Religiosity, citizenship and attitudes to social policy issues.
Saunders, Peter
Introduction
There has always been a close affinity between religious belief and
social goals. Whether it is the emphasis given to the importance of
compassion for the poor in traditional Western Christian teaching or the
role of Confucian philosophical values in shaping the role of the family
under the different forms of welfare orientalism, these influences have
been profound. The impact of religion has been reinforced in a country
like Australia, where religious organisations have long played a major
role in the provision and delivery of welfare services across the life
cycle, from schooling and health care to aged care. This role has been
increasing under a contracting-out agenda that has extended the
long-established service delivery role of religious organisations into
new areas, including employment and disability services. This trend, in
part, reflects the view that these organisations are run economically
and have an institutional ethos of care and compassion that makes them
better providers of social services in an era characterised by the twin
imperatives of economic efficiency and personal responsibility.
Despite the strong links between religious beliefs and social
policy goals, the relationship between them remains relatively
under-researched. A recent review of the role of religion in the
development of social policy concluded that its role 'has not
always been understood properly or appreciated particularly well'
(van Kersbergen & Manow 2010: 265). Much of the existing work on the
topic has been narrowly focused on the experience of Western countries,
yet the changing relationship between religion and social policy in all
countries makes the topic of growing global significance. Van Kersbergen
and Manow argue that religion can impact on social policy through three
main channels; first, religious values (such as the Protestant work
ethic or the 19th century papal encyclicals) provide a framework that
can have a profound impact on the assumptions and objectives t-hat shape
social policy; second, religious movements can exert pressure by
advocating in support of specific issues and/or groups through the
political process; third, religious organisations can influence (and be
influenced by) policy outcomes in their role as service delivery agents.
The balance between these three roles will influence how religious
organisations are structured, shape their relationships with the state,
and affect how religious individuals engage with social policy issues
and debates.
The traditional alignment of religious values with the underlying
goals of the welfare state has been matched by a closer relationship
between church and state in the ways in which welfare benefits (broadly
defined) are delivered. The positive impacts of this relationship are
assumed to work both ways: the state is freed from responsibility for
the minutiae of service provision as a result of contracting out, while
religious organisations become more economically powerful and socially
and politically influential. There are many facets of these changes that
warrant detailed examination from the perspective of their impact on
social policy, but this paper focuses on the degree to which the
attitudes held by those who describe themselves as religious are
consistent with welfare state goals, programmes and policies. (1) It
examines the extent to which some of the conclusions drawn recently by
Putnam and Campbell (2010) for the United States apply to Australia,
focusing specifically on whether it is the case that when religiosity is
identified and defined on the basis of practice as opposed to belief,
those who are identified as being religious are 'better
citizens' in relation to the kinds of attitudes and behaviours that
suggest an affinity with the principles of the welfare state.
There is, of course, no reason why having a more compassionate
attitude towards those who are less economically fortunate should be
associated with a higher level of support for state interventions
designed to improve their circumstances. It may instead lead to a
conviction that 'charity should begin (and end) at home' and
that state programmes undermine the individualist values on which such
charity relies. However, when it comes to attitudes or beliefs about the
causes of poverty and inequality (as opposed to what actions should be
taken to address them) there is a presumption that those who are
religious should lie at the compassionate end of the spectrum,
expressing values that are more consistent with the underlying social
objectives of fairness, need satisfaction and poverty relief.
Background to the approach
There is a vast literature--much of it comparative--that examines
how public attitudes in general affect support for, and the design of,
different forms of social policy intervention (for example, Taylor-Gooby
1985; van Oorschot et al. 2008; Svallfors 1997, 2010). Many of these
studies highlight the fact that people's attitudes are not only
hard to shift ('coins that do not readily melt' as Schumpeter
(1942: 12) noted), but can also reveal information about how social
problems are perceived and guide the kinds of ameliorative actions that
will attract widespread political support. (2) Such studies also
highlight the difficulty involved in explaining the considerable
attitudinal diversity that exists by a small set of observable
variables.
In relation to attitudes to inequality for example, studies have
shown (for Australia and other countries) that a large proportion of the
population regard the existing level of inequality--as reflected in the
gap between rich and poor as too large. (3) However, the same survey
data often also indicates that this belief does not translate into
support for income redistribution designed to reduce the income gap
(Saunders & Wong 2013). (4) Attitudes matter, but it is also clear
that there is currently an incomplete understanding of the nature of
those attitudes and the underlying values that they represent. In this
context, it is important to get a better understanding of the nature of
the attitudes expressed by different groups in the population in order
to explain the lack of consistency between the attitudes held and the
strength of support for action. (5)
Addressing such issues requires a more thorough examination of the
data and a reduced willingness to accept at face value what the
expressed attitudes appear to represent. A better understanding of
attitudes to such issues as inequality and state social provision can
also assist the policy process because existing attitudes can be an
obstacle to the introduction of new policies--particularly when those
attitudes are mobilised politically. It is also possible to use
attitudinal surveys to 'road test' new policies against public
opinion and modify the policies to better fit the existing patterns of
belief. (6) Attitudes can be changed, but only if they are known and
understood.
