education, cumulative advantage, and health

27

Send correspondence to J. Mirowsky and C.E. Ross at Department of Sociology and PopulationResearch Center, University of Texas at Austin, 336 Burdine Hall, 1 University Station A1700,Austin, TX 78712-0118; Email [email protected] or [email protected].

Ageing International, Winter 2005, Vol. 30, No. 1, pp. 27-62.

Invited Paper

EDUCATION, CUMULATIVEADVANTAGE, AND HEALTH

JOHN MIROWSKY AND CATHERINE E. ROSS

UNIVERSITY OF TEXAS, AUSTIN

Education’s positive effect on health gets larger as people age. The large socioeco-nomic differences in health among older Americans mostly accrue earlier in adult-hood on gradients set by educational attainment. Education develops abilities thathelp individuals gain control of their own lives, encouraging and enabling a healthylife. The health-related consequences of education cumulate on many levels, fromthe socioeconomic (including work and income) and behavioral (including healthbehaviors like exercising) to the physiological and intracellular. Some accumula-tions influence each other. In particular, a low sense of control over one’s own lifeaccelerates physical impairment, which in turn decreases the sense of control. Thatfeedback progressively concentrates good physical functioning and a firm sense ofpersonal control together in the better educated while concentrating physical im-pairment and a sense of powerlessness together in the less well educated, creatinglarge differences in health in old age.

Education’s positive effect on health grows as people age. The large socio-economic differences in health seen among older Americans develop through-out adulthood. Educational attainment sets the course. The health-related con-sequences of education cumulate on many levels, from the socioeconomic(employment, job quality, earnings, income, and wealth) and behavioral (habitssuch as smoking or exercising, beliefs such as perceived control over one’sown life, personal relationships) to the physiological (blood pressure, choles-terol levels, aerobic capacity), anatomical (body fat, joint deterioration, arte-rial fatty plaque) and perhaps even intracellular (insulin resistance, free radi-cal damage). Many consequences of educational attainment influence eachother, creating self-amplifying feedback, or interact with each other, produc-ing cascading results (Mirowsky & Ross, 2003).

28 Ageing International/Winter 2005

Many things in life produce effects that fade over time as individuals ad-just to their circumstances, tastes and times change, enthusiasms wane, andnew experiences intervene. Education’s effects work the opposite way, grow-ing with time. That is because education transforms the person, putting theindividual’s life on a different track. Education acts as a structural element ofthe individual’s life, like a structural beam in a building. Many other elementsof life depend on education and take shape with respect to it. To fully under-stand education’s positive impact on health one must envision that benefitunfolding across the lifetime. Education’s health-related effects are presentthroughout adulthood. Even if they were small in any one year, they accumu-late and compound over a lifetime, producing ever larger health differencesbetween persons with different levels of education.

We begin by describing the dormancy and rebirth of American interest inthe relationship of health to socioeconomic status. Next we show how thelarge old-age differences in health across levels of education develop overadulthood. We illustrate with a new growth-curve model of the changes inphysical impairment. Then we describe the three properties of education’sconsequences that drive the divergence: permeation (wide ranging benefits),accumulation (summing over a lifetime), and self-amplification (mutually re-inforcing consequences). Finally, we summarize our view on a debate in cur-rent research. We argue that the health advantages or disadvantages accumu-lated over a lifetime do not recede in old age. From the economic level to thecellular, the differences that affect health and survival continue growing.

Differences Growing over Recent History

Socioeconomic differences in health are large and increasing in the UnitedStates, as well as in Great Britain, Canada, and elsewhere (Bartley, Blane, &Smith, 1998; Ross & Mirowsky 2000). For decades American health sciencesacted as if social status has no great bearing on health. The ascendence ofclinical medicine within a culture of individualism partly accounts for thatomission. The clinical setting and the physician’s role as agent for the indi-vidual patient obscures the role of social status in regulating the risk, severity,and consequence of disease. Realistically, clinical medicine cannot changethe social statuses or personal histories that generate and regulate the flow ofdisease and disability. Medicine’s traditional focus on the distinct proximatecauses of specific diseases also deflects attention from forces that create dis-eases of many kinds. Over the last half of the twentieth century, though, chronicdisease research forced science to think more broadly, looking for elementalfactors that contribute to many health problems.

American sociology, epidemiology, and public health also said surpris-ingly little about the effects of social status on health for decades. Partly that’sbecause the effects are so pervasive that socially oriented health scientiststake them for granted. Researchers studying the effects of risk factors such as

Mirowsky and Ross 29

cigarette smoking or obesity or environmental exposure to carcinogens gen-erally find them more common in lower status individuals, households, andneighborhoods. To avoid mistaking the effects of other risks associated withlow status for the particular one under study, researchers typically make statisti-cal adjustments based on education, occupation, or income. There is nothing wrongwith this. Indeed, good scientific practice demands it. Unfortunately, the habit ofadjusting those effects away may have obscured their powerful implication:social status affects just about everything that affects health.

Perhaps American health scientists had an additional reason for payinglittle attention to the effects of social status on health: the unexamined as-sumption that those effects soon would vanish. During the twentieth centurythe advanced industrial nations made enormous progress in public health pro-grams that benefit all citizens, especially workers and the poor. Everyone ben-efits from public supplies of monitored and treated water, the testing and regu-lating of private wells, public sanitary sewers and sewage treatment, regulationof septic systems, removal of trash and garbage to sanitary landfills or incin-erators, rat control, mosquito control, fire control, flood control, safety stan-dards for buildings, environmental and occupational health and safety stan-dards and programs, transportation safety standards and agencies, the regulationof food purity and vitamin content, the evaluation and regulation of productsafety, the evaluation and regulation of dangerous medical interventions, pro-grams that mandate or promote vaccination against childhood infectious dis-eases, and agencies that scan ceaselessly for the outbreaks of epidemics, com-bating them as early as possible. Health scientists know the value andeffectiveness of these systems. Perhaps that knowledge encouraged a com-placent assumption that the disparities in health across social strata were fad-ing and soon would vanish.

In the United States it took a long while before researchers began to ques-tion the disappearance of socioeconomic differences in health. The resultswere a surprise and a wake-up call. Although mortality rates are going down,the differences in mortality rates across social strata are growing (Lauderdale,2001). The size of the increase is larger for newer cohorts. Not only haseducation’s correlation with survival grown, but the growth appears to beaccelerating.

At first the American researchers suspected that the absence of a U.S. na-tional medical care system might explain the growing disparities. When theAmerican scientists turned to the British literature it disabused them of thatidea. The Black report and its offspring showed the same growing socioeco-nomic disparities in morbidity and mortality in Great Britain as in the UnitedStates (Bartley, Blane, & Smith, 1998; Black et al., 1982.) Studies in Canadaand Sweden found it too. Clearly, providing basic medical care to all citizensdid not avert the trend.

Modern nations have instituted increasingly successful programs for pub-lic, occupational and environmental health that protect all members of their


populations, but particularly the persons of lowest status who otherwise wouldbe most at risk. Ironically, the success of those programs creates a residualand growing association between status and health mediated by behaviorswith a strong and irreducible element of personal choice and self-determi-nation. Societies increasingly face health problems with solutions requir-ing personal knowledge, choice, effort, and effectiveness. This does notimply wayward self-destructiveness as the primary cause of modern healthproblems. Rather, it implies that too many individuals lack the tools neededto gain effective control of their own lives. Given those tools they wouldseek health as willingly and effectively as others do (Mirowsky & Ross,1998).

Differences Growing across Adulthood

The socioeconomic differences in health grow within generations as theyage as well as between generations as the historical trends progress. Curi-ously, the developments within generations and the trends between them canobscure each other. In any given year, the oldest segments of the populationhave had the longest for their socioeconomic differences in health to develop,but also lived in times when those differences were smaller and grew slower.Likewise, the youngest segments have had the least time for the differences todevelop, but show faster development of differences over time. Analyses needto address both phenomena together to show either one most clearly (Lauder-dale, 2001; Lynch, 2003).

A growth-curve model of rising physical impairment shows how the largesocioeconomic differences among older Americans develop over the lifecourse. To illustrate, we modeled the six-year changes in physical impairmentin our three-wave survey of Aging, Status, and the Sense of Control withinterviews in 1995, 1998, and 2001 (Mirowsky & Ross, 2003). We measuredimpairment by asking individuals how much trouble they have climbing stairs,kneeling or stooping, lifting or carrying objects under ten pounds like a bagof groceries, preparing meals or cleaning and doing other household work,shopping or getting around town, seeing even with glasses, and hearing (Nagi,1976). For each item we asked, “Would you say no difficulty, some difficulty,or a great deal of difficulty?” We coded responses 0, 1, or 2 respectively, andtook the mean score across the items to make the index.

Our analyses divided the sample into three groups by level of education:less than high school degree (382 persons), high school degree but no collegedegree at the bachelors level or higher (1,533), and college degree or higher(660). Persons age 60 and up were oversampled so that the sample’s medianage is near 55. Within each level of education, a growth-curve model predictsimpairment at the beginning and change in it over the follow-up from age atthe beginning. The age effects can be any combination of linear, quadratic,and cubic functions with significant coefficients. The predictions fit the ob-


servations quite well. (The probability is .721 that the observed deviationsfrom the model’s predictions occurred purely by chance.)

Figure 1 shows a vector graph of the growth curve model. It is a new typeof graph designed to compare changes across age groups in order to see howthe changes might add up (Mirowsky & Kim, 2004). It illustrates vectors asarrows representing the predicted origin and change in level of impairment.To simplify, the figure shows the arrow for every third one-year age group. Italso shows only the arrows for those with a college degree or higher com-pared to those with less than a high school degree.

The impairment vectors in Figure 1 illustrate several important observa-tions about education and health. First, persons with college degrees havelower levels of impairment in every age group. The differences are quite large.At all ages, the persons with less than a high school degree report impairmentlevels roughly equal to those of college educated persons who are 20 to 30years older. (The high school graduates, not shown, report impairment levelssimilar to those reported by college graduates who are 15 to 20 years older.)Second, impairment increases faster over the follow-up for those with lessthan a high school degree than for those with college degrees. At all ages, thearrow above has a steeper upward slope than the arrow below. (The same istrue for high school graduates compared to college graduates.) This is most

Figure 1

Origin and change in physical impairment predicted for Americans with a collegedegree or higher (N = 663) compared to those with less than a high school degree(N = 383), based on a growth-curve model of self reports in 1995, 1998, and 2001.

Less than high school degree

College degreeor higher

Age

20 30 40 50 60 70 80 90

Phy

sica

l Im

pai

rmen

t

0.0

0.2

0.4

0.6

0.8

1.0


pronounced among the young and middle-aged. Each group shows an essen-tially constant rate of increase from the mid-20s up to the mid-60s, with thecollege educated on a flatter slope. This difference in the growth rate progres-sively enlarges the differences in impairment. Third, beyond age 65 the age-specific slopes get more similar, but they get added to very different baselevels of impairment. The differences in impairment still grow, but by pro-gressively smaller amounts.

