Singh-Manoux, Archana PhD; Marmot, Michael G. FRCP; Adler, Nancy E. PhD
The inverse gradient between different measures of socioeconomic status (SES) and health has been widely reported (1–3). In addition to the objective measures of SES (occupational position, education, income, etc.), there have been attempts to explore the links between subjective SES and health. Interest in subjective SES is driven by two different strands of research. First, there is evidence showing it to be a composite measure of SES, including aspects such as education, occupation, and income (4,5). More recently, a paper used 16 variables categorized in four groups (indicators of objective SES, wealth measures, life satisfaction measures, and measures of psychological functioning) to predict subjective SES (6). That study concluded that subjective SES represents a cognitive average of standard markers of SES, including elements representing an assessment of current and future prospects. This “averaging hypothesis” would suggest that subjective SES is both a social and an economic phenomenon (7) and perhaps a better measure of SES at the individual level than any single indicator of SES.
The second source of interest in subjective SES stems from research into the link between income inequality and population health (8,9). Wilkinson (10) has argued that this association is mediated through perceptions of place in the social hierarchy. Such perceptions are believed to produce negative emotions that then translate into poorer health via neuroendocrine mechanisms. Perception of place in the social hierarchy is thus seen to be critically linked to health.
As subjective SES encompasses a wide range of socioeconomic phenomena, it is not surprising that recent studies have shown it to be a good predictor of health (6,11–13). However, it remains unclear if subjective status is a better predictor of health when compared with a comprehensive measure of objective SES. Individual measures of SES are likely to capture only one aspect of the relationship between socioeconomic circumstances and health (14). Consequently, a valid comparison of the relationship between health and subjective and objective measures of SES can only be made if the latter is also a global, composite measure of SES. We were able to examine this issue using data from the Whitehall II study, a longitudinal study on British civil servants. Employment grade, a measure of occupational position in the British civil service, is a global measure of social position, and it reflects levels of education and income, as well as work responsibilities and occupational prestige (15).
The primary purpose of this investigation was to assess the nature of the relationship between health outcomes and an objective and a subjective measure of SES. The first step involved an exploration of the association between these measures of SES with other indicators of SES. This exercise in examining the convergent validity of the two measures was aimed at verifying that both were global measures of SES. The specific aims of this study were as follows:
1. To examine the relationships between objective and subjective measures of SES and health among middle-aged adults and to further examine if these relationships are independent of each other.
2. To examine if objective and subjective measures of SES predict change in health status and to assess if they predict change in health independently of each other.
Data for the current analyses come from the Whitehall II study. The target population for the Whitehall II study was all London-based office staff, aged 35 to 55, working in 20 Civil Service departments (15). With a response rate of 73%, the final cohort consisted of 10,308 participants (6895 men and 3413 women) at the first phase of data collection between 1985 and 1988. Although mostly white collar, respondents covered a wide range of employment grades, from office support to permanent secretary, with salaries in 1995 from £4995 to £150,000.
The screening at baseline (Phase 1) involved a clinical examination and a self-administered questionnaire containing sections on demographic characteristics, health, lifestyle factors, work characteristics, social support, and life events. Since baseline screening, five further data-collection rounds have been completed. Successive phases alternate between collecting data by self-administered questionnaire only and collecting data via a clinical screening in addition to questionnaire completion. Data on measures of SES are from Phase 5 (1997–1999), when the measure of subjective social status was introduced to the study, and that on health outcomes are from Phases 5 and 6 (2000–2001). The mean follow-up period between Phases 5 and 6 was 3 years.
Measures of SES
Objective SES: Grade
Objective SES was assessed via civil service employment grade in the Whitehall II study. All participants were asked to give their current or last known civil service grade title. The civil service identifies 12 nonindustrial grade levels, which have been regrouped to form 6 employment grades overall. Grade 1 reflects high SES; grade 6, low SES. People in different grades differ with respect to salary, social status, and level of responsibility. The annual full-time salaries in 1995 ranged from £4995 to £150,000.
