This chapter summarizes research on the importance of deeper learning and “21st century skills” to success in education, work, and other areas of adult responsibility. The first section focuses on educational achievement and attainment, the second section on work, the third on health and relationship skills, and the fourth on civic participation. Overall, the research reviewed in these sections finds statistically significant, positive relationships of modest size between various cognitive, intrapersonal, and interpersonal competencies and desirable adult outcomes. However, these relationships are based on correlational research methods.
We also reviewed evidence on the role of formal schooling in adult success, which we include in the sections on work and health. We found statistically significant, positive relationships between years of educational attainment and labor market success, not only in research using correlational methods, but also in studies using stronger research methods (see discussion below). Measured cognitive, intrapersonal, or interpersonal competencies appeared to account for surprisingly little of these relationships between years of educational attainment and labor market success. In the fifth section, we show that the benefits of additional years of formal schooling for individuals include not only higher wages but also somewhat greater adaptability to changes in workplace technology and in jobs.
The literature discussed in this chapter comes from a variety of disciplines, including industrial-organizational psychology, developmental psychology, human resource development, and economics. Researchers in these disciplines have investigated the relationship between a range of different skills and abilities and later outcomes, using a variety of methods and data
sets. Some of the evidence we present is correlational in nature, and we call these “simple correlations.” Other evidence is longitudinal, in which competencies and other capacities measured at one point are related to outcomes measured years later, often after adjusting for individuals’ differences in family backgrounds. We call these “adjusted correlations” and view this evidence as more suggestive of causal connections than the evidence from simple correlations, but still prone to biases from a variety of sources. The strongest causal evidence, particularly the evidence of the impacts of years of completed schooling on adult outcomes, comes from statistical methods that are designed to approximate experiments.
IMPORTANCE TO EDUCATIONAL SUCCESS
Many more studies of school success have focused on the role of general cognitive ability (IQ) than specific interpersonal and intrapersonal competencies (see Table 3-1). Economists tend to lump all competencies other than IQ into the category of “noncognitive skills.” Personality and developmental psychologists have developed a much more refined taxonomy of them.
Most personality psychologists have centered their work on the “big five” personality traits—conscientiousness, openness, agreeableness, emotional stability, and extroversion—plus general cognitive ability. Although these traits have traditionally been viewed as relatively stable across the life span, a growing body of evidence indicates that that personality traits change in response to general life experiences (e.g., Roberts, Walton, and Viechtbauer, 2006; Almlund et al., 2011) and to structured interventions (see Chapters 4 and 5).
Developmental psychologists have a dynamic view of competence and behavioral development, with children’s competencies and behaviors determined by the interplay between their innate abilities and dispositions and the quality of their early experiences (National Research Council, 2000). Both groups have investigated associations among cognitive, intrapersonal, and interpersonal competencies and children’s success in school.
Personality Factors and School Success
The comprehensive Almlund et al. (2011) study of personality and attainment offers the following summary of “prediction” evidence on correlations and, in some cases, adjusted correlations between personality traits and educational attainment (see also Table 3-1):
Measures of personality predict a range of educational outcomes. Of the Big Five, Conscientiousness best predicts overall attainment and achieve-
ment. Other traits, such as Openness to Experience, predict finer measures of educational attainment, such as attendance and course difficulty. Traits related to Neuroticism also affect educational attainment, but the relationship is not always monotonic. Conscientiousness predicts college grades to the same degree that SAT scores do. Personality measures predict performance on achievement tests and, to a lesser degree, performance on intelligence tests. (p. 127)
It is important to note that while these associations are large enough to pass conventional thresholds of statistical significance, they almost never account for more than a nominal amount of the variation in the educational attainment outcomes under study.
The most noteworthy meta-analysis of these kinds of data is by Poropat (2009), who examined studies of the simple correlations between personality factors and school grades in primary, secondary, and higher education.1 He found a significant positive association between conscientiousness and grades in primary school through college (see top half of Table 3-2). The simple correlations between conscientiousness and grades in high school and college were in the 0.20-0.25 range, about as high as the correlations between measures of general cognitive ability and grades in high school and college.2 In comparison with other correlates of grades identified in previous studies, these two correlations are at approximately the same level as socioeconomic status (Sirin, 2005) and slightly lower than the correlations found for conscientiousness in industry training programs (Arthur et al., 2003).
In elementary school, general cognitive ability is the strongest correlate of grades, although all five personality factors are positively correlated with grades. Correlations between personality factors and grades generally fall over the course of high school and college. In higher education, among the five personality factors, only conscientiousness is correlated with grades.
Three studies of the correlations between “big five” personality traits and completed schooling have included at least some regression controls (Goldberg et al., 1998; van Eijck and de Graaf, 2004; Almlund et al., 2011). All find positive adjusted associations for concientiousness that range from 0.05 to 0.18, and all find modest negative adjusted associations for extroversion, agreeableness, and neuroticism.
1The Poropat (2009) analysis included many more studies focused on grades in secondary (24-35 studies) and higher education (75-92 studies) than in elementary school (8 studies).
2In social science research, such correlations are generally interpreted following rules of thumb developed by Cohen (1988), in which a correlation of 0.20 is considered small, a correlation of 0.50 is considered medium, and a correlation of 0.80 is considered large.
TABLE 3-1 Key Studies Cited in Chapter 3: The Importance of Deeper Learning and 21st Century Skills
|Reference||Key Findings/Conclusions||Research Methods||Measures of Skills|
|Studies of Personality Factors|
|Almlund et al. (2011)||Conscientiousness has strong correlations with outcomes from a number of adult domains.||Research synthesis||“Big five” personality traits measured using a variety of direct and indirect methods|
|Studies of the Relationship Between Skills and Educational Attainment|
|Duncan et al. (2007)||Reading, math, and attention skills at school entry predict subsequent reading and math achievement. Neither behavior problems nor mental health problems were associated with later achievement, holding constant achievement as well as child and family characteristics.||Formal meta-analysis of standardized regression coefficients emerging from the 236 individual study regressions analyzing the relationship between school-entry reading and math achievement and noncognitive skills and later reading and math achievement. Controls for general cognitive ability, behavior and temperament and parent education and income were included in the regressions.||Cognitive Skills: Measures of school-entry reading and math achievement
Interpersonal and Intrapersonal Skills: The six longitudinal data sets included measures of attention (intrapersonal), antisocial behavior (both intrapersonal and interpersonal), and mental health (intrapersonal).
|Duncan and Magnuson (2011)||Although school-entry reading and math achievement skills predicted later school achievement, single point-in-time assessments of primary school skills were relatively weakly predictive of later outcomes. Children with persistent math or behavior problems were much less likely to graduate from high school or attend college and those with||Review of theory and empirical studies of the relationship between young children’s skills and behaviors and their later attainments. The studies included measures of individual students’ skills at multiple points in time to identify persistent patterns.||Cognitive Skills: Measures of school-entry reading and math achievement
Interpersonal and Intrapersonal Skills: The studies included measures of attention (intrapersonal), antisocial behavior (both intrapersonal and interpersonal), and mental health (intrapersonal).
|persistent behavior problems were much more likely to be arrested or jailed.|
|Poropat (2009)||At the elementary school level, cognitive ability is the strongest predictor of grades. At the high school and college levels, cognitive ability is a weaker predictor of grades and conscientiousness is the only personality factor that predicts grades. Where tested, correlations between conscientiousness and academic performance were largely independent of measures of cognitive ability. Studies controlling for secondary academic performance found conscientiousness predicted college grades at about the same level as measures of cognitive ability.||Meta-analysis of studies of the correlation between personality traits and academic performance. Most of the studies came from higher education, with a smaller sample from primary education.||Cognitive Skills: Some of the studies included tests of general cognitive ability.
