Subjective financial well-being, income and health inequalities in mid and later life in Britain

By Sara Arber1*, Kirsty Fenn2 and Robert Meadows1

Affiliations:
1
Department of Sociology, University of Surrey, Guildford, Surrey, UK
2
No institutional affiliation.

* Corresponding author:
Professor Sara Arber
Co-Director, Centre for Research on Ageing and Gender (CRAG),
Department of Sociology,
University of Surrey,
Guildford GU2 7XH, Surrey, UK
Tel: 00-44-(0)1483-686973; Fax: 00-44-(0)-1483-689551
Email: [email protected]

Accepted by Social Science and Medicine, 16th October 2013

Research Highlights

• Both low income and financial problems are associated with poor health in mid-life. In old
age, subjective financial wellbeing rather than income is linked to health.

• The poor health of divorced people is mediated by subjective financial wellbeing.

• Both economic strain and perceived material deprivation may adversely affect health.
Subjective financial wellbeing, income and health inequalities in mid and later life in Britain

Abstract
The relationship between health and income is well established, but the link between subjective
financial wellbeing and self-assessed health has been relatively ignored. This study investigates
the relationship between health, subjective financial wellbeing and income in mid-life and later life
in Britain. Analysis of the General Household Survey for 2006 examined these relationships at
ages 45-64 (n=4639) and 65 and over (n=3104). Logistic regression analysis was used to adjust for
income and other socioeconomic factors linked to self-assessed health. Both income and
subjective financial wellbeing are independently associated with health in mid-life; those with
lower incomes and greater subjective financial difficulties had higher risk of reporting ‘less than
good’ health. In contrast in later life, subjective financial wellbeing was associated with health, but
the effect of income on health was mediated entirely through subjective financial wellbeing. The
poorer health of the divorced/separated was also mediated entirely by differences in subjective
financial wellbeing. Research on health inequalities should pay greater attention to the link
between subjective financial hardship and ill-health, especially during periods of greater economic
difficulties and financial austerity.
Subjective financial well-being, income and health inequalities in mid and later life in Britain

Introduction
Since the Black Report (DHSS, 1980) demonstrated social class inequalities in health in Britain
there has been a plethora of research on the relationships between different measures of socio-
economic status and health, especially social class and education (Bartley, 2004; Scambler, 2012).
Research has also established a positive and fairly linear relationship between income and health
(Ettner, 1996; Benzeval, Judge, & Shouls, 2001; Wilkinson, 1996). However, the relationship
between subjective financial well-being (SFW) and health has been relatively ignored, particularly
in the UK.

Subjective financial well-being (SFW) refers to the individual’s self rating of their income
adequacy to meet their general needs. It is thought to be associated with perceptions of financial
strain and stress. The lack of research in this area is surprising given that Angel, Frisco, Angel and
Chiriboga (2003) argued the importance for both researchers and policy makers of differentiating
whether it is income and poverty alone, or an individual’s perception of their financial situation
(SFW) that impacts on health. This paper examines the relationship between SFW and health while
adjusting for income. As both health inequalities and economic activity vary over the life course
and levels of ill-health increase with age, we compare this relationship in mid and later life.

Health and income

Numerous studies have reported an income gradient with poor health (Ettner, 1996; Mackenbach,
Kunst, Cavelaars, Groenhof, & Geurts, 1997; Mackenbach, Meerding, & Kunst, 2007; Blane,
Bartley, & Smith, 1997; Hart, Smith, & Blane, 1998). The material explanation is largely used to
account for the relationship between income and poor health, arguing that income affects housing,
neighbourhood environments, diet and access to facilities for exercise and health care, which in
turn all affect health (DHSS, 1980; Bartley, 2004; Link and Phelan, 2005). Income and earning
potential are also related to power which can influence health (Bartley, 2004). Wilkinson (1996;
Wilkinson and Pickett, 2010) show the importance of relative income and position in the income
hierarchy for health, arguing that those with lower incomes are more likely to experience stress,
social and psychological deprivation that have a detrimental impact on health. The relationship
between income and health is argued to be bi-directional, since health can impact on the ability to
work and level of earnings. However, longitudinal research in the US (Muennig, 2008) and UK
(Beckett, 2000) suggests that income has a greater impact on health, than health on income.

The relationship between income and health varies with age. In European samples, income and
health are more strongly associated in mid-life (45-64 years old) than in younger age groups and in
later life (65+) (Mackenbach et al., 2007). A US longitudinal study also found income was a strong
predictor of health, particularly below age 65 (McDonough, Duncan, Williams, & House, 1997).
These age differences may be due to reverse causation, as individuals in mid-life are more likely to
depend on income from paid work, whilst in later life income primarily comes from pensions.
Since health can impact on the ability to work, health is more likely to influence income in mid
than later life, leading to a stronger relationship between income and health in mid-life (Muennig,
2008).

US research finds that income is a more important determinant of health than other measures of
socio-economic status (SES), such as education and social class (Duncan, Daly, McDonough, &
Williams, 2002). Given changes in the structure of labour markets and the possibility that social
class may be less central for health now than in the past (Scambler, 2012), it is pertinent to
examine the relative importance of social class, education and employment status, as well as
income, in patterning health. Health inequalities research has paid less attention to subjective
measures of status, such as subjective financial well-being, although ‘people’s sense of their social
positioning is salient for their health’ (Scambler, 2012: 133).

Health and subjective financial well-being

Despite the established relationships between SES and health, and between income and health, few
studies have explored the relationship between subjective financial well-being (SFW) and health.
Angel et al. (2003) argue that it is important to distinguish between SFW and income due to their
distinct meanings and potentially individual consequences for health. Objective measures of
income do not capture the meaning of income adequacy to individuals (Hazelrigg & Hardy, 1997;
Mirowsky & Ross, 1999) with people on low incomes not always reporting financial strain, which
indicates that these two measures are different and therefore may differentially impact on health
(Angel et al., 2003). Kahn and Fazio (2005) highlight that income and financial strain (a form of
SFW) are not the same; a level of income that may be sufficient to meet one individual’s needs
may be insufficient for another individual (Zimmerman & Katon, 2005). Social comparison and
expectation may lead to differences in the perception of the adequacy of income (Angel et al.,
2003). Perceptions of income adequacy are likely to be related to comparisons with an individual’s
reference groups, with these perceptions varying across different local and societal contexts
(Whelan, Layte, Maitre & Nolan, 2001; Whelan & Maitre, 2013).