Many studies have shown that different dimensions of religious
belief and behaviour are correlated with attitudes to social objectives
(such as equality), social policies (such as welfare provision), and
social activities (such as volunteering). The key findings to emerge
from these studies include that religious belief and participation in
religious activities is associated with conservative attitudes to social
issues such as abortion (Huber 2005), and that there is an inverse
relationship between religious participation and support for welfare
spending after controlling for the other effects of modernisation (Gill
& Lundsgaarde 2004). (7) Chang (2010) used data from the Taiwan
Social Change Survey to examine a range of factors affecting social
policy preferences and concluded that 'Religion plays a significant
role in shaping people's beliefs about fairness, causes and
consequences, the reward for hard work, individual responsibility,
altruism, and the relationship between personal interests versus the
collective well-being of society' (2010: 83). Despite these
seemingly powerful conclusions, there is no agreement from empirical
studies about precisely how religiosity and social attitudes are
related. Differences exist in how to measure the strength (or even the
existence) of religious belief, and studies differ in the policy goals
of interest and the sophistication of the statistical methods used to
identify the underlying relationships.
One important recent study examines data from the United States
Faith Matters survey that was conducted specifically to examine the role
of religion on a range of social attitudes and practices (Putnam &
Campbell 2010). The study provides a detailed examination of the
relationships between religiosity and different aspects of social
behaviour in the United States, focusing on what the authors refer to as
'the three B's of religiosity'--believing (adherence to
religious principles and values), belonging (to a religious community)
and behaving (through attendance at religious services and events). (8)
The authors draw on a broad body of evidence to show that it is behaving
and belonging (which are closely linked) rather than believing that
matter when it comes to understanding the impact of religion. This is an
important finding, because it suggests that those earlier studies that
have focused on the role of religious beliefs may have missed the more
important mechanisms through which the practice of involvement in
religion and engagement with like-minded others influences support for
specific goals or policies.
The key variable used in the Putnam and Campbell study is a
composite religiosity index (RI) which captures the frequency of
attending religious services, the frequency of praying outside of
religious services, the importance of religion in daily life, the
importance of religion in one's sense of who one is, the strength
of religious belief and the strength of belief in God (see Putnam &
Campbell 2010:18-23 and Appendix 1). The RI was derived for each
individual and the sample was split into quintiles, which were then
compared in order to identify the underlying attitudinal gradients.
The resulting evidence reveals a number of important findings,
including that for the most part religiously observant Americans are
more civic and in some respects simply 'nicer' (Putnam &
Campbell 2010: 44). The results also indicate that religiously observant
men and women are more philanthropic--especially towards organisations
that serve the needy. In relation to volunteering, the authors found
that '[the] primary predictor of generosity is the strength of
one's religious commitment, regardless of one's religious
tradition' (2010: 453), and that: 'religious people are more
satisfied with their lives mostly because they build religious social
networks, thus reinforcing a strong sense or religious identity'
(2010: 491). Combining these findings, the study concludes that:
'For happiness as for neighbourliness, praying together seems to be
better than either bowling together or praying alone' (2010: 492).
The remainder of this paper examines whether similar conclusions can be
drawn (at least tentatively, given the available data) in the Australian
context.
Data and methods
The data used in the analysis is drawn from two surveys, the
Community Understanding of Poverty and Social Exclusion (CUPSE) survey,
conducted in 2006, and the Poverty and Exclusion in Modern Australia
(PEMA) survey, conducted in 2010. Both surveys were distributed by mail
to a random sample of 6,000 adults drawn from the Australian electoral
roll and achieved response rates of around 46 per cent. (9) The PEMA
survey was accompanied by a follow-up survey of 1,000 of the CUPSE
respondents as a way of providing an insight into the dynamics of social
disadvantage, and the resulting two-period panel is also used in the
analysis that follows.
Comparisons with official Australian Bureau of Statistics data
reveal that the CUPSE and PEMA samples are broadly representative of the
general population, although the following groups are under-represented:
males; those who have never been married; those who live alone;
Indigenous Australians; those with lower levels of education; those in
private rental accommodation; and those with incomes over $1,000 a week
(in 2006) and over $2,000 a week (in 2010). The main area where the
samples in both years differ from the population is in relation to age
structure: as with many social surveys of this type, older people (aged
50 and over) are disproportionately represented among the respondents
compared to younger people (aged under 30). A weighting system has been
derived to adjust for any resulting bias, although the results presented
below are based on the raw (un-weighted) data because the focus here is
on comparing groups within the total sample and weighting by age should
not affect these comparisons. A third data source used later is the 2009
Australian Survey of Social Attitudes (AuSSA), which attracted 3,243
respondents equivalent to a response rate of around 35 per cent. (10)
The PEMA, CUPSE and PEMA follow-up surveys included a set of core
questions which included the following: Would you describe yourself as a
religious person? (11) Participants were given the three response
options shown below and these were used to segment the sample into those
who were religious in practice (RP); those who were religious in belief
only (or nominally religious) (RB); and those and those who said they
were not religious (NR).
Yes--but I do not regularly attend church or other places of
religion (RB)
Yes--I regularly attend church or other places of religion (RP)
No--I do not follow any religion (NR)
The question asked in AuSSA 2009 asked about the frequency of
attendance at religious services (How often do you attend religious
services?). This presents problems in deciding how frequently one must
attend services before being regarded as a religious in practice,
although the following classification has been used: (12)
Attend more than once a month (RP)
Attend up to once a month (RB)
Never attend (NR)
The use of the reported incidence or frequency of service
attendance to identify those who are religious in practice can be
criticised as being somewhat crude; however, it is all that is possible
given the available data. It is relevant to note here that Putnam and
Campbell conclude from their study (2010: 19) that: 'Church
attendance is ... an excellent measure of religious commitment in most
cases'--an assessment which suggests that the religiosity measures
used here capture a central aspect of the concept.