Subjective SES: Ladder
A self-anchoring scale (16) in the form of a 10-rung ladder was introduced to the study at Phase V in order to measure subjective SES (11). Participants were given the drawing of a ladder with the following instructions: “Think of this ladder as representing where people stand in society. At the top of the ladder are the people who are best off—those who have the most money, most education and the best jobs. At the bottom are the people who are worst off—who have the least money, least education and the worst jobs or no job. The higher up you are on this ladder, the closer you are to people at the very top and the lower you are, the closer you are to the bottom. Where would you put yourself on the ladder? Please place a large ‘X’ on the rung where you think you stand.” Details of the validation of this instrument can be found elsewhere (6).
Other Measures of SES
Childhood SES was assessed with a latent variable made up of two measures: father’s social class and socioeconomic circumstances in childhood. Father’s social class was assessed using Registrar General’s Social Class classification. In order to assess socioeconomic circumstances in childhood, respondents were asked to recall family conditions before they were 16 years of age. A 4-item scale was used: father/mother unemployed when they wanted to be working, family had continuing financial problems, family did not have an inside toilet, and family did not have a car. Participants responded either yes or no, and the yes responses were summed so that a high score indicated poor socioeconomic circumstances in childhood. Principal component analysis on these two items revealed 1 factor explaining 68.2% of the variance.
Education was measured as the highest level of education achieved, with the respondent choosing 1 of 11 categories. This was regrouped into five standard hierarchic levels: (1) no formal education, (2) lower secondary education, (3) higher secondary education, (4) degree, (5) higher degree.
Income was assessed using the question on “amount received annually from salary or wages, or pensions, benefits and allowances before deduction of tax.” There were eight categories in all, ranging from “less than £9999” to “more than £70,000.”
Household income was assessed by asking the respondents about “total annual household income from any source, including personal income,” The 11 response categories ranged from “less than £999” to “more than £200,000.”
Household wealth was measured using a question where respondents were asked to assess their total assets (“amount of money the respondent would have if he or she cashed in all household assets—house, car, caravan, boat, jewelry—and paid off all the debts”). The 6 categories measuring household wealth ranged from “less than £4999” to “more than £500,000.”
Feeling of financial security was assessed through response on a 4-point Likert scale to the question: Thinking of the next 10 years, how financially secure do you feel?
Measures of Health
General health functioning was assessed using the Short Form 36 (SF-36) General Health Survey Scales (17). The SF-36 is a 36-item questionnaire that covers issues relating to physical, psychological, and social functioning. It is coded into eight scales: physical functioning, social functioning, role limitations due to physical problems, role limitations due to emotional problems, vitality, bodily pain, general health perception, and general mental health. All scales ranged from 0 (worst health) to 100 (best health). These eight scales of the SF-36 can be summarized into physical and mental components scores (PCS and MCS) using factor analysis (18,19). Low scores imply poor functioning, with 50 being the mean in the general US population.
Psychological distress was measured using the 30-item General Health Questionnaire (GHQ) (20). The GHQ is a well-established screening questionnaire for minor psychiatric disorder suitable for use in the general population samples. GHQ items were scored on 4-point Likert scales (0–3) in order to assign each individual a score reflecting the degree of intensity of psychological distress reported by the individual; the higher the score, the higher the distress felt by the individual (possible score range: 0–90).
Self-rated health was assessed on a 5-point scale using the question: In general, would you say your health is excellent/very good/good/fair/poor? Low scores reflect better health.
First, correlation analysis was carried out to examine the interrelationships between the different measures of SES. Then the association between objective and subjective SES with the other measures of SES (childhood SES, education, income, household wealth, and feeling of financial security) was compared by converting the correlation coefficient into a z-score using Fisher’s r-to-z transformation (21). Then linear regression models were used to examine the age-adjusted relationship between subjective SES (ladder) and the health outcomes. This was repeated using the objective measure of SES (grade). Subsequently, objective and subjective measures of SES were entered simultaneously, along with age, in the regression model to allow an examination of the independence of their associations with health. In these analyses, the SES measures come from Phase 5 and health measures from Phase 6, with a mean follow-up period of 3 years between these phases.