Taken together, the observations suggest that the large socioeconomic dif-ferences in impairment among older Americans accumulated earlier in life.By definition, a cumulative advantage is a benefit acquired by successiveaddition (O’Rand, 1996). Education’s cumulative health advantage rests onthree underlying phenomena: permeation, accumulation, and amplification.Permeation means that education affects virtually all aspects of life. A rangeof things influenced by education in turn affect health, including habits, inter-personal relationships, family responsibilities, occupational exposures andopportunities, economic sufficiency and security, neighborhood qualities,autonomous and creative activities, and a sense of controlling one’s own life.Accumulation means that education’s effects add up over time. Amplificationmeans that many of education’s consequences influence each other or regu-late each-others effects in ways that enlarge differences (Mirowsky & Ross,2003; Ross, Mirowsky, & Pribesh, 2001).

In the coming sections we describe the phenomena of permeation, accu-mulation, and amplification in the effects of education on health. We alsosummarize and discusses three current and interrelated scientific questionspertaining to cumulative advantage: Do the positive effects of education onhealth continue to grow throughout life, or do they diminish in old age? Doeseducation slow the physical decline at every age or simply delay the inevi-table until a sudden precipitous decline in old age? Does education reduce theamount of the lifetime spent with serious chronic disease and impairment, ordoes it increase that period of burden by keeping alive those who otherwisewould die of their diseases and infirmities? We argue that the differences inhealth across levels of education grow throughout life. The cumulative ben-efits do not suddenly disappear at the end. As a result, the well educated enjoya longer healthy life with less time spent sick or impaired near the end.

Permeation: Education’s Broad Health Benefits

Education acts as a root cause of good health (Mirowsky & Ross, 2003;Phelan et al., 2004). It gives people the resources to control and shape theirown lives in ways that protect and foster health. Schooling builds skills andabilities that individuals can use to achieve a better life, including better health.It also develops a sense of personal control over one’s own life that motivatesattention and effort (Mirowsky & Ross, 1998). To the extent that people wantgood health, education develops the means toward creating that end through


a lifestyle that promotes health. A large part of the reason the well-educatedexperience good health is that they engage in a lifestyle that includes walking,exercising, drinking moderately, and avoiding being overweight and smok-ing (Ross & Wu, 1995; Ross & Mirowsky, 1999). A strong sense of personalcontrol among the well-educated accounts for much of the reason they en-gage in a healthy lifestyle (Mirowsky & Ross, 1998). Education has otherconsequences that link it to health too, and all are beneficial. In particular,education also provides access to full-time, fulfilling work and to adequate,secure income that helps prevent economic hardship (Ross & Van Willigen,1997). Education helps individuals to build and maintain supportive relation-ships and to avoid divorce and single parenthood. Socioeconomic advantagesand supportive relationships also link education to good health. Because allthe pathways between education and health are positive, we could eliminateany one mediator of the relationship between education and health and theassociation would still be a positive one (Mirowsky & Ross, 2003).

Learned Effectiveness

The skills learned in school increase effective agency. Learned effective-ness is the opposite of learned helplessness. Those who feel helpless see littleconnection between their actions and important outcomes in their lives. Theysee little they can do to improve the present or the future. Education helpspeople avoid helplessness by developing personal resources that increase ef-fectiveness.

Formal education builds “human capital”—the productive capacity devel-oped, embodied, and stocked in human beings themselves (Schultz, 1962;Becker, 1964). Formal education develops skills and abilities of general value,as distinct from “firm-specific” ones of value to a particular employer (Becker,1964). An individual who acquires an education can use it to solve a widerange of problems. Some are the problems of productivity that concern em-ployers and economists. Some are problems in which economic prosperity isone of several means toward a more basic end. Health is one of those basicends.

Schooling builds skills and abilities that ultimately shape health in a vari-ety of ways. On the most general level, education teaches people to learn(Hyman, Wright, & Reed, 1975). It develops the skills and habits of commu-nication: reading, writing, inquiring, discussing, looking things ups, and fig-uring things out (Hyman, Wright, & Reed, 1975; Kohn & Schooler, 1982;Nunn et al., 1978; Spaeth, 1976). It develops analytic skills of broad use suchas mathematics, logic, and, on a more basic level, observing, summarizing,synthesizing, interpreting, experimenting, and so on. The more years of school-ing the greater the cognitive development, characterized by flexible, rational,complex strategies of thinking (Pascarella & Terenzini, 1991; Spaeth, 1976).Education teaches people to think logically and rationally, see many sides of


an issue, and analyze problems and solve them. Education also developsbroadly effective habits and attitudes such as dependability, good judgment,motivation, effort, trust, and confidence (Kohn & Slomczynski, 1993). More-over, in school one encounters and solves problems that are progressivelymore difficult, complex, and subtle. The process develops persistence as wellas skill. It instills the habit of meeting problems with attention, thought, ac-tion, and persistence. Thus, education increases both effort and ability, thetwo fundamental components of problem-solving effectiveness (Wheaton,1980). By developing the ability to gather, interpret, and use information,education increases control over things not covered in the curriculum, includ-ing things far in one’s future and perhaps as yet unimagined.

Individual responsibility and structured disadvantage often get contrastedas rival explanations of differences in health. Contemporary health scienceand popular culture alike put heavy emphasis on behaviors and lifestyles.Researchers who see health as a function of a social structure that allocatesresources unequally (e.g., Crawford, 1986) sometimes criticize the view thathealth is determined by lifestyle characteristics such as exercise and smoking(e.g., Knowles, 1977). Such critics rightly argue that emphasizing insight andchoice while ignoring means and circumstance blames individuals for theirown health problems while ignoring the systematic limitations and disadvan-tages that endlessly generate suffering. Neither individual choice nor struc-tured limitation can be ignored. An effective theory of health must recognizethat many essential elements of health require the personal knowledge, in-sight, and will of the individual. No one can decide for another whether or notto smoke cigarettes, exercise, or eat a proper diet. Some medical scientistsview free will as a quaint but misleading concept. None of them has yet in-vented the drug that will make individuals do what those individuals mustchoose to do themselves.

Education is a resource that inheres in the person (Mirowsky & Ross, 2003).The ability to learn, to persist, to communicate, to seek and use information,to figure out the cause of a problem, and to solve it exists within the indi-vidual. As a result, education’s consequences are present in all aspects of lifethroughout the entire lifetime. Those consequences are uniformly positivewith regard to health.

Designing and Implementing a Healthy Lifestyle

Educated, instrumental people merge otherwise unrelated habits and waysinto a healthy lifestyle that consequently behaves as a coherent trait. Educa-tion encourages individuals to acquire information with intent to use it. Thuseducation helps individuals assemble a set of habits and ways that are notnecessarily related except as effective means toward health. Purposeful indi-viduals coalesce a healthy lifestyle from otherwise incoherent or diametricpractices allocated by subcultural forces. Individuals generally tend to do


whatever others like them do, particularly if it distinguishes the people theyidentify with from the ones they do not. Some of those things make healthbetter and some make it worse. For example, men exercise more frequentlythan women; women restrict body weight more closely than men (Hayes &Ross, 1987; Ross & Bird, 1994). Likewise young adults smoke more thanolder adults, but also exercise more (Ross & Wu, 1995). Individuals puttingtogether a healthy lifestyle must adopt the healthy habits of men and women,young and old. In doing so they create positive correlations among traits thatotherwise are uncorrelated or even negatively correlated (Mirowsky & Ross,1998).

Compared to those with little schooling the well-educated are more likelyto exercise, are more likely to drink moderately rather than abstain or drinkheavily, and are less likely to smoke or be overweight (Ross & Wu, 1995;Ross & Mirowsky, 1999). Education encourages and enables people to createa healthy lifestyle from diverse sources. The health behaviors associated withhigher education show little consistent relationship to other sociodemographictraits. Only education consistently correlates positively with healthy behav-iors (Mirowsky & Ross, 1998).

Economic Well-Being and Creative Work

In addition to encouraging and enabling a healthy lifestyle, education alsohelps individuals avoid the stress of economic hardship and oppressive work.Throughout adulthood persons with higher incomes generally have better health.Even so, health is not a commodity that money can buy. Income bolstershealth largely because it helps people avoid stressful economic hardship. Thehealth benefits of greater income occur mostly in the bottom third of socio-economic strata. They essentially disappear in the top third. Economic hard-ship, in the form of difficulty paying bills and buying food, clothes, medicalcare, or other household necessities, mediates much of lower income’s asso-ciation with poorer health. Some of economic hardship’s effect on health re-flects the impact of extreme or prolonged material privation. Most of it, though,reflects the behavioral and physiological responses to threatening and dispir-iting situations. Education moderates the effect of low income on economichardship by improving the ability to manage household resources. The welleducated avoid economic hardship better than others when household incomeis low (Mirowsky & Ross, 1999). For this and similar reasons, education alsomoderates the association between low income and poor health (Mirowsky &Hu, 1996; Mirowsky & Ross, 2003; Schnittker, 2004).

Jobs also are important to health, but not for the reasons previously thought.Lower status workers no longer generally face dangerous or harmful workenvironments, but they often must do work that is oppressive. Education cre-ates good health partly because learned effectiveness helps individuals ac-quire autonomous and creative paid work. Work is physical or mental effort


or activity directed toward the production or accomplishment of something.Employment is paid work. Employment almost always trades some degree offreedom for income. In a market economy everyone needs money to get thingsthey require or want, and most people must work for the money. The balancein that trade depends as much on the amount of freedom given up, and theburden of the work, as it does on the pay.

Often when people think of the burden of work they think of time spent,physical and mental strain endured, risk taken, and harm suffered. The trueburden lies in the denial of self-expression and the inhibition of autonomousaction—the stifling of free will. The burden of employment results from theloss of independent choice and self-generated action. Education lifts this bur-den. It minimizes the loss of independence, maximizes the opportunity forcreative self-expression, and transforms pay from compensation for surren-dered freedom to reward for productive accomplishment. Creativity is theproduction of favorable or useful results in an original and expressive man-ner. Creative work allows and encourages the individual to do a number ofdifferent kinds of things, to do things the individual enjoys, to develop andlearn, and to figure out how to solve problems. The more creative a person’swork or daily activities the better their health. Currently in the United States,adults with full-time jobs have more creative activities, and education increasesthe probability of full-time employment. In addition, whether paid or not, thecreativity of work and daily activities increases greatly with the level of edu-cation (Mirowsky & Ross, 2003).

Accumulation: Summing Effects over Time

Accumulation refers to gathering many smaller effects into a larger one.Some accumulations benefit health and others harm it. Education tends tospeed or advance the beneficial accumulations and slow or delay the detri-mental ones. Accumulation occurs when consequences, once present, tend tostay present. The health-related consequences of education accumulate onmany levels from the socioeconomic down to the cellular (Mirowsky & Ross,2003).