The initial approach used to examine the impact of religion
involves comparing the mean values of a range of wellbeing, behavioural
and attitudinal variables that are relevant to social status and social
policy concerns across the RP, RB and NR sub-samples. These variables
include several indicators of subjective well-being (SWB), of the
incidence of community and social participation (CSP) and of attitudinal
support for a range of social policy issues (ASSP). The mean values of
these variables are compared across the three religious categories (RP,
RB & NR) and the following null hypotheses are tested: (13)
[H.sub.1]: SWB (or CSP or ASSP)/RP > SWB (or CSP or ASSP)/RB
[H.sub.2]: SWB (or CSP or ASSP)/RB = SWB (or CSP or ASSP)/NR
The vexed issue of causality is examined using the linked panel of
respondents to the CUPSE and PEMA follow-up surveys. Finally, a
multivariate (regression) approach is used to assess the extent to which
the relationships between religiosity and attitudes to social policy
issues are robust after controlling for other factors that exert an
influence on social behaviour and attitudes.
Results
Because much of the initial analysis is based on the 2010 PEMA
survey, Table 1 provides a breakdown of the demographic composition of
the three sub-samples of the PEMA sample that are classified according
to the reported religiosity of the respondent. Comparisons between the
compositions of the religious in practice sub-sample and the sample as a
whole indicate that the following groups are over-represented in the
actively religious sample: females, those aged 65 and over, those born
in an overseas non English-speaking country, those who are married,
those with a tertiary qualification, those who are not employed, those
who have a low income, and those who (in 2006) described themselves as a
Liberal/National party voter. In contrast, the religious in practice
group contains an under-representation of people who are aged under 30,
Australian-born, never married, employed and ALP voters.
The upper panel of Table 2 suggests that those who practise
religion as defined here are more satisfied with their standard of
living, are somewhat happier, lead more active social lives, and report
a higher level of participation in community activities than the other
two groups. Against this, they also say that they have less control over
their own lives and the things that happen to them. In terms of the null
hypotheses specified earlier, the values of two of the five wellbeing
variables shown in the upper panel of Table 2 are significantly higher
for the religious in practice group than for the religious in belief
group, while all but one of the differences for the last two groups are
not significantly different. This initial evidence thus provides some
support for the two hypotheses (particularly for the latter) although
there are a number of cases that do not conform to the expected pattern.
The lower panel of Table 2 compares rates of participation in
specific community activities between the three religious sub-groups.
(14) The results indicate that those who practise religion are more
likely to participate in education-based, voluntary and cultural
activities, are (not surprisingly) much more likely to take part in
church activities, and have the highest participation rates for all of
the other activities. In contrast, only a small proportion (6.6 per
cent) of those who say they are religious but do not attend a place of
religion regularly participate in church groups or activities. The one
area where there are very small between-group differences is in relation
to participation in neighbourhood activities. With this exception, all
of the specific participation rates are significantly higher in the RP
than in the RB group, and while there are also some significant
differences between the RB and NR groups (contrary to the second
hypothesis specified above), overall the results in Table 2 are
consistent with the findings reported by Putman and Campbell for the
United States in showing that those who actively practise their
religious beliefs seem more content with their lives and more engaged as
citizens.
One possible criticism of the results presented so far is that the
correlation between religiosity and wellbeing or community participation
does not imply causality. Are religious people more content and more
actively engaged socially because of their religious practices, or is it
that more content social beings are drawn to religion? This issue was
addressed by Putnam and Campbell in two ways. First, they examined
whether the simple relationships between religiosity and indicators of
good neighbourliness still existed after controlling for other factors
such as gender, education, race, location, home ownership, length of
residence, marital and parental status, and ideology. They found that
the correlation was robust. Second, they examined longitudinal data
extracted from the two waves of the Faith Matters survey and found that
changes in religiosity (though infrequent and small) gave rise to
changes in philanthropic generosity and community involvement that were
consistent with causation running from religiosity to these other civic
variables. (15)
The role of confounding factors is examined later using a
regression framework, but before that the causality issue is examined
using data from the linked panel data. Panel members have been separated
into three groups: those who were actively religious in both years using
the definition applied above; those who were not actively religious in
either year; and those who were actively religious in 2006 but not in
2010. (16) If it is the case that being actively involved in religion
(i.e. by frequently attending religious services) is a causal factor
explaining the other variables, then one should find that those who
stopped being religious between 2006 and 2010 should look much like
others who were religious in 2006, but by 2010 should look more like
those who were not actively religious in 2010.
The results in the upper panel of Table 3 provide little evidence
to support the causality hypothesis: in fact, contrary to expectations,
those who stopped being actively religious between 2006 and 2010 report
an increased level of satisfaction with their standard of living, no
change in their happiness, and a sharp rise in control over their lives.
They also experienced a substantial fall in the level of their social
life activity--although so too did the other two groups. However, none
of the changes for the group that stopped being actively religious are
statistically significant. The changes in community participation in the
lower panel of Table 3 are (at least in some regards) more consistent
with the causality hypothesis, although even here the results are not
compelling. While those who ceased being actively religious experienced
sharp declines in their participation in education-based, voluntary,
church-related and cultural activities, their participation in
neighbourhood activities increased sharply. Again, however, the reported
changes in participation are not statistically significant, aside from
the reduced participation in church activities.
Table 3 provides some support for the hypothesis that those who
stopped being actively religious between 2006 and 2010 shifted from
being like those who were actively religious in 2006 to being like those
who were not actively religious in 2010. However, the hypothesis that
the mean values for all three groups are the same cannot be rejected,
reinforcing the view that these results cannot be claimed to support the
hypothesis under review. The only exception to this pattern occurs for
participation in church activities, where the differences between the
first and third column values and between the fourth and sixth column
values are both statistically significant (although these test outcomes
are not shown in Table 3). Thus, participation in church activities in
2006 among those who stopped being actively religious after 2006 was
significantly below that of those who remained active after 2006, but in
2010 was significantly above that for the group that was not religious
in that year. This suggests that the middle group of
'shifters' differs from the other two groups in terms of their
church attendance, rather than revealing anything about causality as
such.