Interpersonal Skills: Measures of agreeableness and extroversion
Intrapersonal Skills: Measures of conscientiousness, emotional stability, and openness
|Reference||Key Findings/Conclusions||Research Methods||Measures of Skills|
|Studies of the Relationship Between Skills and Income/Earnings/Job Performance|
|Autor, Levy, and Murnane (2003)||From 1970 to 1988, across the U.S. economy, computerization reduced routine cognitive and manual tasks and increased nonroutine cognitive and interactive tasks. This model explains 60% of the growth in college-educated labor from 1970-1988. Conclusion: Demand is growing for nonroutine problem-solving and complex communication skills.||Paired representative data on job task requirements from the Dictionary of Occupational Titles (DOT) with samples of employed workers from the Census and CPS to create a consistent panel of industry and occupational task input from 1960 to 1998.||Cognitive: DOT measures of: nonroutine cognitive tasks: (1) level of direction, control, and planning of activities; and (2) quantitative reasoning
Manual Tasks: DOT measures of routine manual tasks: finger dexterity and nonroutine tasks: eye-hand-foot coordination
Interpersonal and Intrapersonal: No direct measures
Cognitive: No measures
Interpersonal: Measures of extroversion, agreeableness
Intrapersonal: Measures of emotional stability, conscientiousness, and openness to experience
Cognitive Skills: Tests of mathematics and reading recognition
Interpersonal and Intrapersonal: Several subscores of the Behavioral Problems
|Barrick, Mount, and Judge (2001) (job performance)||Conscientiousness is a valid predictor of job performance across all performance measures in all occupations studied, with average correlations ranging from the mid .20s to low .30s.||Second-order meta-analysis of the results of 11 prior meta-analyses of the relationship between Five Factor Model personality traits and job performance.|
|Cunha and Heckman (2008) (earnings and high school graduation)||Increased parental investments in their children’s skills impact adult earnings and high school graduation rates through effects on both cognitive and noncognitive||Dynamic factor model used to address endogeneity of inputs and multiplicity of parental inputs relative to instruments. Estimated the scale of the factors by estimating|
|skills. Improvements in noncognitive skills raised both cognitive and noncognitive skills.||their effects on high school graduation and earnings at age 23.||Index were combined into a single measure of noncognitive skills.|
|Lindqvist and Vestman (2011)||Conclusion: Noncognitive ability is considerably more important than cognitive ability for success in the labor market.||Data: Sample of 1,053 white males from the CNLSY/79 data set Multiple regression analysis. Authors used ordinary least squares to estimate the effect of cognitive and noncognitive skills on wages, earnings, and unemployment. They matched a dataset on socioeconomic outcomes for a representative sample of the Swedish population with data from the military enlistment.||Measures of Parental Investments: Number of books, number of musical instruments, newspaper subscriptions, special lessons, trips to the museum, trips to the theater
Cognitive Skills: Test of general intelligence
Intrapersonal and Interpersonal Skills: Authors used the overall score and the sum of the subscores assigned by a certified psychologist on the basis of a semi-structured, 25-minute interview. The interview is designed to measure the ability to function during armed combat. A high score reflects both intrapersonal and interpersonal skills
|Studies of the Relationship Between Skills and Health|
|Cutler and LlerasMuney (2010a)||The effect of education on health increases with increasing years of education and appears to be related to critical thinking and decision-making patterns.||1990, 1991, and 2000 waves of the National Health Interview Survey, National Death Index||Completed years of schooling|
SOURCE: Created by the committee.
TABLE 3-2 Correlations and Regression-Adjusted Associations Among Skills, Behaviors, and School Performance
|Concurrent (simple) Correlations||Longitudinal (simple) Correlations||Regression-Adjusted Correlations|
|Personality Factors||Outcome is school grades.|
|Skills and Behaviors||Outcome is reading achievement.||Outcomes are later reading and math achievement.|
NOTE: Concurrent correlations for personality factors and cognitive ability come from Poropat (2009). Concurrent correlations for skills and behaviors in kindergarten and fifth grade come from Duncan and Magnuson (2011). Longitudinal and regression-adjusted correlations are from Duncan et al. (2007). Regression controls in the final column include family background, child temperament, and IQ.
SOURCE: Created by the committee.
Skills, Behaviors, and School Success
There are many ways that developmental psychologists classify competencies in the cognitive, intrapersonal, and interpersonal domains, and some of their categories correspond to some of the “big five” personality traits. One recent review classified important competencies into four groups: achievement, attention, behavior problems, and mental health (Duncan and Magnuson, 2011).
Achievement, in the cognitive domain, refers to concrete academic competencies such as literacy (e.g., for kindergarteners, decoding skills such as beginning to associate sounds with letters at the beginning and end of words) and basic mathematics (e.g., ability to recognize numbers and shapes and to compare relative sizes). Although scores on tests of cognitive ability and achievement tend to have substantial correlations, there is an important conceptual difference between cognitive ability as a relatively stable trait and the concrete achievement competencies that develop in response to schooling and other environmental inputs.
Attention, in the intrapersonal domain, refers to the ability to control impulses and focus on tasks (e.g., Raver, 2004). Developmental psychologists often distinguish between two broad dimensions of behavior problems that reflect the domains of interpersonal and intrapersonal competencies—externalizing and internalizing. Externalizing behavior refers to a cluster of related behaviors, including antisocial behavior, conduct disorders, and more general aggression (Moffitt, 1993; Campbell, Shaw, and Gilliom, 2000). Internalizing behavior refers to a similarly broad set of mental health constructs, including anxiety and depression as well as somatic complaints and withdrawn behavior (Bongers et al. ,2003).3
Many studies have established simple and, in some cases, adjusted correlations between this set of intrapersonal and interpersonal competencies and academic outcomes in the early grades (e.g., Vitaro et al., 2005, and Currie and Stabile, 2007, for attention; Pianta and Stuhlman, 2004, for antisocial behavior; and Fantuzzo et al., 2003, for depressive symptoms). Duncan and Magnuson (2011) use nationally representative data on kindergarteners and fifth graders to compute the simple correlations shown in the bottom left panel of Table 3-2. Since letter grades are rarely recorded in the early grades, the table shows correlations between reading achievement and measures of attention, antisocial behavior and mental health. All are substantial by fifth grade, with the expected positive achievement
3Cutting across the attention and externalizing categories is the idea of self-regulation, which current theory and research often subdivides into separate cognitive (cool) and emotional components (hot) (Raver, 2004; Eisenberg et al., 2005; Raver et al., 2005). Cognitive self-regulation fits into our “attention” category while emotional self-regulation fits into our “behavior problems” category.
associations for attention and negative associations for antisocial behavior and mental health problems. All of these associations are smaller in kindergarten, which, in contrast with the research on personality factors (Poropat, 2009), suggests increasing correlations as children grow older.