The expected association between subjective financial well-being and health can be theorised
through two main mechanisms. Firstly, in terms of feelings of ‘relative material deprivation’
(Pantazis, Gordon & Levitas, 2006), and to what extent individuals feel they have insufficient
income to participate in ways seen as customary within their community or peer group, such as
being able to afford a week’s annual holiday. This mechanism relates to reference group theory
(Whelan and Maitre, 2013) and the role of social comparisons as a potential mediating factor
between SFW and health. Secondly, subjective financial well-being may be linked to health
through perceptions of ‘financial strain/economic stress’ because of inability to manage on their
income, which involves psycho-social processes associated with stress, anxiety and helplessness
(Bartley, 2004: Kahn & Fazio, 2005). Both of these mechanisms are likely to have longer-term
negative impacts on health and well-being due to feelings of lack of a sense of control,
hopelessness, demoralisation, and reductions in self-worth and self-confidence (Angel et al. 2003;
Pearlin, Menaghan, Lieberman & Mullan, 1981).

Previous studies have not distinguished these two conceptual approaches to the link between
subjective financial well-being and health. This is despite Whelan et al. (2001) showing that
‘perceived material deprivation’ and ‘economic stress’ represent two distinct dimensions of
subjective financial well-being, which are both independently related to income. However, Whelan
et al. (2001) do not examine the relationship of each of these two indicators of subjective financial
well-being with health.

Most studies of SFW and health have focused only on older people. These have found that older
people who experienced periods of financial inadequacy throughout the life course report poorer
health (Kahn & Pearlin, 2006) and older people reporting current financial strain or subjective
financial inadequacy had worse subjective health (Cheng, Chi, Boey, Ko, & Chou, 2002;
Nummela, Sulander, Heinonen & Uutela, 2007; Angel et al., 2003). However, these studies did not
adjust for level of income. US research on women aged 70-79 found that those who reported
subjective financial strain were 60% more likely to die within five years when absolute income and
socio-economic status were adjusted (Szanton et al., 2008). Some studies have researched SFW
and health in other age groups, but not using nationally representative samples. For example,
Tucker-Seeley and colleagues (2013) report a positive association between perceived financial
hardship and self-reported health in a US study of low income housing residents, when socio-
economic factors and psychological distress were controlled; and Szanton and colleagues (2010)
found that African American twins aged 25-89 years who reported financial strain in adulthood
were more likely to have a physical disability and report depressive symptoms, but neither study
adjusted for income.

Previous studies have not examined the relationship between SFW and health at different stages of
the life course. It is important to contrast this relationship in mid and later life, because economic
position and income sources vary at these two life stages, with income generally becoming more
fixed after retirement (Muennig, 2008), and health inequalities are at their greatest in mid-life
(House, Kessler, Herzog, Mero, Kinney, & Breslow, 1990; House, Lepkowski, Kinneyt, Mero,
Kessler, & Herzog, 1994). Most previous research on SFW has not adjusted for income, therefore
the independent relationship between SFW, income and health has not been well characterised in
nationally representative samples. In addition, previous research has focused on a single measure
of SFW and has not examined the independent effects of subjective ‘material deprivation’ and of
‘economic/financial strain’ on health.

The aims of this paper are to examine the relationship between subjective financial well-being
(SFW) and health in mid and later life using nationally representative British data whilst adjusting
for income. The research questions are: (i) How are two distinct measures of SFW associated with
health in Britain; (ii) What role does income play in these relationships; and (iii) Do these
relationships differ in mid and later life.

Methodology
This study used data from the General Household Survey (GHS) for 2006 (Office for National
Statistics, 2008), which is representative of private households in England, Scotland and Wales. A
stratified, two-stage probability sample selected 576 postal sectors and addresses from the Postcode
Address File. The selected sample consisted of 12,562 eligible households, with interviews
achieved in 9,731 households. Interviews were conducted with all household members aged 16 and
over. The survey was administered using face-to-face computer-assisted personal interviewing
(CAPI) and achieved a 76% response rate (ONS, 2008). Interviews (excluding proxies) were
undertaken with 5654 adults aged 45-64 and 3813 aged 65+. The analysed data set was fully
anonymised and supplied by the UK Data Archive.

Measuring self-rated health

The analysed self-report measure of general health has been extensively used in health research.
Studies have found that an individual’s subjective assessment of their health is related to objective
measures of health and mortality (Farmer & Ferraro, 1997; Mackenbach et al., 2002). Self-
reported health was measured by the question ‘How is your health in general? Would you say it
is.... very good, good, fair, bad, or very bad?’ Responses were dichotomised: ‘very good/good
health’ recoded as ‘good’; and ‘fair’, ‘bad’ and ‘very bad health’ recoded as ‘less than good’
health’, as in previous studies (Kunst et al., 2005; Mackenbach et al., 2002, 2007).

Measuring subjective financial well-being

Subjective financial well-being (SFW) was analysed using two self-report variables measured at
the household level which indicate perception of their household’s SFW.

(i) Households’ ‘ability to make ends meet’ provides a subjective measure of ‘economic/financial
strain’. Respondents were asked ‘Thinking of your household’s total monthly or weekly income, is
your household able to make ends meet, that is pay your usual expenses…. with great difficulty,
with difficulty, with some difficulty, fairly easily, easily, or very easily’. We combine ‘great
difficulty’/‘difficulty’ and ‘easily’/‘very easily’ to yield four categories. This is similar to the
measure used by Angel et al. (2003).

(ii) ‘Number of problems with household expenditure’ provides an indicator of SFW linked to
‘perceived material deprivation’ (Whelan et al., 2001; Whelan and Maitre, 2013). The question
asked ‘Looking at this card, can I check whether your household could afford the following?:
To pay for a week's annual holiday away from home?
To eat meat, chicken or fish (or vegetarian equivalent) every second day?
To pay an unexpected, but necessary, expense of £500?
To keep your home adequately warm?’
Principal Components Analysis found these four dichotomous ‘Yes/No’ items loaded onto a single
factor and were internally consistent (Cronbach’s alpha=0.654). We use a three category scale:-
reporting ‘no problems’ with any of the four items; one problem; and reporting 2-4 problems.