Table 4 draws on the cross-sectional CUPSE and PEMA data to examine
differences in attitudes to a number of aspects of inequality and
redistribution among the three religious sub-groups identified in Table
1. The survey questions relate to different aspects of inequality,
focusing on outcomes (in the income space) and fairness (in relation to
taxes, government welfare spending and the relationship between taxes
paid and benefits received). The response categories provided were
'strongly agree', 'agree', 'neither agree nor
disagree', 'disagree', and 'strongly disagree',
although only the percentages in the first two categories are shown in
Table 4. The results indicate that, relative to the other two groups,
the religious in practice group are less convinced that poverty is a big
problem, are more likely to agree that the income gap is too high and
should be reduced, regard it as fair that taxes should be used to
support those in need, but also believe that those on higher incomes
should be free to purchase better health care and education for their
children.
Only one of the attitudinal differences between the religious in
practice and religious in belief groups is statistically significant,
while two of the differences between the last two groups (which were
hypothesised to be the same) are significantly different. There is also
no clear attitudinal gradient across the three religious sub-groups,
with the highest values appearing for the middle group in several
instances. The evidence in Table 4 does not therefore suggest any clear
patterns between religiosity and attitudes to inequality, fairness and
redistribution.
Table 5 complements the results in Table 4 by comparing attitudes
to a broader range of social policy issues across religious sub-groups
using data from the 2009 Australian Survey of Social Attitudes where
religiosity is defined, as noted earlier, by the frequency of attendance
at religious services. These results indicate that, compared to the
other two groups, those who attend a religious service at least once a
month (identified here as the religious in practice group) are more
supportive of government efforts to reduce the income gap, and are more
concerned about the negative impact of cutting welfare benefits for the
unemployed. They also express more confidence in both charities and the
welfare system (which suggests an affinity with a 'mixed economy of
welfare' approach), and for providing support that allows sole
parents to stay at home to raise their children. All three groups
express similar levels of support for sole parents as for all families
with children, with such support well below that for providing a decent
minimum for the unemployed. Again, however, most of the group
differences in Table 5 are not statistically significant, a finding
which provides little evidence to support the null hypotheses under
examination.
The analysis conducted so far has focused on bivariate
relationships between religiosity and a number of attitudinal variables
that capture support for the goals of social policy, or with the
mechanisms used to achieve those goals. The focus now shifts to
examining whether the findings are robust when a multivariate approach
is taken that controls for the effects of other variables that are
likely to influence the relationship between religiosity and social
attitudes. A number of alternative dependent variables that capture
different aspects of attitudes to social goals and policies have been
specified for this purpose. The first two relate to attitudes to overall
income inequality and fairness and are based on responses to the
following survey questions: (17)
1. The gap between rich and poor is too great and incomes should be
reduced
2. Incomes are too high at the top and should be reduced
3. Incomes are too low at the bottom and should be increased
4. It is fair that taxes paid by the majority help to support those
in need
5. It is not fair that some people pay a lot of tax and hardly use
the services their taxes pay for
6. It is not fair that people benefit from services that they
haven't helped to pay for
7. It is fair that people with higher incomes can buy better health
care than those with lower incomes
8. It is fair that people with higher incomes can buy better
education for their children than those with lower incomes
In each case, the response categories provided (strongly agree,
agree, neither agree nor disagree, disagree and strongly disagree) were
assigned a score of 1 through to 5, where the higher score indicated
greater support for either income redistribution or for the fairness
criterion implied in the question. These scores were then summed for
questions 1-3 and for questions 4-8 and these aggregate scores represent
the first two dependent variables. (18) The third (zero-one) dependent
variable reflects one of the assumed characteristics of good civic
citizens highlighted in the earlier discussion, and refers to whether or
not the respondent acted as a volunteer in health and community services
over the last 12 months. The final variable is designed to capture
compassion in attitudes to the plight of the poor, and is based on
responses to the following 2 questions:
1. People are poor because they have been unlucky in life
2. People are poor because they have not had the opportunities that
other people have
In this case, a scale was constructed by assigning a strongly agree
response to each question a score of 5, down to a score of 1 for
strongly disagree responses; these scores were then summed across the
two questions (so that a higher score indicates greater agreement with
the propositions in each question and thus a greater willingness to see
the poor as the victims of external circumstances).
The same set of explanatory variables was included in each
regression model and these are defined in the Appendix--the results
themselves are presented in Table 6. When reviewing these results, it is
important to note that the dependent variables have been defined so that
a positive coefficient implies that the relevant explanatory variable is
associated with a more egalitarian attitude, greater civic engagement,
or a more compassionate attitude towards the causes of poverty and the
poor. The first point to note about the results is that the explanatory
power of all four models is very low and relatively few of the
explanatory variables are statistically significant. This reflects the
difficulty in explaining the variability in attitudes to complex social
issues using a small set of observable socio-demographic variables.
With these caveats in mind, the results in Table 6 indicate that
younger people, those born in a non English-speaking country and those
on higher incomes tend to be less egalitarian (in the income space). On
the question of fairness more generally (specifically in relation to
access to basic services), wage and salary earners are less egalitarian,
as are those born in a non English-speaking overseas country, while
females and graduates are more egalitarian. Broadly similar patterns
relate to those engaged in volunteering. The final column of estimates
show that graduates, those who regard themselves as poor, and those on
lower incomes, hold more compassionate views towards the poor.
