The role of labour unions in explaining workers’ mental and physical health in Great Britain. A longitudinal approach

Article (PDF Available)inSocial Science & Medicine 247:112796 · February 2020with 105 Reads 
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
DOI: 10.1016/j.socscimed.2020.112796
Cite this publication
Abstract
Objectives: To assess whether there are mental and physical health benefits of being employed in a workplace where there is a union or staff association recognized by the management or being a member of such a union. Methods: Using four waves [W2 (2010-11), W4 (2012-13), W6 (2014-15), W8 (2016-18)] from Understanding Society (UKHLS), we use a propensity score matching method and apply a latent growth modeling on the original dataset and on the matched dataset to estimate the impact of change in union presence and union membership between wave 2 and wave 4 for the employed population on the change in mental health (Mental Component Summary - MCS) and physical health (Physical Component Summary - PCS), after controlling for socioeconomic characteristics, age and sector of activity. Results: Collective negotiation within the workplace plays a statistically significant role in supporting workers' mental and, to a greater degree, physical health. Being unionized does not add up significant physical health benefits but a slight positive effect on mental health is observed. Conclusion: About 50 per cent of the employed population is not represented by a labour union at company level and this has negative effects on health. A major health policy issue is also about promoting collective negotiation at the workplace and more research is needed about the impact of implementing such type of negotiation. The study shows the benefits of using a longitudinal approach when analysing the impact of union presence and union membership on workers' health.
Figures - uploaded by Jacques Wels
Author content
All content in this area was uploaded by Jacques Wels
Content may be subject to copyright.
Advertisement
https://doi.org/10.1016/j.socscimed.2020.112796
1
The Role of Labour Unions in Explaining Workers’ Mental and Physical Health in
Great Britain. A longitudinal approach.
Jacques Wels, PhD1,2
1 Université libre de Bruxelles, Centre Metices
2 University of Cambridge, department of Sociology
wjacques@ulb.ac.be
Abstract:
Objectives. To assess whether there are mental and physical health benefits of being
employed in a workplace where there is a union or staff association recognized by the
management or being a member of such a union.
Methods. Using four waves [W2 (2010-11), W4 (2012-13), W6 (2014-15), W8 (2016-18)]
from Understanding Society (UKHLS), we use a propensity score matching method and apply
a latent growth modeling on the original dataset and on the matched dataset to estimate the
impact of change in union presence and union membership between wave 2 and wave 4 for
the employed population on the change in mental health (Mental Component Summary
MCS) and physical health (Physical Component Summary PCS), after controlling for
socioeconomic characteristics, age and sector of activity.
Results. Collective negotiation within the workplace plays a statistically significant role in
supporting workers’ mental and, to a greater degree, physical health. Being unionized does
not add up significant physical health benefits but a slight positive effect on mental health is
observed.
Conclusion. About 50 per cent of the employed population is not represented by a labour
union at company level and this has negative effects on health. A major health policy issue is
also about promoting collective negotiation at the workplace and more research is needed
about the impact of implementing such type of negotiation. The study shows the benefits of
using a longitudinal approach when analysing the impact of union presence and union
membership on workers’ health.
Key words: Trade Union; Latent Growth Curve; Unionization; Occupational Health;
Understanding Society
Pre-print version, please quote as:
Wels J. (2020). The Role of Labour Unions in Explaining Workers’ Mental and Physical
Health in Great Britain. A longitudinal approach. Social Science & Medicine, Volume
247, February 2020, 112796. https://doi.org/10.1016/j.socscimed.2020.112796
https://doi.org/10.1016/j.socscimed.2020.112796
2
Introduction
Are there any health benefits of being unionized or being employed in a workplace where there
is a union or staff association recognized by the management and does it reflect over the life
course? The question is worth asking as unionization rates have reached lower levels in 60 years
in the United Kingdom: in 2016, the OECD estimates a ratio of 23.7 per cent of unionized
workers against 40.5 per cent in 1960. Union density has decreased in most OECD countries
over the last 60 years. For instance, it was 35 per cent in Japan in 1960 against 17.3 per cent in
2016, 41.7 per cent in 1960 in the Netherlands against 17.3 in 2016, 19.6 in 1960 in France
against 7.9 in 2014. Only a few exceptions can be observed: it has slightly increased in Sweden
and in Italy (respectively from 64.6 and 29.7 per cent in 1960 to 66.8 and 35.7 per cent in 2015).
Despite such a sharp global decline, the United Kingdom remains one of the liberal countries
where union density is relatively high. By comparison, union density was 30.9 per cent in 1960
and 10.3 per cent in 2017 in the US and respectively 50.2 and 29.2 in Australia.
Though, collective negotiation is very different in the United Kingdom compared with the
United States. Whilst collective bargaining is organized based on workers' occupational
categories in the US, collective negotiation is mainly company-based in the UK with little
negotiation at sector level. There is still industry level bargaining in some industries (e.g. textile
and furniture industries) but a clear move towards bargaining at local level was observed during
the 1980s. Even though some sectors are more traditionally unionized (particularly in the public
sector) and some have very low unions densities (such as the banking sector), most sectors of
activity have companies that have a collective negotiation (see table 1, below). From a political
point of view, there is no straightforward connection between political parties and trade unions
even though some major organizations have recently supported the Labour Party. But such a
support has nothing in common with a more direct support that could be observed in other
European countries (such as in Belgium) and the political power of labour unions has been
https://doi.org/10.1016/j.socscimed.2020.112796
3
considerably reduced between 1980 and 1993 (Addison and Siebert, 2002). The relationship
between politics and the world of work is traditionally perceived under the idea of ‘free
collective bargaining’ (Hyman, 2006) according to which industrial relations are largely
autonomous from the political sphere. Therefore, it can be assumed that instead of taking a
sector point of view as it could be the case in the US, the company level is the key level in the
understanding of British labour unions.
This is particularly true regarding health and safety. In Britain, the 1974 Health and Safety at
Work act gave to trade unions the statutory right to be involved in regulating health and safety
at the workplace (Walters, 1987). But the role played by self-regulation and goal setting has
been seen as contributing making health and safety issues more vulnerable to deregulatory
initiatives (Beck and Woolfson, 2000) and reforms are needed to ensure workers representation
in health and safety, in a context that is characterized by a reduced number of workplaces where
trade unions are recognized (James and Walters, 2002). Looking at the association
between workplace managers’ perceptions of the health and safety risks faced by workers and
the degree to which workers have control over those risks in Britain, Bryson (2016) shows that
union density is positively associated with risks, which could indicate that the demand for union
representatives rises with health and safety risks. Though, the presence of a labour organization
within the workplace is not associated with workers’ control over risks.
Surprisingly, even though the workplace has been seen as playing a central role in explaining
health variations, epidemiological research on the association between collective bargaining
and workers’ health is almost non-existent. Most of the scientific literature looking at labour
organizations focuses either on the role played by unions in explaining economic change such
as wages (at company level such as in Bryson, 2007; or distinguishing unionized from non-
unionized employees such as in Hildreth, 2000), social solidarity (Rosetti, 2019) and
privatization (Zoogah, 2009) or on the determinants explaining union membership (Mason and
https://doi.org/10.1016/j.socscimed.2020.112796
4
Bain, 1993) particularly when looking at the young workforce (Cregan et al., 1992; Cregan
and Johnston, 1990). The larger part of this literature has used cross-sectional data, applying
multilevel modeling and paying particular attention to the differences across sectors of activity
or across countries (see, for instance, Zoogah 2009; Fazekas 2011; Rosetti 2019)
An interesting point to look at is the factors influencing individuals’ choice to join a labour
union. This has been done using either an individualistic approach (multilevel modeling on
micro-data) or a holistic approach (principally using econometrics). From a holistic perspective,
union decline during the 19980s and 1990s is rather endogenous to labour market changes
(Checchi and Visser, 2005). From the individualistic point of view, Toubøl and Jensen
demonstrate that workplace union density remains the main predictor of whether or not an
employee is going to join a union (2014), which is similar to what was shown by Fazekas
(2011). Similarly, Cregan (2013) has shown using a rough change score analysis (longitudinal
data) that industrial actions are associated with union joining (but not members’ leaving)
behaviours. Looking particularly sixteen-year-old school leavers in London (1979-1980),
Cregan and Johnston (Cregan and Johnston, 1990) show that personal values are highly
significant in explaining why respondents joined a labour union, additionally to workplace
factors.