Socioeconomic Accumulation

On the socioeconomic level, individuals generally stay in the same line ofwork, accumulating experience and seniority and gaining access to the moredesirable and lucrative positions. Generally the norms and rules in an organi-zation, occupation, or profession lay out a sequence of stages and discourageeither skipping ahead or going backward in line (Rosenbaum, 1984). As aconsequence, the extrinsic and intrinsic work rewards that influence healthtend to accumulate. Some rewards almost take on a life of their own, becom-ing accumulations of accumulations. For example, pay generally increases


the longer someone works, with each raise a multiple of the previous amount.That means the raises generally get larger as pay increases. As income goesup larger amounts get used to acquire wealth in the form of durable goods,real estate, savings, and investments, which accumulate over time (Crystal &Shea, 1990). Some of that wealth generates income or capital gains, increas-ing the rate of accumulation.

The rising levels of pay, income, and wealth all insulate against economichardship and enhance the sense of control over one’s own life, thereby im-proving health. The relatively small education-based differences in prosperityat the beginning of adulthood grow over the life course. Meanwhile the baselevel of health risk grows with age, so the impact of economic hardship onhealth also tends to grow. A year of economic troubles might multiply the riskof serious impairment or chronic disease by roughly the same percentage forolder persons as for younger ones, but a higher base level of serious healthproblems makes the amount of the increase much greater for older persons.By these and other mechanisms, the accumulating socioeconomic differencesacross levels of education become accumulating health differences (Ross &Wu, 1996; Mirowsky & Ross, 2003).

Behavioral Accumulation

Accumulation happens on the behavioral level too. Humans naturally formhabits, which are recurrent and often unconscious patterns of behavior ac-quired through frequent repetition, established dispositions of the mind orcharacter, and customary manners or practices. Habits relevant to health in-clude physical and social activities, diet, smoking, drinking, and so on. Edu-cation influences many habits and activities by increasing the sense of controlover one’s own life, which is itself a learned and stable world view summinga lifetime of experience (Ross & Mirowsky, 2003). The sense of control influ-ences a person’s habitual approach to risks and problems. The more controlsomeone feels they have the more effort they make to detect and avoid risksand the more actively and practically they respond to problems that arise.That tends to make individuals more effective. In addition, a higher sense ofcontrol generally reduces the psychophysiological distress associated withcrises by reducing the sense of helplessness and hopelessness. The stability ofactivities, health-related behaviors and sense of control create persistent ef-fects present throughout long portions of life (Mirowsky & Ross, 2003).

Humans also form sustaining personal relationships, which can be viewedas habits of association and interaction. Relationships tend to be stable overtime. Education generally improves the quality of relationships, making themseem to the participants more fair, equal, respectful, and sustaining. Partlybecause of that, education makes marital relationships more enduring, increas-ing the prevalence of marriage and reducing the risk of exposure to the stressof divorce. Marriage, marital quality, and social support all contribute to health.


To the extent that education’s health-promoting interpersonal consequencespersist they contribute to its accumulating health benefits (Berkman et al.,2000; Mirowsky & Ross, 2003).

Biological Accumulation

The socioeconomic and behavioral accumulations necessarily influencehealth through biological mechanisms, many of which accumulate too. Healthscience knows of many biological accumulators that affect health, and prob-ably will discover many more. The accumulation of body fat may be the mostobvious one, and the best known because of its place in popular culture. Itserves as a good example of biological accumulators in general.

Body Fat: A Typical Bio-Accumulator. The buildup of body fat happensslowly over a period of years. An individual who eats ten calories a day morethan he or she burns will accumulate a pound of fat a year. Start doing that atage 20 and it adds up to ten pounds of excess fat by age 30, and 30 pounds ofexcess fat by age 50. Ten calories is the number in one LifeSaver candy orone-fifth of an Oreo cookie. It is roughly the amount of energy burned ineight minutes of sitting or sleeping. Make that two LifeSavers a day or two-fifths of an Oreo and it adds up to 60 extra pounds by age 50. Make it anentire Oreo a day and it adds up to 150 extra pounds by age 50. That’s morethan enough body weight for an entire second person.

The ugly fact is that, once gained, body fat persists. There’s no quick wayto get rid of it. It takes about eight hours of jogging to burn one pound of fat.It takes over a year of jogging an hour a day to burn 50 pounds of fat. On theother hand, not accumulating the fat takes much less intensive effort. Twoextra minutes a day of gardening, light bicycling, or brisk walking will burnthe ten calories of excess food a day that adds up to 50 pounds of fat by age50. So will an extra minute and a half of jogging, racquetball, tennis, moder-ate bicycling, swimming, or stair climbing. So will an annual one-week vaca-tion with a couple hours of hiking each day rather than lounging. Seeminglysmall differences in habits add up greatly over time. Two candy bars a weekrather than two apples adds up to six pounds of extra fat in a year. It takesabout 4,700 minutes of brisk walking to burn those extra calories. That is 78hours, or almost ten eight-hour days, but it is only about 13 minutes a dayover the 365 days of the year. A person who habitually parks an extra fiveminutes away, and takes stairs instead of elevators and escalators, easily burnsan extra six pounds of fat a year instead of accumulating it. Lifestyle differ-ences may look small on any given day, but they add up greatly over theyears.

Desirable and Undesirable Accumulators. The human body has many lessobvious biological accumulators that influence health (Marieb, 1994; Memler,Cohen, & Wood, 1996; Thibodeau & Patton, 1997). Many of the desirableones contribute to aerobic capacity, which is the ability to burn oxygen. The


desirable biological accumulators include vital capacity of the lungs, the numberand sensitivity of insulin receptors on cell membranes, the ratio of slow to fastmuscle fibers, the number of mitochondria and nuclei in muscle cells, and thenumber of small arteries supplying the heart and skeletal muscles. Some un-desirable biological accumulations get defined as diseases or medical condi-tions when they progress beyond a clearly dangerous point. They includehigh resting blood pressure, buildup of fatty plaque in arteries, a low ratio ofhigh-density to low-density lipoprotein in the blood, cellular resistance to in-sulin and the resulting increase in blood glucose, declining bone density, de-posits of uric acid crystals in the soft tissues of joints, softening and fraying ofcartilage, and the calcification of ligaments.

Accumulations and Critical Events: The Heart Attack Example. Some un-desirable accumulations eventually provoke damaging and deadly crises suchas embolism, fibrillation, heart failure, infarction, hemorrhage, stroke, shock,or respiratory arrest. A heart attack provides a good example (Libby, 2002;Marmot & Mustard, 1994; Memmler, Cohan, & Wood, 1996; Rozanski,Blumenthal, & Kaplan, 1999). The slow accumulation of fatty plaque in anartery feeding blood to the heart diminishes the amount of blood that can flowthrough it to feed the heart tissue. The narrower the artery the more likely thata situational demand for the heart to pump more blood will cause a crisis. Asthe heart pumps harder and faster to supply blood to the rest of the body itneeds more blood itself. The occluded coronary artery cannot supply it fastenough. The heart tissue gets starved for oxygen, resulting in the death ofsome heart tissue. Sometimes the fatty plaque ruptures, releasing substancesthat clot the blood (Libby, 2002; Marmot & Mustard, 1994). The resultingthrombus breaks away and clogs a downstream artery supplying part of theheart, starving that portion of oxygen and resulting in its death. Either way,the damaged heart has less muscle to meet future demands, and a tendency tobecome uncoordinated and unable to pump at all. Dangerous and damagingmedical events, like a heart attack, often result from the decline of desirableaccumulations such as aerobic capacity and the development of undesirableones such as the buildup of fatty plaques in arteries.

Stress, Allostatic Load, and Neuroendocrine Accumulators. Most of thebetter understood biological accumulators influenced by education reflect el-ements of health lifestyle such as smoking, diet, and exercise. However dif-ferences in the levels of stress over the lifetime probably also influence bio-logical accumulators directly, apart from health lifestyle. As used here, theword “stress” refers to a specific neuroendocrine reaction, called the stressresponse, to threatening events or conditions, called stressors. The stress re-sponse is a two-phase activation of the sympathetic nervous system and thehypothalamus, pituitary, and adrenal glands. The short-term alarm phase re-leases epinephrine (adrenaline) and norepinephrine, which increases heart rate,respiration rate, blood pressure, and blood glucose, and also dilates the smallpassageways of the lungs. The longer resistance phase releases aldosterone,


which increases blood volume and blood pressure, and also cortisol, whichincreases blood glucose and depresses inflamation and immune function(Marieb, 1994).

Much current biobehavioral research examines “allostatic load,” which isthe impact of intense, recurring, or chronic stress on neuroendocrine accumu-lators that influence health (McEwen, 1998; Taylor, Repetti, & Seeman, 1997).Allostasis refers to the fluctuation in physiological systems to meet demandsfrom external events or exposures. The nervous and endocrine systems regu-late those responses. Allostatic load refers to persistent and potentially harm-ful changes in the regulatory system itself in response to its own history ofactivity. On a biological level the hypothetical accumulators involve changesin the size or layout of structures in the nervous and endocrine systems, andchanges in the receptivity or sensitivity of various tissues to neurotransmitters orhormones. These alterations change the individual’s characteristic physiologicalresponse to stressors. The harmful changes include hair-trigger activation, failureto relax in a normal amount of time, failure to adapt to a stressor with experience,or abnormal suppression of response (Evans, Hodge, & Pless, 1994; McEwen,1998; Rozanski, Blumenthal, & Kaplan, 1999; Sapolsky, 1998). Some of thechanges represent learned, habitual responses of parts of the brain that activatethe sympathetic nervous system and control the hypothalamic-pituitary-adre-nal axis, and some represent deterioration of neurons in one of the controlcenters. Both changes increase the entire body’s exposure to hormones such asepinephrine, aldosterone and cortisol. That exposure over time affects the state ofother accumulators such as resting blood pressure, body fat, insulin resistance,and so on. We think that education reduces allostatic load by giving individualsthe skills, resources, standing and confidence to master their own lives and copewith its challenges effectively and efficiently (Mirowsky & Ross, 2003).

Because biological research on the cumulative effects of stress over thelifetime is fairly new, most of it addresses well-known accumulators such asblood pressure, body fat, and insulin resistance. Current and future researchprobably will find other significant biological accumulators. Some of the mostinteresting current research looks at the effects of chronic stress on function-ing of the immune system (Cohen et al., 1999; Cohen & Herbert, 1996; Glaseret al., 1999; Herbert & Cohen, 1993; Irwin et al., 1997). That research sug-gests two candidate accumulators: a decreased ability to produce the naturalkiller lymphocytes that detect and destroy cells showing abnormalities of thesort produced by genetic transcription errors or viral invasions, and decreasedability to produce antibodies and T-lymphocytes keyed to the detection anddestruction of specific invaders.