When it comes to the role of the religiosity variables, the
parameter estimates in Table 6 indicate that those who are actively
religious hold more egalitarian views when it comes to access to basic
services and are more likely to have volunteered over the last year. In
contrast, those who are not religious hold even more egalitarian views
on service fairness and are more compassionate when it comes to views
about how responsible the poor are for their plight. The patterns of
size and significance of the religiosity variables do not conform to the
hypotheses set out earlier, but the important point is that religiosity
is shown to have an impact on social attitudes even after the influence
of other factors is taken into account. In fact, religiosity shows up as
statistically significant in at least as many cases as any of the other
broad classes of variables included in the modelling. The results are
thus not definitive, but they do suggest that religiosity exerts an
influence on social attitudes and has the potential to influence support
for different social goals and policies.
Conclusions
Religious beliefs reflect and reinforce people's wider value
system and one might expect that those values influence how people feel
about the goals of social policy and the methods used to achieve them.
However, there is no simple relationship between personal values and
attitudes towards state action in the social policy field--in part
because of the lack of agreement about what those goals are, and because
of the uncertainty and confusion that surrounds debates about the impact
and effectiveness of different policy choices. This paper has presented
a range of data on Australian attitudes to social policy issues and
explored whether they differ systematically with religiosity, measured
in terms of the frequency of attending religious services.
The focus here has been on examining the impact of religious
practice rather than of religious belief itself, because other studies
have shown that the former variable is a better predictor of social
attitudes than the latter. It is, however, important to acknowledge that
the variable used to capture the strength of religious practice is
rather crude and can only be expected to map the actual differences that
exist in general terms. Further research is needed (along the lines of
that conducted for the United Sates by Putnam and Campbell (2010)) that
allows a more sophisticated specification of the religiosity variable
and importantly, is also able to breakdown the responses by type of
religion.
Despite these limitations, the results suggest that some of the
attitudes expressed about the role, design and delivery of social policy
programs vary systematically across religious groups defined on the
basis of both religious service attendance and reported non-belief.
Future research in this field will clearly benefit from better data on
which to identify a person's religiosity and from samples that are
large enough to allow its overall (and diverse) effects to be better
isolated and quantified. Although there is some evidence that those who
are actively religious have more egalitarian, pro-welfare attitudes,
there is no consistent pattern linking pro-welfare attitudes (such as
aversion to inequality, support for redistribution, compassion for the
poor) or more civically-minded behaviour (such as in relation to the
incidence of volunteering) systematically to religious activity or
non-belief. Those involved in religious activity appear to hold a
diversity of views about the importance of the ends and the
effectiveness of the means of social policy.
APPENDIX
Specification of the independent variables used in the regression
analysis (reference categories shown in italics)
Variable name Specification
Gender 1 if female, 0 if male
Age Under 30 years; 30-64 years; 65 and over
Main source of income Wages or salaries (W or S); interest,
last week dividends, superannuation, etc.; social
security payment (SSP); other
Subjectively poor Would you describe yourself as poor? 1 if
response is yes, 0 otherwise
Country of birth Australia; another English-speaking country;
another non English-speaking country
Indigenous status 1 if yes, 0 otherwise
Disability status 1 if yes, 0 otherwise
Housing status Home owner/purchaser; public or private
renter; other
Educational attainment Primary/some secondary; completed secondary
school; trade certificate or similar;
degree/postgraduate degree
Gross weekly income Under $600; $600-$999; $1,000-$1,999; $2,000
and over
Labour force status Employed: unemployed; other
Religiosity Religious in practice, religious in belief
only/nominally religious; not religious
Acknowledgements
The author acknowledges the helpful comments provided on the
original version of this paper by John Nevile and two anonymous
referees. Statistical support was provided by Melissa Wong. Financial
assistance from the Australian Research Council under Linkage project
grants LP0560797 and LP100100562 is also acknowledged.
References
Chang, W.C. (2010) 'Religion and preferences for
redistributive policies in an East Asian country', Poverty &
Public Policy, 2 (4), 81-109.
Chesters, J. & Western, J. (2010) 'Evidence and
perceptions of inequality in Australia', CEPR Discussion Paper No.
635, Centre for Economic Policy Research, Research School of Economics,
Australian National University.
Gill, A. & Lundsgaarde, E. (2004) 'State welfare spending
and religiosity: a cross-national analysis', Rationality and
Society, 16 (4), 339-436.
Huber, J.D. (2005) 'Religious belief, religious participation,
and social policy attitudes across countries', mimeo, Department of
Political Science, Columbia University.
Kelley, J., Evans, M. & Sikora, J. (2004) 'Is there too
much inequality in Australia?', Australian Social Monitor, 7 (3/4),
92-100.
Meagher, G. & Wilson, S. (2008) 'Richer, but more unequal:
perceptions of inequality in Australia 1987-2005', The Journal of
Australian Political Economy, 61 (June), 220-243.
Oorschot, W.V., Opielka, M. & Pfau-Effinger, B. (eds.) (2008)
Culture and the Welfare State. Values and Social Policy in Comparative
Perspective, Cheltenham, Edward Elgar.
Osberg, L. & Smeeding, T.M. (2006) '"Fair"
inequality? Attitudes toward pay differentials: the United States in
comparative perspective', American Sociological Review, 71 (30),
450-73.
Pusey, M. & Turnbull, N. (2005) 'Have Australians embraced
economic reform?'. In S. Wilson, G. Meagher, R. Gibson, D. Denemark
& M. Western (eds), Australian Social Attitudes. The First Report,
Sydney, UNSW Press, 161-81.
Putnam, R.D. & Campbell, D.E. (2010) American Grace. How
Religion Divides and Unites Us, New York, Simon & Schuster.
