Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Aug 5.
Published in final edited form as: Soc Sci Med. 2012 Aug 10;75(11):2007–2012. doi: 10.1016/j.socscimed.2012.07.036

Glitch in the gradient: Additional education does not uniformly equal better health

Anna Zajacova a,*, Richard G Rogers b, Vicki Johnson-Lawrence c
PMCID: PMC5544818  NIHMSID: NIHMS886859  PMID: 22920276

Abstract

While the relationship between education and general health has been firmly established in the literature, surprisingly little research has analyzed individual components of the global health judgments, such as chronic conditions or pain. We present a systematic account of the health gradient for multiple health outcomes by detailed educational categories among U.S. working-age adults. Using the 1997–2010 National Health Interview Surveys (N = 204,764), we analyze individual health outcomes ranging from cardiovascular disease to vision problems with a series of logistic regression models. The results at the presecondary and baccalaureate levels are consistent with the health gradient. An unexpected finding occurs among adults with some college but no degree, and those with technical/vocational associate degrees: these groups report more pain and a higher prevalence of a broad range of conditions than high school graduates who never attended college. We discuss several explanations for the observed patterns. The findings challenge the broadly accepted educational gradient in health; additionally, the lower postsecondary groups comprise a quarter of American adults. Jointly, there is a clear research and policy impetus to understand the source of this ‘glitch’ in the health gradient.

Keywords: Education, Health gradient, Adults, Health outcomes, USA

Introduction

An extensive body of research has documented the association between education and health. Using primarily self-rated health and all-cause mortality as outcomes, researchers have shown repeatedly that more schooling is linked to better health (Adler et al., 1994; Cutler & Lleras-Muney, 2008; House et al., 1990; Ross & Wu, 1995; Smith, 2004), slower health declines (Lantz et al., 2001; Ross & Wu, 1996), and longer life (Elo & Preston, 1996; Rogers, Everett, Zajacova, & Hummer, 2010; Zajacova & Hummer, 2009). The results have been so pervasive and persuasive that researchers have summarized them as follows: “Health, by any definition and by any measure, increases with the level of education” (Mirowsky & Ross, 2003).

In recent years, the literature on social determinants of health began focusing on understanding the causal relationships between education and health. Researchers have studied mediators like income and health behaviors through which education may affect health (Chandola, Clarke, Morris, & Blane, 2008; Conti & Heckman, 2010; Cutler & Lleras-Muney, 2008), as well as potential confounders like parental socioeconomic status or personality characteristics (Eide & Showalter, 2011; Fujiwara & Kawachi, 2009; Haas & Fosse, 2008). Another line of inquiry aimed to describe the precise shape of the education-health association to discern whether each additional year of schooling matters equally or whether the years associated with earning credentials are more important (Ross & Mirowsky, 1999; Schnittker, 2004; Zajacova, Hummer, & Rogers, 2012). These studies, using newly available data with detailed educational categories, still typically used global measures of health such as self-rated health and generally also found a uniformly positive relationship between schooling and health status.

An important aspect of the education-health research has been curiously neglected: the examination of individual components of the global health status, such as chronic conditions or pain. Health is a complex multidimensional construct and it is critical to understand which specific health problems drive the gradient, both to posit causal explanations and ultimately to design and implement policies to decrease the health inequalities in the population. In the epidemiological literature, there is a sizeable body of research on socioeconomic status (SES) and individual outcomes, such as prevalence rates for specific conditions. Many of these studies, however, used occupation, neighborhood poverty level, income, or other measures to capture the respondents’ place in the SES hierarchy. Moreover, the studies typically targeted a clinically-meaningful population such as the elderly or patients with some illness, rather than the general population.

Among the relatively few studies on the association between education and specific health conditions or problems, most reported the expected inverse link whereby higher education was associated with lower prevalence rates of diabetes (Mezuk, Eaton, Golden, & Ding, 2008; Smith et al., 2011), hypertension (Grotto, Huerta, & Sharabi, 2008; Liu et al., 2011), cardiovascular disease (Winkleby, Kraemer, Ahn, & Varady, 1998), or pain measures such as back pain or migraines (Latza, Kohlmann, Deck, & Raspe, 2004; Le, Tfelt-Hansen, Skytthe, Kyvik, & Olesen, 2011). There were exceptions where the SES gradient was weak or positive, as in some studies of emphysema (Lovasi et al., 2011), melanoma (Harrison, Haque, Roseman, & Soong, 1998), and breast cancer (Vanloon, Goldbohm, & Vandenbrandt, 1994).

Nearly all of these epidemiological analyses, however, either forced a linear specification of education or trichotomized it as presecondary, secondary, and postsecondary. The linear specification only shows the general trend across all levels of schooling. Trichotomizing, especially for the postsecondary level, is problematic at a time when the majority of the population falls into this single broadly-defined category (Planty et al., 2009). Doing so obscures differences between adults with just a few college credits and those with professional/doctoral degrees.