The few studies available on the association between labour unions and workers’ health mainly
use multilevel models. Looking at the direct association between health and unionization,
Taylor (1987) has demonstrated that greater union coverage is associated with fewer serious
workplace accidents. But the association between unionization and health might also be
negative. Appleton and Baker have shown that union mines experience more disabling injuries
per year than non-union mines (Appleton and Baker, 1985). Similarly, blue-collar union
members are more likely to report health conditions caused by job accidents compared with
non-unionized blue-collar workers (Worrall and Butler, 1983). As a matter of fact, the
https://doi.org/10.1016/j.socscimed.2020.112796
5
association between unionization and health is not only about health as such but also about
whether workers declare work accidents or not (Morse et al., 2003). It might also be assumed
that some sectors of activity, more likely to employ a unionized workforce, are particularly
affected by occupational health issues (Laden et al.. 2007; Stern et al.. 2000 ; Stern et al.. 1997).
Finally, looking at institutional changes it has been shown that high rates of unionization lead
to improved workers’ health indicators. For instance, comparing hotel workers in Canada and
the US. Siddiqi and Hertzman mentioning a doctoral study carried on by Zuberi (2006) have
pointed out that the ability of unionized hotel workers to secure collective agreements in Canada
provided them with “higher wages, improved benefits, better working conditions, and increased
job security than their US counterparts” (Siddiqi and Hertzman, 2007). Unions contribute to
securing professional pathways and may be a factor explaining health variations among the
workforce (Kim et al., 2012).
One of the main disadvantages of using cross-sectional data is that causality cannot be assessed
properly. A few studies have focused on the benefits of using a longitudinal approach to assess
the association between union membership and wages (for references, please see Freeman
1984) but the association between health and collective bargaining is still to investigate using
such an approach. Over the past few years, an extensive amount of research was published
about the impact of professional transitions on workers’ health focusing both on different types
of transitions (Wels, 2016) such as retirement (Di Gessa et al., 2016; Di Gessa and Grundy,
2014), return to employment (Hyde and Dingemans, 2017), bridge jobs (Ruhm, 1990) or care
activities (Benson et al., 2017; Evandrou and Glaser, 2004) , and on different types of health
measurements such as the self-reported health (Di Gessa and Grundy, 2014), the physical
exposure (Platts et al., 2013), quality of life (Hyde et al., 2003) or depression (Börsch-Supan et
al., 2009; Croda, 2010), to name just a few. Several articles have highlighted the benefits of
remaining in paid work, but the topic remains controversial as one needs to account for the
https://doi.org/10.1016/j.socscimed.2020.112796
6
heterogeneity of the workforce (Hashimoto, 2015) such as for the length of the time sequence
that is taken into consideration (Wels, 2019). The sector of activity, the level of education,
financial wealth, gender, and the socioeconomic status also play a key role in explaining the
association between employment and health (Annesley, 2007; Di Gessa et al., 2016; Di Gessa
and Grundy, 2014; Lain, 2015; Schaap et al., 2018). That is also the case of the ethnicity
(Sundquist and Johansson, 1998) that is an emerging but relevant dimension to look at in
the United Kingdom (Bécares et al., 2012; Mathur et al., 2013). One major feature of these
approaches is to take a longitudinal point of view on health, looking at health as a process
changing over time.
Collective bargaining remains absent from these perspectives. In a recent article, I have
attempted to assess the specific effects unionization could have on older workers’ health in the
United States (Wels, 2018). Using a rough change score analysis (i.e. the dependent variable
was the difference in health outcomes over time), the article demonstrates the benefits of being
unionized when retiring. These are partly due to the fact that unionized workers tend to retire
slightly earlier. Although, this previous study did not specifically look at the full working
population and did not account for all employment trajectories. Additionally, no distinction was
made between union membership and union presence. The current study raises a much wider
question and uses a more suitable methodology to deal with it.
Using a longitudinal perspective, this study assesses whether and to what extent union presence
and union membership could play a role in explaining health variations, after controlling for
other cofounders and pre-transition health. More precisely, the article aims to answer two
research questions. On the one hand, what is the impact of transitioning from a workplace in
which there is no collective organization to a workplace in which there is such an organization,
and vice versa? On the other hand, what is the specific impact of working in a unionized
workplace and being member of a labour organization versus not being a member?
https://doi.org/10.1016/j.socscimed.2020.112796
7
Methods
Understanding Society
Data come from Understanding Society the UK Household Longitudinal Study (UKHLS)
dataset (former British Household Panel Survey. BHPS). Understanding Society is a
longitudinal survey of the members of approximately 40,000 households (at Wave 1) in the
United Kingdom and provides longitudinal and cross-sectional data at individual or household
unit. Information about union presence and union membership is only available in waves 2, 4,
6 and 8. To match the baseline sample with information about unions, the study uses waves 2,
4, 6 and 8. The following analyses focus on the working-age population (18-65 years of age at
the baseline) that is employed in wave 2 and in wave 4 (N=9,755) in order to account for the
change in union presence and union membership between wave 2 to wave 4 and its impact over
the following waves.
Dependent variables
The study looks at two outcome variables: the Physical Component Summary (PCS) and the
Mental Component Summary (MCS). Both the Physical Component Summary and the Mental
Component Summary convert valid answers to twelve origin questions into a single
physical/mental functioning score, resulting in a continuous scale with a range of 0 (low
functioning) to 100 (high functioning) (Ware et al., 2002). These scores are calculated based
on several variables on a five-point Likert scale: the general health, whether health limits
moderate activities, whether health limits several flights of stairs, whether physical health limits
the amount of work, whether mental health limits the accomplishment of tasks, whether mental
health means less care in working, whether pain interferes with work, whether one feels calm
and peaceful, has a lot of energy, feels downhearted and depressed, and whether physical or
mental health interfere with social life. Except for the general health and whether the health
https://doi.org/10.1016/j.socscimed.2020.112796
8
limits moderate activities, information is collected for the last four weeks only. Both PCS and
MCS scales are normalized (Booker and Sacker, 2011).
Independent variable
UKHLS contains two types of information about labour unions: union presence and union
membership. Union presence distinguishes workers that are working in a workplace where there
is a trade union or a similar body, such as a staff association, recognized by the management
for negotiating pay or conditions for people doing the same sort of job. The variable is coded
‘yes’ or ‘no’. Union membership distinguishes, among respondents working in a place where
there is a union presence, those who are members of such a union or association from those
who are not. The study combines information about union membership and union presence and
change in employment status between wave 2 (all respondents aged 18 to 65) and wave 4 (all
employees). There are 11 possible types of economic activities at the baseline: ‘family care’,
‘maternity/paternity leave’, ‘sickness’, ‘study or training’, ‘unemployment’, ‘retirement’,
‘employment and union membership’, ‘employment and union presence’, ‘employment and no
union presence’, ‘self-employed’, and ‘other’ (0.58 per cent of the original sample). As non-
employed respondents account for 10.8 per cent of the sample, the results section only focuses
on the population that was employed in waves 2 and 4.