Measuring income
Income was based on the GHS derived variable ‘equivalised gross household income’, which
summed the following self-reported income from each household member: ‘their usual gross
weekly pay, weekly income from state benefits, gross weekly income from other sources, gross
bonus weekly rate, gross weekly income from other jobs, self employed gross weekly earnings and
other regular payments’. Household income was equivalised to adjust for differences in size and
composition of households, using the McClements Scale (McClements, 1977). This equivalisation
takes into account that a large family needs a higher income than a single person household in
order for households of different size and composition to have a similar standard of living.
Household members were weighted with the following equivalised values; head of household with
a partner=0.50, partner=0.50, first additional adult in couple household=0.42, each additional
adult=0.36, head of household without partner=0.61, first additional adult in household=0.46,
second additional adult=0.42, each additional adult=0.36, child aged 16-17=0.36, child aged 13-
15=0.27, child aged 11-12=0.25, child aged 8-10=0.23, child aged 5-7=0.21, child aged 2-4=0.18,
child aged 0-1=0.09. These values were summed and total household income was divided by this
weighted measure of household composition to produce equivalised gross household income.

Equivalised weekly household income quartiles were determined separately for the two age groups.
Income quartiles for the 45-64 age group are: <£301, £301<£531, £531<£804, >£804.01, and for
the 65+ age group are: <£205, £205<£290, £290<£435, >£435.01.

Covariates
Covariates were analysed as potential confounders or mediators of the relationship between
income, SFW and health. The following covariates were included in the models: Gender
(male/female); Marital status comprised ‘never married’, ‘divorced/separated’, ‘widowed', and
‘married’ (including legal civil partnerships); Age coded into five year age groups, with the highest
group aged ‘85 and over’; Ethnicity coded as ‘White’ and ‘Non-White’; Education measured by
highest reported level of education attainment and classified as ‘no qualifications’, ‘intermediate
qualifications’ and ‘higher education’ (degree level and above); Social class based on the
individual’s current or last occupation and coded ‘higher’, ‘middle’ and ‘lower’ social class;
Employment status coded ‘Employed’ (including self-employed) and ‘Not Employed’; and
Smoking behaviour categorised as ‘current smoker’, ‘ex-smoker’ and ‘never smoked’.

Statistical analyses
First, bivariate analysis examines the proportions rating their health as ‘less than good’ for each
covariate, income and the two measures of SFW, using chi-squared (two tailed) significance levels
for age groups 45-64 (n=5651) and 65+ (n=3807) (Table 1). Second, logistic regression modelling
is used to analyse the relationship between SFW and health after adjusting for income and the
covariates. Separate models are presented for the 45-64 and 65+ age groups in order to compare
how SFW and income are associated with health in mid and later life. The sample size for all
logistic models is 4639 (age 45-64) and 3104 (age 65+).

To understand the relative contribution of SFW with health the analysis used models with three
steps (Tables 2 and 3). Model 1 analysed subjective health according to gender, age, marital status,
ethnicity, employment status, education level, social class and smoking. Model 2 added income
quartiles. Model 3 (final model) introduced the two SFW measures: ‘ability to make ends meet’
and ‘number of problems with household expenditure’. By including SFW in the final step of the
model the relationship between SFW and health can be explored independently of income and
other factors known to influence health. Table 4 presents the two SFW measures separately when
introduced into the final model in order to assess their independent contribution. Analysis was
conducted using the SPSS programme, version 16. Results are presented as odds ratios (ORs) with
95% confidence intervals.

Regression diagnostics were performed on all models. Analysis of residuals identified no cases
with a studentized residual >2.5. Multicollinearity returned VIF and tolerance figures within an
acceptable range, and the pearson correlation of the two SFW measures was 0.532 in the 45-64 age
group and 0.426 for ages 65+. The Hosmer-Lemeshaw test for both age groups was significant,
but since the sample size is very large, even very small divergencies would lead to significance.
We therefore report Nagelkerke R2 and the AUC to provide an indication of predictive power.
Missing value analysis was conducted using SPSS. Results indicate that missing data is spread
evenly across categories and that it is missing at random.

The GHS interviews all household members over age 16, resulting in some clustering of
respondents within households. Within our analysis sample, 1263 households include two
respondents in the 45-64 age group (out of 4639 cases) and 687 households include two
respondents in the 65+ age group (out of 3104 cases). Multilevel models were run on the final
models to examine clustering and whether results alter if the higher (household) level is taken into
account. Results from the multilevel and single level logistic models were comparable, so we only
report the single-level logistic regression analysis.

Results
The proportions in mid and later life reporting ‘less than good’ health are shown in Table 1. No
significant association is found between gender and self-reported health. In both age groups, the
‘not employed’ report poorer health than the ‘employed’. In mid-life, there is a gender/employment
status interaction with employed women reporting the best health and non-employed men reporting
the poorest health. Therefore, the Models for Mid-life (Table 2) include a four category
gender/employment status variable. Married individuals are least likely to report poor health in
both age groups. Respondents categorised as ‘White’ report better health than the ‘Non-white’.
Statistically significant associations were found between each socio-economic measure and
reporting ‘less than good’ health. In both age groups, fewer respondents with ‘higher
qualifications’, in higher social classes and higher income groups report ‘less than good’ health. A
higher proportion of current smokers report poor health than ex-smokers and people who had never
smoked.

There are strong associations between the two measures of subjective financial well-being (SFW)
and self-rated health in both age groups. A higher proportion who report ‘difficulty making ends
meet’ report ‘less than good’ health than respondents who find it ‘easy/very easy’ to make ends
meet. Respondents who report two or more problems with household expenditure are more likely
to have poor health than those reporting no problems.

Overall the findings in Table 1 are in line with previous research, namely there is poorer health
among respondents who are non-employed, divorced, non-white, with no educational
qualifications, in the lowest social class, in low income quartiles, and who currently smoke. There
are also strong associations between reporting ‘less than good’ health and having difficulties
‘making ends meet’ and reporting ‘problems with household expenditure’. Since low income, low
social class, and several other variables are associated with SFW, it is important to adjust for these
variables when analysing the relationship between SFW and health. Logistic regression is
undertaken to examine the independent effects of SFW and income on health, after adjusting for
other social variables known to be associated with health.

Multivariate analysis: Mid and later life models

Odds ratios for ‘less than good’ health are presented in three sequential logistic regression models
for ages 45-64 (Table 2) and 65 and over (Table 3). Covariates known to be associated with self-
reported health were introduced in Model 1 and income in Model 2.

In mid-life, the gender/employment status variable is highly significant with non-employed men
having the highest odds of ‘less than good’ health (OR=5.88) compared with employed women
(reference category). These odds are only reduced marginally across the models, falling for non
employed men to OR=4.28 in Model 3. Above age 65, the non-employed have significantly higher
odds of poor health, which remains largely unchanged across the models.