The motivation for this study was to analyze the gradient for multiple health measures—from cardiovascular and chronic respiratory conditions to vision problems and functional limitations—across detailed educational-attainment levels. We examined health gradients for 10 specific outcomes, as well as self-rated health, across a set of nine educational categories in a large, nationally representative sample of working-age adults. We thus contribute both to the epidemiological literature by adding detailed educational attainment information to the analysis of specific health outcomes, as well as to the social determinants literature by focusing on multiple dimensions of the general health status.

Data and methods

Data

The analyses were based on data from the 1997–2010 National Health Interview Surveys (NHIS). The NHIS is an ongoing annual household survey conducted through face-to-face interviews by the National Center for Health Statistics. The NHIS uses a complex multistage stratified sampling design to obtain a sample representative of the civilian non-institutionalized U.S. population. We obtained the data through the Integrated Health Interview Surveys (IHIS), a consolidated source of NHIS data compiled by the Minnesota Population Center (2008). Since a major redesign of NHIS in 1997, the sample design and the core questions have remained identical from year to year. The unconditional response rates for Sample Adults, the group used for the analyses, which comprised a single randomly-selected individual from each sampled household, exceeded 80% in 1997 although it declined to 61% in 2010 (NCHS, 2011).

The analysis sample was defined as U.S.-born adults age 30 to 64 who provided valid education information (N = 204,764). For the lower age threshold, 30 was optimal because over 10% of adults age 25–29 are still enrolled in school so their attainment information is unavailable, whereas fewer than 5% are enrolled past age 30 (Planty et al., 2009). The upper boundary is the typical age of transition to retirement; jointly, they define working-age adults with completed schooling. Among U.S.-born respondents aged 30–64, only 1664 (0.8%) were missing education information, so their exclusion is unlikely to bias the results.

Variables

Educational attainment

Information about schooling was collected identically in all interview waves as the highest completed year up to the 12th grade and as educational credentials for those who completed at least high school. During the interview, respondents were handed a card with a list of educational categories and asked to select the one that best represented their level. Due to the relatively small numbers, we collapsed master’s and doctoral degrees as MA+, grades 0–8 as grade-school level and grades 9–10 and 11–12 as high school dropout categories. The latter group included adults with the General Educational Development (GED) diploma, a group of adults who have been found comparable to high school dropouts in terms of health outcomes (Zajacova, 2011). All other credentials were retained in the original detail (high school = referent). “Some college” includes adults who attended college but did not earn any postsecondary credential. There are two types of associate degrees, both requiring about 60 credit-hours of study. The first is a technical/vocational degree that prepares students directly for specific occupations, such as paralegal, computer or lab technician, medical transcriptionist, or a teacher’s aide. The second type is academic, sometimes referred to as a transfer associate degree, designed to provide the first two years toward a bachelor’s degree.

Health measures

All measures were self-reported by the respondents. We included 10 specific health outcomes and self-rated health. All were dichotomized and coded as 0 if respondent did not report the outcome and 1 if they reported it. The health outcomes included: 1) cardiovascular conditions comprising angina pectoris, coronary heart disease, heart attack, or other heart conditions; 2) respiratory conditions, which included chronic bronchitis and emphysema; 3) hypertension; 4) diabetes; 5) back or neck pain; 6) facial pain or severe migraine; 7) acute conditions comprising a cold or an intestinal illness during the previous two weeks; 8) current vision problems asked in terms of “having any trouble seeing, even when wearing glasses of contact lenses; ” 9) other conditions that included kidney problems, liver disease, and stomach problems or ulcer; and 10) functional limitations, including any difficulty pushing or pulling large objects, shopping or going to the movies or other events, participating in social activities, doing other things for leisure, walking up to 10 steps without resting, or standing for 2 h. Self-rated health was dichotomized and coded 0 if the respondent indicated excellent, very good, or good health, and 1 for fair or poor health. Most conditions were ascertained with the following question: “Have you been told by a doctor or other health professional that you had [this condition]?” The pain questions were phrased as “During the past 3 months, have you experienced pain in [this body part]?” Some conditions (i.e., cardiovascular) were combined because of their relatively low prevalence and to simplify the presentation of findings.

Tetrachoric correlations among all 55 pairs of outcomes were low to moderate, averaging r = 0.32. Only a single correlation—between functional limitations and self-rated health—exceeded 0.6 (tetrachoric r = 0.65; all correlation results available on request). This indicates that each outcome carried unique information about the respondents’ health, justifying the estimation of the multiple models.

Controls

All logistic models controlled for age (in single years, centered on the sample mean of 46 years), sex, race (non-Hispanic white = referent, non-Hispanic black, Hispanic, and other), and survey year (centered on the mean year of 2003).