Averaging across six longitudinal data sets, Duncan et al. (2007) calculate the bivariate correlations shown in the “longitudinal correlations” column of Table 3-2. Shown here are simple correlations among kindergarten entry achievement, attention and behavioral competencies, and math and reading test scores measured 2-8 years later. Correlations between later achievement and the three measures of attention, antisocial behavior, and mental health problems are similar to what was found for corresponding correlations with kindergarten achievement shown in the first column. As might be expected, correlations between math and reading competencies at school entry and later in the elementary school years are quite high.
To more accurately assess the importance of any one of these competencies and behaviors for school and career success, some studies have gone beyond these simple correlations to account for the fact that children with different levels of a given competency or behavior are likely to differ in many other ways as well. Children with, say, higher math scores may also have higher IQs, be better readers, exhibit less antisocial behavior, or come from more advantaged families. When adjustments for differences in these other conditions are made, the size of the relationship between early competencies and behaviors and later outcomes tends to shrink. This is shown in the fifth and sixth columns of numbers in Table 3-2. A clear conclusion from these columns of numbers is that only three of the five school-entry competencies have noteworthy adjusted correlations with subsequent reading and math achievement: reading, math, and attention. Neither behavior problems nor mental health problems demonstrated a statistically significant positive correlation with later achievement, once achievement and child and family characteristics are held constant.4
Studies estimating bivariate correlations between high school completion and measures of early competencies and behaviors—including achievement, attention, behavior problems, and mental health—find them to be quite modest (.05 to .10; Entwisle, Alexander, and Olson, 2005; Duncan and Magnuson, 2011, Appendix Table 3.A9). Even when these competencies and behaviors are measured at age 14, none of the correlations with high school completion is stronger than .20.
Much larger correlations are observed for early indications that children have persistent deficits in some of these competencies and behaviors. In particular, children with persistently low mathematics achievement and
4A replication and extension analysis by Grissmer et al. (2010) also found predictive power for measures of fine motor skills.
persistently high levels of antisocial behavior across elementary school were 10-13 percentage points less likely to graduate high school and about 25 percentage points less likely to attend college than children who never have these problems (Duncan and Magnuson, 2011). In contrast, persistent reading and attention problems had very low adjusted correlations with these attainment outcomes.5
IMPORTANCE TO WORKPLACE SUCCESS
Technological advances, globalization, and other changes have fueled demand for more highly educated workers over the past four decades. Across much of the 1980s, the inflation-adjusted earnings of high school graduates plunged by 16 percent, while the earnings of college-educated workers rose by nearly 10 percent. In the following two decades, low-skill worker earnings continued to fall, while the earnings of college-educated workers continued their modest rise.6
How these occupation and education-related changes in the labor market affect the demand for cognitive, intrapersonal, and interpersonal competencies is the subject of this section. We begin with a brief review of the large literature on the economic payoff to years of formal education, and of the remarkably modest extent to which prior cognitive, intrapersonal, and interpersonal skills account for that payoff. We then turn to a more detailed discussion of trends in demand for 21st century competencies.
Educational Attainment and Employment Outcomes
From the pioneering work in the 1960s and 1970s of Schultz (1961), Becker (1964), and Mincer (1974) to the present, studies have shown that investments in education produce rates of monetary return that are comparable or higher than market rates on investment in physical capital. Remarkable in this literature is that the estimates have changed little as increasingly sophisticated studies have eliminated likely sources of bias in the estimation of the economic payoff to education, the most prominent of which is the self-selection of more able or motivated into higher levels of completed schooling.7
5These results come from an analysis in which the predictive power of any given skill or behavior was assessed after adjusting for the others and for family background characteristics.
6Autor, Katz, and Kearney (2008, Table 1). Data are based on weekly earnings for full-time workers with 5 years of experience. Earnings of high school dropouts fell even more than the earnings of high school graduates (see also Levy and Murnane, 2004).
7An overview of the efforts to address these bias issues is provided in Card (1999). One strategy for reducing bias from genetic factors is to use siblings or even identical twins to relate earnings and employment differences to schooling differences pairs of otherwise ¨similar¨
In most studies, the so-called private rate of return to added years of schooling (which relates the after-tax earnings benefits enjoyed by workers to the portion of the education costs they have borne) for the United States has varied between 7 and 11 percent, with even higher rates in many other countries (Psacharoupoulos and Patrinos, 2004). The social rate of return tends to be lower than the private rate of return because it includes the full resource costs of schooling provision, much of which is paid through government subsidies rather than the students themselves.
Barrow and Rouse (2005) have concluded that each additional year of schooling generates additional income of about 10 percent, a return that is about the same across the races. And Autor, Katz, and Kearney (2008, Figure 2A) estimate that the earnings advantage for college as opposed to high school graduates rose from about 50 percent higher in the mid-1970s to close to twice as high in 2005. In their summary of evidence on education curriculum, Altonji, Blom, and Maghir (2012) find greater labor market returns to more advanced high school courses and to engineering, business, and science majors in college.
Looking beyond earnings, Oreopoulos and Salvanes (2011) find that workers with higher educational attainment enjoy more nonmonetary employment advantages, including a higher sense of achievement, work in more prestigious occupations, and greater job satisfaction than comparable workers with lower levels of education. Those with more formal education are more likely to be selected for jobs that require further training and that merit training investment. Presumably, the rationale for basing selection decisions on the candidate’s level of education is that the costs of training for reaching job proficiency are reduced when more educated persons are chosen for training programs (Thurow, 1975; Lynch, 1994). Finally, evidence suggests that one person’s added years of schooling benefits others by raising the productivity of other workers at all levels of education (Moretti, 2004).8
In short, the economic importance of a highly educated workforce is impressive and, if anything, increasing. Since the schooling process
individuals. For example, using Norwegian data, Oreopoulos and Salvanes (2011) find that, in comparison with their siblings, siblings with 1 additional year of education have annual incomes that are about 5 percent higher and lower probabilities of being unemployed or on welfare. Another is to use instrumental variable strategies based on, for example, compulsory schooling laws, where the obligatory age of school attendance determines the number of years and the permissible date at which students can leave. Since years of schooling under the compulsory attendance requirements are not subject to voluntary choice, differences in education are exogenous to other influences that might affect the amount of education obtained. None of these strategies is free from all potential biases, however.
8Using a different estimation strategy that focuses only on the returns to secondary schooling for individuals subject to compulsory school attendance laws, Acemoglu and Angrist produce a smaller, but still positive, estimate of external returns than Moretti (2004).
presumably imparts the competencies and behaviors that are responsible for these productivity advantages, it is important to know how cognitive, intrapersonal, and interpersonal competencies are connected to education’s high rates of return.
Test Scores, Education, and Employment Outcomes
Cognitive competencies (as measured by standardized test scores) have the potential to play an important role in accounting for the links between schooling and earnings. First, since smarter people are more likely to acquire more schooling, failure to control for differences in prior cognitive competencies may bias estimates of the role of education per se. But second, even if two graduating high school seniors with identical cognitive competencies make different decisions about whether to attend college, the college experience itself might develop capabilities that command higher earnings from employers.
