. The results suggest no change in the relation between contact frequency and bias; however, the relation between contact quality and bias increases from ages 10–14 and then levels off. Differences between Catholics, the historic minority group, and Protestants, the historic majority group, also emerged. The article concludes with implications for future research and interventions for youth growing up amid conflict.

Dynamic, Yet Stable: Separating Within- and Between-Person Components of Collective Action in Support of a Disadvantaged Outgroup and its Antecedents

Article

Dec 2022

Despite an increasing interest in the drivers of intergroup solidarity, the within-person longitudinal relationships between advantaged group members’ engagement for disadvantaged groups and its postulated antecedents remain scarcely tested. In the context of the refugee crisis following Russia’s invasion of Ukraine, we conducted a three-wave longitudinal survey ( N T1 = 804, N T2 = 702, and N T3 = 624) assessing Poles’ (the advantaged group) willingness to act for Ukrainians (the disadvantaged group), together with three hypothesized predictors—moral convictions, intergroup contact, and politicized identity. Employing a random intercept cross-lagged panel model that separates between- from within-person variance, we found that within-person changes in moral convictions and friendship contact directly predicted subsequent action intentions. Contrary to past theorizing, politicized identity emerged as consequence rather than an antecedent of collective action. Superficial intergroup contact indirectly predicted engagement intentions by facilitating cross-group friendship. We discuss the implications of our findings for current models of collective action.

The Longitudinal Relationship Between Youth Intergroup Contact and Social Cohesion Outcomes in Two Divided Societies

Abstract

Recommended publications

The Longitudinal Effects of Intergroup Contact on Youth Attitudes Towards Ethnic Minorities and Cons...

Replicating and Extending : Contact Predicts No Within-Person Longitudinal Outgroup-Bias Change

Perceived peer and school norm effects on youth antisocial and prosocial behaviours through intergro...

Fluctuations in Prejudice Do Not Track Fluctuations in Ordinary Contact in Three 5-Wave “Shortitudin...