Ethnicity is a significant predictor of health in both age groups, with ‘non-white’ respondents
having higher odds of ‘less than good’ health. In mid-life, this association becomes non-significant
in the final model, suggesting that subjective financial well-being may mediate the relationship
between ethnicity and poor health in mid-life but not in later life. Marital status is a significant
predictor of poor health in both age groups in Models 1 and 2, with divorced/separated individuals
reporting the highest odds of poor health. However, once SFW is adjusted in the final model,
being divorced/separated is no longer a significant predictor of health. These findings suggest that
SFW mediates the relationship between being divorced/separated and self-reported poor health in
both age groups (Tables 2 and 3).

Education is a highly significant predictor of self-rated health at all stages in both age groups. The
odds ratio for those with ‘no qualifications’ decreases at each stage, but remains significant in
Model 3. The high odds of reporting poor health for those with no qualifications is similar in both
mid and later life. Social class, for both age groups, is a significant predictor of ‘less than good’
health at Model 1, with reductions in significance when income is adjusted in Model 2 and further
reductions when SFW is adjusted (Model 3). Although, those in the lower social class have the
highest odds of poor health, in later life the intermediate class report better health than the higher
class, but these differences are non-significant. In both age groups, current smoking is a significant
predictor of poor health, with odds ratios only marginally reduced when SFW is adjusted in Model
3.
Equivalised household income is a significant and linear predictor of ‘less than good’ health in
both Models 2 and 3 in mid-life. After adjusting for SFW, although there is a reduction in
statistical significance, those in the lowest income quartile remain most likely to report poor health.
In contrast in later life, income is a significant predictor of poor health only in Model 2 (p<.05),
and is non-significant in the final model when SFW is adjusted. This indicates firstly, a weaker
relationship between income and self-reported health in later life than in mid-life at Model 2,
which is in line with past research. Secondly, that the whole relationship between low income and
poor health appears to be mediated by SFW in later life, but this is not the case in mid-life.

In both age groups, ‘ease of making ends meet’ (economic/financial strain) is a highly significant
predictor of ‘less than good’ health in Model 3. Number of problems with household expenditure
(perceived material deprivation) is also a strong predictor of health in the final model. Individuals
reporting two or more household expenditure problems are more likely to report poor health in
mid-life (OR=1.87) and later life (OR=2.28) than those reporting no problems.

To compare the relative association of each SFW measure with self-rated health, Table 4 presents
Model 3a which adds only ‘ability to make ends meet’ and Model 3b adds only ‘number of
household expenditure problems’. In mid-life, both SFW measures make a similar contribution to
the model, as evidenced by the identical Nagelkerke R2 of 0.235 and other model parameters. In
contrast in later life, the ‘material deprivation’ measure makes a somewhat greater contribution to
the model (R2=.133) than the ‘economic strain’ measure (R2=.124).

In Tables 2 and 3, the AUC for the final models is 0.76 in mid-life and 0.69 in later life.
Comparison of the Nagelkerke R2 suggests that Models 1 and 2 improve upon the null model more
in mid-life than in later life. This would be expected given that employment status, education and
income are stronger predictors of health in mid-life than in later life. In contrast, Model 3 shows a
marginally greater change in R2 in later life than mid-life, suggesting that SFW may be particularly
salient for older people’s health.

In summary, the main difference between the mid-life and later life models is that income is more
strongly associated with self-reported health in mid-life than in later life. In mid-life, income and
SFW are independently associated with health, while in later life the relationship between income
and health is mediated by SFW. Measures of SFW, especially related to material deprivation,
appear to be more strongly associated with subjective health above age 65 than at ages 45-64. We
also reanalysed our data using ordinal regression (self-reported health was measured as a Likert
scale), and found the results for both the 45-64 and 65+ age groups were substantively similar. For
example, in the final ‘midlife’ model both income and the two measures of SFW had statistically
significant effects, whereas in the ‘older’ age group only the two measures of SFW had significant
effects, but income was non-significant. In addition, multilevel logistic models produced
comparable results as the reported final single-level logistic models, enhancing confidence in our
findings.

Discussion
This study explored the relationship between subjective financial well-being (SFW) and self-rated
health, while adjusting for income, based on nationally representative British data and compared
this relationship in mid and later life. SFW and income were independently associated with health
in mid-life; those with low incomes and difficulty coping on their incomes were more likely to
report poor health. Whilst in later life, only SFW was directly associated with health; people who
found it difficult to cope on their income regardless of its level had increased odds of ‘less than
good’ health. In mid-life, income remained a significant predictor, the risk of poor health
decreasing sequentially as income increased from the lowest to the highest income quartiles, even
when SFW was adjusted. However, in later life income is no longer a significant predictor of
health after adjusting for SFW. This suggests that the relationship between income and health is
entirely mediated by perceptions of income adequacy in later life. The finding that income is
positively related to health in mid-life is in line with previous research (Ettner, 1996; Blane,
Bartley, & Smith, 1997; McDonough et al., 1997). Whereas, our finding that among older people,
income is only related to health through the pathway of SFW has not previously been reported.

The paper also reports that divorced/separated individuals had the highest odds of ‘less than good’
health in Models 1 and 2 in both age groups, but this was mediated entirely following adjustment
for SFW in Model 3. Although, past research suggests that widowhood may be a significant
predictor of poor health (Hughes & Waite, 2009; Manzoli et al., 2007); this was not found in our
study. Despite widows being more likely to experience low income than their married counterparts
(Arber, 2004), this may not detrimentally affect their health, if they view their income as adequate
for their needs.
Being outside the labour market and having no educational qualifications are robustly associated
with poor health; these associations only marginally diminish following adjustment for income and
SFW in both age groups. In contrast, social class has a weak association with poor health that is
largely mediated by income and SFW, suggesting social class may be less centrally linked to health
inequalities in Britain than in the past (DHSS, 1980; Barclay, 2004).

Our finding that income does not predict health when SFW is adjusted in later life supports the
argument of Nummela et al. (2007: 39) that perception of the adequacy of income may be an ‘even
better predictor of self-rated health than income.’ However, this is not the case in mid-life when
both income and SFW have independent effects on health. The elevated odds of poor health for
those who report struggling to cope on their income are broadly comparable for both age groups,
but somewhat greater in later than middle life. This supports Cheng et al. (2002: 1416) who argued
that the relationship between self-reported financial well-being and health is ‘particularly true’ in
later life, and supports previous research in Finland and the US (Nummela et al., 2007; Angel et
al., 2003; Szanton et al., 2008).