Analysis

We estimated a series of logistic regression models of each health outcome on education, controlling for basic covariates. The estimation adjusted for the complex sampling design using the svy suite in Stata 11 (StataCorp, 2009).

In addition to models shown here, we also estimated a comparable series of models stratified by sex, age (two age groups, 30–44 and 45–64), race (white and nonwhite), and time period (1997–2003 and 2004–2010). The aim was to ensure that the findings were consistent across groups, which could be obscured in the total-sample models. These auxiliary analyses showed that the findings and conclusions presented below held for all major population subgroups. The results are available on request.

Results

Table 1 shows the distribution of the educational attainment in the sample, as well as the prevalence of all health outcomes and basic sample characteristics by education. The modal educational attainment was a high school diploma (26%); nearly 31% of the sample was classified as ‘subbaccalaureate’ segment, encompassing some college or an associate degree. Another 30% earned at least a bachelor’s degree. The prevalence ranged from about 6% for respiratory conditions, comprising chronic bronchitis and emphysema, to 36% for back or neck pain.

Table 1.

Sample characteristics and prevalence of health conditions, by education.

Total Less than
high school		 High school
diploma		 Some college
or AA		 Bachelor’s
or more
Education
  0–8 years 2.2%
  9–10 years 3.6%
  11–12 years or GED 7.6%
  High school diploma 26.1%
  Some College 19.3%
  AA – vocational/technical 7.5%
  AA – academic 3.9%
  BA/BS 19.1%
  MA/MS or higher degree 10.7%
Health conditions
  Cardiovascular 10.0% 15.4% 9.8% 10.2% 7.7%
  Respiratory 6.2% 11.9% 6.3% 6.5% 3.3%
  Hypertension 26.0% 35.5% 28.1% 25.9% 19.9%
  Diabetes 7.7% 12.9% 8.5% 7.7% 4.6%
  Back or neck pain 36.1% 45.4% 35.8% 38.5% 29.7%
  Migraine or facial pain 19.9% 26.2% 18.7% 21.9% 16.2%
  Acute conditions 16.5% 20.8% 15.8% 16.8% 14.9%
  Vision problems 9.9% 16.5% 9.8% 10.2% 6.8%
  Other conditions 10.8% 17.8% 10.7% 11.4% 6.9%
  Functional limitations 33.2% 49.3% 34.5% 34.6% 23.6%
  Fair or poor SRH 12.4% 31.2% 13.8% 10.9% 4.3%
Control variables
Age – mean (s.e.) 46.0 (0.03) 47.3 (0.07) 46.7 (0.05) 45.5 (0.05) 45.4 (0.05)
Female 54.0% 52.0% 53.5% 56.6% 52.6%
Race
  White 81.1% 69.7% 80.4% 80.1% 87.9%
  Black 12.8% 20.4% 14.1% 13.5% 7.6%
  Hispanic 4.4% 8.0% 4.2% 4.7% 2.6%
  Other 1.7% 1.9% 1.4% 1.7% 1.8%
Open in a new tab

Note: AA = associate degree, BA/BS = bachelor’s degree, MA/MS = master’s degree. Cardiovascular conditions include angina pectoris, heart attack, coronary heart disease and other heart conditions.

Respiratory conditions include chronic bronchitis and emphysema.

“Other” conditions include kidney or liver problems, or stomach problems or ulcers. Source: NHIS 1997–2010, U.S.-born adults age 30–64, N = 204,764. Adjusted for sampling design.

Table 2 displays results for demographics-adjusted logistic models of ten health outcomes and self-rated health. For parsimony, the table omits precise p-values and confidence interval information, which are available on request. The reference category is high school diploma. Results can be easily recalculated for a different referent (for instance, master’s degree or more) by dividing each odds ratio by the odds ratio of the desired referent (i.e., for cardiovascular disease, OR = 0.79 for the MA + category in the original models so adults with 0–8 years of schooling have 2.16/0.79 = 2.75 times the odds of reporting a cardiovascular condition, compared to adults with the highest education.)

Table 2.

Logistic models of health outcomes on education.