Prevention and Correction

Education’s health benefits flow primarily from the “ounce of prevention”taken regularly, which forestalls damaging and deadly crises, and also im-


proves recovery from the injuries and crises that do happen. However, educa-tion also increases the likelihood of sensible and effective responses amongthose who experience a health crisis (Pincus, 1996; Pincus et al., 1998). Forexample, among middle-aged cigarette smokers who have a heart attack, theprobability of quitting goes up dramatically with the level of education (Wrayet al., 1998). Only about a third of smokers with less than a high school de-gree quit after a heart attack, compared to over 80% of those with four yearsof college or more. Higher education improves the odds of a healthy responseat every step. Education reduces the likelihood of ever smoking. Among thosewho smoke, education increases the likelihood of quitting before a healthcrisis occurs. Among smokers who have a health crisis such as a heart attack,higher education increases the odds of quitting in response to the crisis.

Undesirable accumulations typically can be reversed, even after a crisis,but generally only over a period of time as a result of concerted and multifac-eted effort. Education helps individuals to avoid undesirable accumulationsthat need correction, to correct undesirable accumulations before they pre-cipitate a crisis, and failing that to heed the implications of the crisis and takethe difficult but necessary corrective action.

Amplification: Self-Reinforcing Feedback and Interaction

Education’s amplifying effects form the third element of cumulative ad-vantage. We’ve seen that education has pervasive and accumulating effects.Many of those consequences either influence each other or regulate each other’seffects on health. Scientific research invariably uses analytic methods thatconceptually isolate specific consequences or pathways of effect. That allowsscientists to understand and describe the parts and how they work. It is impor-tant to remember, though, that all of these analytic elements exist together,each within a context of the others. The relationships among the parts producephenomena of their own. In particular, mutual effects progressively concen-trate various desirable outcomes together in more educated individuals andprogressively concentrate various undesirable outcomes together in less edu-cated individuals. Feedback and structural amplification are the two genericforms of mutual effect that concentrate advantage in some and disadvantagein others over time. Together with education’s pervasive and accumulatingeffects they produce a cumulative advantage in health for the better educated(Mirowsky & Ross, 2003).

Feedback Amplification

Feedback occurs when the current state of a system produces effects thatlead to a change in its state. All feedback takes one of two forms: deviationsuppressing or deviation amplifying. In deviation suppressing feedback achange in one direction has consequences that lead to corrective changes in


the opposite direction. The body’s homeostatic mechanisms provide the clas-sic examples of deviation suppressing feedback. For example, the hypothala-mus monitors body temperature, keeping it in an acceptable range. When thebody starts to get too cold (below 36.1°C, 97° F) the hypothalamus constrictsthe blood vessels near the skin, reducing the loss of heat to the environment,which brings body temperature back up, which signals the hypothalamus torelax the constriction of the blood vessels, which keeps the body from gettingtoo hot (above 37.8°C, 100° F). Unlike deviation suppressing feedback, whichkeeps things balanced, static, proportional, or similar, deviation amplifyingfeedback makes things change, develop, differentiate, or diverge.

Deviation Amplifying Feedback. In deviation amplifying feedback a changein one direction has consequences that lead to more changes in the samedirection. Rather than bringing things back toward the starting or target state,deviation amplifying feedback propels them away from the original state inthe direction of the deviation. Body fat again provides a good example. Themore body fat individuals have the less they feel like exercising, and the lessthey exercise the more fat they add. On the other hand, the less body fatindividuals have the more they feel like exercising, and the more they exer-cise the less fat they add. Over time that deviation amplifying feedback hastwo effects. First, the differences among individuals in body weight growover time, and so do the differences in amount of exercise. Second, two ben-eficial accumulations (low body fat and high aerobic capacity) get increas-ingly paired in some individuals, whereas their detrimental opposites (highbody fat and low aerobic capacity) get increasingly paired in others. Amplifi-cation works both ways, accumulating the advantages in some individualsand the disadvantages in others.

The Sense of Control and Physical Functioning. The feedback betweenphysical functioning and the sense of control over one’s own life amplifiesone of the most important links between education and health (Mirowsky &Ross, 2003). In doing so it magnifies over the life course the advantage inboth of those accumulators enjoyed by the better educated. The feedbackhelps enlarge the education-based differences that develop in levels of bothphysical impairment and the sense of control. It also concentrates good physi-cal functioning and a firm sense of personal control together in the bettereducated, while concentrating physical impairment and a weak sense of con-trol together in the less well educated.

Psychologists studying adaptation in old age emphasize the self-reinforc-ing dynamics of function and morale. Rodin’s (1986) ideas exemplify thepoint of view. According to Rodin, old age increases environmental chal-lenges to the sense of control and physiological vulnerability to the effects ofhelplessness. The combination multiplies dysfunction and disease, which fur-ther degrades the sense of control and physiological competence. The mutualreinforcement of physical functioning and the sense of control intensify thelong-run impact of conditions such as low income, widowhood, and chronic


disease, and of events such as forced retirement, death of a spouse, diagnosiswith a serious disease, or occurrence of a health crisis such as a stroke. Ac-cording to Rodin, interventions that bolster the personal sense of control amongolder persons reduce the rate and amounts of biological decline by counter-acting the downward spiral.

Rodin’s research and ideas primarily look at short-run effects among per-sons in nursing homes over periods ranging from several months to severalyears. They necessarily treat educational attainment and its influence acrossthe life course as immutable history at that point. Nevertheless, the researchand ideas have clear implications for long-run development of effects. Thoseimplications generally find support in the results of research looking at adultsof all ages living in the general community, and comparing across age groupsor following cohorts as they age (Mirowsky, 1995; Wolinsky & Stump, 1996.)Figure 2 illustrates one such analysis. It uses structural equation modeling toestimate the strength of the deviation-amplifying feedback between impair-ment and the sense of control, and to measure the fit between the patternspredicted by the model and those actually observed in the data.

The results corroborate the hypothesized amplifying feedback. They im-ply that a higher sense of control decreases the probable level of impairment

Figure 2

Structural equation model estimating the strength of the deviation-amplifyingfeedback between impairment and the sense of control. Fit indexes above 99.4%

indicate that the model reproduces observed variances and covariances well.

Minority

Female

(Age-18)3

Parents'Education

Respondent'sEducation

D

D

C

ControlGood

e g

ControlBad

eb

ek e

w

Sense ofControl

-.547-.614

P

Difficultywith Stairsor Walking

DifficultyCarrying

or Kneeling

.850

.852

.54 3

.742

-.056

.039

.239

.052

.216

PhysicalImpairment


and a lower sense of control increases it. Likewise, they imply that higherimpairment decreases the sense of control and lower impairment increases it.Together these create a “double-negative” feedback. Anything that boosts thesense of control thereby decreases the level of impairment, which furtherimproves the sense of control. Likewise, anything that decreases the senseof control thereby increases the level of impairment, which further de-grades the sense of control. If this model is correct, then it and the data implythat the feedback amplifies the effects of other things on impairment and onthe sense of control by around 50%. High model fit statistics indicate that theimplications of the model correspond to the patterns observed in the dataquite well.

The feedback between physical impairment and the sense of control hasthree important implications. First, it amplifies the long-term effects of struc-tural variables on each of the two accumulators (sense of control and physicalimpairment). Structural variables include unchanging sociodemographic at-tributes such as sex, race, and year of birth and persistent socioeconomic onessuch as education, occupation, and wealth. The effects of such long-presentattributes get enlarged by the feedback. Second, it amplifies the effects ofshort-term random shocks to each of the two accumulators. The effects ofpsychosocial crises such as layoffs, bouts of unemployment, or episodes ofeconomic hardship get enlarged when a weaker sense of control slackens thebrakes on the accumulation of physical impairment, further degrading thesense of control. Likewise, the effects of health crises such as injuries, infec-tions, flare ups of chronic diseases, and critical events such as heart attacksand strokes get enlarged when the increased physical impairment underminesthe sense of control, thereby undermining efforts to lessen or reverse the re-sulting physical impairment. Third, structural effects and random shocks oneither side can set the feedback in motion. Psychosocial conditions and eventsproduce amplifying health effects, and vice versa. Disadvantage or loss onone side produces reverberating effects on both sides.

The amplifying feedback between the sense of control and physical im-pairment help create growing differences across levels of education in bothaccumulators with increasing age (Mirowsky & Ross, 2003). The changesobserved over a period reinforce that view. Figure 3 shows the average changein sense of control between 1995 and 1998 by age and level of education atthe beginning, adjusting for sex and initial sense of control (Mirowsky &Ross, 1991). The adjustment means that the figure compares the differentchanges in sense of control over the period observed among persons in differ-ent age or education categories but otherwise similar in terms of sex and ini-tial sense of control. Figure 3 shows a net decrease in sense of control amongpersons with less than a high school degree in every age group. The size ofthe decrease gets larger in older age groups. In contrast, the young and middle-aged adults with four years of college or more have substantial net increasesin sense of control over the period. Among the college educated only the


oldest persons ages 75 and up have a net decrease in sense of control (Mirowsky& Ross, 2003).

Taken together, the models and results support the idea that one’s sense ofcontrol and level of physical impairment affect each other, producing feed-back amplification of differences among individuals in both as they accumu-late over the life course. American adults with low education begin adulthoodwith a relatively low sense of control and relatively high impairment for youngpersons. Each disadvantage leads to subsequent undesirable changes in theother, producing parallel degradation in sense of control and physical func-tioning throughout adulthood. In contrast, adults with high education beginwith a relatively high sense of control and low impairment, and manage todelay the deterioration in control and functioning until old age. The firm senseof control among the well educated helps them slow and delay the rise ofphysical impairment, as illustrated earlier in Figure 1. Good physical func-tioning in turn reinforces their sense of control (Mirowsky & Ross, 2003).

Feedback amplification among education’s consequences magnifies itseffects on health. In the process it concentrates advantageous outcomes to-gether in those with high education and disadvantageous outcomes together

Figure 3

Comparison of mean three-year change in sense of control by age group for personswith a college degree and those with no high school degree, adjusting for initial sense

of control and sex.

Net

Ch

ange

in S

ense

of

Con

trol

Ove

r 3

Yea

rs

< h. s. degree

college degree

75 & older

55 - 74

35 - 54

18 - 34

.2

.1

-.0

-.1

-.2

-.3


in those with low education. As a consequence, education-based differencesin health grow in adulthood, particularly in young adulthood and middle age.The next section describes the distinct process of structural amplification, inwhich low education worsens the consequences of difficult situations that italso makes more likely, giving extra impetus to the downside disadvantages.

Structural Amplification

Problems faced by the poorly educated more often than by the well edu-cated frequently also degrade health more severely for the poorly educated.This is an instance of what we call structural amplification (Ross, Mirowsky,& Pribesh, 2001). Structural amplification exists when the factors that make asituation less damaging also are less common among those in the situation.Corrosive situations and effective traits create structural amplification. In thefirst case, a difficult situation corrodes the traits or resources that protect indi-viduals against its harmful effects. Resources as varied as accumulated wealth,perceived control, marital commitment, emotional support, and cardiorespira-tory fitness that protect individuals in difficult situations also get diminishedor strained in those situations. In the second case, a stable personal character-istic that makes a situation less damaging also helps individuals to avoid orescape the situation. As a result, the effective trait is relatively uncommonamong persons in the situation, amplifying the situation’s harmful effect(Harnish, Aseltine, & Gore, 2000). By not developing effective traits, such asperceived control over one’s own life, the poorly educated disproportionatelyfall into stressful situations such as unemployment, single parenthood, eco-nomic hardship, or neighborhood disorder, and also suffer worse consequencesin those situations (Mirowsky, Ross, & Reynolds, 2000).