Saunders, P., Eardley T. & Evans, C. (2000) 'Community
attitudes towards unemployment, activity testing and mutual
obligation', Australian Bulletin of Labour, 26 (3), 211-35.
Saunders, P., Naidoo, Y. & Griffiths, M. (2008) 'Towards
new indicators of disadvantage: deprivation and social exclusion in
Australia', Australian Journal of Social Issues, 43 (2), 175-94.
Saunders, P. & Wong, M. (2012) Promoting Inclusion and
Combating Deprivation: Recent Changes in Social Disadvantage in
Australia, Social Policy Research Centre, University of New South Wales.
Saunders, P. & Wong, M. (2013) 'Examining Australian
attitudes to inequality and redistribution', Journal of Australian
Political Economy, forthcoming.
Schumpeter, J. (1942) Capitalism, Socialism and Democracy, New
York, Harper.
Sikora (2003), 'Tastes for privatisation and tastes for
subsidisation: have Australians been getting cold feet?, Australian
Social Monitor, 6(2), 34-40.
Stilwell, E & Jordan, K. (2007) Who Gets What? Analysing
Economic Inequality in Australia, Cambridge, Cambridge University Press.
Svallfors, S. (1997) 'Worlds of welfare and attitudes to
redistribution: A comparison of eight Western nations, European
Sociological Review, 13(3), 331-52.
Svallfors, S. (2010) 'Public attitudes'. In E G. Castles,
S. Leibfried, J. Lewis, H. Obinger & C. Pierson (eds), The Oxford
Handbook of the Welfare State, Oxford, Oxford University Press, 241-51.
Taylor-Gooby, P. (1985) Public Opinion, Ideology, and State
Welfare, London, Routledge & Kegan Paul.
van Kersbergen, K. & Manow, P. (2010) 'Religion'. In
EG. Castles, S. Leibfried, J. Lewis, H. Obinger & C. Pierson (eds),
The Oxford Handbook of the Welfare State, Oxford, Oxford University
Press, 265-77.
Endnotes
(1.) It is acknowledged that this is only one aspect of what is a
much broader set of issues relating to the actual interaction between
religion and the welfare state that includes, for example, working
conditions, the notion of a 'just wage', the ownership of
private property and relations between the church and the state more
generally.
(2.) The source of the quote from Schumpeter is Svallfors (2010:
241).
(3.) Among the many recent contributions to the Australian
literature are Sikora (2003); Kelley and colleagues (2004); Pusey and
Turnbull (2005); Stilwell and Jordan (2007); Meagher and Wilson (2008);
Chesters and Western (2010); and Saunders and Wong (2013, forthcoming).
(4.) Data from the 2005 Australian Survey of Social Attitudes
(AuSSA) indicates that although over 86 per cent agreed that the income
gap was too high, support for income redistribution from better-off to
less well-off was less than half of this, at below 42 per cent (Saunders
& Wong 2013: Table 1). In other words, a majority of the 86 per cent
who thought that the income gap is too large were either opposed to
redistribution or were ambiguous about it, with almost one-third of them
opposed to redistribution from high to low income. Part of the
contradiction apparent in these findings may reflect the use of
undefined terms like 'rich' and 'poor' to gauge
people's attitudes, as Osberg and Smeeding (2006) have observed, or
the (unknown) impact of the context in which a particular question is
embedded.
(5.) This apparent lack of consistency may also reflect a lack of
faith in the effectiveness of the redistributive policies proposed to
deal with the problem, or a concern that they may give rise to
unacceptable side effects (e.g. on incentives).
(6.) An example of this occurred in the late 1990s when surveys of
public attitudes to the introduction of mutual obligation requirements
(MOR) on social security recipients were used to better tailor MOR to
those groups where support for them was strongest (see Saunders et al.
2000).
(7.) Huber argues that the relationship between religious belief
and religious participation (e.g. in services) varies cross-nationally,
and is strongest in those countries that are most economically and
politically developed.
(8.) The need to distinguish between religious belief and church
attendance has been emphasised by Huber (2005: Figures 4 and 5), who
notes that a 'disconnect' exists between religious belief and
church attendance in a cross-country sample derived from the World
Values Survey, which reveals that many who attend church weekly do not
adhere to standard religious beliefs (for example, about the existence
of heaven, or being very spiritual), while many believers do not attend
church regularly. Gill and Lundsgaarde (2004: Tables 2, 4 and 5) find
that there is a statistically significant cross-national relationship
between welfare spending and church attendance, the percentage of the
population who describe themselves as non-religious, and the percentage
who express comfort in religion.
(9.) Further details of the CUPSE and PEMA surveys can be found in
Saunders and colleagues (2008) and Saunders and Wong (2012),
respectively.
(10.) Further details of AuSSA (including the 2009 questionnaire)
can be found at www.aussa.anu.edu.au. The AuSSA response rate varied
slightly between the two versions of the questionnaire that were applied
in 2009 and the figure of 35 per cent is the average.
(11.) Information on the form of religion practised by respondents
was not requested, making it impossible to examine the impact of type of
religion on the issues examined.
(12.) Despite the differences in formulation of the religiosity
variable, the sample sizes shown in Tables 1 and 5 below indicate that
the boundary used to distinguish between the first two groups in the
AuSSA sample produces a breakdown of the sample that is very close to
that produced by the three-way classification derived from the PEMA
survey.
(13.) Note that a one-tailed test is appropriate when testing the
first hypothesis.
(14.) The survey asked about more activities than those shown in
Table 2 and the 'no participation' variable refers to all of
the specified activities, not just those identified in Table 2.