Next, we examined the age-adjusted change in health status between Phases 5 (mean age of participants = 55.90 years, SD = 6.06) and 6 (mean age of participants = 58.39 years, SD = 6.07). We further examined whether grade and ladder were associated with the change in health status from Phase 5 to Phase 6. These analyses were carried out using health outcomes at Phase 6 as an outcome, controlling for the relevant health score at Phase 5 (conditional model for change), as well as age. Here again, first the effect of the measures of SES were examined individually and then simultaneously. All analyses were carried out separately in men and women.
Ten thousand three hundred eight individuals were examined at baseline (Phase 1, 1985–1988). A total of 355 participants died between Phases 1 and 5. The number of participants at Phases 5 and 6 was 7830 and 7358, respectively. Complete data on all the variables of interest in this study were available on 5486 of the participants. Missing data were more common among the older (p < .001) and the lower-SES individuals (p < .02).
Table 1 shows the distribution of the study population on the objective and subjective measure of SES. In this white-collar occupational cohort, 75.7% of the men chose to place themselves in the top five rungs of a 10-rung ladder against 55.1% of women. This reflects the objective distribution by gender of occupational grade as more men occupy the higher grades when compared with women; 54.1% of men occupy the top two grades, whereas 52.1% of women occupy the bottom two grades. Few people chose the bottom three positions on the 10-rung ladder (3.7% men and 7.0% women). This, too, reflects their objective position in society since all were employed and not likely to be at the very bottom of the SES ladder.
Table 2 presents correlation analyses aimed at verifying whether the objective and subjective measures of SES are broadly similar. Results indicate that that both measures of SES are significantly correlated with all other indicators of SES. Education, personal and household income are more strongly related to grade, the measure of objective SES, in both men and women. Feeling of financial security is more strongly related to the ladder (measure of subjective SES) in both sexes, and household wealth is more strongly related to the ladder among women.
The top half of Table 3 presents age-adjusted standardized regression coefficients showing the association between health outcomes and the two measures of SES. These coefficients are calculated from standardized data and reflect the impact on the outcome variable of a change of 1 SD in the predictor variables. Both measures of SES were significantly related to all health outcomes in the expected way, in that high SES was associated with good health. The ladder was more strongly related to all the health outcomes except physical functioning in women (t = 0.86, p = .39). The lower half of Table 3 shows results where both objective and subjective SES were entered simultaneously in the regression model in order to allow comparison of their unique effects. The partial regression coefficients in Table 3 clearly show that subjective SES continues to show a significant relationship with all health outcomes, whereas the relationship between objective SES and many of the health outcomes, particularly in men, becomes nonsignificant.
Table 4 presents the mean health scores in Phases 5 and 6 in men and in women. The raw difference scores (Phase 6 − Phase 5) in the health outcomes is also shown. A paired t test was used to compare the difference in means for the health outcomes (Table 4). In men, there was a significant decline in physical functioning but significant improvement in mental functioning and level of psychological distress; there was no change in self-rated health. The results in women were similar, with an additional significant improvement in self-rated health. However, these global analyses do not allow an assessment of whether these changes are similar across the socioeconomic spectrum.
Results from analyses aimed at assessing socioeconomic patterning in the change in health status between Phases 5 and 6 are shown in Table 5. These analyses also examine if objective and subjective measures of SES predict change in health independently of each other (Table 5, bottom half). Negative change scores on the PCS and MCS and positive change scores on GHQ and self-rated health denote declines in health status. The top half of Table 5 shows both measures of SES to be significantly associated with change in health status, the only exception being psychological distress (GHQ score) among women as a function of grade. Results show that low SES is associated with greater decline in health. For example, a unit increase in subjective SES (denoting a decrease in SES as l denotes high status and 10, low status) is associated with a quarter-point decrease in physical functioning in men (regression coefficient = −0.26, 95% confidence interval = −0.38 to −0.13; p < .0001). Similarly, a unit increase in the ladder in men is associated with an increase in psychological distress (regression coefficient = 0.30, 95% confidence interval = 0.19, 0.41; p < .0001).