Surprisingly, empirical studies show that cognitive competencies are able to account for only a small fraction of the association between education and earning. Bowles, Gintis, and Osborne (2001) summarized 25 studies conducted over four decades, which yielded 58 estimates of earnings functions that incorporated test scores. They found that the estimated effect of schooling on earnings retained about 82 percent of its value, on average, after accounting for prior test scores, suggesting that most of the impact of years of educational attainment on earnings was attributable to determinants other than the cognitive competencies.
A second, more direct, approach to investigating the role of cognitive competencies on labor market outcomes does not involve the intervening role played by schooling. An extensive literature, including meta-analyses (e.g., Schmidt and Hunter, 1998, 2004) has examined the simple, unadjusted correlations between cognitive ability, personality factors, and job performance. Schmidt and Hunter (2004) reviewed several studies and meta-analyses, finding that measures of general cognitive ability were strongly correlated (the magnitude of these correlations was higher than 0.53) with occupational level, income, job performance, and job training performance. Comparing these correlations with those found in studies of the association between personality traits and job outcomes, they concluded that general cognitive ability was more important for later job success than conscientiousness or any other intrapersonal or interpersonal competency.
It is worth noting that an NRC committee (1989) reanalyzed the data from over 700 criterion-related studies of the concurrent correlations between scores on a test of general cognitive ability and measures of job performance (typically supervisor ratings, but in some cases, grades in a training course) in about 500 jobs. They found that, despite claims of
much higher predictive validities (i.e., correlations) in the literature (U.S. Department of Labor, 1983), the average correlation in studies that had been conducted since 1972 was about .25 after correction for sampling error. Cognitive test scores explained about 6 percent of the variance in performance, leaving 94 percent to be explained by other factors. Estimates of predictive validities in one subsequent review of the empirical literature also reflected this modest range (Sackett et al., 2001).
Economists have favored prospective longitudinal studies of the relationship between cognitive competencies and earnings (Hanushek and Woessman, 2008). In their examination of the associations between earnings and the cognitive skills of 15-18-year-olds as measured by the Armed Forces Qualifying Test, Neal and Johnson (1996) found that, with no controls for family background, a one-standard deviation increase in test scores was associated with roughly a 20 percent increase in earnings for both men and women. Using data from the National Child Development Survey (NCDS), which has followed a cohort of British children born in 1958 through midlife, Currie and Thomas (1999) related scores on reading and math tests administered at age 7 to wages and employment at age 33. Even in the presence of extensive family background controls, their models show 10-20 percent earnings differentials when comparing both males and females in the top and bottom quartiles of the two test score distributions. Murnane, Willett, and Levy (1995) show links between the mathematics tests scores of two cohorts of high school seniors and their wages at age 24.
Intrapersonal and Interpersonal Competencies and Employment Outcomes
In an effort to understand the large amount of variation in earnings and other employment outcomes that cannot be attributed to cognitive competencies, researchers have begun to examine the role of a variety of intrapersonal and interpersonal competencies. As with our earlier review of the determinants of achievement and attainment, research divides into a focus on personality factors and on other competencies and behaviors.
Almlund et al. (2011) summarize their review of correlational evidence on the role of “big five” personality traits for labor market outcomes as follows:
Personality measures also predict a variety of labor market outcomes. Of the Big Five traits, Conscientiousness best predicts overall job performance but is less predictive than measures of intelligence. Conscientiousness,
however, predicts performance and wages across a broad range of occupational categories, whereas the predictive power of measures of intelligence decreases with job complexity. Additionally, traits related to Neuroticism (e.g. locus of control and self-esteem) predict a variety of labor market outcomes, including job search effort. Many traits predict sorting into occupations, consistent with the economic models of comparative advantage…. Personality traits are valued differentially across occupations. (p. 127)
A key study in this literature is Barrick, Mount, and Judge (2001), which conducts a second-order meta-analysis of the results of 11 prior meta-analyses of the simple associations between Five Factor Model personality traits and job performance. They find that conscientiousness is a valid correlate of job performance across all performance measures studied, with average correlations ranging from the mid .20s to low .30s. Emotional stability was correlated with overall work performance although not with all of the work performance criteria examined. The remaining factors—extroversion, openness and agreeableness—failed to correlate consistently with overall work performance.
Skills, Behaviors, and Earnings
The literature on links between earnings and specific achievement and behavioral skills has employed prospective longitudinal data and well-controlled regression models, yielding stronger evidence than that provided by studies of simple correlations. For example, Heckman, Stixrud, and Urzua (2006), using data from the National Longitudinal Study of Youth (NLSY) estimate substantial adjusted correlations between earnings and a scale combining adolescent self-esteem and sense of personal effectiveness.
Carneiro, Crawford, and Goodman (2007) use data from the British NCDS to relate a wide variety of achievement and behavioral measures assessed when the sample children were 11 years old to later earnings. The diversity of their behavioral measures is reflected in their names: “anxiety for acceptance,” “hostility toward adults,” “withdrawal,” and “restlessness.” When summed into a single index, a standard deviation increase in this collection of antisocial skills and behaviors is found to be associated (net of parental background) with a 3.3 percent decrease in age-42 earnings, about one-fifth of the estimated positive association for a one standard-deviation increase in achievement tests scores. Ironically, an examination of the social and behavioral subscales found the greatest explanatory power for “inconsequential behavior”—a heterogeneous mixture of items related to inattention (“too restless to remember for long”), antisocial behavior (“in informal play starts off with others in scrapping and rough play”), and inconsistency (“sometimes eager, sometimes doesn’t bother”).
In more recent work, Cunha and Heckman (2008) used longitudinal data to study cognitive and noncognitive development over time as it affects high school completion and earnings. They developed a battery of noncognitive scores focused on an antisocial construct using student anxiety, headstrongness, hyperactivity, and peer conflict to go along with cognitive test scores in this analysis. Based upon the psychological, neurological, social, and other aspects of child development, they modeled the developmental path and estimated the impact of investments in cognitive and noncognitive competencies on high school graduation and earnings (at age 23) at three different periods during the age span from 6 to 13. The parental investments studied included purchases of books and musical instruments, newspaper subscriptions, special lessons, trips to the museum, and trips to the theater.
The authors found that the impact of investment returns shifts markedly as the child ages, from cognitive competencies at the earlier ages (6 and 7 to 8 and 9) to noncognitive competencies during the later period (9-13). They also found evidence that noncognitive outcomes contribute to cognitive test results, but little evidence that test scores affect noncognitive outcomes. This finding suggests that investments in noncognitive competencies may contribute to economic productivity not only directly but also by increasing cognitive achievement.
One difficulty in research evaluating and comparing the relative associations between labor market outcomes and both cognitive and noncognitive competencies is the lack of strong measures of noncognitive competencies. Cognitive competencies are measured using well-established and validated standardized testing methods. By contrast, noncognitive competencies are almost always measured by ratings rather than tests—either self-ratings or ratings by observers who are not experts.
Better measurement methods, for example, by trained psychologist observers, might result in more valid measurement and therefore an increase in the estimated importance of noncognitive competencies. This apparently is the finding of a study by Lindqvist and Vestman (2011), which analyzed data on military enlistees in Sweden, where enlistment is compulsory for male 18-year-olds. These individuals complete a cognitive ability test and an extensive questionnaire. A trained psychologist combined the latter with results from a 30-minute clinical interview to assess the individual’s noncognitive competencies, particularly, responsibility, independence, outgoingness, persistence, emotional stability, and initiative. The researchers examined a Swedish database and were able to match labor market outcomes of 14,703 32- to 41-year-olds who had earlier been tested through the enlistment. Comparing the impact of cognitive and noncognitive measures on wages, unemployment, and annual earnings, they found that, in general, the adjusted correlations between these outcomes and their noncognitive
variable were larger than the correlations of earnings with their cognitive variable. Men who did poorly in the labor market were especially likely to lack noncognitive abilities. In contrast, cognitive ability was a stronger correlate of wages and earnings for workers with earnings above the median.