(15.) Putnam and Campbell acknowledged that these latter tests do
not 'prove' that religious change 'caused' civic
change, although they argue that the panel results 'make it less
plausible to suppose that some enduring personal trait, such as generic
niceness or generic activism, explains the correlation' (2010:
462).
(16.) A fourth group consisting of those who were not actively
religious in 2006 but were in 2010 contains only seven individuals and
is not large enough to conduct statistical analysis.
(17.) Questions 1-3 were included in both the CUPSE and PEMA
surveys, while questions 4-8 were included in CUPSE only. The following
results are based on the PEMA survey for questions 1-3 and on CUPSE for
questions 4-8.
(18.) Reliability tests indicate that the resulting two scales are
robust, with Cronbach alpha values of 0.726 and 0.678, respectively.
Table 1: Demographic composition of the three religiosity sub-samples
(unweighted percentages)
Characteristic Religious Religious
in practice in belief
(16.2%) (42.5%)
Gender
Female 62.5 60.1
Age
Under 30 9.6 11.7
Aged 65+ 37.1 23.6
Country of birth
Australian-born 64.9 75.2
Born in a NES country (a) 20.9 11.5
Marital status
Never married 11.2 14.4
Married 74.1 67.8
Educational attainment
Did not complete high school 30.6 33.9
Bachelor degree or higher 36.6 24.6
Labour force status
Employed 43.5 54.1
Income
Family income (equiv.) < $600 66.4 61.2
Political affiliation (b)
ALP voter 23.5 33.8
Liberal/National voter 47.7 37.5
Characteristic Not Share
religious of total
(41.3%) sample
Gender
Female 50.3 56.4
Age
Under 30 13.7 12.2
Aged 65+ 17.0 23.1
Country of birth
Australian-born 78.9 75.1
Born in a NES country (a) 6.7 11.1
Marital status
Never married 17.1 15.0
Married 67.4 68.7
Educational attainment
Did not complete high school 28.3 31.1
Bachelor degree or higher 30.5 29.0
Labour force status
Employed 59.5 54.6
Income
Family income (equiv.) < $600 54.0 59.0
Political affiliation (b)
ALP voter 36.5 32.5
Liberal/National voter 25.5 35.8
Notes: (a) NES = non English-speaking country; (b) The information
on voting intention was not available in 2010 and refers to 2006.
Source: PEMA survey (supplemented by the CUPSE survey for
political affiliation).
Table 2: Subjective wellbeing, autonomy and participation by
religiosity (unweighted percentages)
Variable Definition
Satisfaction with % 'very satisfied'
Standard of Living or 'satisfied'
Happiness %'very happy'
Autonomy % scoring of 8-10
on a 10-point scale
Social life %with 'fairly' or 'very'
active' social life
Overall level of % scoring 9-10 on
Community Participation a 10-point scale
Participation in (a)
Education/school % that participated
activities
Volunteer in health or % that participated
community services
Church groups/activities % that participated
Cultural activities % that participated
Neighbourhood activities % that participated
No Participation % that did not participate
in any of 8 identified
activities
Religious Religious Not
Variable in practice in belief religious
Satisfaction with 71.2 * 65.9 69.5 *
Standard of Living
Happiness 89.2 88.2 88.2
Autonomy 39.7 45.2 48.0
Social life 61.1 55.5 51.4
Overall level of 17.6 * 13.8 13.7
Community Participation
Participation in (a)
Education/school 28.4 22.3 22.8
activities
Volunteer in health or 26.0 17.8 15.4
community services
Church groups/activities 59.5 6.6 1.7
Cultural activities 25.3 17.4 21.2
Neighbourhood activities 18.8 17.2 17.4
No Participation 21.2 *** 33.7 31.5
Notes: (a) Data refer to the percentage that participated in each
activity over the last 12 months; (b) The asterisks (*/**/***)
indicate that the differences between those shown and those in
the middle column are statistically significant
([rho] - 0.10/0.05/0.01).
Source: PEMA survey.
Table 3: Changes in religiosity, wellbeing and community/social
participation, 2006 to 2010 (linked panel, unweighted percentages)
Activity Actively religious in
both years (n = 94)
2006 2010
Satisfaction with 73.4 73.4
standard of living
Happiness 94.6 93.6
Autonomy 45.7 43.6
Social life 70.1 67.4
Participation rate
(previous 12 months):
Education/school activities 25.0 22.6
Volunteer in health or 23.9 28.0
community services
Church groups/activities 68.2 65.6
Cultural activities 25.0 23.7
Neighbourhood activities 18.2 15.1
No participation 12.5 15.1
Stopped being actively
religious (n = 33)
2006 2010
Satisfaction with 75.8 78.1
standard of living
Happiness 90.9 90.6
Autonomy 43.8 53.1
Social life 61.3 54.6
Participation rate
(previous 12 months):
Education/school activities 32.3 24.2
Volunteer in health or 16.1 6.1
community services
Church groups/activities 32.3 * 12.1
Cultural activities 25.8 15.2
Neighbourhood activities 9.7 24.2
No participation 19.4 27.3
Not actively religious
in both years (n = 372)
2006 2010
Satisfaction with 70.8 73.8
standard of living
Happiness 92.4 89.9
Autonomy 46.6 49.9
Social life 61.5 ** 53.4
Participation rate
(previous 12 months):
Education/school activities 28.7 ** 21.8
Volunteer in health or 20.0 19.1
community services
Church groups/activities 2.5 3.3
Cultural activities 20.0 18.5
Neighbourhood activities 23.1 21.0
No participation 21.7 26.4
Note: (a) The activity variables are defined in Table 2; (b).