The two measures of SFW are both independently associated with health, after adjusting for
income and other covariates. However, in later life, there is a suggestion that perceptions of
‘material deprivation’ may have a stronger association with health than ‘economic/financial strain’.
These findings reinforce the value of analysing more than one indicator of SFW.

The greater importance of income for health in mid-life compared to later life may be due to three
factors. First, in later life levels of income vary less between individuals than in mid-life (Brown &
Prus, 2006). This was confirmed in separate analyses that found the Gini coefficient for equivalised
household income in mid-life was 0.42 and in later life was 0.33. Thus in mid-life income may
have a stronger impact due to greater income inequalities at this life course stage. Second, more
people in later life live on a fixed income whilst those in mid-life are more likely to experience
income changes associated with changes in paid employment and periods out of the labour market.
Third, the relationship between income and health is likely to be more bi-directional in mid-life
than later life, since income is primarily derived from work, and ability to work may be affected by
ill-health. Fewer respondents in later life are economically active, therefore income is less likely to
be affected by health (Muennig, 2008). Due to the cross-sectional nature of this study the direction
of causation cannot be determined and longitudinal research is needed.
Our analyses found strong associations between both measures of subjective financial well-being
(‘perceived material deprivation’ and ‘economic/financial strain’) and self-reported health lending
support to more than one potential mechanism linking SFW to health. First, ‘perceived material
deprivation’ is associated with poor health pointing to the role of reference groups (Whelan et al.,
2001; Whelan & Maitre, 2013) and ability to enjoy customary standards of living as critical for
health. Second, subjective feelings of income inadequacy and ‘financial strain’ are likely to
produce stress, anxiety and helplessness with negative consequences for health (Bartley, 2004;
Kahn & Fazio, 2005). Both relative ‘material deprivation’ and ‘financial strain’ can impact on
health through feelings of demoralisation, and reductions in self-confidence, self-worth and sense
of control (Angel et al., 2003; Pearlin et al., 1981). Inadequate financial resources can also reduce
social participation and increase the likelihood of social exclusion, representing a further source of
stress (Bartley, 2004).

Research has shown that worry and anxiety associated with financial problems and debt are
associated with sleep disturbances in the UK (Dregan & Armstrong, 2009; Kumari, Green, &
Nazroo, 2010) and US (Hall et al., 2008). Therefore poor sleep may be implicated in the link
between SFW and health. Strong associations have also been found between diverse indicators of
low socio-economic status and poor sleep (Arber, Bote, & Meadows, 2009; Arber & Meadows,
2011), and between disturbed sleep and poor self-reported health (Kumari et al., 2010; Arber &
Meadows, 2011). We suggest that poor sleep may represent a potential mechanism linking SFW
with health, since pathways between subjective financial problems and poor health are likely to
include worry and anxiety resulting in disrupted sleep. Future health researchers should examine
the relationships between measures of socio-economic status (including income), subjective
financial well-being, sleep and health in order to clarify these mechanisms. As Dregan and
Armstrong (2009) showed that sleep loss through worry was greater during periods of economic
downturn, researchers should also consider the health implications of changes in welfare policies
that may be creating greater financial difficulties for families.

Methodological Considerations
A strength of this study is that it analysed high quality data from the large nationally representative
General Household Survey (ONS, 2008). However, the GHS represents a sample of private
households and excludes people who live in institutions, such as nursing homes. Therefore findings
can be generalised to all adults aged over 45 living in private households in Britain. The study was
restricted to Britain, which Whelan and colleagues (2001) shows is in an intermediate position
across European countries regarding our measures of SFW (economic stress and perceived material
deprivation) and level of income inequality. This ‘average’ positioning may enhance the
generalisability of our findings to other European societies, but further research is required on
SFW, income and self-rated health in other national contexts.

The study used a cross-sectional design, therefore causation cannot be determined. Longitudinal
studies are required to clarify the direction of causation of identified relationships, for example,
whether the poor health of the divorced/separated is because being married ‘protects’ against ill-
health or ill-health ‘selects’ people out of marriage (Koball et al., 2010). We use a measure of
income based on self-reports of all household members, which may lack reliability. Future studies
should employ more objective income measures, such as based on registry data.

Conclusions
Since the early 1980s extensive research has documented the influence of social factors on health
(Scambler, 2012). Within this corpus of research, a key overlooked issue is the relative
contributions of subjective financial well-being and income for health. This paper examined the
independent association of both income and SFW with self-rated health, comparing these
relationships in mid and later life. Two indicators of SFW (economic strain and perceived material
deprivation) were strongly associated with health, especially in later life. Income remained
independently associated with health only in mid-life, while in later life the association of income
with health was mediated entirely through SFW. By adopting an age-stratified approach this study
has demonstrated that different factors are associated with health at these two life course stages.
The study also suggests that the poor health of the divorced/separated is mediated entirely by SFW.

As one of the first studies to use two measures of SFW and adjust for income when examining the
relationship between SFW and self-rated health, it has thrown light on the need to adopt an
approach to understanding health that examines both measures of SFW and income and how these
relationships vary at different points in the life course.
References
Angel, R., Frisco, M., Angel J., & Chiriboga, D. (2003). Financial strain and health among elderly
Mexican-origin individuals. Journal of Health and Social Behavior, 44(4), 536-551.

Arber, S. (2004). Gender, marital status, and ageing: Linking material, health, and social resources.
Journal of Aging Studies, 18(1), 91-108.

Arber, S., Bote, M., & Meadows, R. (2009). Gender and socio-economic patterning of self-reported
sleep problems in Britain. Social Science and Medicine, 68(2), 281-289.

Arber, S., & Meadows, R. (2011). Social and health patterning of sleep quality and duration. In S.
McFall, & C. Garrington (Eds.), Understanding Society: Early Findings from the first wave of the
UK’s Household Longitudinal Study. Colchester: ISER, University of Essex.

Bartley, M. (2004). Health Inequality: An Introduction to Theories, Concepts and Methods.

Cambridge: Polity.

Beckett, M. (2000). Converging health inequalities in later life—an artefact of mortality selection.
Journal of Health and Social Behaviour, 4, 106-119.

Benzeval, M., Judge, K., & Shouls, S. (2001). Understanding the relationship between income and
health: How much can be gleaned from cross-sectional data? Social Policy & Administration, 35,
376-396.

Blane, D., Bartley, M., & Smith, G. (1997). Disease aetiology and materialist explanations of
socioeconomic mortality differentials. European Journal of Public Health, 7(4), 385–391.

