Elsevier

Intelligence

Volume 44, May–June 2014, Pages 93-102
Intelligence

Is health literacy an example of construct proliferation? A conceptual and empirical evaluation of its redundancy with general cognitive ability

https://doi.org/10.1016/j.intell.2014.03.004Get rights and content

Highlights

  • The distinction between Health Literacy (HL) and cognitive ability is tested.

  • Current conceptualizations of HL are indistinguishable from cognitive ability.

  • Empirically, HL measures fail to capture unique construct variance.

  • HL failed to show incremental validity with respect to health outcomes.

  • Totality of the evidence suggests health literacy is a redundant construct.

Abstract

During the last 20 years, health literacy has been promoted as an important determinant of individual and group differences in health outcomes. Despite a definition and pattern of associations with health outcomes highly similar to ‘g’ (i.e., the general cognitive ability factor), health literacy has been conceptualized as a distinct construct. This study evaluates the conceptual and empirical distinctiveness of health literacy. A sample of 167 students from a southeastern urban university (117 females and 50 males) between the ages of 18 and 53 (M = 21.31, SD = 5.61) completed a cognitive ability battery, three health literacy tests, two knowledge tests, and a questionnaire assessing 12 health behaviors and health outcomes. Across 36 tests of criterion-related validity, cognitive ability had an effect in all 36 cases, where the health literacy tests only showed measureable incremental validity in 6 of 36 cases. Factor analysis revealed only three factors defined by the traditional ability tests with the health literacy measures loading on the ability factors as predicted by the content validity analysis. There was no evidence of a health literacy factor. The combined results from a comparative content analysis, an empirical factor analysis, and an incremental validity analysis cast doubt on the uniqueness of a health literacy construct. It is suggested that measures of health literacy are simply domain-specific contextualized measures of basic cognitive abilities. Implications for linking these literatures and increasing our understanding of the influence of intellectual factors on health are discussed.

Introduction

Health literacy has become a topic of significant interest among health and medical researchers during the past two decades, particularly in regard to its potential explanatory role in health disparities. For example, low health literacy has been shown to be associated with a variety of health outcomes including increased risk of chronic health problems and decreased utilization of health care services (Berkman et al., 2004). Findings such as these have compelled health researchers to consider the elucidatory role of cognitive factors with respect to individual and group differences in health outcomes. At the same time, researchers in the science of mental abilities have amassed evidence that individual differences in basic cognitive abilities, in particular g (the general mental ability construct), are associated with a variety of health behaviors and health outcomes. For example, it has been shown that measures of cognitive abilities predict health behaviors such as amount of physical activity, eating fruits and vegetables, taking vitamins, and smoking (e.g., Anstey et al., 2009, Batty et al., 2007).

Despite the apparent similarity between these streams of research, there has been little effort to understand how the health literacy construct fits within the broader nomological network of intelligence. In fact, it would appear that evidence demonstrating the influence of cognitive abilities on health outcomes is still relatively unknown health psychology, medicine and associated fields. To date, only a few studies have investigated how cognitive abilities and health literacy are associated (e.g., Mõttus et al., 2013, Murray et al., 2011, Wolf et al., 2012). For example, Mottus et al recently showed that health literacy tests have somewhat limited incremental validity after accounting for cognitive ability and education. As such, it remains unclear to what extent health literacy measures are assessing construct variance that is distinct from basic cognitive abilities.

We believe this is a potentially important oversight as it creates the conditions for construct redundancy and construct proliferation. Construct proliferation and redundancy has been noted as a major problem in psychology (e.g., Le et al., 2010, Morrow, 1983, Schwab, 1980) and can be viewed as a major failure to adhere to the canon of parsimony in science. As these authors have noted, “a science that ignores the mandate for parsimony cannot advance its knowledge base and achieve cumulative knowledge.” (Le et al., 2010, p. 112). We believe a clear understanding of the role of cognitive differences in health outcomes will emerge more quickly with an accurate understanding of the construct space being assessed by purported measures of health literacy. As such, our purpose is to evaluate the conceptual and empirical uniqueness of the health literacy construct compared to known cognitive abilities.

Within the larger process of construct conceptualization and validation, it is imperative to demonstrate the uniqueness of a construct by either its complete or partial independence from other comparable constructs (Cronbach and Meehl, 1955, Messick, 1989). The first step of that process requires the construct definition and nomological network of the target construct to be conceptually distinct from existing validated constructs. To avoid construct proliferation and redundancy, it is incumbent upon the newer proposed construct to distinguish itself from known constructs. In cases of where concepts compete for the same construct space, the parsimonious model is accepted as the superior model.