Control variables
The model controls for several fixed and time-varying covariates. Fixed covariates include the
gender (ref.: male), the race/ethnicity (ref.: white English. Scottish. Welsh or Northern Irish), a
quadratic function of the age at the baseline, the geographical area based on the Government
Office Region nomenclature (ref.: South-East of England), the highest level of education
obtained, the sector of activity, the type of occupation, job satisfaction (that is coded on a Likert
scale from completely dissatisfied (1) to completely satisfied (7)), whether the job was
temporary or permanent and the size of the company in wave 4. Time-varying covariates are a
https://doi.org/10.1016/j.socscimed.2020.112796
9
logarithm function of the daily incomes, the employment status, whether employed respondents
are unionized or work in a unionized company in waves 6 and 8 and household composition.
The total income summarizes information about monthly income from labour (gross income,
monthly), income from benefits and other sources (monthly) as well as income from savings
and investment (annual). The variable is post-calculated on a daily basis. Finally, employment
status distinguishes paid employment, self-employment, unemployment, retirement, maternity
leave, those looking after family or home, full-time students, long-term sick or disabled
respondents, people on a government training scheme, unpaid workers in a family business and
those working in an apprenticeship. The employment status is controlled for waves 6 to 8 as
only employed respondents are selected in wave 4.
Propensity score matching
Transition of union presence at the workplace could be endogenously determined by the reasons
that could affect the health, such as the age, the income level or the type of occupation. To
ensure that this is not the case (i.e. to avoid endogenous selection on union presence
availability), we are using a propensity score matching method. Propensity score matching
(PSM) is a statistical technique that estimates the effect of treatment, policy, or other kind of
intervention by accounting for the covariates that predict receiving the treatment. In this article,
we analyse the impact of being employed in a unionized workplace versus being employed in
a non-unionized workplace in wave 4 using several covariates: age, gender, ethnicity, education
level, income level, area of residence, type of occupation, sector of activity and whether the job
is temporary or not. First, we run an ordered logit regression to calculate the propensity scores
as such (in this case, the odd ratios). Second, we use the R-package ‘matchit’ to create a matched
dataset (Zhang et al., 2019). In our case, the purpose of the matching technique is to find one
(or more) employed respondents working in a non-unionized workplace for every employed
respondent working in a unionized workplace with similar observable characteristics. To do so,
https://doi.org/10.1016/j.socscimed.2020.112796
10
we use a Nearest neighbour matching method that selects best control matches for each
individual in the unionized group using a distance (in logit) measure. The size of the matched
dataset (m(d)) is 4,704 respondents against 9,755 in the original dataset (d).
Latent growth curve
A latent growth curve is used to assess the association between change in union presence
(versus no union within the company) and PCS and MCS on the one hand, and, on the other
hand, looking at the population that is employed in a unionized company, the association
between union membership (versus no membership) and PCS and MCS.
[Please, insert figure 1]
The latent growth curve is a multivariate approach (Masyn et al., 2013) in which intra-
individual change is captured by the measurement model for the growth factors, and
interindividual differences are captured by the structural model (i.e. the mean and
variance/covariance structure of the growth factors). The linear latent growth curve contains
two latent factors, an intercept and a slope. The dependent variable is the observed outcome Y
for individual i at time t. The intercept (random intercept factor) is the expected outcome on y
for respondent i at baseline). The slope (random linear slope factor) is the change in the
expected outcome of the dependent variable for respondent i for one-unit increase in time.
Figure 1 shows the main features of the model where ‘H’ is the dependent variable (PCS and
MCS) calculated in all waves, ‘c’ is the set of fixed-covariates, ‘v’ are the time varying
covariates and ‘T’ is the variable of interest, i.e. the type of change in union status from wave
2 to wave 4.
Four models are tested. First (models 1 to 4), we look at the association between union presence
and union absence and PCS and MCS in the original dataset (d) and in the matched dataset
(m(d)). Second (models 1’ to 4’), we assess the same association but taking into consideration
https://doi.org/10.1016/j.socscimed.2020.112796
11
information about whether respondents moved from one company to another between wave 2
and 4. Third (models 5 and 6), we selected only the population working in a unionized
workplace in both waves 2 and 4 and assess the association between union membership and no
membership and PCS and MCS. Fourth (models 5’ and 6’), we replicate the third model
considering potential change in employer over the sequence. All models control for fixed and
time-varying covariates. Attrition is controlled using UKHLS longitudinal weighting and non-
responses are controlled using full information maximum likelihood estimation. Analyses were
carried out using the R-package Lavaan.
Results
Descriptive statistics
Table 1 shows the total number of respondents transitioning from wave 2 (all labour market
statuses) to wave 4 (employees only) depending on whether they are members of a trade union
or work in a unionized or non-unionized workplace. For instance, it can be observed that among
those who were in family care in wave 2 and employees in wave 4 (1.7 per cent of the sample),
10 were members of a labour union, 40 were working a unionized workplace and 119 where
neither member of a labour union nor working in a unionized workplace. As in most
longitudinal studies, what can be observed is that the majority of the sample was employed in
wave 2 and is employed in wave 4 (89.2 per cent). 2.7 per cent of the sample was in education
in wave 2 and employed in wave 4 and 2.5 per cent was unemployed at the baseline. The bottom
line indicates the percentage of employees that are unionized, work in a unionized workplace
or work in a non-unionized workplace in wave 4. The unionization rate is 31.5. Interestingly,
19.8 per cent of the sample is working in a unionized workplace but is not unionized and 48.7
per cent of the sample worked in a non-unionized workplace and is, by definition, not unionized
(there is no such thing as community unions (Royle and Urano, 2012) that would allow non-
workers to join a union outside of the labour market). Overall, it can be estimated that the
https://doi.org/10.1016/j.socscimed.2020.112796
12
percentage of workers working in a unionized workplace (independently of whether they are
unionized or not) was 51.3 per cent in 2012-2013 (when data were collected). Unionized
workers are slightly overrepresented (by 5 percentage points) compared with OECD figures
(about 26 per cent in 2013). Three reasons could explain this difference. First, it can be imputed
to the original age group that was selected at the baseline. Second, OECD calculation is based
on administrative data (for the UK) and accounts for independent unions (without membership
in the Irish Republic but with members of Irish Unions working in Northern Ireland) whilst
UKHLS collects information on a declarative basis. Finally, OECD calculation accounts for
retired workers, self-employed and unemployed workers (but total membership is deflated by
10 per cent for these categories of workers).
[Please, insert table 1]
Table 2 shows the percentage of respondents that are working in a workplace where there is no
collective negotiation, respondents that are working in a workplace where there is collective
negotiation and are not members of such an organization and respondents who are working in
a unionized workplace and are members of a trade union or staff association, by type of industry
(2007 nomenclature). What can be clearly observed is that, except in the case of two industry
types (trading agents and private households) that only account for less that 2 per cent of the
total workforce all industries have a minimum percentage of unionized companies. As
expected, high union densities can be observed in the public sector or in sectors that used to be
public such as social security, train industry, public administration, health service and
education.