But while this body of research on intrapersonal and interpersonal competencies is growing rapidly, there is little consensus emerging from it. The prospective studies reviewed above capitalize on the haphazard availability of measures in their data sets. Much further investment is needed to specify such competencies and measure them in a streamlined way. Such specification will be useful in understanding how best to teach noncognitive skills to students (Durlak and Weissberg 2011; see Chapter 6) and how mastery of such competencies may, in turn, affect employment, earnings, and other adult outcomes. The European Commission has begun to examine how noncognitive competencies and personality traits contribute to workplace success (Brunello and Schlotter, 2010).
Trends in Demand for 21st Century Competencies
Clearly, labor market demand for increased years of schooling has risen over the past four decades. There is also some evidence that employers currently value and reward a poorly identified mix of cognitive, intrapersonal and interpersonal competencies. As noted in previous chapters, the committee views 21st century skills as dimensions of human competence that have been valuable for many centuries, rather than skills that are suddenly new, unique, and valuable today. One change from the past may lie in society’s desire that all students now attain levels of mastery—across multiple areas of skill and knowledge—that were previously unnecessary for individual success in education and the workplace. Another change may lie in the pervasive spread of digital technologies to communicate and share information. Although the underlying communications and information-processing competencies have not changed, they are applied at an increasing pace to accomplish tasks across various life contexts, including the home, school, workplace, and social networks. According to recent press reports, over half of the estimated 845 million Facebook users around the globe log on daily; among those aged 18 to 34, nearly half check Facebook within minutes of waking up and 28 percent do so before getting out of bed (Marche, 2012). An estimated 400 million people use Twitter to send or receive brief messages. Even in the world of print media, the pace of communication has quickened, as newspapers adopt a “digital first” strategy and publish fresh information online as news stories break (Zuckerman, 2012). Here, we review research addressing the question of whether such changes are increasing demand for cognitive, intrapersonal, and interpersonal competencies, and, if so, whether this will continue in the future.
The economy’s need for different kinds of worker competencies has shifted over time due to a variety of factors, including shifts in the distribution of occupations. Blue collar jobs have shrunk dramatically over the past 40 years, declining from nearly one-third of all jobs in 1979 to only one-fifth of all jobs in 2009. Over the same time period, white collar administrative support jobs, such as filing clerks and secretaries, also declined. This rapid decline in middle-skill, middle-wage jobs has been accompanied by rapid growth at the top and bottom of the labor market, with a trend toward increasing polarization in wages and educational requirements (Autor, Katz, and Kearney, 2008).
The growth jobs at the top and bottom of the labor market is illustrated by Bureau of Labor Statistics (BLS) data, which organizes all occupations in 10 large clusters, three of which—professional/related, service, and sales—constitute fully half of the labor force. The two largest clusters—professional/related (e.g., computer science, education, healthcare professions) and service (e.g., janitorial, food service, nursing aids, home healthcare workers)—are at the opposite ends of the spectrum in terms of education and wages. These two clusters are projected to create more new jobs than all of the other 8 occupational clusters combined over the period 2008 to 2018 (Lacey and Wright, 2009).
Autor, Levy, and Murnane (2003) conducted a study that analyzed not only the mix of occupations but also the competencies demanded within occupations. Drawing on the Dictionary of Occupational Titles (a large catalogue of occupations and their characteristics), they developed measures of the routine and nonroutine cognitive tasks and routine and nonroutine manual tasks required by various occupations. Comparing tasks over time, from 1960 to 1998, they concluded that beginning in 1970 computers reduced routine cognitive and manual tasks and increased nonroutine cognitive and interactive tasks. Their model explained 60 percent of the growth in demand for college-educated labor over the period from 1970-1988. The authors concluded that computers substitute for workers in performing routine tasks and complement workers in performing nonroutine tasks.
Building on this study, Levy and Murnane (2004) argued that demand is growing for expert thinking (nonroutine problem solving) and complex communication competencies (nonroutine interactive skills). Levy and Murnane (2004) also proposed, that demand is growing for verbal and quantitative literacy. They view reading, writing, and mathematics as essential enabling competencies that supported individuals in mastering tasks that require expert thinking and complex communication production processes. Predicting that jobs requiring low or moderate levels of competence will continue to decline in the future, the authors recommended that schools teach complex communication and nonroutine problem-solving competencies, along with verbal and quantitative literacy, to all students.
More recently, Autor, Katz, and Kearney (2008) analyzed data on wages and education levels from 1962 to 2005. The analysis supports the argument that computers complement workers in performing abstract tasks (nonroutine cognitive tasks) and substitute for workers performing routine tasks. However, it also suggests that the continued growth of low-wage service jobs can be explained by computers’ lack of impact on nonroutine manual tasks. Noting that these tasks, performed in service jobs such as health aides, security guards, cleaners, and restaurant servers, require interpersonal and environmental adaptability that has proven difficult to computerize, Autor, Katz, and Kearney (2008) suggest that low-wage service work may grow as a share of the labor market.
Goos, Manning, and Salomons (2009) reached a similar conclusion, based on an analysis of occupational and wage data in Europe. They concluded that technology was the primary cause of polarization in European labor markets, eliminating routine tasks concentrated in mid-level manufacturing and clerical work while complementing nonroutine tasks in both high-wage professional jobs and low-wage service jobs.
These two studies both suggest that low-wage service work involves nonroutine tasks that cannot be readily replaced by computers. There is debate in the literature about the level of cognitive, intrapersonal, and interpersonal competencies required to perform such work. Some case studies and surveys suggest that successful performance in low-wage service jobs requires complex communications skills and nonroutine problem solving (Gatta, Boushey, and Appelbaum, 2007). However, the low levels of education required to enter these jobs, together with their low wages and a plentiful supply of unskilled labor, suggests that their competency demands are—and will remain—low (Autor, 2007). Yet another view is that the competencies required by these and other jobs depend largely on management decisions about how the job is structured and the level and type of training provided (National Research Council, 2008).
Borghans, ter Weel, and Weinberg (2008) studied the role of interpersonal competencies in the labor market and concluded that “people skills” are an important determinant of occupations and wages. They argue that interpersonal competencies vary both with personality and across occupations, and that individuals are most productive in jobs that match their personality. They also found evidence that youth sociability affects job assignment in adulthood, and that interpersonal interactions are consistent with the assignment model. This study built on earlier, unpublished work which suggested that technological and organizational changes have increased the importance of interpersonal competencies in the workplace (Borghans, ter Weel, and Weinberg, 2005).