The steep gradient in health across levels of income near the bottom existsin part because the low education that results in low income also makes indi-viduals more susceptible to low income’s damaging consequences. Low in-come would not harm the poor as much as it does if they had more educationand the other kinds of resources higher education provides. In reality, though,the effect of education on income means that persons with low income gener-ally lack the education that would help them avoid the harmful effects of lowincome. Similarly, low education worsens the impact on subjective health ofpsychosocial strains and of risky health behaviors such as smoking and lackof exercise that are more common among the poorly educated (Cohen, Kaplan,& Salonen, 1999).

Resource Substitution. Education makes individuals more adept at resourcesubstitution, which means using one thing in place of another, and findingways to achieve ends with whatever materials, relationships, and circumstancespresent themselves. Resource substitution and the structural amplification ofdisadvantage are, in many ways, opposite faces of the same phenomenon.Higher education makes individuals better at acquiring whatever they need,


and better at turning to use whatever they find available. As a result, highereducation tends to increase an individual’s store of the society’s standard re-sources while improving the individual’s ability to improvise resources. Agreater capacity for resource substitution makes the absence of any one stan-dard resource less harmful for the better educated. Conversely, lower educa-tion leaves individuals less adept at acquiring and inventing resources, in-creasing the individual’s dependence on each standard resource (Mirowsky,Ross, & Reynolds, 2000; Ross & Mirowsky, 2004).

The impact on impairment of having someone who helps when you aresick is a typical example of resource substitution. The extra amount of impair-ment associated with not having someone who helps depends on the level ofeducation. Persons with college degrees tend to have little physical impair-ment even if they have no one who nurses them when sick. Persons with lessthan a high school degree tend to have relatively high levels of impairment,especially if they have no one who helps them when sick. Compared to thosewith college degrees, individuals without high school degrees apparently de-pend much more heavily on having someone to help them when sick in orderto maintain physical functioning. However, those with less than high schooldegrees are twice as likely to lack that kind of help. They need it more, butthey have it less (Mirowsky & Ross, 2003).

One of the most important resource substitutions occurs between educa-tion and household income in their effects on economic hardship, which inturn shapes the association between income and health. At any given level ofincome, people with higher levels of education are more successful at avoid-ing trouble paying the bills or buying things the household needs such asfood, clothing, housing, and medical care (Mirowsky & Hu, 1996; Mirowsky& Ross, 1999; Ross & Huber, 1985). Figure 4 illustrates the interaction. Itgraphs the results of a regression analysis predicting economic hardship fromincome at three levels of education, adjusting for sex, race/ethnicity, and age.The lines represent the regression slopes and the flowers represent the num-bers of persons at the various levels of income. The higher the level of educa-tion the less that economic hardship rises as income falls. A poorly-educatedperson needs more money to fend off economic hardship than does a well-educated person. Education does two things. It reduces the risk of economichardship at every level of income. It also reduces the amount that the risk ofeconomic hardship increases at successively lower levels of household in-come. Persons with low education are doubly disadvantaged. Low educationrestrains income and also increases the difficulty of getting by on low income(Mirowsky & Ross, 1999, 2003).

Cascading Sequences. Structural amplifiers often stack up in cascadingsequences. The capacity for resource substitution helps the better educated toavert problems or ameliorate outcomes at each step. In contrast, the relativelack of resources and resourcefulness among the poorly educated exacerbatesthe outcomes at each step. To a large extent, resource substitution and struc-


tural amplification are positive and negative faces of the same phenomenon.Structural amplification concentrates poor health in a minority of persons withmultiple related disadvantages.

Ineffective individuals often move through a cascading sequence of corro-sive situations made worse at each step by the predisposing traits and condi-tions that led into those situations. Imagine a teenage girl from a low-incomehousehold who might not have started having sex except that her family couldonly afford to live in a neighborhood with a lot of unemployed young menhanging out on the streets. Sexual activity probably would not have led her tobecome an unwed mother if her family was college educated, but none ofthem had finished high school. Being an unwed mother probably would nothave caused her to drop out of school if she was from a middle-class family,but she was not. She might have stayed in school if she had been doing well,but no one ever taught her good study habits, and home was often too crowdedor noisy to think. She might have stayed in school if there was a program forpregnant girls, but the district had no money for it and the principal didn’t likethe idea of having pregnant girls around. Being a dropout might not havemade her chronically unemployed if she was not an unwed mother, but jobs

Household Income in $ Thousands

1501251007550250

Pre

dict

ed E

cono

mic

Har

dshi

p S

core

1.00

0.75

0.50

0.25

0.00

College degree

High school degree

Less than high school degree

Figure 4

Sunflower plot of economic hardship score predicted from household income withinthree levels of education. Each sunflower center or petal represents five cases.


were hard to find near home and when she had one she missed work a lot. Thechronic unemployment might have given her time for exercise, but she washome with her child a lot, watching mostly television for entertainment. Shemight have exercised more when her child was older, but by that time she hadput on a lot of weight, and had aches and pains in her joints too. Besides, shedidn’t know of any gyms or pools in her neighborhood, and the streets andparks didn’t look safe. She didn’t have any friends who exercised. Most of thewomen she knew got heavy, so she figured it was normal. The inactivity mightnot have caused her to have high blood pressure and too much cholesteroland glucose in her blood if she ate more fruits, vegetables, and grains, less fat(particularly the saturated kind) and sugar, and fewer calories overall. Theserum cholesterol might not have made her coronary artery occlude if she hadexercised more. Without that occlusion she would not have shed a throm-boembolism when she was served an eviction notice for nonpayment of rent,creating an infarct and fibrillation that she might have survived if she had notbeen obese, diabetic, and out of condition.

Cascading structural amplification creates a grim reality, but one with causefor hope. In theory, the chain can be broken at any step. If the absence of aparticular resource magnifies the harm at a specific step, then averting thatstep or providing that resource may break the chain. Realistically, the mosteffective strategies probably avoid the risky situations and supply the protec-tive resources at many points. This underscores the importance of educationalattainment as a critical point in the chain, and the importance of formal educa-tion as a system for developing abilities with pervasive, cumulative and self-amplifying benefits.

Decline Slowed, Not Just Postponed

The pervasive, accumulating, and amplifying effects of education producecumulative health advantages for the better educated that grow with age. Thatraises a question. Do the differences in health across levels of education growwithout limit, or do they reach a peak at some point in the life cycle and thendiminish? One thing is certain. Everyone dies, no matter how well educated.Death abolishes any advantages in health an individual might have had. Evenso, the differences in health among those who remain alive may continue togrow, as may differences in the risk of dying anytime soon. Many currenttopics in research on socioeconomic differences in health pertain to whetheraccumulated advantages dwindle in the general retrenchment of old age(Beckett, 2000; Hamil-Luker, 2004; Herd & House, 2004; House et al., 1994;Lynch, 2003; Quesnel-Vallee, 2004; Singh-Manous et al., 2004; Willson andShuey, 2004) . In this section we summarize a number of the related issuesand give our views on them. In brief, we think the health advantages associ-ated with higher education continue to grow throughout life, including theretirement years (Ross & Wu, 1996). The evidence remains equivocal, how-


ever, and might go against our view. No matter how this debate turns out, theresearchers on both sides generally agree on three important things. The dif-ferences in health across socioeconomic strata favor the well-educated andthe well-to-do in all age groups. The differences grow in adulthood at leastuntil late middle age, and they remain considerable in old age.

On the surface of it, there seem to be several reasons to expect thateducation’s effects on health might diminish in old age after having grownuntil sometime in late middle age or early old age (Beckett, 2000; House etal., 1994). The various arguments all refer to one or both of two underlyingforces. One set of arguments refers to the socioeconomic life-cycle in moderndemocratic welfare states. Those arguments assume that the socioeconomicadvantages associated with a higher level of education diminish in old age,suggesting that the health advantages do too. Another set refers to biologi-cally programmed senescence. They assume that the physical decline inevi-table in old age diminishes the health advantages accumulated over a lifetime.The following sections argue against those views, questioning whether socio-economic differences diminish in old age, and pointing out that current theo-ries of biological aging imply larger cumulative advantages or disadvantagesin old age, not smaller ones.

The Socioeconomic Life Cycle

Education’s effects on health might wane if its effects on income, wealth,and access to medical care do. Modern democratic states use the competitionfor position and wealth as an engine driving continuous economic, technical,political, and cultural development, but do not want those who fare poorly inthe competition to suffer unduly, particularly after the game is over. Individu-als coming of age together start out with relatively few differences in personalincome, wealth and influence. Those differences grow enormous by latemiddle-age. Some individuals accumulate little or nothing, working for mini-mum wages when they can find a job, possessing no savings, property orpension, known by perhaps a dozen other human beings but significant atmost to one or two who are equally destitute and powerless. Others direct thelabor and actions of thousands, earn millions of dollars a year, have personalwealth rivaling that of entire nations, or shape enduring aspects of science,technology, politics, or culture. What happens in old age to this enormousvariance in prosperity?

Clearly, the average levels of income and wealth in the United States de-cline in old age, following peaks sometime around ages 50 and 60 respec-tively. This leads some individuals to assume that the enormous middle-ageinequality of income and wealth shrinks in old age too. However, decline inaverage income and wealth does not necessarily produce greater equality. Inthe United States the inequality of income (measured by the Gini coefficient)grows steadily after age 59 (Crystal & Waehrer, 1996). This implies an even


larger growth in wealth inequality, because wealth provides an increasing frac-tion of household income in successively older age groups, and because in-equality of wealth exceeds inequality of income at all ages (Davern & Fisher,2001). At any given level of income, the oldest Americans have the greatestamount of wealth and the greatest inequality of wealth. These patterns areespecially marked for wealth in the form of stocks, bonds, mutual funds, andthe like. In the United States, Social Security income exerts some equalizingeffect, but not enough to overcome the differences based on savings, invest-ments, private pensions (often keyed to peak lifetime earnings), and govern-ment-sponsored plans that allow tax-deferred saving (Crystal, Shea, &Krishnaswami, 1992; Dannefer, 1987). It should be no surprise that the U.S.social security system does not reduce economic inequality in old age. It wasnot designed to do that. Social Security mostly redistributes income from adultswho are employed to those who are retired, not from retirees who are wealthyto those who are poor.