Cardio-
vascular		 Respiratory Hyper-
tension		 Diabetes Back or
neck pain		 Migraine/facial
pain		 Acute
conditions		 Vision
problems		 Other
conditions		 Function.
limitations		 Poor or
fair SRH
Education
  0–8 years 2.16*** 2.75*** 1.46*** 1.85*** 1.71*** 2.24*** 1.60*** 2.38*** 2.34*** 2.64*** 4.92***
  9–10 years 1.71*** 2.16*** 1.38*** 1.44*** 1.64*** 1.69*** 1.52*** 1.78*** 1.88*** 2.09*** 3.19***
  11–12 years 1.46*** 1.84*** 1.26*** 1.33*** 1.44*** 1.47*** 1.33*** 1.60*** 1.62*** 1.63*** 2.05***
High school
  Some college 1.14*** 1.12*** 0.97 0.98 1.15*** 1.16*** 1.07*** 1.09*** 1.13*** 1.07*** 0.84***
  AA - technical 1.18*** 0.99 0.97 0.94 1.13*** 1.18*** 1.04 1.07 1.09* 1.06** 0.78***
  AA - academic 0.99 0.86* 0.92** 0.85** 0.99 1.06 1.02 0.92 1.00 0.88*** 0.54***
  BA/BS 0.84*** 0.55*** 0.71*** 0.59*** 0.74*** 0.79*** 0.92*** 0.72*** 0.65*** 0.63*** 0.33***
  MA/MS or more 0.79*** 0.47*** 0.65*** 0.54*** 0.74*** 0.81*** 0.92** 0.67*** 0.61*** 0.56*** 0.25***
Control variables
Age 1.06*** 1.03*** 1.07*** 1.07*** 1.01*** 0.98*** 0.99*** 1.03*** 1.03*** 1.05*** 1.05***
Female 0.98 1.85*** 0.86*** 0.93*** 1.22*** 2.55*** 1.17*** 1.38*** 1.09*** 1.52*** 1.13***
Race
  Black 0.95* 0.77*** 1.96*** 1.80*** 0.70*** 0.79*** 0.94* 1.14*** 0.89*** 0.94*** 1.81***
  Hispanic 0.76*** 0.66*** 1.02 1.75*** 0.84*** 0.90*** 0.92** 1.07* 0.90** 0.89*** 1.29***
  Other 1.16* 1.24** 1.41*** 2.03*** 0.96 1.10* 1.05 1.44*** 1.24*** 1.20*** 1.69***
Survey year 1.00 1.00 1.03*** 1.05*** 1.00 1.01*** 1.00 1.01*** 0.99** 1.01*** 1.03***
Open in a new tab
*

p < .05

**

p < .01

***

p < .001.

Note: AA = associate degree, BA/BS = bachelor’s degree, MA/MS = master’s degree. Source: NHIS 1997–2010, U.S.-born adults age 30–64, N = 204,764. Adjusted for sampling design.

In all 11 models, adults with less than high school diploma were more likely to report the given health problem than high school graduates. There was a step-like pattern where the odds ratios for adults with 0–8 years of schooling were greater than the odds ratios for adults with 9–10 years, which were in turn greater than the odds ratios for adults with 11 or 12 years of schooling. This pattern was consistent with the health gradient. At the baccalaureate and higher educational levels, results also corroborated the expected gradient. For all 11 outcomes, respondents with at least a bachelor’s degree reported significantly lower odds of health problems than high school graduates.

The results were quite different at the subbaccalaureate level, which included adults with some college but no diploma and those with associate degrees. College dropouts were significantly more likely or at least equally likely to report all conditions except fair or poor self-rated health, compared to high school graduates. For instance, the odds of reporting a cardiovascular condition and back or neck pain were 14% and 15% higher, respectively, among college dropouts than the odds for high school graduates. The only exception was self-rated health: compared to the odds for high school graduates, the odds of reporting poor or fair SRH were 16% lower for those with some college.

Adults with a technical/vocational associate degree were fairly similar to college dropouts in that they also reported more or at least a similar level of health problems as high school graduates. Interestingly, adults with an academic associate degree tended to report a lower prevalence of health problems than their technical/vocational associate degree peers. For about half of the outcomes, the odds were comparable between the academic associate degree level and the high school reference; for another half of the outcomes, the respondents with an academic associate degree had lower odds than high school graduates.

We also estimated sex-, age-, race-, and period-stratified models that are not shown but are available on request. Excepting self-rated health, there were 80 models of health outcomes (10 outcomes estimated for eight groups), each with two coefficients associated with the ‘some college’ and technical/vocational associate degree levels. Of those 160 coefficients, only a single one (for hypertension in race-stratified models among white adults with ‘some college’ education) was consistent with the gradient in that it showed adults with some college to have lower odds of reporting this condition than high school graduates.

Discussion

This study determined the health gradient for multiple specific outcomes across detailed educational categories. While numerous articles have studied the association between education and global health status, typically measured with self-rated health, surprisingly little research has examined the gradient for individual health conditions. This omission is problematic if we want to truly understand the links between schooling and health. Health is a complex construct, so disaggregating the global measures into their individual components, such as chronic conditions or pain, is necessary to understand what drives the overall patterns.

For most educational levels, our findings corroborated literature and corresponded to the health gradient. Adults who did not complete high school reported uniformly more health problems than high school graduates. Adults with a bachelor’s or master’s degree reported considerably fewer health problems than high school graduates.