Brown, R.L., & Prus, S.G. (2006). Income inequality over the later-life course: A comparative
analysis of seven OECD countries. Annals of Actuarial Science, 1, 307-317.

Cheng, Y., Chi, I., Boey, K., Ko, L., & Chou, K. (2002). Self-rated economic condition and the
health of elderly persons in Hong Kong. Social Science and Medicine, 51, 1416-1424.

DHSS (1980). Inequalities in Health: Report of a Working Group, London: Department of Health
and Social Security. [The Black Report].

Dregan, A., & Armstrong, D. (2009). Age, cohort and period effects in the prevalence of sleep
disturbances among older people: The impact of economic downturn. Social Science Medicine, 69,
1432-38.

Duncan, G., Daly, M., McDonough, P., & Williams, D. (2002). Optimal indicators of
socioeconomic status for health research. American Journal of Public Health, 92, 1151-1157.

Ettner, S. (1996). New evidence on the relationship between income and health. Journal of Health
Economics, 15(1), 67-85.

Farmer, M., & Ferraro, K. (1997). Distress and perceived health: Mechanisms of health decline.
Journal of Health and Social Behavior, 39, 298-311.
Hall, M.H., Buyusse, D.J., Nofzinger, E.A., Reynolds, I.C.F., Thompson, W., Mazumdar, S., et al.
(2008). Financial strain is a significant correlate of sleep continuity disturbances in late-life.
Biological Psychology, 77(2), 217-22.

Hart, C., Smith, G., & Blane, D. (1998). Inequalities in mortality by social class measured at three
stages of the life course. American Journal of Public Health, 88(3), 471–4.

Hazelrigg, L., & Hardy, M. (1997). Perceived income adequacy among older adults. Research on
Aging, 19, 69-108.

House, J., Kessler, R., Herzog, H., Mero, R., Kinney, A., & Breslow, M. (1990). Age,
socioeconomic status, and health. Milbank Quarterly, 68(3), 383-411.

House, J., Lepkowski, J., Kinney, A., Mero, R., Kessler, R., & Herzog, A. (1994). The social
stratification of aging and health. Journal of Health and Social Behaviour, 35(3), 213-234.

Hughes, M.E., & Waite, L. J., (2009). Marital health at mid-life. Journal of Health and Social
Behaviour, 50, 244-358.

Kahn, J., & Fazio, E., (2005). Economic status over the life course and racial disparities in health.
Journal of Gerontology, 60B,76-84.

Kahn, J., & Pearlin, L. (2006). Financial strain over the life course and health among older adults.
Journal of Health Social Behaviour, 47, 17-31.

Koball, H., Moiduddin, E., Henderson, J., Goesling, B., & Besculides, M. (2010). What do we
know about the link between marriage and health? Journal of Family Issues, 31, 1019-1040.
Kumari, M., Green., & Nazroo, J. (2010). Sleep duration and sleep disturbance. In J. Banks, C.
Lessof, J. Nazroo, N. Rogers, M. Stafford, & A. Steptoe (Eds.), Financial circumstances, health
and well-being of the older population in England. London: Institute for Fiscal Studies.

Kunst, A.E, Bos, V., Lahelma, E., Bartley, M., Lissau, I., Regidor, E. et al. (2005). Trends in
socioeconomic inequalities in self-assessed health in 10 European countries. International Journal
of Epidemiology, 34, 295-305.

Link, B., & Phelan, J. (1995). Social conditions as fundamental causes of disease. Journal of
Health and Social Behavior, 36, 80–94.

Mackenbach, J., Kunst, A., Cavelaars, A., Groenhof, F., & Geurts, J. (1997). Socioeconomic
inequalities in morbidity and mortality in Western Europe: the EU Working Group on
socioeconomic inequalities in health. Lancet, 34, 1655-1659.

Mackenbach, J.P., Meerding, W.J., & Kunst, A.E (2007). Economic implications of socio-
economic inequalities in health in the European Union, Luxembourg: European Commission.
http://ec.europa.eu/health/ph_determinants/socio_economics/documents/socioeco_inequalities_en.
pdf. (accessed 22 July 2010)
Mackenbach, J.P., Simon, J.G., Looman, W.N. & Joung, I.M.A. (2002). Self-assessed health and
mortality: Could psychosocial factors explain the association? International Journal of
Epidemiology, 31, 1162-8.
Manzoli, L., Villari, P. M., Pirone, G., & Boccia, A. (2007). Marital status and mortality in the
elderly: a systematic review and meta-analysis. Social Science and Medicine, 6, 77-94.

McClements, L. (1977). Equivalence scales for children. Journal of Public Economics, 8(2): 191-
210.

McDonough, P., Duncan, G., Williams, D., & House, J. (1997). Income dynamics and adult
mortality in the United States, 1972 through 1989. American Journal of Public Health, 87(9),
1476–1483.

Mirowsky, J., & Ross, C. (1999). Economic hardship across the life course. American Sociological
Review, 64, 548-569.

Muennig, P. (2008). Health selection vs. causation in the income gradient: what can we learn from
graphical trends? Journal of Health for the Care Poor and Underserved, 19(2), 574-579.

Nummela, O., Sulander, T., Heinonen, H., & Uutela, A. (2007). Self-rated health and indicators of
SES among the ageing in three types of communities. Scandinavian Journal of Public Health,
35(1), 39-47.

Office for National Statistics (2008). General Household Survey 2006. Appendix B: Sample Design
and Response. Newport: Office for National Statistics. http://www.ons.gov.uk/ons/rel/ghs/general-
household-survey/2006-report/index.html (accessed: 10th April 2013)

Pantazis, C., Gordon, D., & Levitas, R. (2006) Poverty and Social Exclusion in Britain, Bristol:
Policy Press.

Pearlin, L.I., Menaghan, E.G., Lieberman, M.A., & Mullan. J.T. (1981). The stress process.
Journal of Health and Social Behavior, 22(4), 337-356.

Scambler, G. (2012). Health inequalities. Sociology of Health and Illness, 34(1), 130-146.

Szanton, S.L., Allen, J.K., Thorpe, J.R., Seeman, T.M., Bandeen-Roche, K & Fried, L.P. (2008).
Effect of financial strain on mortality in community-dwelling older women. Journal of
Gerontology, 63B(6), 369-374.

Szanton, S.L., Thorpe, R.J., & Whitfield, K. (2010). Life-course financial strain and health in
African-Americans. Social Science and Medicine, 71, 259-265.Tucker-Seeley, R.D., Harley, A.E.,
Stoddard, A.M., & Sorensen, G.G. (2013). Financial hardship and self-rated health among low-
income housing residents. Health, Education and Behavior, 40(4), 442-448.