[Please, insert table 2]
Propensity scores
Propensity scores (in odds ratios) calculated for wave 4 are shown in figure 2. Non statistically
significant results are excluded from the figure. Several interesting trends flow from the figure.
https://doi.org/10.1016/j.socscimed.2020.112796
13
Firstly, the quadratic function of age indicates that the odds of being employed in a unionized
workplace increase by 9 units for one unit increase in years of age. In other words, young works
are less likely to work in a unionized workplace independently from the sector of activity or the
type of occupation. Secondly, there are regional variations that are independent from the
distribution across sectors of activity. For instance, the odds of working in a unionized
environment are higher in Scotland and Wales than in the South-East of England. Thirdly, the
company size plays a significant role. Fourthly, the propensity scores also demonstrate the
heterogeneity when looking at the ethnicity. Employees from a Chinese background (no
statistically significant results were found for other ethnic groups) have lower odds of being
employed in a unionized company. Finally, it can be observed that female workers have higher
odds of working in a unionized workplace than men.
[Please, insert figure 2]
Latent growth curve
Estimates in table 3 exhibits the intercept (Int.) and slopes from the latent growth curve. The
intercept controls for the health level at the baseline (wave 2) whilst the slope gives information
about the change in health (latent curve) following the selected transition (from wave 4 to wave
8). The topside of the table shows the results from the model looking at the association between
union status transitions between wave 2 and wave 4 and, physical and mental health, after
controlling for fixed and time-varying covariates. Models 1 and 3 were performed on the
original dataset whilst models 2 and 4 were performed on the matched dataset. The bottom side
of the table takes accounts of whether respondents moved from one employer to another over
the sequence.
Three main findings pop out from table 3. First, there are clear health benefits of being
employed in a unionized workplace in both waves (2 and 4) compared with being employed in
https://doi.org/10.1016/j.socscimed.2020.112796
14
a non-unionized workplace over the sequence. The slopes of the latent curve clearly indicate
that, compared with those who work in a non-unionized environment, those who do
(independently from whether they actually are members of a labour union) have a more positive
change in both physical (0.187, significant at 99 per cent) and mental health (0.356, significant
at 99 per cent). Estimates from the matched dataset (m(d)) have a lower intensity and are
slightly less significant (at 90 per cent), but still positive (respectively, 0.056 and 0.209).
Second, there are mental health benefits of working in a unionized workplace at the baseline
and in a non-unionized workplace in wave 4 versus not working in a unionized workplace in
both waves. Those who started from being employed in a unionized workplace have an increase
in slope that is higher by 0.451 and 0.552 by units of time compared with those who worked in
a non-unionized workplace in both waves, respectively for the original and the matched dataset,
both statistically significant at 99 per cent.
Finally, taking into consideration potential changes in employers reduces the significance levels
but it can be assumed that those moving from a unionized workplace to a non-unionized
workplace and those moving to a non-unionized workplace to a unionized-workplace benefit
from a more positive change in mental health compared with those who keep being employed
in the same company where there is no collective negotiation.
[Please, insert table 3]
Finally, table 4 looks at one specific question: does health benefits differ depending on whether
one is a member of labour unions or work in a unionized company without being a member of
such a union? As can be observed in table 4, little evidence can be found that being a member
improves the benefits of working in a unionized workplace. Though, when looking at the
change in mental health for those who keep working in the same workplace and are members
of a labour union versus those who keep working in the same workplace but did not unionized,
https://doi.org/10.1016/j.socscimed.2020.112796
15
one can observe a slight and barely significant positive association (0.216, significant at 90 per
cent).
[Please, insert table 4]
Limitations
This study has some important limitations particularly due to the use of longitudinal micro-
data that could feed further research on this matter. First, the impact of labour unions depends
on a larger environment in which they operate (Hipp and Givan, 2015). The latent growth curve
focuses on the intra- and inter-individual changes in health but does not look at structural factors
that could potentially affect mental and physical health at the workplace level. This study uses
an individualist approach but ignore holistic dimensions beyond the workplace level. Second,
the use of longitudinal analyses when looking at labour unions has been criticized by Freeman
(Freeman, 1984), particularly because longitudinal methods tend to extrapolate
misclassification. The effects of misclassification when looking at union membership could be
important, even if the number of misclassified cases is low. However, as both union
membership and union presence are collected on a declarative basis, the risk of
misclassification is quite limited in Understanding Society. Additionally, the non-response rate
is relatively low (overall, less than 0.02 per cent of the total sample in all waves). One way to
deal with both issues would be to use a longer sequence in further research, not only focusing
on the transition from wave 2 to wave 4 and controlling for further transitions types as it is done
in the current study. Finally, a gender perspective is cruelly missing when looking at industrial
relations. Overall, female workers tend to be more unionized than men, but they face other
challenges over the life course that are just controlled and not analysed in depth in this study.
https://doi.org/10.1016/j.socscimed.2020.112796
16
Discussion
Social epidemiology and, particularly, research focusing on employment transitions have
ignored the contribution of industrial relations in explaining health outcomes. Similarly, health
issues are quite absent from industrial relation studies that are much more focused on wages
and social solidarity except in the case of Health and Safety that has been extensively
discussed in Britain (see above) but with no use of longitudinal data. This study aimed to put
new emphasis on the potential relationship between industrial relations and workers’ health.
There are several reasons for why longitudinal methods should be used to deal with the
association between collective negotiation and health. First, the association between health and
union membership raises causation issues. By controlling for baseline health, the latent growth
curve can tackle such a problem. Second, unionization and union membership potentially vary
over time and there is a clear need to assess the long-term effects of being unionized at a certain
point in time as the health effects labour market transitions go beyond the moment when the
transition occurs (Wels, 2019). Finally, the use of a latent model that accounts for the change
in slope over time is a good way to control for external factors that may affect health at a certain
point in time (e.g. unemployment). This raised many methodological issues among which the
risk of selection bias (that is the reason why we used a propensity score matching method), the
necessity to look at the potential change in employer over time and the necessity to look at the
long-term change in health following the transition.
The study shows that unions and, more particularly, collective negotiation at the workplace play
a role in supporting workers’ physical and mental health, with a stronger and more statistically
significant impact when looking at mental health. The study did not find physical health benefits
of being a member of a union (versus working in a unionized environment) but slight mental
health benefits could be considered. Although, the study was not able to investigate the potential
impact of implementing collective negotiation in companies that have not had a union before.
https://doi.org/10.1016/j.socscimed.2020.112796
17
This is particularly due to the small size of respondents benefiting from such a change (n=231)
and should be investigated in further research.
What could be the implications of such findings? About half of the British working population
is employed in companies where there is no collective negotiation and, consequently, cannot
join a union. This has a particular impact in terms of health and safety, despite new regulations
aiming at promoting workers’ participation in companies where there is no union
representation. One thing for sure is that favouring collective negotiation in British companies
is a clear way to improve workers’ mental and – to a lesser extent physical health. Therefore,
securing recognition of labour unions could be seen as a lever for improving workers health
but this requires paying attention to several factors included in the agreement such as which
unions will be recognized by the employer, which issues will be negotiated and the way they
will be negotiated.
More research is needed on this matter, particularly using cross-national comparison to assess
the impact of collective bargaining systems (and their content) on workers’ health (Hyman,
2001). Combining cross-national comparison and longitudinal data would be the ideal. Though,
many surveys that allow looking at the association between employment transitions and health
(e.g. ELSA in the UK, SHARE in Europe or JSTAR in Japan) do not contain information about
collective bargaining. Collective negotiation is fully a part of working lives and there are
obvious research benefits of integrating such a dimension in further research on this matter.
https://doi.org/10.1016/j.socscimed.2020.112796
18
Bibliography
Addison, J.T., Siebert, W.S., 2002. Changes in Collective Bargaining in the UK. IZA 562, 57.
Annesley, C., 2007. Lisbon and social Europe: towards a European “adult worker model”
welfare system. J. Eur. Soc. Policy 17, 195205.
https://doi.org/10.1177/0958928707078363
Appleton, W.C., Baker, J.G., 1985. The Effect of Unionization on Safety in Bituminous Deep
Mines: Comment. J. Labor Res. 6, 209210.