While these studies propose that demand for cognitive, intrapersonal, and interpersonal competencies has grown in recent decades and will
continue to grow in the future, some experts disagree. For example, Bowles, Gintis, and Osborne (2001) analyzed longitudinal studies that presented 65 different correlational estimates of the relationship between cognitive test scores and earnings over a 30-year period. The authors found no increase in the estimates over time, indicating that labor market demand for cognitive competencies had not grown. Based on responses to a new national survey of skills, technology, and management practices, Handel (2010) argues that, for most jobs in the U.S. economy, education and academic skill demands are low to moderate, noting that large numbers of workers report educational attainments that exceed the requirements of their jobs.
All efforts to predict future competency demands are, of necessity, based on past trends. For example, BLS has often been criticized for using past trends to project detailed occupational requirements and competency needs a decade into the future (National Research Council, 2000). Similarly, Levy and Murnane (2004) call for schools to teach complex communications skills and nonroutine problem solving based on the assumption that the trends identified by Autor, Levy, and Murnane (2003) will continue for decades.
IMPORTANCE TO HEALTH AND RELATIONSHIP SKILLS
Education, Competencies, and Health Outcomes
There is a long history of research on the associations between education and health. Researchers statistically analyze data from self-reports on health status, behavior, and challenges in terms of explanatory variables, including gender, race, age, education, and income. Based on these analyses, they construct a health gradient demonstrating the conditional relation between education and health status. The overwhelming finding is that general health status, specific health outcomes, and healthy behaviors are strongly and positively correlated with educational attainment.
Cutler and Lleras-Muney (2010a) summarized the literature in which educational attainment is linked both statistically and substantively to health outcomes and behaviors. They found higher levels of educational attainment were associated with an array of reductions in adverse health events and increases in healthy eating and exercise. For example, the age-adjusted mortality rate of high school dropouts was found to be about twice that of those with some college in the 25-64-year-old age group in 1999.
Although these findings are widely accepted, two important questions dominate the literature. The first is to what degree is this relation causal as opposed to the explanation that those with better health are more likely to succeed educationally? That is, to what degree is the coefficient or gradient for health by level of educational attainment biased upward by
reverse causation or omitted determinants of both education and health. The second question refers to the mechanism by which education improves health results. While the simplest explanation is that more educated persons are more knowledgeable about how to improve and maintain their health status and are better able to respond to health problems, there are other explanations. These include the effects of education on access to the healthcare system (for example, through higher income) or effects of education on increasing consideration for the long-run consequences of present behavior and taking preventative measures.
To answer the first question, health economists have relied increasingly on the use of instrumental variables techniques to isolate the exogenous effects of education on health outcomes. Following the studies on education and labor market outcomes, they have used externally imposed differences in compulsory schooling such as changes in compulsory attendance requirements that affect the amount of education attained. To control for genetic factors and family backgrounds, they have also compared the health of siblings who have different educational attainments. Lochner (2011) provides a recent review of the latest set of studies employing these sophisticated methodologies. His preferred set of 39 estimates shows a wide range of estimates of education effects on mortality, self-reported health, and disability, as well as two health-related behaviors—smoking and obesity. Not all of the estimates are statistically significant, and some have the wrong signs. By and large, the links tend to be stronger in U.S. than European studies.
With respect to trying to isolate the mechanisms by which education influences health outcomes and behavior, the relations are less clear. There is some evidence that both the general cognitive capabilities of more educated persons as well as specific knowledge contributes to this relation. Cutler and Lleras-Muney (2010b) have also attempted to decompose the education-health nexus into major components including differences associated with education, socioeconomic status and income, and access to social networks. They find that about 30 percent of the education-health gradient is due to a combination of the advantages of income, health insurance, and family background associated with more education; 10 percent is due to the advantages of social networks; and about 30 percent is due directly to education. They also explore the educational mechanisms that might account for the relationship. They conclude that it may not be the specific health knowledge conferred by education as much as greater interest and trust of science and general skills such as critical thinking and decision-making abilities, analytic abilities, and information processing skills that enable educated individuals to make better health-related decisions. Such mechanisms as risk aversion and longer-range time considerations (low time discount rate) do not seem to have substantial support in explaining the health gradients.
A few studies have attempted to estimate links between health and cognitive, intrapersonal, and interpersonal competencies. The Almlund et al. (2011) review reaches the following conclusions regarding personality traits:
All Big Five traits predict some health outcomes. Conscientiousness, however, is the most predictive and can better predict longevity than does intelligence or background. Personality measures predict health both through the channel of education and by improving health-related behavior, such as smoking. (pp. 127-128)
Many of these conclusions are based on the meta-analysis of Roberts et al. (2007), who review evidence from 34 different studies on links between longevity and the “big five” personality traits. They find that conscientiousness was the strongest predictor among the “big five” traits and a stronger predictor than either IQ or socioeconomic status. openness to experience and agreeableness were also associated with longevity, while neuroticism was associated with shorter life spans.
Among individual studies, Conti, Heckman, and Urzua (2010a, 2010b) estimate a multifactor model of schooling, earnings, and health outcomes using data from the British Cohort Study. They find that cognitive ability is not a very important determinant of smoking decisions or obesity but that noncognitive competencies are generally more important for smoking, obesity, and self-reported health. More recently, Hauser and Palloni (2011) studied the relationship between high school class ranking, cognitive ability, and mortality in a large sample of American high school graduates. They found that the relationship between cognitive ability (IQ) and survival was entirely explained by a measure of cumulative academic performance (rank in high school class) that was only moderately associated with IQ. Moreover, the effect of class ranking on survival was three times greater than that of IQ. The authors’ interpretation of these findings is that higher cognitive ability improves the chances of survival by encouraging responsible, well-organized, timely behaviors appropriate to the situation—both in terms of high school academics and in later-life health behaviors.
COMPETENCIES AND HEALTHY RELATIONSHIPS IN ADULTHOOD
Insights into the importance of transferable competencies for healthy marriages and other relationships in adulthood can be gleaned from the literature in a number of areas. Our review concentrates on three: (1) studies of couple satisfaction and marriage duration, (2) programs designed to promote healthy marriages, and (3) programs targeting teen relationship building.
A literature review by Halford et al. (2003; see also Gonzaga, Campos, and Bradbury, 2007) suggests four broad classes of variables that impact the trajectory of relationship satisfaction over time: couple interaction, life events impinging upon the couple, enduring individual characteristics of the partners, and contextual variables. Most relevant to the committee charge are the enduring individual characteristics and interactions.
Behavioral genetic studies show substantial heritabilities for divorce in adulthood (McGue and Lykken, 1992; Jockin, McGue, and Lykken, 1996). A handful of studies have examined early childhood correlates of adult relationship stability. Two of the most relevant drew data from the Dunedon birth cohort study. Newman et al. (1997) found that undercontrolled temperament observed at age 3 predicted greater levels of conflict in romantic relationships at age 21. Relatedly, Moffitt et al. (2011) found that childhood self-control predicts the likelihood of being a single parent.
Most personality traits are not very predictive of relationship satisfaction (e.g., Gottman, 1994; Karney and Bradbury, 1995). However, low neuroticism (i.e., high ability to regulate negative affect) as an adult has been found to predict high relationship satisfaction (Karney and Bradbury, 1997). In addition, Davila and Bradbury (2001) find that low anxiety over abandonment and comfort with emotional closeness are also predictive.
Among the elements of couple interaction, effective communication competencies has predicted relationship satisfaction in numerous studies although, interestingly enough, prospectively and not concurrently (Karney and Bradbury, 1995).