Some individuals also assume that Medicare and similar programs reducehealth inequalities by assuring access to needed medical care for all retirees.Supposedly that somehow counteracts the cumulative differences in healthresulting from a lifetime of differences in exposures, behaviors, incidents,and stress. That expects a lot, and there is little reason to think Medicare doesor ever could meet those expectations. Sole reliance on government-spon-sored medical insurance, including Medicare, appears to accelerate declinesin health (Ross & Mirowsky, 2000). Universal medical care systems in theUnited Kingdom, Canada, and other countries did not reduce socioeconomicdisparities in health for their populations, even though those countries pro-vide care over the whole lifetime. Partly that’s because no country provides itspoorest and neediest citizens the level of medical services available to its rich-est ones (although some small, wealthy countries such as Kuwait might comeclose). Mostly, though, it is because medicine generally does not make peoplehealthy (Evans, 1994; Evans & Stoddart, 1994). Most often medicine simplymanages the course of a disease, attempting to slow the deterioration andforestall damaging and perhaps deadly events. Too often medicine’s danger-ous interventions create additional pathology from side effects, errors, andaccidents. Systems such as Medicare and Medicaid in the United States andthe National Health in the United Kingdom do what they can. The practicallimits of welfare politics and medical science combine to assure that suchsystems do not undo the health disadvantages of a lifetime (Ross & Mirowsky,2000).

Biological Senescence, Life Span and Healthy Life Expectancy

As humans get old they become increasingly frail and susceptible. No matterhow healthy individuals have been throughout life, eventually they age physi-cally and die. This fact of human existence can mislead observers into assum-


ing that old age eliminates the health advantages associated with higher edu-cation. Most young persons enjoy good health, and most old ones eventuallybear disease and disability, seemingly regardless of their socioeconomic sta-tus. While true in a sense, this does not mean that health differences diminish.It is true that, at every level of education, most older adults were healthierwhen they were younger. Nevertheless, among seniors of the same age, theones with lower levels of education have declined in health longer and far-ther, if they survive at all. That gets more and more true at successively olderages. High death rates among the unhealthiest individuals, who are dispropor-tionately poorly educated, may compress the health differences among thesurvivors somewhat. The healthiest of the poorly educated survive, makingthe poorly educated group a little healthier on average than it would havebeen. Even so, we think the differences in health continue to grow in old age,as individuals with lower levels of education acquire more numerous andvaried diseases and disabilities that have progressed to more serious levels.

Current biological theories and studies of senescence imply that the differ-ences among individuals in biological decline increase with the age of thesurvivors. Several biological accumulators increase the variability of organicdecline as individuals born about the same time age. The cellular aging clockticks at a speed determined by the rate at which cells must divide to repair orreplenish tissues, and it ticks only a limited number of times. The neuroendo-crine aging clock ticks at a speed determined by damage from stress to neu-rons that regulate the stress response. It seems likely that biological agingclocks tick faster for the poorly educated, making them biologically olderthan the well-educated who were born about the same time. We briefly de-scribe the biological aging clocks below. However, it is more important forsocial scientists to understand the implications than the mechanisms. If thebiological theory described below is correct, then the process has three prop-erties: it acts as an accumulator on the level of molecules, cells and tissues; itis advanced by stress and disease and in turn eventually increases susceptibil-ity to disease, thereby creating deviation-amplifying feedback; and it makesindividuals with a history of exposures and problems more susceptible to theacquisition, progression, and debilitation of disease, thereby structurally am-plifying health disadvantages.

Biologists say that every species has an upper limit on the longest durationof life that members of the species can achieve (Perez-Campo et al., 1998).Many biologists think that limits on the repair and replacement of cells en-force the lifespan limit. Some tissues have cells that do not get replaced whenthey die. In humans the muscle cells, fat cells, and most neurons do not getreplaced. A mature human can lose those cells more or less quickly, but can-not replace them once lost. More commonly, tissues have cells that can divideonly a limited number of times (Hayflick, 1998). Chromosomes have endscalled telomeres that get shorter each time the cell divides. Once a telomeregets too short its cell no longer can divide and replenish its tissue. The more


cell lines that reach their limit, the more the tissue degenerates and loses itsability to function.

Damage to mitochondrial DNA in cells and cell lines also adds to the de-generation of tissues (Hayflick, 1998; Wallace, 1997). Mitochondria containspecialized DNA inherited only from one’s mother. Every cell contains hun-dreds of mitochondria with several loops of DNA. The mitochondrial DNAcontains the code needed to make special proteins that form parts in a struc-ture that manufactures a molecule called adenosine triphosphate (ATP). Thatmolecule, ATP, provides the energy needed for cellular processes. The food aperson eats gets made into glucose that circulates throughout the body andenters cells when insulin unlocks the ports. Once inside a cell, the glucosegets broken into parts that the mitochondria use to make ATP, creating heatand carbon dioxide as byproducts. Apparently the process that makes the ATPalso creates toxic byproducts called oxygen free radicals, which are highlyreactive molecules that can damage components of the cells, including theproteins making ATP and the DNA templates for making those proteins. Themore damage to the proteins making ATP the more free radicals get produced,doing even more damage. The resulting molecular disorder spreads through-out the cell and accumulates within it. The division of a cell over many gen-erations concentrates the damaged mitochondria in subsets of the offspring,creating concentrations of impaired cells within the organs they compose. Incells that do not divide, like those of the muscles, shortages of ATP in a regionserved by a damaged mitochondrion stimulate the replication of that mito-chondrion, adding to the accumulation of damaged energy-producing units.Both types of accumulation progressively impair cells and thus the tissues andorgans they compose.

The third biological aging accumulator operates on the level of the brainand endocrine system, and may work as an amplifier of the other two. Re-search on animals shows that exposure to severe or chronic stress createspersistent anomalies in the structure and function of the endocrine system,and also in parts of the brain that regulate endocrine response (Sapolsky, 1998;Selye, 1976). Research on humans suggests similar effects of severe or chronicstress. The structures involved activate and regulate the body’s response tothreats, called the fight or flight syndrome. Apparently the system also re-shapes itself in response to chronic stress, somewhat like muscles getting largerand more efficient from exercise. Unlike greater strength and stamina, though,a beefed-up stress response produces undesirable effects, particularly in thecontext of modern society, where neither fighting nor fleeing are useful oreven feasible.

A part of the brain called the pituitary initiates the stress response in theendocrine system, which releases cortisol and other hormones. According tocurrent theory, that initiation is regulated by two other parts of the brain(LeDoux, 1996; McEwen, 1998; Sapolsky, 1998). The amygdala signals thepituitary to activate the hormonal stress response and the hippocampus sig-


nals it to deactivate the response. Together the balance of “go” and “stop”signals regulate the phasing (rise and fall) and intensity of the stress response.

The hippocampus happens to have many receptors for cortisol, and thusserves a thermostat-like control function. When blood cortisol levels are highor rising rapidly the hippocampus signals the pituitary to cut back the releaseof cortisol by the adrenal glands. Unfortunately, the high sensitivity to cortisolmeans that high levels of it can damage the hippocampal neurons. Cortisolincreases each neuron’s need for fuel and oxygen. Under peak load a neuroncan suffer a deficit of fuel or oxygen and die. To protect themselves fromfrequent exposure to high cortisol, the neurons cut the filaments of their den-drites that receive signals from other neurons. Over time the trimming of den-drites lowers the responsiveness of the hippocampus to high or rising cortisol.(It also impairs some memory functions.) While that protects the neurons inthe short run, it reduces the ability of the hippocampus to counteract the “go”signals from the amygdala during exposure to stress. As a result the levels ofepinephrine, norepinephrine and cortisol rise faster with less provocation, peakhigher, and stay high longer. That in turn puts the hippocampal neurons ateven greater risk. Eventually the hippocampus fails to shut off the productionof cortisol even in the absence of stressors.

The loss of hippocampal dendrites and neurons reflects the cumulativelifetime exposure to the fight-or-flight hormones. It also increases current ex-posure to those hormones, particularly cortisol. Too much circulating cortisolproduces fatigue, thinning muscles, adult-onset diabetes, hypertension, os-teoporosis, reproductive decline, and immune suppression (Sapolsky, 1998).The diabetes and hypertension degrade the circulatory system and kidneys,which in turn degrades or puts at risk other organ systems. The stress hor-mones mobilize the body’s energy resources for intense physical activity. Indoing so they delay the allocation of the body’s resources to the repair oftissues. In addition, the mobilization of energy for action that never occursmay accelerate the accumulation of mitochondrial free-radical damage men-tioned previously. Current theory focuses on how the free-radical damageaccumulating from the production of ATP makes the life span of a speciesinversely proportional to its metabolic rate, ultimately limiting how long any mem-ber of the species can live. Among modern humans, though, differences in therate of at which free-radical damage accumulates may reflect lifetime differ-ences in the frequency and intensity of the stress response. A sedentary lifestylemight magnify the rate of accumulation, by failing to use mobilized energy.

Each organ system has a limited capacity for regeneration, sometimes calledorgan reserve. Infections, toxins, injuries and exposures to physical extremestend to use up that capacity. So does chronic damage from high blood glu-cose, high blood pressure, poor circulation, psychophysiological stress, andso on. As an organ system nears the end of its reserve, its susceptibility toproblems increases. The biological aging clocks measure the declining abilityto shut off cortisol release, the accumulation of molecular disorder in cells,


and the depletion of limited cell divisions, rather than time itself. The accumu-lating atrophy, errors and divisions inevitably increase with time, but morerapidly for some individuals than others.

Education helps individuals to live longer because it helps make themhealthier and biologically younger for their chronological age. Many personsmistakenly believe that humans are living longer because medical advanceskeep the diseased and impaired alive longer (Evans & Stoddart, 1994). On thecontrary, people are living to older ages being physically younger, and havinghad fewer health problems throughout life (Crimmins & Saito, 2001; Freed-man & Martin, 1988). The active, healthy, and disability-free life expectancygenerally increases along with overall life expectancy itself. (The chief ex-ception occurs when measuring the average length of life free of diagnosedchronic disease. Chronic diseases can appear to start earlier and last longer onaverage because of deliberate efforts to find cases at earlier stages and be-cause of the broadening of diagnostic criteria and the proliferation of diagnos-tic categories.) Education increases the expected number of healthy years anddecreases the expected number of unhealthy years, as well as increasing thetotal years of life expectancy (Crimmins & Saito, 2001).

The Trend toward Bigger Effects of Education

The differences in health and life expectancy among persons of the sameage but different levels of education continue to expand. We think that’s be-cause the better educated persistently lead in incorporating advances in knowl-edge into their own choices and ways of life. Advances in science and tech-nology favor society as a whole, but disproportionately favor those who knowhow to find, evaluate, and use information to best advantage. The more thor-ough and accurate society’s store of knowledge becomes, the more effectiveeducation becomes. As one indication, the correlation of education with theodds of surviving from one decennial census to the next has been gettingstronger (Lauderdale, 2001). For example, in 1960 the women in their 50swho had gone beyond high school had a 4.6% better odds of surviving an-other ten years than did the ones who had never finished high school. By1980 the survival advantage for women in their 50s who had gone beyondhigh school was 44.5%. Although the age-specific odds of continued survivalimproved at all levels of education, they improved most rapidly for the bettereducated. That trend got stronger in successive generations.