The major finding was a ‘glitch’ in the health gradient for adults with some college education, as well as recipients of technical/vocational associate degrees. Although these groups had more schooling than high school graduates, they reported at least as many health problems and conditions—ranging from cardiovascular and respiratory conditions to acute illness, pain, and functional limitations—as the high school graduates. This unexpected result prompts several questions. First, why do these lower-postsecondary groups fail to conform to the health gradient? And second, why do they report a similar or higher prevalence of all the studied conditions than high school graduates, but better self-rated health?

One reason why adults with some postsecondary education report more conditions than high school graduates might be due to reporting bias: adults with additional education may be more likely to visit a doctor, receive a diagnosis, and remember it. However, pain measures were among the strongest outcomes in terms of the ‘glitch,’ and the respondents do not need a doctor’s diagnosis to report pain. An additional argument against the reporting bias as an explanation for the ‘glitch’ is that if the results were due to a differential knowledge about conditions, we could expect adults with less than high school to visit doctors less and be less knowledgeable than high school graduates. This would imply a lower prevalence rate of the reported conditions at the pre-secondary levels, while we found the opposite, a significantly higher prevalence, fully consistent with the health gradient.

Assuming the ‘glitch’ does not result from reporting differences, we may want to understand how much of this pattern is causal from schooling to health, versus confounded by unobserved characteristics of the subbaccalaureate groups, or even resulting from reverse causality whereby health problems prevented the college dropouts from reaching their educational goals or impacted the choice to earn a technical/vocational associate degree. Although the data and space constraints do not permit analyzing these questions in a systematic way, we draw on existing literature to posit several explanations, which can be tested in future studies.

When considering the possibility that the subbaccalaureate educational attainment somehow causes the worse-than-expected health outcomes, we would want to understand whether the ‘glitch’ may be explained by differences in the major pathways known to convert educational attainment into health. Extensive literature (Adler & Ostrove, 1999; Chandola et al., 2008; Cutler & Lleras-Muney, 2008; Eide & Showalter, 2011; Lantz et al., 2001; Mirowsky & Ross, 1998, 2003; Ross & Wu, 1995) has documented that health behaviors such as smoking, economic characteristics including employment, income, and wealth, as well as psychosocial factors all contribute to explaining the links between education and health. While a thorough examination of such pathways is beyond the scope of the current paper, some preliminary analyses suggest that the ‘glitch’ is not explained by smoking, obesity, alcohol use, employment status, family income, and a small number of additional potential explanatory factors. We strongly urge further research in this direction.

We do not expect reverse causality to be a driving explanatory factor of the observed ‘glitch’ for two reasons. First, while poor childhood health may impact educational attainment (Haas & Fosse, 2008; Jackson, 2009), the effects are relatively weak (Case, Fertig, & Paxson, 2005; Palloni, Milesi, White, & Turner, 2009; Smith, 2009). Additionally, extensive educational literature spanning decades has identified critical determinants of educational attainment to be family background and cognitive and noncognitive skills; however, early health does not appear in this literature at all (Cameron & Heckman, 1999, p. 89; Kao & Thompson, 2003; Marjoribanks, 2005; Ou & Reynolds, 2008; Portes & Wilson, 1976; Sewell, Haller, & Ohlendorf, 1970). Most importantly, it is unlikely that the reverse causation would act selectively on only two educational groups (college dropouts and those with technical/vocational associate degrees). Second, the glitch appears for acute conditions like colds or intestinal problems experienced during two weeks preceding the interviews, and for pain experienced during the three months preceding the interview. We realize, of course, that we cannot definitively exclude the possiblitity of long-term compromised health status that would first curtail the respondents’ educational trajectories and later also cause the presence of these health outcomes. The reverse-causality explanation could be tested if the data included the age of onset for the conditions or information about childhood and adolescent health, which we encourage NCHS to collect in the future.

Finally, some of the ‘glitch’ could be due to confounding, meaning that unobserved characteristics drive both educational attainment and health outcomes. Compared to students attending 4-year colleges, subbaccalaureate students come from more disadvantaged backgrounds in terms of their parents’ lower income and education and they perform worse on standardized tests, suggesting lower cognitive skills (Bailey et al., 2003). Interestingly, although we found differences between the two types of associate degrees, the literature suggests that students pursuing these two types of credentials are rather similar in sociodemographic characteristics (Hudson, Kienzl, & Diehl, 2007). The influence of skills and background, and possibly of the institutions where the degrees are earned, may be explored using data sources like the National Longitudinal Surveys of Youth.