Bécares, L., Nazroo, J., Albor, C., Chandola, T., Stafford, M., 2012. Examining the
differential association between self-rated health and area deprivation among white
British and ethnic minority people in England. Soc. Sci. Med. 74, 616624.
https://doi.org/10.1016/j.socscimed.2011.11.007
Beck, M., Woolfson, C., 2000. The regulation of health and safety in Britain: from old Labour
to new Labour. Ind. Relations J. 31, 3549. https://doi.org/10.1111/1468-2338.00145
Benson, R., Glaser, K., Corna, L.M., Platts, L.G., Gessa, G. Di, Worts, D., Price, D.,
Mcdonough, P., Sacker, A., 2017. Do work and family care histories predict health in
older women? Eur. J. Public Health online, 16. https://doi.org/10.1093/eurpub/ckx128
Booker, C.L., Sacker, A., 2011. Health over the life course: associations between age,
employment status and well-being. Underst. Soc. Early Find. from first wave UK’s
Househ. Longitud. study 7586.
Börsch-Supan, A., Brugiavini, A., Croda, E., 2009. The Role of Institutions and Health in
European Patterns of Work and Retirement. J. Eur. Soc. Policy 19, 341358.
https://doi.org/10.1177/1350506809341515
Bryson, A., 2016. Health and safety risks in Britain’s workplaces: where are they and who
controls them? Ind. Relations J. 47, 547566. https://doi.org/10.1111/irj.12162
https://doi.org/10.1016/j.socscimed.2020.112796
19
Bryson, A., 2007. The effect of trade unions on wages. Reflets Perspect. la Vie Econ. 46, 33
45.
Checchi, D., Visser, J., 2005. Pattern persistence in European trade union density: A
longitudinal analysis 1950-1996. Eur. Sociol. Rev. 21, 121.
https://doi.org/10.1093/esr/jci001
Cregan, C., 2013. Does workplace industrial action increase trade union membership? An
exchange relationship approach to union joining and leaving behaviour. Int. J. Hum.
Resour. Manag. 24, 33633377. https://doi.org/10.1080/09585192.2013.775956
Cregan, C., Johnston, S., 1990. An Industrial Relations Approach to the Free Rider Problem:
Young People and Trade Union Membership in the UK. Br. J. Ind. Relations 28, 84104.
https://doi.org/10.1111/j.1467-8543.1990.tb00355.x
Cregan, C., Rudd, C., Johnston, S., 1992. Young People and Trade Union Membership: An
International Comparative Study. Econ. Labour Relations Rev. 3, 165180.
https://doi.org/10.1177/103530469200300209
Croda, E., 2010. The Role of Institutions and Health in European Patterns of Work and
Retirement. J. Eur. Public Policy 19, 341358.
https://doi.org/10.1177/1350506809341515.The
Di Gessa, G., Corna, L.M., Platts, L.G., Worts, D., McDonough, P., Sacker, A., Price, D.,
Glaser, K., 2016. Is being in paid work beyond state pension age beneficial for health?
Evidence from England using a life-course approach. J. Epidemiol. Community Health
onlinefirs, jech-2016-208086. https://doi.org/10.1136/jech-2016-208086
Di Gessa, G., Grundy, E., 2014. The relationship between active ageing and health using
longitudinal data from Denmark, France, Italy and England. J. Epidemiol. Community
Heal. 68, 261267. https://doi.org/10.1136/jech-2013-202820
Evandrou, M., Glaser, K., 2004. Family, work and quality of life: changing economic and
https://doi.org/10.1016/j.socscimed.2020.112796
20
social roles through the lifecourse. Ageing Soc. 24, 771791.
https://doi.org/10.1017/S0144686X04002545
Fazekas, Z., 2011. Institutional effects on the presence of trade unions at the workplace:
Moderation in a multilevel setting. Eur. J. Ind. Relations 17, 153169.
https://doi.org/10.1177/0959680111400897
Freeman, R.B., 1984. Longitudinal Analyses of the Effects of Trade Unions. J. Labor Econ. 2,
126. https://doi.org/10.1086/298021
Hashimoto, H., 2015. Impacts of Leaving Paid Work on Health, Functions, and Lifestyle
Behavior: Evidence from JSTAR panel data. RIETI Discuss. Pap. Ser. 15-E-114 15-E-
114, 18.
Hildreth, A.K.G., 2000. Union wage differentials for covered members and nonmembers in
Great Britain. J. Labor Res. 21, 133147. https://doi.org/10.1007/s12122-000-1008-1
Hipp, L., Givan, R.K., 2015. What do unions do? A cross-national reexamination of the
relationship between unionization and job satisfaction. Soc. Forces 94, 349377.
https://doi.org/10.1093/sf/sov051
Hyde, M., Dingemans, E., 2017. Hidden in Plain Sight? Does Stricter Employment Protection
Legislation Lead to an Increased Risk of Hidden Unemployment in Later Life? Work.
Aging Retire. 3, 231242. https://doi.org/10.1093/workar/wax013
Hyde, M., Wiggins, R.D., Higgs, P., Blane, D.B., 2003. A measure of quality of life in early
old age: The theory, development and properties of a needs satisfaction model (CASP-
19). Aging Ment. Heal. 7, 186194. https://doi.org/10.1080/1360786031000101157
Hyman, R., 2006. Trade union research and cross-national comparison [online]. LSE Res.
Online.
Hyman, R., 2001. Understanding European Trade Unionism. Between Market, Class &
Society. Sage, London.
https://doi.org/10.1016/j.socscimed.2020.112796
21
James, P., Walters, D., 2002. Worker representation in health and safety: options for
regulatory reform. Ind. Relations J. 33, 141156. https://doi.org/10.1111/1468-
2338.00225
Kim, I.H., Muntaner, C., Vahid Shahidi, F., Vives, A., Vanroelen, C., Benach, J., 2012.
Welfare states, flexible employment, and health: A critical review. Health Policy (New.
York). 104, 99127. https://doi.org/10.1016/j.healthpol.2011.11.002
Laden, F., Hart, J.E., Smith, T.J., Davis, M.E., Garshick, E., 2007. Cause-specific mortality in
the unionized U.S. trucking industry. Environ. Health Perspect. 115, 11921196.
https://doi.org/10.1289/ehp.10027
Lain, D., 2015. Work Beyond Age 65 in England and the USA, in: Scherger, S. (Ed.), Paid
Work Beyond Pension Age. Comparative Perspectives. Palgrave Macmillan UK,
London, pp. 3156. https://doi.org/10.1057/9781137435149
Mason, B.O.B., Bain, P., 1993. The Determinants of Trade Union Membership in Britain : A
Survey of the Literature. ILR Rev. 46, 332351.
Masyn, K.E., Petras, H., Liu, W., 2013. Growth Curve Models with Categorical Outcomes.
Encycl. Criminol. Crim. Justice 20132025. https://doi.org/10.1007/978-1-4614-5690-
2_404
Mathur, R., Grundy, E., Smeeth, L., 2013. Availability and use of UK based ethnicity data for
health research. Natl. Cent. Res. Methods Work. Pap. Ser. 130.