Researchers sometimes mistake the historical trend toward bigger effectsof education on health for a decline in education’s effect in old age. The old-est among us lived their lives when humankind knew less about staying pros-perous and healthy. Under those circumstances education had less of a posi-tive effect on health throughout life. That is one reason why the effect ofeducation on health can appear to diminish in old age when it actually in-creases in persons born about the same time as they get older.


No one knows for certain whether the patterns and trends of the twentiethcentury will continue through the twenty-first. Certain things follow logicallyif they do. If it remains true that education’s correlation with health and sur-vival gets larger as a cohort ages, then those correlations will explode for theoldest age groups. Disease, impairment, and mortality hazard among olderAmericans will concentrate increasingly among those with lower education.

Conclusions and Recommendations

Summary of Education’s Cumulative Advantage

The advantages or disadvantages in health associated with level of educa-tion accumulate during all of adulthood, over many areas of life and at alllevels of organization from the socioeconomic down to the molecular. Overshort periods the increments or decrements in any one accumulator may bequite small and seemingly insubstantial. Individuals might not notice the changes,and even scientists might have difficulty measuring them against the backgroundof random fluctuations and measurement errors. Even so, the accumulated differ-ences grow substantial over long periods of time—the three to four decades be-tween school and retirement and two or three decades after retirement.

Some of the positive or negative accumulations affect each other, creatingfeedback that amplifies the effects of persistent attributes such as educationand also the effects of events such as job losses, disease episodes, economiccrises, and personal losses. Benefit begets benefit and deficit begets deficit,accelerating the growth of differences among individuals who began adult-hood about the same time, concentrating various good outcomes together insome of them and various bad outcomes together in others.

Cascading sequences of structural amplification multiply the accumula-tion and feedback. The same traits that make the well educated less likely tobe in risky situations make them less susceptible to the risks if in those situa-tions. In particular, a greater capacity for resource substitution makes the welleducated less dependent on any one heath resource such as income or mar-riage. The obverse is also true. The same traits that make the poorly educatedmore likely to be in risky situations make them more susceptible to thoserisks. A smaller capacity for resource substitution magnifies the harm fromlacking or losing a standard resource. Education’s pervasive, accumulating,and amplifying effects create the cumulative advantages and disadvantages inhealth that emerge and grow over the lifetime.

Implications for Policy and Intervention

As basic-science researchers we have tried to understand how the largeand growing socioeconomic differences in health come to exist. Profession-als, policymakers, providers, and educators will want to consider the implica-


tions of what we have found. Those considerations involve areas of expertisebeyond ours that are the province of this journal’s readers. We close with ourthoughts about the implications for policy and practice that the readers maywant to consider.

Start Young. The health inequalities and disadvantages that develop over65 years of childhood, youth, and middle age cannot be erased in the 20 or 30years of old age. This may be the hardest implication for gerontologists toaccept. Those who work with seniors, or care for them, or have responsibilityfor their well-being want to do whatever can improve the lives of seniors.Many things can be done, but changing an individual’s past is not amongthem. The problems of cumulative disadvantage cannot be solved near theend of life. They must be solved over life’s entirety. This realization forms thenucleus around which a new understanding of gerontology is crystallizing.Aging happens at every age. The study of aging, and the regulation of it tohuman benefit, requires that science, policy, and practice transcend the artifi-cial boundary at age 65. Society will continue to need specialists in the prob-lems of those who are old. Those specialists increasingly must take responsi-bility for understanding and improving the entire process of aging.

Enable Effective Self-Direction. Solutions to the problems of health andaging increasingly require effective self-direction by ordinary individuals. Thisin turn requires a reorientation of sciences, professions and policies attempt-ing to solve the problems of health and aging. Health sciences increasinglymust develop and design science for public use. Health professions increasinglymust advise, encourage and aid rather than prescribe, intervene and remedy. Healthpolicies increasingly must enable rather than regulate. This does not mean aban-doning the technocratic institutions developed over the nineteenth and twentiethcenturies. Human health and longevity will continue to benefit from traditionalforms of medicine, public health, and regulation. The irony of their success isthat the residual problems increasingly require a different approach.

Begin at the Bottom. The worst cases can make the greatest gains. Thephenomena of feedback and structural amplification seem to condemn someindividuals to an ever steeper descent. The unfavorable economic, social, be-havioral, and biological accumulations of a lifetime make untoward eventsmore likely and their consequences more grim. Oddly, though, that same sus-ceptibility and fragility multiplies the effects of resources gained, abilities ac-quired, or shocks and threats avoided. Take the example of adding ten min-utes of physical activity to the daily routine of two persons: one who previouslyran five miles a day and now runs six, and the other who previously remainedin a bed or chair all day and now walks around the block each afternoon. Thefirst person will be healthiest by far, but the second will gain the most from theadditional activity. This principle applies equally to the citizens of a nation asto the residents of a retirement village. The greatest opportunities for improv-ing the health, function, and longevity of a population occur among individu-als on the worst trajectories.


Think Classrooms, Not Operating Rooms. Ask people what they think canimprove the health and longevity of a population and they talk about buildinghospitals and clinics, funding emergency services, training and paying doc-tors and nurses, and subsidizing or outright providing prescribed drugs andinterventions. Some mention protecting the safety of homes, workplaces, prod-ucts and the environment. A few think of the water and sewage systems thatmust handle ever denser populations. No one mentions building schools, orputting classrooms wherever people will use them, or paying more people toteach, or subsidizing learning throughout the lifetime. This needs to change.The health of a population rests on the knowledge, ability, and effectivenessof its people. This is not just a matter of health education, although that has apart. It is not just a matter of telling people the facts about a healthy life andexhorting them to live it. It is a matter of developing in all individuals thecapacity for controlling and directing their own lives. Given the capacity forself-direction, people will direct their lives toward health.

Biographical Notes

John Mirowsky is a professor in Sociology and in the Population Research Center. He studies socialaspects of health and well-being, particularly as they develop over the life course. His recent booksinclude Education, Social Status, and Health, and the second edition of Social Causes of Psycho-logical Distress, both co-authored with Catherine E. Ross and published by Aldine (2003). “Ageat First Birth, Health, and Mortality” will be published in the Journal of Health and Social Behaviorin 2005.

Catherine E. Ross is a professor in the Department of Sociology and the Population Research Centerat the University of Texas. She studies the effects of socioeconomic status, work, family, andneighborhoods on men’s and women’s physical and mental health, and their sense of control versuspowerlessness. Recent publications include “Does Medical Insurance Contribute to Socioeco-nomic Differentials in Health?” Milbank Quarterly 2000 (with John Mirowsky), “NeighborhoodDisadvantage, Disorder, and Health.” Journal of Health and Social Behavior 2001(with JohnMirowsky), and “Powerlessness and The Amplification of Threat: Neighborhood Disadvantage,Disorder, and Mistrust” American Sociological Review 2001 (with John Mirowsky and ShanaPribesh). With John Mirowsky, she published two books in 2003: Social Causes of PsychologicalDistress (second edition), and Education, Social Status and Health (Aldine).

Acknowledgment

This research was funded by two grants from the National Institute onAging: “Aging, Status, and the Sense of Control” (RO1-AG12393) to JohnMirowsky (p.i.) and Catherine E. Ross (co p.i.) and “Education, ResourceSubstitution, and Health (RO1-AG023380) to Catherine Ross (p.i.) and JohnMirowsky (co p.i.).

References

Bartley, M, Blane, D., & Smith, G.D. (1998). Beyond the Black report. Sociology of Healthand Illness, 20, 563-577.


Becker, G.S. (1964). Human capital. New York: Columbia University Press.Beckett, M. (2000). Converging health inequalities in later life—an artifact of mortality

selection? Journal of Health and Social Behavior, 41, 106-119.Berkman, L.F., Glass, T, Brisette, I., & Teresa E. Seeman, T.E. (2000). From social integra-

tion to health: Durkheim in the new millennium. Social Science and Medicine, 51,843-857.

Black, Morris, Smith, & Townsend. (1982). Inequalities in health. The Black report, ed-ited by P. Townsend & N. Davidson. Middlesex: Penguin.

Cohen, F., Kemeny, M.E., Kearney, K.A., Zegans, L.S., Neuhaus, J.M., & Conant, M.A.(1999). Persistent stress as a predictor of genital herpes recurrence. Archives of InternalMedicine, 159, 2430-2436.

Cohen, S., & Herbert, T.B. (1996). Health psychology: Psychological factors and physicaldisease from the perspective of human psychoneuroimmunology. Annual Review ofPsychology, 47, 113-142.

Cohen, S., Kaplan, G.A., & Salonen, J.T. (1999). The role of psychological characteristicsin the relation between socioeconomic status and perceived health. Journal of AppliedSocial Psychology, 29, 445-468.

Crawford, R. (1986). Individual responsibility and health politics. In P. Conrad & R. Kern(Eds), The Sociology of Health and Illness, 2d ed (pp. 369-377). New York: St. Martin’s.

Crimmins, E.M., & Saito, Y. (2001). Trends in healthy life expectancy in the United States,1970-1990: Gender, racial, and educational differences. Social Science and Medicine,52, 1629-1641.

Crystal, S., & Shea, D. (1990). Cumulative advantage, cumulative disadvantage, and in-equality among elderly people. The Gerontologist, 10, 437-443.

Crystal, S., Shea, D., & Krishnaswami, S. (1992). Educational attainment, occupationalhistory, and stratification as determinants of later-life economic outcomes. Journals ofGerontology Social Sciences, 47, S213-221.

Crystal, S. & Waehrer, K. (1996). Later-life economic inequality in longitudinal perspec-tive. Journals of Gerontology Social Sciences, 51, S307-318.

Dannefer, D. (1987). Aging as intracohort differentiation: Accentuation, the Matthew ef-fect, and the life course. Sociological Forum, 2, 211-237.

Davern, M.E., & Fisher, P.J. (2001). Household net worth and asset ownership: 1995. U.S.Census Bureau current population reports, household economic studies, series P70-71. Washington, D.C.: U.S. Government Printing Office.

Evans, R G. (1994). Introduction. In R.G. Evans, M.L. Barer, & T.R. Marmor (Eds.) Why aresome people healthy and others not? (pp. 3-16) New York: Aldine de Gruyter.

Evans, R.G., Hodge, M., & Pless, I.B. (1994). If not genetics, then what? Biological path-ways and population health. In R.G. Evans, M.L. Barer, & T.R. Marmor (Eds.) Why aresome people healthy and others not? (pp. 161-188) New York: Aldine de Gruyter

Evans, R.G., & Stoddart, G.L. (1994). Producing health, consuming health care. In R.G.Evans, M.L. Barer, & T.R. Marmor (Eds.) Why are some people healthy and others not?(pp. 27-64) New York: Aldine de Gruyter

Freedman, V.A., & Martin, L.G. (1998). Understanding trends in functional limitationsamong older Americans. American Journal of Public Health, 88, 1457-1462.