The second question raised by the unexpected ‘glitch’ pertains to the discrepancy between self-rated health and all other outcomes: why do the adults with some college and technical associate degrees report more health problems but better self-rated health? The self-rated health judgment likely includes conditions and health problems that we did not study, such as mental-health indicators. If the omitted measures had lower prevalence in the subbaccalaureate groups, that could help explain the discrepancy. Future studies could analyze additional measures, like depression scores, to test this possibility. Alternatively, the lower-postsecondary groups may form health judgment in a fundamentally different way than other groups. Several recent studies (Dowd & Zajacova, 2010; Zajacova & Dowd, 2011) suggested that SES groups differ in the way they assess their health but they do not provide detail that would allow us to resolve the discrepancy. This could also be a fruitful avenue for future research.

We mention three limitations. First, all conditions were self-reported. Self-reports of health status are widely accepted in population health research; nonetheless, they may vary across respondents with different levels of education. Future studies could address the possible reporting issues by using such objective outcomes as biomarkers, which are not available in NHIS. A second limitation is that the data did not include information on the quantity or quality of schooling. If we knew how many college credits the dropouts earned, for instance, we could determine whether more schooling at this level was associated with fewer health problems. Quality of schooling information would also be desirable; perhaps the quality of education among the college dropouts was lower than among those who graduated, compounding the disadvantage of fewer years by lesser quality. And finally, the data did not allow us to explore why the academic and technical/vocational associate degrees differ. An academic associate degree is sometimes considered a ‘transfer’ degree, with recipients expected to continue to a 4-year degree. In that respect, they are more similar to college dropouts in that they did not complete the intended bachelor’s degree; however, our findings showed that they were better off than recipients of the technical/vocational degree. Alternatively, technical/vocational associate degree may train for more physically demanding careers, compared to academic associate, which could lead to chronic back pain and more functional limitations. Occupational data could be used in future studies to explore whether the discrepancy in the two types of associate degree can be explained by different employment paths.

Conclusion

Although higher education is generally associated with better health, our study uncovered an unexpected glitch in the gradient: adults with lower postsecondary educational attainment report more or at least no fewer health problems than high school graduates who never attended college. This lower-postsecondary group represents about a quarter of the total adult population and is expected to grow over time (Crissey, 2009). Our findings challenge the broadly accepted educational gradient in health and underscore the need for both researchers and policymakers to examine detailed educational categories in public-health research.

Acknowledgments

We thank the Eunice Kennedy Shriver NICHD-funded University of Colorado Population Center (grant R24 HD066613) for administrative and computing support, and the National Center for Health Statistics (NCHS) for collecting the data and making the linked files available to the research public. We also thank the Minnesota Population Center for creating and maintaining the Integrated Health Interview Series. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of NIH, NICHD, or NCHS.