The lower half of Table 5 shows that when both objective and subjective measures of SES are entered simultaneously in the model to predict change in health status, it is only the latter that continues to be significantly associated with changes in health. In men, each point increase in subjective SES (denoting a decrease in SES) is associated with a decline in physical functioning (regression coefficient = −0.30, 95% confidence interval = −0.45 to −0.14; p < .0001) and mental functioning (regression coefficient = −0.40, 95% confidence interval = −0.58 to −0.22; p < .0001). It is also associated with increase in psychological distress (regression coefficient = 0.33, 95% confidence interval = 0.20 to 0.46; p < .0001) and poorer self-rated health (regression coefficient = 0.04, 95% confidence interval = 0.03 to 0.06; p < .0001). The results in women were similar (Table 5), and again the objective measure of SES was no longer significantly associated with change in health once subjective status was taken into account.
This study replicates other studies in demonstrating the existence of a gradient between objective (1–3) and subjective (6,11,12) measures of SES and health. High SES, whatever the measure, is associated with better health status. The specific research question addressed in this paper is whether subjective SES predicts health better than objective SES. We looked at this issue by examining the association between the measures of SES and health and change in health status over a mean follow-up period of 3 years. Our results show that although both measures of SES predict health and change in health status; it is only subjective SES that does so when both measures of SES are entered simultaneously in the model.
One of the issues examined in this paper is whether the measures of objective and subjective SES were composite or global measures of SES. Correlation analysis using measures of SES from different points in the life course show this to be the case. Education and income are clearly more strongly related to objective SES and the feeling of financial security to subjective SES. Childhood SES and wealth are almost equally related to objective and subjective SES. The results show both these measures to tap into various aspects of socioeconomic circumstances. The stronger association between subjective SES and the feeling of financial security in this middle-aged cohort suggests that subjective SES provides a better assessment of a person’s future prospects, opportunities, and resources.
The results obtained here show that subjective status is strongly associated with all measures of health and functioning examined. PCS and MCS are measures of general health functioning using an instrument that has been extensively tested for reliability and validity. They were designed to be both conceptually and empirically distinct (18). Ware and colleagues (18) see PCS as reflecting physical morbidity and etiology, whereas MCS reflects psychological or mental morbidity and etiology. The association between subjective status and mental health status is further evident in its relationship with the GHQ. Self-rated health is considered to be an excellent measure of general health and is a strong predictor of mortality (22,23).
The overall analysis of change in health status (not adjusted for age or SES) showed a decline only in physical functioning over the 3-year follow-up period. In fact, the other measures show improved health (mental functioning, psychological distress, and self-rated health in women) over the 3-year period. However, this is not a true picture of the change in health over the 3 years. Analyses, adjusted for age, reveal that change in health over the 3-year period is socially patterned. Lower SES is associated with increasing declines in health status, whereas higher-SES individuals show either an improvement or no deterioration, depending on the health outcome. Despite the global nature of both measures, our results show subjective SES to be a better predictor of decline in health over time.
The precise mechanisms through which subjective status influences health remain unclear; as such, the results showing subjective SES to be a better predictor of health could be explained by three competing hypotheses.
1. Subjective SES Is a More Precise Measure of Social Position
If subjective status represents the cognitive average of various markers of SES (5,6), it might reflect an individual’s social position more accurately, perhaps by taking into account the past and the perceived future prospects. We have argued earlier that it is likely that subjective status reflects an individual’s sociocultural circumstance more fully than objective measures of social class (6). It is likely that subjective status has a multidimensional quality similar to that of self-rated health (22) in that it is a better synthesis of the different elements of SES at the individual level. Subjective status would thus present an individual’s socioeconomic trajectory better than a global measure of objective SES.