Glaser, R., Rabin, B., Chesney, M., Cohen, S., & Natelson, B. (1999). Stress-inducedimmunomodulation: Implications for infectious diseases? JAMA, 281, 2268-2271.

Hamil-Luker, J. (2004, August). Education-based health inequalities over time and acrosscohorts. Paper presented at the American Sociological Association meetings, San Fran-cisco.


Harnish, J.D., Aseltine, R.H. Jr., & Gore, S. (2000). Resolution of stressful experiences as anindicator of coping effectiveness in young adults: An event history analysis. Journalof Health and Social Behavior, 41, 121-136.

Hayes, D., & Ross, C.E. (1987). Concern with appearance, health beliefs, and eating habits.Journal of Health and Social Behavior, 28, 120-130.

Hayflick, L. (1998). A brief history of the mortality and immortality of cultured cells. KEIOJournal of Medicine, 47, 174182.

Herbert, T.B., & Cohen, S. (1993). Stress and immunity in humans: A meta-analytic review.Psychosomatic Medicine, 55, 364-379.

Herd, P. & House, J.S. (2004, Aug.). Convergence or divergence: Functional status declineamong the 1931 to 1941 birth cohort. Paper presented at the American SociologicalAssociation meeting, San Francisco.

House, J.S., Lepkowski, J.M., Kinney, A.M., Mero, R.P., Kessler, R.C., & Herzog, A.R.(1994). The social stratification of aging and health. Journal of Health and SocialBehavior, 35, 213-234.

Hyman, H.H., Wright, C.R., & Reed, J.S. (1975). The enduring effects of education. Chi-cago, IL: University of Chicago Press

.Irwin, M., Hauger, R., Patterson, T.L., Semple, S., Ziegler, M., & Grant, I. (1997). Alzheimercaregiver stress: Basal natural killer cell activity, pituitary-adrenal cortical function,and sympathetic tone. Annals of Behavioral Medicine, 19, 83-90.

Knowles, J.H. (1977). The responsibility of the individual. In J.H. Knowles Doing betterand feeling worse. Health in the U.S. (pp 57-80), New York: W.W. Norton.

Kohn, M., & Schooler, C. (1982). Job conditions and personality: A longitudinal assess-ment of their reciprocal effects. American Journal of Sociology, 87, 1257-1286.

Kohn, M., & Slomczynski, K.M. (1993). Social structure and self-direction. A compara-tive analysis of the United States and Poland. Cambridge, MA: Blackwell.

Lauderdale, D.S. (2001). Education and survival: Birth cohort, period, and age effects.Demography, 38, 551-561.

LeDoux, J. (1996). The emotional brain: The mysterious underpinnings of emotional life.New York: Touchstone

Libby, P. (2002). Atherosclerosis: The new view. Scientific American, 286, 47-55.Lynch, S.M. (2003). Cohort and life-course patterns in the relationship between education

and health: A hierarchical approach. Demography, 42, 309-331.Marieb, E.N. (1994). Essentials of human anatomy and physiology. Redwood City, CA:

Benjamin/Cummings Publishing Co.Marmot, M.G., & Mustard, J.F. (1994). Coronary heart disease from a population perspec-

tive. In R.G. Evans, M.L. Barer, & T.R. Marmor (Eds.) Why are some people healthy andothers not? (pp. 189-214). New York: Aldine de Gruyter.

McEwen B.S. (1998). Stress, adaptation, and disease: Allostasis and allostatic load. An-nals of The New York Academy of Sciences, 840, 3344.

Memmler, R.L., Cohen, B.J., & Wood, D.L. (1996). The human body in health and disease,Eighth Edition New York: Lippincott, Williams and Wilkins.

Mirowsky, J. (1995). Age and the sense of control. Social Psychology Quarterly, 58, 31-43.Mirowsky, J., & Hu, P.N. (1996). Physical impairment and the diminishing effects of In-

come. Social Forces, 74, 1073-1096.Mirowsky, J., & Kim, J. (2004, August). Age trajectories of depression: Growth-curve

vectors, synthetic cohorts, and cross-sectional profiles. Paper presented at the AmericanSociological Association meetings, San Francisco.

Mirowsky, J., & Ross, C.E. (1991). Eliminating defense and agreement bias from measuresof sense of control: a 2x2 Index. Social Psychology Quarterly, 54, 127145.


Mirowsky, J., & Ross, C.E. (1998). Education, personal control, lifestyle, and health: Ahuman capital hypotheses. Research on Aging, 20, 415-449.

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

Mirowsky, J., & Ross, C.E. (2003). Education, social status, and health. New York: Aldinede Gruyter.

Mirowsky, J., Ross, C.E., & Reynolds, J.R. (2000). Links between social status and healthStatus. In C.E. Bird, P. Conrad, and A.M. Freemont (eds.) The handbook of medicalsociology, 5th Edition, (pp. 47-67). Upper Saddle River, NJ: Prentice Hall.

Nagi, S.Z. (1976). An epidemiology of disability among adults in the United States. MilbankMemorial Fund Quarterly, 54, 439–468.

Nunn, C. A., Crockett, Jr. H.J., & Williams, Jr., J.A. (1978). Tolerance for nonconformity.San Francisco, CA: Jossey-Bass.

O’Rand, A.M. (1996). The precious and the precocious: Understanding cumulative disad-vantage and cumulative advantage over the life course. The Gerontologist, 36, 230-238.

Pascarella, E.T. & Terenzini, P.T. (1991). How college affects students. San Francisco, CA:Josey-Bass.

PerezCampo R., LopezTorres, M., Cadenas, S., Rojas, C., & Barja, G. (1998). The rate offree radical production as a determinant of the rate of aging: Evidence from the com-parative approach. Journal of Comparative Physiology, 168, 149158.

Phelan, J.C., Link, B.G., Diez-Roux, A., Kawachi, I., & Levin, B. (2004). Fundamentalcauses of social inequalities in mortality: A test of the theory. Journal of Health andSocial Behavior, 45, 265-285.

Pincus, T. (1996). Importance of self-care in the association between health and socioeco-nomic status: Implications for health policy. The Bulletin, 40, 5-20.

Pincus, T. Esther, R., DeWalt, D.A., & Callahan, L.F. (1998). Social conditions and self-management are more powerful determinants of health than access to care. Annals ofInternal Medicine, 129(5), 406-411.

Quesnel-Vallee, A. (2004, August). Early socioeconomic disadvantage and the cumula-tive impact of socioeconomic status over the life course on adult health. Paper pre-sented at the American Sociological Association meetings, San Francisco.

Reynolds, J.R. &. Ross, C.E. (1998). Social stratification and health: Education’s benefitbeyond economic status and social origins Social Problems, 45, 221-247.

Rodin, J. (1986). Aging and health: Effects of the sense of control. Science, 233,1271-1276.

Ross, C.E., & Bird, C.E. (1994). Sex stratification and health lifestyle: Consequences formen’s and women’s perceived health. Journal of Health and Social Behavior, 35, 161–178.

Ross, C.E., & Huber, J. (1985). Hardship and depression. Journal of Health and SocialBehavior, 26, 312-327.

Ross, C.E., & Mirowsky, J. (1999). Refining the association between education and health:Effects of quantity, credential, and selectivity. Demography, 36, 445-460.

Ross, C.E., & Mirowsky, J. ( 2000). Does medical insurance contribute to socioeconomicdifferentials in health? Milbank Quarterly, 78, 291-321.

Ross, C.E., & Mirowsky, J. (2003). Social structure and psychological functioning: Dis-tress, perceived control and trust. In J. DeLamater Handbook of social psychology (pp.411-447). NY: Kluwer-Plenum.

Ross, C.E., & Mirowsky, J. (2004, Aug.). Sex differences in the effect of education ondepression: Resource multiplication or resource substitution? Paper presented at theAmerican Sociological Association annual meeting. San Francisco.


Ross, C.E., Mirowsky, J., & Pribesh, S. (2001). Powerlessness and the amplification ofthreat: Neighborhood disadvantage, disorder, and mistrust. American SociologicalReview, 66, 568-591.

Ross, C E., & Van Willigen, M. (1997). Education and the subjective quality of life.Journal of Health and Social Behavior, 38, 275-297.

Ross, C.E., & Wu, C-l. (1995). The links between education and health. American Socio-logical Review, 60, 719-745.

Ross, C.E., & Wu, C-l. (1996). Education, age, and the cumulative advantage in health.Journal of Health and Social Behavior, 37, 104-120.

Rozanski, A.J., Blumenthal, A., & Kaplan, J. (1999). Impact of psychological factors on thepathogenesis of cardiovascular disease and implications for therapy. Circulation, 99,2192-2217.

Sapolsky, R.M. (1998). Why zebras don’t get ulcers: An updated guide to stress-relateddiseases, and coping. New York: W.H. Freeman.

Schnittker, J. (2004). Education and the changing shape of the income gradient in health.Journal of Health and Social Behavior, 45, 286-305.

Schultz, T. (1962). Reflections on investment in man. Journal of Political Economy,70,1–8.

Selye, H. (1976). The stress of life (Revised Edition). New York: McGraw-Hill.Singh-Manoux, A., Ferrie, J.E., Chandola, T., & Marmot, M. (2004). Socioeconomic trajec-

tories across the life course and health outcomes in midlife: Evidence for the accumu-lation hypothesis? International Journal of Epidemiology, 33, 1072-1079.

Spaeth, J. L. (1976). Cognitive complexity: A dimension underlying the socioeconomicachievement process. In W.H. Sewell, R.M. Hauser, and D.L. Featherman (eds.) School-ing and achievement in American society (pp. 103131). New York: Academic Press.

Taylor S.E., Repetti, R.L., & Seeman, T. (1997). Health psychology: What is an unhealthyenvironment and how does it get under the skin? Annual Review of Psychology, 48,411447.

Thibodeau, G.A., & Patton, K.T. (1997). The human body in health and disease, SecondEdition New York: Mosby.

Wallace, D.C. (1997). Mitochondrial DNA in aging and disease. Scientific American, 276,40-47.

Willson, A.E., & Shuey, K. (2004, Aug.). Cumulative advantage and trajectories of Health.Paper presented at the American Sociological Association meetings, San Francisco.

Wheaton, B. (1980). The sociogenesis of psychological disorder: An attributional theory.Journal of Health and Social Behavior, 21,100124.

Wolinsky, F.D., & Stump, T.E. (1996). Age and the sense of control among older adults.Journal of Gerontology: Social Sciences, 51B, S217-220.

Wray, L.A., Herzog, A.R., Willis, R.J., & Wallace, R.B. (1998). The impact of education andheart attack on smoking cessation among middle-aged adults. Journal of Health andSocial Behavior, 4, 271-294.

* Invited paper. Revised manuscript accepted for publication in December, 2004. Actioneditor: P.S. Fry.

education, cumulative advantage, and health

Documents