Insights into needed skills can also be gleaned from the curricula of effective adult couple relationship education programs. Many such programs attempt to boost couples’ positive communication, conflict management, and positive expressions of affection (Halford et al., 2003). In contrast, curricula for teen relationship programs promote positive attitudes and beliefs rather than skills, although, as with adult programs, some also target relationship behavior (Karney et al., 2007).
IMPORTANCE TO CIVIC PARTICIPATION
Civic engagement is variously understood to include involvement in activities focused on improving one’s community, involvement in electoral activities (voting, working on campaigns, etc.), and efforts to exercise voice and opinion (e.g., protests, writing to elected officials, etc.) (Zukin et al., 2006). Academics, foundations, and policy makers have expressed concern about decreasing levels of political engagement in the United States, particularly among youth. For example, political scientist Robert Putnam (2000) drew attention to Americans’ lack of connection through clubs, civic associations, and other groups in his influential book Bowling Alone.
In response to these concerns, there has been a resurgence of interest in the development of the knowledge, skills, and dispositions that facilitate civic engagement—this cluster of knowledge, skills, and dispositions is sometimes referred to as “civic literacy.” Studies are looking at the roles played by peers, schools, the media, and other factors in civic literacy and engagement (Delli Carpini and Keeter, 1997; Niemi and Junn, 1998). A recent review of this literature (Garcia Bedolla, 2010) finds that schools have a greater impact on civic literacy than was previously thought, and it has also pointed to the importance of parents and neighborhoods. However, these studies have focused on young people’s attitudes, dispositions, or intentions about future political behavior, and have not linked school-based civics programs with later voting behavior and other civic activities in adulthood.
Prevalence of Civic Participation
Recent survey data suggest that some forms of engagement are fairly widespread (e.g., voting in general elections, volunteerism, consumer boycotts). A majority of young people report that they regularly follow public affairs (Lopez et al., 2006). But upward of 60 percent of young people are unable to describe activities that they can attribute to civic or political engagement, and a significant percentage is “highly disengaged.” These young people do not generally believe their civic or political actions are likely to make much difference. Another type of civic participation is direct political action—protest, work on political campaigns, and the like. Overall, just 13 percent of young people are reported as being intensely involved in politics at this level—survey data indicate they are motivated by a desire to address a social or political problem.
Factors Associated with Civic Participation
Studies have shed light on the factors that correlate with political engagement, focusing on the role of family, schools, and peers in the development of children’s political attitudes and behaviors. Early studies found that families tend to be more important than schools, as political orientations and other attitudes and perspectives appeared to be socially inherited from parents to children (Abramowitz, 1983; Achen, 2002). Indeed, research over four decades has demonstrated that socioeconomic status (SES) is a strong predictor of engagement and participation (Garcia Bedolla, 2010). More recent studies underscore the importance of parents and neighborhoods in the socialization process; they also indicate that schools can play a more important role than was previously believed (Niemi and Junn, 1998; Kahne and Sporte, 2008).
The literature linking years of schooling with civic outcomes is extensive. However, as with labor market and health outcomes, studies providing convincing causal estimates are relatively rare. Lochner (2011) provides a review of these rigorous studies and concludes that this literature suggests important effects of completed schooling on a wide range of political behaviors in the United States, but not in the United Kingdom or Germany. The U.S. impacts are found for voting registration and behavior, political interest, and the acquisition of political information.
Smith (1999) examined the effects of early investments in young people’s social capital on political involvement and “civic virtue” in young adulthood. Using longitudinal data, she examined parental involvement, youth religious involvement, and participation in voluntary associations. She found that early extensive connections to others, close family relationships, and participation in religious activities and extracurricular activities during adolescence were significant predictors of greater political and civic involvement in young adulthood.
EDUCATIONAL ATTAINMENT AND TRANSFER IN THE LABOR MARKET
A general theme of the evidence presented in this chapter is that measurable cognitive competencies, personality traits, and other intrapersonal and interpersonal competencies developed in childhood and adolescence are, at best, modestly predictive of adult successes, particularly labor market productivity. Cognitive ability does appear to matter and, among personality traits, so, apparently, does conscientiousness. But, in the research to date, their predictive power is modest. In terms of “transfer,” we are unable to point to a particular set of competencies or behaviors that have been shown to transfer well to the labor market. (Boosting these skills may increase educational attainment, however, as discussed in the following chapters.)
Education attainment, in contrast, is strongly predictive of labor market success, even in research approaches designed to approximate random assignment experiments. Measurable cognitive, intrapersonal, and interpersonal competencies account for surprisingly little of the impact of education on future productivity. But even if we do not know exactly what it is about spending an additional year in school that makes people more productive, a policy approach designed to promote attainment might be promising, particularly if it can be shown that attainment promotes competencies that are transferable across jobs or across an individual’s entire career.
Prior to the human capital revolution of the 1960s, the manpower planning approach assumed that each job and occupation required a specific level and type of education. Education policy planners produced projections
of economic output by sector multiplied by a fixed formula of occupational requirements per unit of output that was further translated into a rigid formula of educational needs of a future labor force. Needless to say, the manpower forecasts failed, largely because of the rigid assumptions relating educational requirements to occupation and occupational requirements to economic output. Changes in technology, organization, and the market prices of labor and capital, and error-prone projections of sectoral output all undermined the accuracy of the projections of educational need.9
Becker’s (1964) early work on human capital took a more general approach by distinguishing between general and specific human capital. He proposed that education developed “general” human capital that was valuable across different firms, while training and experience within a firm work developed “specific” human capital, valuable only in a particular firm. Becker’s (1964) human capital model depended upon market dynamics in which adjustments would take place through responses to the costs and productivity of different kinds of labor. Labor supply and demand were expected to adapt, as any changes in demand for human capital resulting from changes in the firm’s organization, technology, and mix of outputs would be met by individual and company investments in education, job training, and on-the-job learning.
There is considerable evidence that labor supply, allocation, and productivity are widely adaptable to changes in the economy, especially over the long run. This is because education increases the capacity of workers to learn on the job, benefit from further training, and respond to productive needs as they arise. Workers with more education are generally able to learn their jobs more quickly and do them more proficiently. They can work more intelligently and with greater precision and can accomplish more within the same time period. Greater levels of education increase their ability to benefit from training for more complex job situations, and this is evidenced in the literature on training.10 The research demonstrating the overall impact of education on productivity and economic outcomes did not address precisely what competencies were developed by educational investments. However, an important insight was established by Nelson and Phelps (1966), who suggested that a major contribution of education was to enable workers to adapt to technological change.
Welch (1970) and Schultz (1975) generalized this insight to suggest that investments in more educated workers had an even greater impact on a firm’s ability to adapt to technological change. They argued that hiring more educated workers can improve a firm’s productivity not only because, relative to less educated workers, these workers are more productive in
9See Blaug (1975) for a trenchant critique of this type of approach.
10See Lynch (1992); Leuven and Oosterbeek (1997); Blundell et al. (1999).
their current jobs and can be more quickly and easily trained for complex jobs, but also because they can allocate their time and other resources more efficiently in their own jobs and in related jobs in ways that increase the overall productivity of the firm. In this way, the contributions of more educated workers go beyond their own job performance to impact the overall performance of the organization. For both Welch and Schultz, these benefits represent the greatest opportunity for investments in more educated workers to pay off for the firm.