It is also probable that subjective assessments of social status allow a more nuanced judgment of objective indicators of social status and related life chances. For example, individuals may be given the same objective rating based on college graduation, but those graduating from more prestigious universities might rate their subjective status as being higher, possibly due to better life chances. This hypothesis falls very clearly under the domain of social causation where social structure is seen to influence health outcomes. Thus, it is likely that an individual’s assessment of their own SES allows them to account for their unique circumstances, reflecting not only current socioeconomic circumstances but also an assessment of the their past (socioeconomic, educational, and economic background) along with their future prospects. This interpretation of the results needs to be made in light of the fact that subjective SES was measured on a 10-point scale compared with the 6-point scale used for objective SES, inherently allowing greater differentiation.
2. Subjective SES Measures Hierarchical Rank: The Results Provide Support for the Hierarchy-Health Relationship
This second explanation differs from the first one in that it advances the idea that subjective status is important because it implicates mechanisms related to one’s place in the social hierarchy. Therefore, subjective status reflects a person’s “relative” social position as opposed to “absolute” social position. Conceptually, the discrepancy between objective and subjective status can be seen to be related to the notion of relative social position. Hierarchical rank is seen to influence health in two ways: direct effects on physiological processes and neuroanatomic structures, leading to an increase in biological vulnerability to disease; and indirectly through unhealthy behaviors (8).
3. Reverse Causation/Common Method Variance
The association between subjective SES and health could be spurious. It is possible that an individual’s assessment of their own SES is mediated by their health status or that both subjective status and health ratings are affected by a common underlying variable, such as response bias or other individual variable, for example a personality variable. Other studies have provided evidence that response bias is not a major explanation (12). The current study adds further evidence that the association is unlikely to be primarily due to response bias or reverse causation. The analyses reported in this paper are longitudinal, with status measured at Phase 5 (1997–1999) and health measured at Phases 5 and 6 (2001–2002) of data collection, offering some protection against reverse causation, as well as common method variance.
The first two hypotheses are not mutually exclusive, but they emphasize different aspects of subjective status. Our research design does not allow us to determine which, if either, is more accurate or important. The first interpretation sees subjective SES as measuring SES more comprehensively, whereas the second interpretation is more closely associated with Wilkinson’s (8) analysis on the importance of relative income (hence relative SES) in explaining the link between income inequality and health. The first interpretation of the results is in light of the life-course perspective of inequalities in health (24). The etiological framework within this perspective highlights the importance of risk factors over the course of an individual’s life in order to explain adult health status. There is a growing body of research that childhood socioeconomic circumstances have an independent effect on adult health (25). A life-course perspective of the results showing subjective social position to be a better predictor of health would be based on the fact that subjective status allows the individual to take account of the socioeconomic circumstances of their life better than objective status. It is hypothesized that the causal sequence linking social status to health implicates multiple pathways: social, cultural, psychological, and economic (26). The way in which the conceptualization of social status, as a better measure of social status or a measure of relative social position, influences the exploration of these pathways requires further research. An examination of the factors that predict discrepant subjective and objective status will also shed further light on the relationship between social status and health.
There are obvious caveats to the conclusions drawn in the paper. First, data are from a white-collar cohort that does not represent the entire socioeconomic spectrum in the United Kingdom. Particularities of the Whitehall II cohort are also likely to affect the results; for example, the relative stability of employment of the civil servants. Second, although the health measures used have proven validity, conclusions about the importance of subjective SES to health can only be made with an examination of objective health outcomes. Future waves of data collection in the Whitehall II study will allow this question to be explored. However, confidence in our findings is enhanced by the results being similar for both the prediction of health and the change in health over time.
The authors thank all participating civil service departments and their welfare, personnel, and establishment officers; the Occupational Health and Safety Agency; the Council of Civil Service Unions; all participating civil servants in the Whitehall II study; and all members of the Whitehall II study team.
1. Davey Smith G, Bartley M, Blane D. The Black report on socioeconomic inequalities 10 years on. BMJ 1990;301:373–7.
2. Fox JA. Health Inequalities in European Countries. Aldershot: Gower Publishing Company Limited; 1989.
3. Townsend P, Davidson N, Whitehead M. Inequalities in Health: The Black Report and the Health Divide. London: Penguin Books; 1992.