Morse, T., Punnett, L., Warren, N., Dillon, C., Warren, A., 2003. The relationship of unions
to prevalence and claim filing for work-related upper-extremity musculoskeletal
disorders. Am. J. Ind. Med. 44, 8393. https://doi.org/10.1002/ajim.10234
Platts, L.G., Netuveli, G., Webb, E., Zins, M., Goldberg, M., Blane, D., Wahrendorf, M.,
2013. Physical occupational exposures during working life and quality of life after
labour market exit: results from the GAZEL study. Aging Ment. Health 7863, 3741.
https://doi.org/10.1016/j.socscimed.2020.112796
22
https://doi.org/10.1080/13607863.2013.781120
Rosetti, N., 2019. Do European trade unions foster social solidarity? Evidence from
multilevel data in 18 countries. Ind. Relations J. 50, 84101.
https://doi.org/10.1111/irj.12242
Royle, T., Urano, E., 2012. A new form of union organizing in Japan? Community unions and
the case of the McDonald’s “McUnion.” Work. Employ. Soc. 26, 606–622.
https://doi.org/10.1177/0950017012445093
Ruhm, C., 1990. Bridge jobs and partial retirement. J. Labor Econ. 8, 482501.
Schaap, R., de Wind, A., Coenen, P., Proper, K., Boot, C., 2018. The effects of exit from
work on health across different socioeconomic groups: A systematic literature review.
Soc. Sci. Med. 198, 3645. https://doi.org/10.1016/j.socscimed.2017.12.015
Siddiqi, A., Hertzman, C., 2007. Towards an epidemiological understanding of the effects of
long-term institutional changes on population health: A case study of Canada versus the
USA. Soc. Sci. Med. 64, 589603. https://doi.org/10.1016/j.socscimed.2006.09.034
Stern, F.B., Ruder, A.M., Chen, G., 2000. Proportionate mortality among unionized roofers
and waterproofers. Am. J. Ind. Med. 37, 478492. https://doi.org/10.1002/(SICI)1097-
0274(200005)37:5<478::AID-AJIM4>3.0.CO;2-8
Stern, F.B., Sweeney, M.H., Ward, E., 1997. Proportionate mortality among unionized
construction ironworkers. Am. J. Ind. Med. 31, 176187.
https://doi.org/10.1002/(SICI)1097-0274(199702)31:2<176::AID-AJIM7>3.0.CO;2-Y
Sundquist, J., Johansson, S.E., 1998. The influence of socioeconomic status, ethnicity and
lifestyle on body mass index in a longitudinal study. Int. J. Epidemiol. 27, 5763.
https://doi.org/10.1093/ije/27.1.57
Taylor, G.S., 1987. A reanalysis of the relation between unionization and workplace safety.
Int. J. Heal. Serv. 17, 443453.
https://doi.org/10.1016/j.socscimed.2020.112796
23
Toubøl, J., Jensen, C.S., 2014. Why do people join trade unions? The impact of workplace
union density on union recruitment. Transf. Eur. Rev. Labour Res. 20, 135154.
https://doi.org/10.1177/1024258913516902
Walters, D., 1987. Health and safety and trade union workplace organization?a case study in
the printing industry. Ind. Relations J. 18, 4049. https://doi.org/10.1111/j.1468-
2338.1987.tb00886.x
Ware, J.E., Kosinski, M., Turner-Bowker, D.M., Gandek, B., 2002. How to score version 2 of
the SF-12 HEALTH Survey. RI: Quality Metric Incorporated, Lincoln.
Wels, J., 2019. Does retirement affect the mental and general health of the older Japanese
workforce? A four waves follow-up using the Japanese Study of Ageing and Retirement
(JSTAR- RIETI). Japan Inst. Labour Policy Train. 126.
Wels, J., 2018. Are there health benefits of being unionized in late career? A longitudinal
approach using HRS. Am. J. Ind. Med. 61, 751761.
Wels, J., 2016. The Statistical Analysis of End of Working Life: Methodological and
Sociological Issues Raised by the Average Effective Age of Retirement. Soc. Indic. Res.
129, 291315. https://doi.org/10.1007/s11205-015-1103-6
Worrall, J.D., Butler, R.J., 1983. Health conditions and job hazards: Union and nonunion
jobs. J. Labor Res. 4, 339347. https://doi.org/10.1007/BF02685341
Wu, W., West, S.G., Taylor, A.B., 2009. Evaluating Model Fit for Growth Curve Models:
Integration of Fit Indices From SEM and MLM Frameworks. Psychol. Methods 14, 183
201. https://doi.org/10.1037/a0015858
Zhang, Z., Kim, H.J., Lonjon, G., Zhu, Y., 2019. Balance diagnostics after propensity score
matching. Ann. Transl. Med. 7, 1616. https://doi.org/10.21037/atm.2018.12.10
Zoogah, D., 2009. A Multilevel Analysis of the Moderating Role of Trade Union Strength in
the Relationship between Privatization and Corporate Governance. Adv. Compet. Res.
https://doi.org/10.1016/j.socscimed.2020.112796
24
17, 53.
Zuberi, D., 2006. Differences That Matter: Social Policy and the Working Poor in the United
States and Canada. Cornell University Press, Cornell. https://doi.org/10.2307/20460675
https://doi.org/10.1016/j.socscimed.2020.112796
25
Table 1. Transitions to employment between wave 2 and wave 4 and union density and union membership in wave 4
(Wave 2 > Wave 4)
Member*
Union**
No Union
Per Cent
Familly Care
10
40
119
1.7
Maternity
47
19
49
1.1
Sick
1
6
18
0.2
Student
46
58
173
2.7
Unemployed
24
56
177
2.5
Retired
2
8
39
0.5
Other
4
2
19
0.2
Self-employed
20
25
146
1.9
Employed
Member
2,637
175
207
29.6
Union
241
1,265
358
18.3
No Union
185
363
3,655
41.3
Per Cent
31.5
19.8
48.7
n=10,187
Source: Understanding Society, wave 4, employees only. Author’s calculation
Note: * total number of respondents among the sample at baseline declaring being a member of a collective organization or trade union that is recognized by the management ;
** Total number of respondents declaring working in a workplace where there is a collective organization or a trade union but not being a member of such an organization.
Information was collected on a declarative basis. Source: BHPS, waves 2 and 4 (population aged 18-64 at the baseline) author’s calculation.
https://doi.org/10.1016/j.socscimed.2020.112796
26
Table 2 Percentage of union members, employees working in a unionized workplace, employees working in a non-unionized workplace by Standard Industrial
Classification (2007 classification) and total contribution by sector (in per cent) in wave 4
Member
Union
No Union
Sector's
contribution
Mining
17.9
10.7
71.4
0.3
Trading Agents
0.0
0.0
100.0
0.0
Chemicals
20.2
13.2
66.7
1.1
Priv. Househld
0.0
0.0
100.0
0.1
Earth/Clay/Ston.
20.5
15.4
64.1
0.4
Wholesale
5.1
7.6
87.4
2.6
Mechanical Eng.
20.5
17.9
61.5
4.1
Service Indust.
6.5
3.2
90.3
0.6
Financial Inst
24.4
34.0
41.7
3.1
Clothing/Text.
7.0
11.4
81.6
1.1
Food Industry
25.5
16.8
57.8
1.5
Restaurants
7.1
9.0
84.0
3.5
Volunt./Church
25.9
16.7
57.4
9.5
Other Services
8.1
10.9
81.0
8.4
Trash Removal
28.9
28.9
42.1
0.4
Electrical Eng
10.7
13.0
76.3
1.2
Other Trans.
35.3
17.4
47.3
3.3
Agric..Forestry
10.8
2.7
86.5
0.4
Energy/Water
36.4
32.3
31.3
0.9
Legal Services
11.8
13.6
74.6
2.7
Iron/Steel
39.0
9.8
51.2
0.4
Wood/Paper/Prit.
12.0
14.6
73.4
1.8
Communication
44.7
17.8
37.5
2.9
Constr. Related
12.8
8.3
78.8
1.5
Educ./Sport
52.5
26.3
21.3
15.4
Construction
14.4
12.6
73.0
2.1
Public Admin.