References

  1. Adler N, Boyce T, Chesney MA, Cohen S, Folkman S, Kahn RL, et al. Socioeconomic status and health: the challenge of the gradient. American Psychologist. 1994;49:15–24. doi: 10.1037//0003-066x.49.1.15. [DOI] [PubMed] [Google Scholar]
  2. Adler NE, Ostrove JM. Socioeconomic status and health: what we know and what we don’t. Annals of the New York Academy of Sciences. 1999;896:3–15. doi: 10.1111/j.1749-6632.1999.tb08101.x. [DOI] [PubMed] [Google Scholar]
  3. Bailey TR, Leinbach DT, Scott M, Alfonso M, Kienzl GS, Kennedy B, et al. The characteristics of occupational sub-baccalaureate students entering the new millennium. Washington, DC: National Assessment of Vocational Education, US Department of Education; 2003. Report produced under U.S. Department of Education Contract No. ED-00-CO-0023. [Google Scholar]
  4. Cameron SV, Heckman JJ. The dynamics of educational attainment for Blacks, Hispanics, and White. Washington, DC: National Bureau of Economic Research; 1999. [Google Scholar]
  5. Case A, Fertig A, Paxson C. The lasting impact of childhood health and circumstance. Journal of Health Economics. 2005;24:365–389. doi: 10.1016/j.jhealeco.2004.09.008. [DOI] [PubMed] [Google Scholar]
  6. Chandola T, Clarke P, Morris JN, Blane D. Pathways between education and health: a causal modelling approach. Journal of the Royal Statistical Society: Series A (Statistics in Society) 2008;169:337–359. [Google Scholar]
  7. Conti G, Heckman JJ. Understanding the early origins of the education–health gradient. Perspectives on Psychological Science. 2010;5:585–605. doi: 10.1177/1745691610383502. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Crissey SR, editor. Educational attainment in the United States: 2007. Washington, DC: U.S.C. Bureau; 2009. [Google Scholar]
  9. Cutler DM, Lleras-Muney A. Education and health: evaluating theories and evidence. In: Schoeni RF, House JS, Kaplan GA, Pollack H, editors. Making Americans healthier: Social and economic policy as health policy. New York, NY: Russel Sage Foundation; 2008. pp. 29–60. [Google Scholar]
  10. Dowd JB, Zajacova A. Does self-rated health mean the same thing across socioeconomic groups? Evidence from biomarker data. Annals of Epidemiology. 2010;20:743–749. doi: 10.1016/j.annepidem.2010.06.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Eide ER, Showalter MH. Estimating the relation between health and education: what do we know and what do we need to know? Economics of Education Review. 2011;30(5):778–791. [Google Scholar]
  12. Elo IT, Preston SH. Educational differentials in mortality: United States, 1979–85. Social Science & Medicine. 1996;42:47–57. doi: 10.1016/0277-9536(95)00062-3. [DOI] [PubMed] [Google Scholar]
  13. Fujiwara T, Kawachi I. Is education causally related to better health? A twin fixed-effect study in the USA. International Journal of Epidemiology. 2009;38:1310–1322. doi: 10.1093/ije/dyp226. [DOI] [PubMed] [Google Scholar]
  14. Grotto I, Huerta M, Sharabi Y. Hypertension and socioeconomic status. Current Opinion in Cardiology. 2008;23:335–339. doi: 10.1097/HCO.0b013e3283021c70. 310.1097/HCO.1090b1013e3283021c3283070. [DOI] [PubMed] [Google Scholar]
  15. Haas SA, Fosse EN. Health and the educational attainment of adolescents: evidence from the NLSY97. Journal of Health and Social Behavior. 2008;49:178–192. doi: 10.1177/002214650804900205. [DOI] [PubMed] [Google Scholar]
  16. Harrison RA, Haque AU, Roseman JM, Soong S-J. Socioeconomic characteristics and melanoma Incidence. Annals of Epidemiology. 1998;8:327–333. doi: 10.1016/s1047-2797(97)00231-7. [DOI] [PubMed] [Google Scholar]
  17. House JS, Kessler RC, Herzog AR, Mero RP, Kinney AM, Breslow MJ. Age, socioeconomic status, and health. The Milbank Quarterly. 1990;68:383–411. [PubMed] [Google Scholar]
  18. Hudson L, Kienzl G, Diehl J. In: Students entering and leaving postsecondary occupational education: 1995–2001. N.C.f.E. Statistics, editor. Washington, DC: US Department of Education, Institute of Education Sciences; 2007. [Google Scholar]
  19. Jackson MI. Understanding links between adolescent health and educational attainment. Demography. 2009;46:671–694. doi: 10.1353/dem.0.0078. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kao G, Thompson JS. Racial and ethnic stratification in educational achievement and attainment. Annual Review of Sociology. 2003;29:417–442. [Google Scholar]
  21. Lantz PM, Lynch JW, House JS, Lepkowksi JM, Mero RP, Musick MA, et al. Socioeconomic disparities in health change in a longitudinal study of US adults: the role of health-risk behaviors. Social Science & Medicine. 2001;53:29–40. doi: 10.1016/s0277-9536(00)00319-1. [DOI] [PubMed] [Google Scholar]
  22. Latza U, Kohlmann T, Deck R, Raspe H. Can health care utilization explain the association between socioeconomic status and back pain? Spine. 2004;29:1561. doi: 10.1097/01.brs.0000131435.56714.15. [DOI] [PubMed] [Google Scholar]
  23. Le H, Tfelt-Hansen P, Skytthe A, Kyvik KO, Olesen J. Association between migraine, Lifestyle and socioeconomic factors: a population-based Cross-Sectional study. The Journal of Headache and Pain. 2011;12:157–172. doi: 10.1007/s10194-011-0321-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Liu SY, Buka SL, Linkletter CD, Kawachi I, Kubzansky L, Loucks EB. The association between blood pressure and years of schooling versus educational credentials: test of the sheepskin effect. Annals of Epidemiology. 2011;21:128–138. doi: 10.1016/j.annepidem.2010.11.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lovasi GS, Roux AVD, Hoffman EA, Smith LJ, Jiang R, Carr JJ, et al. Socioeconomic status is positively associated with percent emphysema on CT scan: the MESA lung study. Academic Radiology. 2011;18:199–204. doi: 10.1016/j.acra.2010.10.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Marjoribanks K. Family background, academic achievement, and educational aspirations as predictors of Australian young adults’ educational attainment. Psychological Reports. 2005;96:751–754. doi: 10.2466/pr0.96.3.751-754. [DOI] [PubMed] [Google Scholar]