4. Segal DR, Segal MW, Koke D. Status inconsistency and self-evaluation. Sociometry 1970;33:347–57.
5. Webster M, Driskell JE. Status generalization: a review and some new data. Am Sociol Rev 1978;43:220–36.
6. Singh-Manoux A, Adler NE, Marmot MG. Subjective social status: its determinants and its association with measures of ill-health in the Whitehall II study. Soc Sci Med 2003;56:1321–33.
7. Jackman MR. The subjective meaning of social class identification in the United States. Public Opin Q 1979;43:443–62.
8. Wilkinson RG. Income distribution and life expectancy. BMJ 1992;304:165–8.
9. Kawachi I, Kennedy BP, Wilkinson RG. Income Inequality and Health: A Reader. New York: New Press; 1999.
10. Wilkinson RG. Unhealthy Societies: The Afflictions of Inequality. London: Routeledge; 1996.
11. Adler N, Epel ES, Castellazzo G, Ickovics JR. Relationship of subjective and objective social status with psychological and physiological functioning: preliminary data in healthy white women. Health Psychol 2000;19:586–92.
12. Operario D, Adler NE, Williams DR. Subjective social status: reliability and predictive utility for global health. Psychol Health 2004;19:237–46.
13. Ostrove JM, Adler NE, Kuppermann M, Washington AE. Objective and subjective assessments of socioeconomic status and their relationship to self-rated health in an ethnically diverse sample of pregnant women. Health Psychol 2000;19:613–8.
14. Bartley M, Sacker A, Firth D, Fitzpatrick R. Understanding social variation in cardiovascular risk factors in women and men: the advantage of theoretically based measures. Soc Sci Med 1999;49:831–45.
15. Marmot MG, Davey Smith G, Stansfeld S, Patel C, North F, Head J, White I, Brunner E, Feeney A. Health inequalities among British civil servants: the Whitehall II study. Lancet 1991;337:1387–93.
16. Cantril H. The Pattern of Human Concerns. New Brunswick, NJ: Rutgers University Press; 1965.
17. Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey: Manual and Interpretation Guide. Boston, MA: The Health Institute, New England Medical Centre; 1993.
18. Ware JE, Kosinski M, Keller SD. SF-36 Physical and Mental Health Summary Scales: A User’s Manual. Boston, MA: The Health Institute, New England Medical Centre; 1994.
19. Ware JE, Kosinski M, Bayliss MS, McHorney CA, Rogers WH, Raczek A. Comparison of methods of scoring the statistical analysis of SF-36 health profile and summary measures: summary of results from the Medical Outcomes Study. Med Care 1995;33:264–79.
20. Goldberg DP. The Detection of Psychiatric Illness by Questionnaire (Institute of Psychiatry Maudsley Monographs No. 21). London: Oxford University Press; 1972.
21. Cohen J, Cohen P. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Hillsdale, NJ: Erlbaum; 1983.
22. Borg V, Kristensen TS. Social class and self-rated health: can the gradient be explained by differences in life style or work environment? Soc Sci Med 2000;5:1019–30.
23. Burström B, Fredlund P. Self rated health: is it as good a predictor of subsequent mortality among adults in lower as well as in higher social classes? J Epidemiol Community Health 2001;55:836–40.
24. Kuh D, Ben-Shomo Y. Introduction: a lifecourse approach to the aetiology of chronic disease. In: Kuh D, Ben-Shlomo Y, editors. A Lifecourse Approach to Chronic Disease Epidemiology. Oxford: Oxford University Press; 1997:3–14.
25. van de Mheen H, Stronks K, Looman CWN, Mackenbach JP. Does childhood socioeconomic status influence adult health through behavioural factors? Int J Epidemiol 1998;27:431–7.
26. Singh-Manoux A. Commentary: psychosocial factors and public health: another point of view. J Epidemiol Community Health 2003;57:563–4.