53.0
34.4
12.6
8.6
Insurance
14.6
24.3
61.2
1.0
Health Service
59.2
21.7
19.1
10.0
Retail
15.9
18.9
65.2
9.8
Social Sec.
61.5
30.8
7.7
0.4
Synthetics
16.1
7.1
76.8
0.5
Train System
63.2
34.2
2.6
0.4
Source: Understanding Society, wave 4, employees only. Author’s calculation
https://doi.org/10.1016/j.socscimed.2020.112796
27
Table 3. Association between change in union presence within the workplace and physical and mental health. Latent growth curve on the original dataset (d) and
the matched dataset m(d).
Physical
Mental
(1)
(2)
(3)
(4)
d
m(d)
d
m(d)
W2 > W4
Int.
Slope
Int.
Slope
Int.
Slope
Int.
Slope
Union > Union
-0.185
0.187***
-0.398*
0.056*
-0.720***
0.356***
-0.413*
0.209*
Union > No Union
0.332
0.060
0.163
-0.145
-0.937***
0.451***
-1.300**
0.552***
No Union > Union
0.040
-0.038
-0.225
0.074
-0.541
0.365**
-0.343
0.286
No Union > No Union
Ref.
Ref.
Ref.
Ref.
Ref.
Ref.
Ref.
Ref.
(1’)
(2’)
(3’)
(4’)
Same
employer
Union > Union
-0.174
0.171**
-0.244
0.023
-0.797***
0.358***
-0.549**
0.189
Union > No Union
0.446
0.071
0.398
-0.114
-0.405
0.238
-0.781
0.256
No Union > Union
-0.001
-0.050
-0.143
0.097
-0.637
0.254
-0.410
-0.012
No Union > No Union
Ref.
Ref.
Ref.
Ref.
Ref.
Ref.
Ref.
Ref.
Different
employer
Union > Union
0.585
-0.010
-0.651
0.137
-0.417
0.453***
-0.000
0.327
Union > No Union
0.337
-0.055
0.101
-0.298
-2.005***
0.868***
-2.582**
1.123***
No Union > Union
0.339
-0.118
-0.001
-0.058
-0.314
0.103**
-0.522
0.874**
No Union > No Union
0.444
-0.219*
0.816**
-0.226
-0.314
0.103
-0.679
0.056
N
9,086
4,704
9,086
4,704
Source: Understanding Society, waves 2, 4, 6 and 8, author’s calculation.
Note: Each model controls for fixed and time-varying covariates. Significance levels as follows: * p< 0.10, ** p< 0.05, *** p< 0.01. Model fit. Standardized root-mean-square
residual (SRMR): (1)=0.003 ; (2)=0.003; (3)=0.009; (4)=0.010; (1)=0.003 ; (2)=0.003; (3)=0.009; (4)=0.019. Tucker-Lewis Index (TLI): (1)=0.967 ; (2)=0.981; (3)=0.726;
(4)=0.711; (1)=0.967 ; (2)=0.981; (3)=0.723; (4)=0.707. For interpretation for the SRMR and TLI, please read Wu et al. (2009).
https://doi.org/10.1016/j.socscimed.2020.112796
28
Table 4. Association between change in union membership within unionized workplaces and physical and mental health.
Physical
Mental
(5)
(6)
W2 > W4
Int.
Slope
Int.
Slope
Member > Union
0.324
-0.253
0.052
0.051
Union > Member
-0.001
-0.042
0.138
-0.215
Member > Member
-0.079
0.037
-0.579
0.206*
Union > Union
Ref.
Ref.
Ref.
Ref.
(5’)
(6’)
Same
employer
Member > Union
0.240
-0.143
-0.121
0.032
Union > Member
-0.204
0.025
0.125
-0.254
Member > Member
-0.088
0.039
-0.666
0.216*
Union > Union
Ref.
Ref.
Ref.
Ref.
Different
employer
Member > Union
0.802
-0.689
0.457
0.182
Union > Member
1.985
-0.627
-0.243
0.228
Member > Member
0.627
-0.019
-0.019
0.230
Union > Union
0.361
-0.032
-0.531
0.122
N
4,318
4,318
Source: Understanding Society, waves 2, 4, 6 and 8, author’s calculation.
Note: Each model controls for fixed and time-varying covariates. Significance levels as follows: * p< 0.10, ** p< 0.05, *** p< 0.01. Model fit. Standardized root-mean-square
residual (SRMR): (5)=0.004; (6)=0.009; (5)=0.004 ; (6)=0.009. Tucker-Lewis Index (TLI): (5)=0.958; (6)=0.738; (5)=0.958 ; (6)=0.736.
https://doi.org/10.1016/j.socscimed.2020.112796
29
Figure 1. Model specifications
https://doi.org/10.1016/j.socscimed.2020.112796
30
Figure 2. Propensity scores for union presence within the workplace at wave 4 (selected variables)
Source: Understanding Society (wave 4).
Note: Binary logit regression comparing respondents working in a unionized workplace with respondents in a non-unionized workplace (reference category). Propensity
scores are in odds ratios (i.e. the exponentials of the logits). The reference category for ‘Area’ is South East of England, the reference for ‘Gender’ is ‘male’. The reference for
‘Education’ is ‘A-level’. The reference for ethnicity is ‘British’. The model also controls for socioeconomic position and sector of activity. Significance levels as follows: * p<
0.10, ** p< 0.05, *** p< 0.01.
ResearchGate has not been able to resolve any citations for this publication.
  • Article
    Full-text available
    Trade unions have been analysed quantitatively primarily in their role as vested interest organisations, attempting to quantify the excludable benefits they provide to members rather than examine their wider impact in an institutional context. Power resource theory acknowledges unions as social agents but assumes the willingness to oppose neoliberalism is constant, limited only by scarce power resources. Whilst true in general terms, this fails to explain trends of increasing labour market dualism in resource‐rich industrial relations regimes. This article examines social solidarity as a union power resource, measuring the impact of trade union membership on social attitudes of solidarity. Data were collected from the 2016 European Social Survey for 18 countries, grouped into five distinct industrial relations regimes. The findings suggest that, at European level, union membership still has a significant effect on all dimensions of social solidarity, but these relationships vary significantly across industrial relations regimes.
  • Article
    Full-text available
    Propensity score matching (PSM) is a popular method in clinical researches to create a balanced covariate distribution between treated and untreated groups. However, the balance diagnostics are often not appropriately conducted and reported in the literature and therefore the validity of the findings from the PSM analysis is not warranted. The special article aims to outline the methods used for assessing balance in covariates after PSM. Standardized mean difference (SMD) is the most commonly used statistic to examine the balance of covariate distribution between treatment groups. Because SMD is independent of the unit of measurement, it allows comparison between variables with different unit of measurement. SMD can be reported with plot. Variance is the second central moment and should also be compared in the matched sample. Finally, a correct specification of the propensity score model (e.g., linearity and additivity) should be re-assessed if there is evidence of imbalance between treated and untreated. R code for the implementation of balance diagnostics is provided and explained.
  • Article
    Full-text available
    Objective To assess whether unionization prevents deterioration in self‐reported health and depressive symptoms in late career transitions. Methods Data come from the Health and Retirement Study (N = 6475). The change in self‐perceived health (SPH) and depressive symptoms (CESD) between wave 11 and wave 12 is explained using an interaction effect between change in professional status from wave 10 to wave 11 and unionization in wave 10. Results The odds of being affected by a negative change in CESD when unionized are lower for unionized workers remaining in full‐time job (OR:0.73, CI95%:0.58;0.89), unionized full‐time workers moving to part‐time work (OR:0.66, CI95%:0.46;0.93) and unionized full‐time workers moving to part‐retirement (OR:0.40, CI95%:0.34;0.47) compared to non‐unionized workers. The same conclusion is made for the change in SPH but with odds ratios closer to 1. Conclusion The reasons for the associations found in this paper need to be explored in further research.