  27. Mezuk B, Eaton WW, Golden SH, Ding Y. The influence of educational attainment on depression and risk of type 2 diabetes. American Journal of Public Health. 2008;98:1480. doi: 10.2105/AJPH.2007.126441. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Minnesota Population Center and State Health Access Data Assistance Center. Integrated health interview series: Version 2.0. Minneapolis MN: University of Minnesota; 2008. [Google Scholar]
  29. Mirowsky J, Ross CE. Education, personal control, lifestyle and health. Research on Aging. 1998;20:415–449. [Google Scholar]
  30. Mirowsky J, Ross CE. Education, social status, and health. New York: Aldine de Gruyter; 2003. [Google Scholar]
  31. NCHS. 2010 National health interview survey (NHIS) public use data release survey description. Centers for Disease Control and Preventions, U.S. DHHS; 2011. [Google Scholar]
  32. Ou SR, Reynolds AJ. Predictors of educational attainment in the Chicago longitudinal study. School Psychology Quarterly. 2008;23:199. [Google Scholar]
  33. Palloni A, Milesi C, White RG, Turner A. Early childhood health, reproduction of economic inequalities and the persistence of health and mortality differentials. Social Science & Medicine. 2009;68:1574–1582. doi: 10.1016/j.socscimed.2009.02.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Planty M, Hussar W, Snyder T, Kena G, KewalRamani A, et al. In: The condition of education 2009 (NCES 2009-081) I.o.E.S. National Center for Education Statistics, editor. Washington, DC: U.S. Department of Education; 2009. [Google Scholar]
  35. Portes A, Wilson KL. Black-White differences in educational attainment. American Sociological Review. 1976;41:414–431. [Google Scholar]
  36. Rogers RG, Everett BG, Zajacova A, Hummer RA. Educational degrees and adult mortality risk in the United States. Biodemography and Social Biology. 2010;56:80–99. doi: 10.1080/19485561003727372. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Ross CE, Mirowsky J. Refining the association between education and health: the effects of quantity, credentials, and selectivity. Demography. 1999;36:445–460. [PubMed] [Google Scholar]
  38. Ross CE, Wu CL. The links between education and health. American Sociological Review. 1995;60:719–745. [Google Scholar]
  39. Ross CE, Wu CL. Education, age, and the cumulative advantage in health. Journal of Health and Social Behavior. 1996;37:104–120. [PubMed] [Google Scholar]
  40. Schnittker J. Education and the changing shape of the income gradient in health. Journal of Health and Social Behavior. 2004;45:286–305. doi: 10.1177/002214650404500304. [DOI] [PubMed] [Google Scholar]
  41. Sewell WH, Haller AO, Ohlendorf GW. Educational and early occupational status attainment process – replication and revision. American Sociological Review. 1970;35:1014–1027. [Google Scholar]
  42. Smith JP. Unraveling the SES: health connection. Population and Development Review. 2004;30(Supplement):108–132. Aging, Health, and Public. [Google Scholar]
  43. Smith JP. The impact of childhood health on adult labor market outcomes. The Review of Economics and Statistics. 2009;91:478–489. doi: 10.1162/rest.91.3.478. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Smith BT, Lynch JW, Fox CS, Harper S, Abrahamowicz M, Almeida ND, et al. Life-course socioeconomic position and type 2 diabetes mellitus. American Journal of Epidemiology. 2011;173:438–447. doi: 10.1093/aje/kwq379. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. StataCorp. Stata statistical software: Release 11.0. College Station, TX: StataCorp LP; 2009. [Google Scholar]
  46. Vanloon AJM, Goldbohm RA, Vandenbrandt PA. Socioeconomic status and breast cancer incidence – a prospective cohort study. International Journal of Epidemiology. 1994;23:899–905. doi: 10.1093/ije/23.5.899. [DOI] [PubMed] [Google Scholar]
  47. Winkleby MA, Kraemer HC, Ahn DK, Varady AN. Ethnic and socioeconomic differences in cardiovascular disease risk factors: findings for women from the third national health and nutrition examination survey, 1988–1994. Journal of the American Medical Association. 1998;280:356–362. doi: 10.1001/jama.280.4.356. [DOI] [PubMed] [Google Scholar]
  48. Zajacova A. Health in working-age Americans: adults with a high school equivalency (GED) diploma are similar to dropouts, not high school graduates. American Journal of Public Health. 2011;102:S284–S290. doi: 10.2105/AJPH.2011.300524. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Zajacova A, Dowd JB. Reliability of self-rated health in US adults. American Journal of Epidemiology. 2011;174:977–983. doi: 10.1093/aje/kwr204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Zajacova A, Hummer RA. Gender differences in education gradients in all-cause mortality for White and Black adults born 1906–1965. Social Science & Medicine. 2009;69:529–537. doi: 10.1016/j.socscimed.2009.06.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Zajacova A, Hummer RA, Rogers RG. Education and health among U.S. working-age adults: a detailed portrait across the full educational-attainment spectrum. Biodemography and Social Biology. 2012;58:40–61. doi: 10.1080/19485565.2012.666122. [DOI] [PubMed] [Google Scholar]

RESOURCES