The asterisks (*/**/***) indicate that the differences between
the two years are statistically significant ([rho] = 0.10/0.05/0.01).
Sources: CUPSE and PEMA follow up surveys.
Table 4: Religiosity and attitudes to inequality and
redistribution (unweighted percentages that strongly
agree/agree with each proposition)
Religious Religious Not
in practice in belief religious
Poverty in Australia today is 26.5 32.5 31.3
a big problem
The rich are getting richer 77.0 74.8 71.2
and the poor are getting
poorer
Large differences in income 15.7 16.3 14.9
are necessary to maintain
Australia's economic
prosperity
The gap between rich and poor 76.0 69.3 70.0
is too great and should be
reduced
Incomes at the top are too 74.3 70.7 69.6
high and should be reduced
Incomes at the bottom are too 81.8 83.8 82.6
low and should be increased
It is fair that taxes paid by 79.8 *** 72.0 75.9 *
the majority help to support
those in need (b)
It is not fair that some 41.0 43.4 35.2 ***
people pay a lot of tax and
hardly use the services that
taxes pay for (b)
It is not fair that people 35.6 39.6 33.4 ***
benefit from services that
they haven't helped to pay
for (b)
it is fair that people with 40.5 38.6 34.9 *
higher incomes can buy better
health care than those with
lower incomes (b)
It is fair that people with 40.2 37.6 34.5
higher incomes can buy better
education for their children
than those with lower
incomes (b)
Notes: (a) The asterisks (*/**/***) indicate that the differences
between those shown and those in the middle column are statistically
significant ([rho] = 0.10/0.05/0.01); (b) These estimates refer to 2006.
Sources: PEMA and CUPSE surveys.
Table 5: Religiosity and attitudes to selected social policy
interventions (unweighted percentages)
Social policy objective Frequency of attending religious services:
Up to once
More than a month Never
once a month (n=1,318; (n=1,295;
(n=485; 15.7%) 42.5%) 41.8%)
Complete/a great deal of 44.3 39.3 30.7 ***
confidence in charities
Complete/a great deal of 30.6 25.4 18.5 ***
confidence in Australia's
social welfare system
Strongly agree/agree that 23.2 20.9 17.9 *
government privatisation
has more benefits than
costs
The gap between high 74.5 74.2 74.3
incomes and low incomes
is much too large/too
large
Strongly agree/agree that 57.8 ** 49.3 51.3
it is the government's
responsibility to reduce
income differences
The government should 56.8 50.5 51.9
provide a decent standard
of living for the
unemployed
The government should 9.0 10.1 9.6
spend less on benefits
for the poor
Cutting welfare benefits 76.8 ** 71.7 74.1
would damage too many
people's lives
Families deserve payments 44.8 45.8 38.5 ***
to help with the costs
of raising children
Single parents deserve 44.1 40.1 38.9
government payments so
that they can be at home
to raise their children
Note: (b) The asterisks indicate that the differences between those
shown and those in the middle column are statistically significant
([rho] = 0.10/0.05/0.01). Source: Author's calculations based on
Au SSA 2009.
Table 6: Regression results
Dependent variable:
Income gaps Service fairness
Independent variable (PEMA) (CUPSE)
Intercept 11.19 *** 16.05 ***
Gender: female 0.15 0.38 *
Age: <30 1.04 *** 0.36
Age: 65+ 0.16 -0.52
Income: W or S 0.61 * 0.84 *
Income: SSP 0.13 -0.70
Income: other -0.12 -1.77 **
Subjectively poor 0.95 *** 0.22
COB: English-speaking -0.18 -0.35
COB: Non English-speaking 053 ** -0.82 *
Indigenous -0.10 -2.09 *
Disability 0.38 * 0-13
Housing: renter -0.08 0.30
Housing: other 0.27 -0.28
Education: primary 0.09 -0.35
Education: trade certificate 0.22 0.43
Education: degree 0.16 1.74 ***
Income: < $600 0.61 ** 0.13
Income: $60041,000 0.44 ** 0.11
Income: $2,000 + -0.78 *** -0.05
LFS: unemployed -0.07 0.17
LFS: other 0.24 0.03
Religious in practice (RP) 0.24 0.47 *
Not religious (NR) 0.15 0.68 ***
Sample size 2,066 2,090
R-squared 0.083 0.055
Dependent variable:
Causes
Volunteer of poverty
Independent variable (CUPSE) (CUPSE)
Intercept -1.10 *** 5.62 ***
Gender: female 0.34 ** 0.13
Age: <30 0.01 0.28 *
Age: 65+ 0.13 0.15
Income: W or S 0.92 *** 0.03
Income: SSP -0.66 ** -0.15
Income: other -0.76 * -0.99 ***
Subjectively poor 0.23 0.52 ***
COB: English-speaking -0.03 -0.16
COB: Non English-speaking -0.60 * -0.23
Indigenous 0.52 0.17
Disability 0.32 0.26 *
Housing: renter 0.19 0.06
Housing: other 0.30 0.17
Education: primary -0.43 * 0.02
Education: trade certificate -0.03 0.02
Education: degree 0.41 * 0.50 ***
Income: < $600 0.27 0.44 **
Income: $60041,000 0.21 0.32 **
Income: $2,000 + 0.48 * 0.01
LFS: unemployed 0.40 0.26
LFS: other 0.25 0.12
Religious in practice (RP) 0.42 * 0.12
Not religious (NR) -0.13 0.22 *
Sample size 2,052 2,062
R-squared 0.045 0.045
Notes: (a) The asterisks indicate that the differences between those
shown and those in the middle column are statistically significant
([rho] - 0.10/0.05/0.01). Sources: COPSE and PEMA surveys.