  • Article
    Full-text available
    Exit from work leads to different effects on health, partially depending on the socioeconomic status (SES) of people in the work exit. Several studies on the effects of exit from work on health across socioeconomic groups have been performed, but results are conflicting. The aim of this review is to systematically review the available evidence regarding the effects of exit from work on health in high and low socioeconomic groups. A systematic literature search was conducted using Pubmed, Embase, Web of Science, CINAHL and PsycINFO. Search terms related to exit from work, health, SES and design (prospective or retrospective). Articles were included if they focused on: exit from work (early/statutory retirement, unemployment or disability pension); health (general, physical or mental health and/or health behaviour); SES (educational, occupational and/or income level); and inclusion of stratified or interaction analyses to determine differences across socioeconomic groups. This search strategy resulted in 22 studies. For general, physical or mental health and health behaviour, 13 studies found more positive effects of exit from work on health among employees with a higher SES compared to employees with a lower SES. These effects were mainly found after early/statutory retirement. In conclusion, the effects of exit from work, or more specific the effects of early/statutory retirement on health are different across socioeconomic groups. However, the findings of this review should be interpreted with caution as the studies used heterogeneous health outcomes and on each health outcome a limited number of studies was included. Yet, the positive effects of exit from work on health are mainly present in higher socioeconomic groups. Therefore, public health policies should focus on improving health of employees with a lower SES, in particular after exit from work to decrease health inequalities.
  • Article
    Full-text available
    Background: Social and policy changes in the last several decades have increased women's options for combining paid work with family care. We explored whether specific combinations of work and family care over the lifecourse are associated with variations in women's later life health. Methods: We used sequence analysis to group women in the English Longitudinal Study of Ageing according to their work histories and fertility. Using logistic regression, we tested for group differences in later life disability, depressive symptomology and mortality, while controlling for childhood health and socioeconomic position and a range of adult socio-economic circumstances and health behaviours. Results: Women who transitioned from family care to either part-time work after a short break from the labour force, or to full-time work, reported lower odds of having a disability compared with the reference group of women with children who were mostly employed full-time throughout. Women who shifted from family care to part-time work after a long career break had lower odds of mortality than the reference group. Depressive symptoms were not associated with women's work and family care histories. Conclusion: Women's work histories are predictive of their later life disability and mortality. This relationship may be useful in targeting interventions aimed at improving later life health. Further research is necessary to explore the mechanisms linking certain work histories to poorer later life health and to design interventions for those affected.
  • Article
    This article examines the influence of national employment protection legislation (EPL) on the likelihood of older workers in Europe being forced into retirement. Data are drawn from 4 waves of the Survey of Health, Ageing and Retirement in Europe (SHARE) covering the period from 2004 to 2013. The sample is restricted to those who were aged between 50 and 80 and exited from paid work during the study period (N = 3,446). EPL was measured using the OECD indicators of employment protection concerning regulations for individual dismissals. Exits from work were defined as forced or unforced based on the respondent’s description of the reason for leaving work. Our cross-national study shows considerable variety in the prevalence of forced career exit across 13 European countries. Furthermore, the results show that career exit through retirement is less likely to have been forced as compared to career exit through non-retirement routes. However, the results also show that with every unit increase in the EPL index, the probability of forced career exit through retirement becomes more likely. Apparently in countries with high levels of employment protection, retirement is a more attractive route to lay off older workers than in countries with low EPL. By forcing older adults to leave their jobs through retirement, these employers are shedding workers who would have preferred to continue their working lives.
  • Article
    The aim of this working paper is to inform researchers of the availability of ethnicity data in population based datasets which are available for use in epidemiological and social science research. The paper begins by introducing the concept of ethnicity and problems associated with definition and classification. Secondly, the paper charts the evolution of ethnicity recording in the UK census and how this has been incorporated across the NHS. Thirdly, the paper focuses down on to the relationship between ethnicity and health and describes electronic NHS databases in which routinely collected ethnicity data are available for research purposes. Finally, the paper briefly reviews existing work on ethnicity and health which has been undertaken using population wide data from the ONS Longitudinal Study and NHS databases.
  • Article
    Introduction Studies that have examined interactions between individual and contextual characteristics have revealed variations in the social gradient in health depending on area-level deprivation, reporting increased health inequality in less deprived areas. The present study examines whether similar variations are found between the environment and other individual characteristics, exploring whether the link between area deprivation and self-rated health (SRH) depends on an individual's ethnicity. Methods Data from the 2007 Citizenship Survey were geocoded to the 2001 UK census, and random effects multilevel logistic regression models were conducted to examine: whether the association between area deprivation and poor SRH differs for ethnic minority groups, as compared to white British people; and whether possible differential associations are mediated by neighbourhood characteristics. Results A detrimental association was found between area deprivation and poor SRH across ethnic groups, but effect sizes were found to be larger for white British than for ethnic minority people. Interaction between area deprivation and ethnicity showed the detrimental association between area deprivation and SRH to be of greater magnitude for white British than for ethnic minority people. This differential association was not mediated by neighbourhood characteristics. Conclusion The association between area deprivation and SRH was found to be less strong for ethnic minority than for white British people, but this was not mediated by neighbourhood characteristics. Other hypothesised explanations include a higher degree of deprivation in ethnic minority neighbourhoods not captured by the deprivation measures used, and habituation effects due to ethnic minority people's cumulative exposure to poverty.
  • Article
    Full-text available
    Background Given the current policy emphasis in many Western societies on extending working lives, we investigated the health effects of being in paid work beyond state pension age (SPA). Until now, work has largely focused on the health of those who exit the labour force early. Methods Our data come from waves 2–4 of the English Longitudinal Study of Ageing, including the life history interview at wave 3. Using logistic and linear regression models, we assessed the longitudinal associations between being in paid work beyond SPA and 3 measures of health (depression, a latent measure of somatic health and sleep disturbance) among men aged 65–74 and women aged 60–69. Our analyses controlled for baseline health and socioeconomic characteristics, as well as for work histories and health in adulthood and childhood. Results Approximately a quarter of women and 15% of men were in paid work beyond SPA. Descriptive bivariate analyses suggested that men and women in paid work were more likely to report better health at follow-up. However, once baseline socioeconomic characteristics as well as adulthood and baseline health and labour market histories were accounted for, the health benefits of working beyond SPA were no longer significant. Conclusions Potential health benefits of working beyond SPA need to be considered in the light of the fact that those who report good health and are more socioeconomically advantaged are more likely to be working beyond SPA to begin with.
  • Article
    Full-text available
    This is the first paper to identify the correlates of workplace managers' perceptions of the health and safety risks faced by workers and the degree to which workers have control over those risks. The risks workers face and the control they have over those risks are weakly negatively correlated. Managerial risk ratings are positively associated with both injury and illness rates, but not with absence rates. The control rating is also positively associated with injury and illness rates, but it is negatively correlated with absence rates. Workers are more likely to be exposed to health and safety risks when their workplace is performing poorly and where it has been adversely affected by the recession. Union density is positively associated with risks but is not associated with worker control over risks. Having on-site worker representatives dealing with health and safety is linked to lower risks than direct consultation between management and employees over health and safety. However, there is no evidence that particular types of health and safety arrangement are related to worker control over health and safety risks.