Whelan, C.T., Layte, R., Maitre, B., & Nolan, B. (2001). Income, deprivation, and economic strain.
An analysis of the European Community Household Panel. European Sociological Review, 17(4),
357-372.
Whelan, C.T., & Maitre, B. (2013). Material deprivation, economic stress, and reference groups in
Europe: An analysis of EU-SILC 2009. European Sociological Review. doi:10.1093/esr/jct006
(accessed on 16th July 2013)

Wilkinson, R. (1996). Unhealthy Societies: The Afflictions of Inequalities. London: Routledge.

Wilkinson, R., & Pickett, K. (2010). The Spirit Level. London: Penguin.

Zimmerman, F., & Katon, W. (2005). Socioeconomic status, depression disparities, and financial
strain: what lies behind the income-depression relationship? Health Economics, 14, 1197-1215.
Table 1. Percentage reporting ‘less than good’ health by covariates, income and subjective financial
well-being (and base numbers) for age 45-64 and age 65+

45-64 65+ Total

% N % N % N
All 26.3% 5651 41.3% 3807 32.4% 9458
Gender
Men 26.2% 2647 40.3% 1766 31.8% 4413
Women 26.4% 3004 42.2% 2041 32.8% 5045
p 0.878 0.217 0.316
Employment Status
Employed 16.3% 3829 25.8% 279 17.0% 4108
Not Employed 47.3% 1820 42.6% 3526 44.2% 5346
p 0.000 0.000 0.000
Gender/Employment Status
Women - Employed 15.8% 1839 27.6% 105 16.4% 1944
Men - Employed 16.9% 1990 24.7% 174 17.5% 2164
Women - Not Employed 43.2% 1163 43.0% 1936 43.1% 3099
Men - Not Employed 54.5% 657 42.0% 1590 45.7% 2247
p 0.000 0.000 0.000
Ethnicity
White 25.7% 5380 40.8% 3720 31.9% 9100
Non-White 37.7% 268 64.4% 87 44.2% 355
p 0.000 0.000 0.000
Marital Status
Married 23.5% 4039 38.1% 2202 28.7% 6241
Never Married 32.0% 485 43.8% 201 35.4% 686
Widowed 34.6% 191 45.3% 1112 43.7% 1303
Divorced/Separated 33.9% 936 48.3% 292 37.3% 1228
p 0.000 0.000 0.000
Education level
Higher Education 16.0% 1594 29.7% 589 19.7% 2183
Intermediate Qualifications 24.2% 2277 33.9% 855 26.9% 3132
No Qualifications 40.7% 1423 48.1% 2026 45.1% 3449
p 0.000 0.000 0.000
Social Class
Higher 17.4% 2113 34.3% 1106 23.2% 3219
Middle 28.1% 2325 40.7% 1686 33.4% 4011
Lower 37.5% 1056 50.9% 896 43.6% 1952
p 0.000 0.000 0.000
Smoking
Never Smoked 20.7% 2883 37.3% 1929 27.4% 4812
Ex-smoker 27.5% 1565 44.5% 1513 35.9% 3078
Current Smoker 38.5% 1202 49.0% 365 40.8% 1567
p 0.000 0.000 0.000
Income quartilesa
Highest quartile 12.4% 1275 29.6% 877
50<75% 18.3% 1279 41.0% 882
25<50% 28.1% 1275 47.3% 880
Lowest quartile 46.9% 1275 46.8% 878
P 0.000 0.000
Ability to make ends meet
Easily/Very Easily 15.1% 1715 31.2% 1107 21.4% 2822
Fairly Easily 23.3% 2159 40.4% 1714 30.9% 3873
Some Difficulty 35.9% 1173 50.0% 674 41.0% 1847
Difficulty/Great Difficulty 50.9% 593 63.8% 304 55.3% 897
P 0.000 0.000 0.000
No. of problems with
household expenditure
None 19.5% 4325 35.4% 2924 25.9% 7249
One 39.3% 560 56.3% 444 46.8% 1004
Two to four 55.3% 761 65.3% 435 58.9% 1196
p 0.000 0.000 0.000

a
Equivalised household income quartiles were determined separately for the two age group (45-64 and
65+), therefore percentages reporting ‘less than good’ health for each income quartile cannot be calculated
for the Total sample.
Table 2. Odds ratios of ‘less than good’ health in mid-life (age 45-64) (n=4639)

Model 1 Model 2 Model 3

OR CI 95% OR CI 95% OR CI 95%
Gender/Employment Status *** *** ***
Women - Employed (reference) 1.00 1.00 1.00
Men – Employed 1.22 (1.00a-1.48) 1.22* (1.00-1.49) 1.22 (1.00a-1.49)
Women - Not Employed 3.75*** (3.06-4.60) 3.10*** (2.51-3.84) 3.06*** (2.46-3.81)
Men – Not Employed 5.88*** (4.63-7.48) 4.58*** (3.55-5.90) 4.28*** (3.30-5.54)
Ethnicity ** * ns
White (reference) 1.00 1.00 1.00
Non-white 1.66** (1.17-2.36) 1.52* (1.06-2.16) 1.24 (0.87-1.78)
Marital status *** ** Ns
Married (reference) 1.00 1.00 1.00
Never Married 1.22 (0.95-1.58) 1.16 (0.89-1.50) 1.12 (0.86-1.46)
Widowed 1.04 (0.71-1.52) 0.96 (0.65-1.40) 0.94 (0.64-1.39)
Divorced/Separated 1.55*** (1.28-1.88) 1.44*** (1.19-1.74) 1.20 (0.98-1.47)
Education level *** *** *
Higher Education (reference) 1.00 1.00 1.00
Intermediate Qualifications 1.30* (1.06-1.59) 1.19 (0.97-1.47) 1.15 (0.94-1.42)
No Qualifications 1.83*** (1.45-2.32) 1.58*** (1.24-2.00) 1.44** (1.12-1.83)
Social class ** Ns ns
Higher (reference) 1.00 1.00 1.00
Middle 1.36** (1.13-1.64) 1.23* (1.01-1.48) 1.21 (0.99-1.46)
Lower 1.46** (1.16-1.84) 1.27* (1.01-1.60) 1.15 (0.91-1.46)
Smoking *** *** ***
Never smoked (reference) 1.00 1.00 1.00
Ex-Smoker 1.35** (1.13-1.60) 1.34** (1.12-1.59) 1.30** (1.09-1.55)
Current Smoker 1.83*** (1.52-2.20) 1.77*** (1.47-2.13) 1.58*** (1.30-1.91)
Household Income Quartiles *** *
Highest (reference) 1.00 1.00
50<75% 1.30* (1.02-1.64) 1.15 (0.90-1.46)
25<50% 1.68*** (1.32-2.12) 1.32* (1.03-1.69)
Lowest Quartile 2.22*** (1.73-2.85) 1.46** (1.11-1.90)
Ability to make ends meet ***
Easily/Very easily (reference) 1.00
Fairly Easily 1.41** (1.16-1.72)
Some Difficulty 1.73*** (1.36-2.19)
Difficulty/Great Difficulty 2.15*** (1.59-2.90)
No. household expenditure problems ***
None (reference) 1.00
One 1.60*** (1.26-2.03)
Two to four 1.87*** (1.46-2.41)