More education, and higher education in particular, appears to develop workers’ abilities to master an understanding of the production process and to tacitly make adjustments to changes in prices, technology, the productivity of inputs, or mix of outputs. These continuous adjustments allow the firm to “return to equilibrium” (in economic terms), maximizing productivities and profits. Neither Welch nor Schultz addressed which specific aspects of schooling contributed to the ability of workers to make the tacit adjustments to production that will increase productivity and profitability. It is possible that schooling develops not only cognitive competencies but also intrapersonal and interpersonal competencies that enable workers to make decisions that benefit the firm.
Welch (1970) and Schultz (1975) provide many examples of how investments in more educated workers may help firms adjust to optimize their productivity and profits, but there are also many examples of adjustments to disequilibria in the overall labor market. During the Second World War, women replaced males in the labor force in what had been male occupations, continuing the high rates of productivity needed to support both the war effort and the economy (Goldin, 1991). Chung (1990) studied vocationally trained workers for particular occupations who had been employed in those occupations or in occupations that were not matched specifically to their training. He found that workers who had received vocational training for a declining manufacturing industry, textiles, were substantially switching to a growing and thriving manufacturing industry, electronics, and were receiving considerably higher earnings in the latter than in the former. That is, the supply of workers was adapting in the short run to the changes in demand, and in the longer run the occupational training choice of workers was adapting too.
The historical evidence suggests that education is transferable across occupations because many occupations require common skills. For example, Gathmann and Schonberg (2010) found that competencies developed at work (which Becker viewed as “specific” and not valuable outside the firm) were more portable than previously thought. Analyzing data on the complete job histories and wages of over 100,000 German workers, along with detailed information on the tasks used in different occupations, they found that workers developed task-specific knowledge and skills and were
rewarded accordingly, with higher wages as they gained experience in an occupation. On average, workers who changed occupations—whether voluntarily or because they were laid off—were more likely to move to an occupation requiring similar tasks (and attendant competencies) to their previous occupation than to a “distant” occupation requiring very different competencies. Laid-off workers who were unable to find work in similar occupations and were forced to move to a distant occupation experienced higher wage losses than those who were able to find work in similar occupations.
The authors found that university graduates appeared to gain more task-specific knowledge and skills than less educated workers and to be rewarded accordingly with higher wages. However, when more highly educated workers were required to move to distant occupations, their wages declined more than did the wages of less highly educated workers who had to move to a distant occupation. This suggests that the deep task-specific competencies developed by the highly educated workers were less transferable than the shallower competencies developed by the less educated workers. Overall, the study suggests that workers are more easily able to transfer competencies developed on the job to a similar occupation, involving similar tasks, than to a dissimilar occupation. This is analogous to research findings from the learning sciences, which have found that transfer of learning to a new task or problem is facilitated when the new task or problem has similar elements to the learned task (see Chapter 4).
Other evidence suggests that even workers with relatively lower levels of education may be able to adapt to the demands of complex jobs. One measure of adaptability is the substitutability among workers with different levels of education. Economists measure employers’ ability to substitute workers at one level of education for jobs that normally are associated with a higher level of education by examining how the mix of more and less educated workers changes as relative wages for different educational levels change. Historical studies in the United States suggest that each 10 percent increase in the labor costs of a higher level of education is associated with a 15 percent decrease in employment at that educational level and increase in workers with less education to replace them (Ciccone and Peri, 2005). This implies that employers view workers as highly adaptable to perform jobs that traditionally require more education, when relative wages encourage such substitution.
CONCLUSIONS AND RECOMMENDATIONS
The research evidence related to the relationship between various cognitive, intrapersonal, and interpersonal competencies is limited and uneven in quality. Some of the evidence reviewed in this chapter is correlational
in nature and should be considered, at best, suggestive of possible causal linkages. Other evidence, from longitudinal studies, is more suggestive of causal connections than the correlational evidence, but it is still prone to biases from a variety of sources. The strongest causal evidence, particularly the evidence of the impacts of years of completed schooling on adult outcomes, comes from statistical methods that are designed to approximate experiments.
- Conclusion: The available research evidence is limited and primarily correlational in nature; to date, only a few studies have demonstrated a causal relationship between one or more 21st century competencies and adult outcomes. The research has examined a wide range of different competencies that are not always clearly defined or distinguished from related competencies.
Many more studies of the relationships between various competencies and outcomes (in education, the labor market, health, and other domains) have focused on the role of general cognitive ability (IQ) than on specific intrapersonal and interpersonal skills (see Table 3-1). Economists who conduct such studies tend to lump all competencies other than IQ into the category of “noncognitive skills,” while personality and developmental psychologists have developed a much more refined taxonomy of them. All three groups have investigated the relationships between cognitive, intrapersonal, and interpersonal competencies and outcomes in adolescence and adulthood.
- Conclusion: Cognitive competencies have been more extensively studied than intrapersonal and interpersonal competencies, showing consistent, positive correlations (of modest size) with desirable educational, career, and health outcomes. Early academic competencies are also positively correlated with these outcomes.
- Conclusion: Among intrapersonal and interpersonal competencies, conscientiousness (staying organized, responsible, and hardworking) is most highly correlated with desirable outcomes in education and the workplace. Antisocial behavior, which has both intrapersonal and interpersonal dimensions, is negatively correlated with these outcomes.
Across the available studies, the relative size of the correlations with the three different domains of skills is mixed. There is some evidence that better measurement of noncognitive competencies might result in a higher estimate of their importance in education and in the workplace.
A general theme of the evidence presented in this chapter is that measurable cognitive skills, personality traits, and other intrapersonal and interpersonal competencies developed in childhood and adolescence are, at best, modestly predictive of adult successes, particularly in the labor market. Educational attainment, in contrast, is strongly predictive of labor market success, even in research approaches designed to approximate random assignment experiments. Measurable cognitive, intrapersonal, and interpersonal competencies account for surprisingly little of the impact of education on future wages (wages, in economic theory, reflect productivity).
Studies by economists have found that more highly educated workers are more productive than those with less years of schooling are because more highly educated workers are better able to accomplish a given set of work tasks and are also more able to benefit from training for more complex tasks. In addition, more highly educated workers have the capacity to allocate resources more efficiently in their own work activities and in behalf of the enterprise in which they work than do workers with fewer years of schooling.
- Conclusion: Educational attainment—the number of years a person spends in school—strongly predicts adult earnings, and also predicts health and civic engagement. Moreover, individuals with higher levels of education appear to gain more knowledge and skills on the job than do those with lower levels of education and they are able, to some extent, to transfer what they learn across occupations. Since it is not known what mixture of cognitive, intrapersonal, and interpersonal competencies accounts for the labor market benefits of additional schooling, promoting educational attainment itself may constitute a useful complementary strategy for developing 21st century competencies.
The limited and uneven quality of the research reviewed in this chapter limits our understanding of the relationships between various cognitive, intrapersonal, and interpersonal competencies and adult outcomes.
- Recommendation 1: Foundations and federal agencies should support further research designed to increase our understanding of the relationships between 21st century competencies and successful adult outcomes. To provide stronger causal evidence about such relationships, the programs of research should move beyond simple correlational studies to include more longitudinal studies with controls for differences in individuals’ family backgrounds and more studies using statistical methods that are designed to approximate