Omnibus χ2 (Model) 689.5*** 43.8*** 104.5***

-2 Log Likelihood 4599.6 4555.8 4451.3
Nagelkerke R2 0.203 0.215 0.243
a
Source: General Household Survey 2006. All models adjusted for 5 year age groups. *p<0.05, **p<0.01, *** p<.001. CI
rounded from 0.999 to 1.00.
Table 3. Odds ratios of ‘less than good’ health among older adults (age 65 and over) (n=3104)

Model 1 Model 2 Model 3

OR CI 95% OR CI 95% OR CI 95%
Gender ns ns ns
Men (reference) 1.00 1.00 1.00
Women 1.04 (0.89-1.23) 1.04 (0.89-1.23) 1.05 (0.89-1.24)
Employment Status *** *** ***
Employed (reference) 1.00 1.00 1.00
Not Employed 2.01*** (1.45-2.78) 1.85*** (1.33-2.58) 1.93*** (1.38-2.70)
Ethnicity ** ** **
White (reference) 1.00 1.00 1.00
Non-white 2.67** (1.50-4.74) 2.58** (1.46-4.59) 2.21** (1.22-4.00)
Marital status * * ns
Married (reference) 1.00 1.00 1.00
Never Married 1.30 (0.93-1.82) 1.28 (0.92-1.80) 1.30 (0.92-1.84)
Widowed 1.00 (0.83-1.20) 0.97 (0.80-1.17) 0.91 (0.75-1.10)
Divorced/Separated 1.43* (1.07-1.90) 1.39* (1.04-1.85) 1.12 (0.83-1.51)
Education level *** *** ***
Higher Education (reference) 1.00 1.00 1.00
Intermediate Qualifications 1.17 (0.90-1.52) 1.11 (0.85-1.45) 1.10 (0.84-1.44)
No Qualifications 1.82*** (1.42-2.34) 1.65*** (1.28-2.14) 1.58** (1.21-2.05)
Social class ** ** *
Higher (reference) 1.00 1.00 1.00
Middle 0.94 (0.77-1.15) 0.90 (0.74-1.10) 0.86 (0.70-1.06)
Lower 1.33* (1.05-1.68) 1.26 (0.99-1.59) 1.13 (0.89-1.44)
Smoking *** *** ***
Never smoked (reference) 1.00 1.00 1.00
Ex Smoker 1.34*** (1.14-1.58) 1.34*** (1.14-1.57) 1.33** (1.13-1.57)
Current Smoker 1.70*** (1.30-2.20) 1.68*** (1.29-2.18) 1.51** (1.15-1.98)
Household Income Quartiles * ns
Highest (reference) 1.00 1.00
50<75% 1.21 (0.97-1.52) 1.11 (0.88-1.40)
25<50% 1.42** (1.12-1.79) 1.10 (0.86-1.41)
Lowest Quartile 1.30* (1.02-1.65) 0.92 (0.71-1.18)
Ability to make ends meet ***
Easily/Very easily (reference) 1.00
Fairly Easily 1.39** (1.15-1.68)
Some Difficulty 1.67*** (1.30-2.14)
Difficulty/Great Difficulty 2.36*** (1.68-3.32)
No. household expenditure probs ***
None (reference) 1.00
One 2.00*** (1.57-2.56)
Two to four 2.28*** (1.74-3.00)

Omnibus χ2 (Model) 207.5*** 8.8* 136.5***

-2 Log Likelihood 4001.9 3993.1 3856.6
Nagelkerke R2 0.087 0.091 0.145
Source: General Household Survey 2006 All models are adjusted for 5 year age groups. *p<0.05, **p<0.01, *** p<.001.
Table 4. Odds Ratios of ‘less than good’ health for income and two Subjective Financial Well-being
indicators entered separately in final models

Midlife (age 45-64) Older (65+)

(n=4639) (n=3104)
Model 3a Model 3b Model 3a Model 3b
OR CI 95% OR CI 95% OR CI 95% OR CI 95%
Household Income Quartiles ** ** ns ns
Highest (reference) 1.00 1.00 1.00 1.00
50<75% 1.14 (0.90-1.46) 1.25 (0.99-1.59) 1.10 (0.87-1.38) 1.19 (0.95-1.50)
25<50% 1.35* (1.06-1.72) 1.51** (1.19-1.92) 1.14 (0.89-1.45) 1.25 (0.99-1.59)
Lowest Quartile 1.58** (1.21-2.06) 1.72*** (1.32-2.27) 0.99 (0.77-1.28) 1.06 (0.82-1.36)
Ability to make ends meet *** ***
Easily/Very easily (reference) 1.00 1.00
Fairly Easily 1.44*** (1.18-1.76) 1.45*** (1.20-1.75)
Some Difficulty 2.08*** (1.67-2.61) 2.18*** (1.72-2.76)
Difficulty/Great Difficulty 2.99*** (2.28-3.92) 3.63*** (2.65-4.97)
No. household expenditure probs *** ***
None (reference) 1.00 1.00
One 1.88*** (1.50-2.37) 2.32*** (1.84-2.94)
Two to four 2.47*** (1.98-3.08) 2.96*** (2.31-3.80)

Omnibus χ2 (Model) 75.6*** 74.4*** 83.2*** 106.5***

-2 Log Likelihood 4480.2 4481.3 3909.9 3886.6
Nagelkerke R2 0.235 0.235 0.124 0.133

Source: General Household Survey 2006. *p<0.05, **p<0.01, *** p<.001.

All models adjusted for 5 year age groups, gender, marital status, ethnicity, employment status, education level, social class and smoking.

26