To begin, we can compare their definitions. Cognitive abilities (or “mental abilities”) in general are defined as the sources of variance in performance on tasks requiring one to mentally process, comprehend, and manipulate information (Carroll, 1993, Reeve and Bonaccio, 2011a). While there are several specific abilities (e.g., quantitative reasoning; visual–spatial perception; cognitive speed), the general cognitive ability underlying all of these, or ‘g’ as it is known, has been defined as the factor reflecting individual differences in the ability to educe relations and correlations since it was first formally proposed by Jensen, 1998, Reeve and Bonaccio, 2011a, Spearman, 1904. As Gottfredson (2009) eloquently interpreted, this means “the generalized capacity to learn, reason and solve problems in essentially any context”. Operationally, this means general ability would manifest behaviorally as the ability to obtain and understand information (i.e., to learn), process information (i.e., reason), and use information to make appropriate decisions (i.e., solve problems in context). Although the definition of health literacy used by researchers has evolved in the past fifteen years, the most commonly cited definition of health literacy is that adopted by the Institute of Medicine (Nielson-Bohlman, Panzer, & Kindig, 2004), the Department of Health and Human Services (Healthy People, 2010) and the World Health Organization (WHO) (1998) which states that health literacy is “the ability to obtain, process, and understand basic information and services need to make appropriate health decisions.” Given the accepted definitions, it would appear that health literacy is essentially a restatement of the extant definition of ‘g,’ as it would be manifest in a specific context. That is, it is well established that general cognitive ability (‘g’) is indifferent to the indicator (Jensen, 1998, Spearman, 1927). This principle states that it is not the surface level features of items or situations that determine how well it measures ‘g’, but rather it is the underlying cognitive process. Any situation, regardless of context, that requires the eduction of relations and correlates and the application of knowledge to novel problem solving situations will measure ‘g’. Thus, it appears that the primary conceptual difference is not in the cognitive processes per se, but in whether a model positing a general information processing capacity is more or less parsimonious than a componential model comprised of numerous domain-specific faculties.

In this respect, it has been predicted and confirmed that individual differences in health literacy scores are associated with health specific outcomes such as lower depression scores (Gazmararian, Baker, Parker, & Blazer, 2000), measures of morbidity, general health status, and use of health resources (DeWalt, Berkman, Sheridan, Lohr, & Pignone, 2004), mortality risk (Baker et al., 2007). However, in keeping with the definition g, Gottfredson (2004) posited that individual differences in g would manifest as the ability to comprehend, interpret and apply information in a health care context just as it has been shown to do so in educational, occupational, and social contexts (Kuncel, Hezlett, & Ones, 2004). This prediction has also been well supported. For example, evidence shows that individual differences in g are associated with lower depression scores, better general health, and significantly lower odds for stroke, congestive heart failure, chronic lung disease, heart problems, and hypertension later in life (Der, Batty, & Deary, 2009), reduced risk of mortality (Batty et al., 2008, Deary et al., 2003), reduced likelihood of smoking (Anstey et al., 2009, Reeve and Basalik, 2010), and higher diet quality and increased physical activity (Anstey et al., 2009, Batty et al., 2007) and other indicators of health (e.g., Mõttus et al., 2013). Thus again, this suggests potential for construct redundancy.

Thus, the conceptual question centers on the strength of the evidential basis for the most parsimonious theory. In this case, the empirical reality and the importance of the g construct have been well documented for more than 100 years. Going back to the early 1900s, a broad array of psychometric, biological, and behavioral genetic evidence (e.g., see Carroll, 1993, Jensen, 1998) has given rise to a broad consensus among experts that cognitive abilities are real psychological constructs (Gottfredson, 1997a, Reeve and Charles, 2008, Snyderman and Rothman, 1987), that they have a significant and meaningful influence on important real-world outcomes (e.g. Gottfredson, 1997b, Gottfredson, 2004, Jensen, 1998, Kuncel et al., 2004, Schmidt and Hunter, 1998), that they have a basis in human biology and physiology (Deary, 2009, Gottfredson, 1997b [special issues of Intelligence]; Lubinski, 2004 [special section of the Journal of Personality and Social Psychology]) and is consistent with predictions from evolutionary psychology (Kanazawa, 2004, Kanazawa, 2010). In contrast, the term “health literacy” did not appear until 1974 and a formal treatment of health literacy as a construct did not emerge until the late 1990s (Mancuso, 2009). Additionally, the proposal of a second domain-specific construct to replace a domain-independent construct is less parsimonious, and lacks a basis in the biology or physiology of the central nervous system. Moreover, it is not possible for a domain-specific adaptation of the brain to have evolved within the last century (arguably the time frame the domain in question has been in existence). Conceptually, it is difficult to see where or how health literacy might carve out a unique space within the existing network known as intelligence.

Recent research from IO psychology demonstrates that contextualized ability measures can yield incremental predictive validity above traditional ability measures despite denoting the same construct as the traditional measure (e.g., Klingner & Schuler, 2004). This suggests that contextualized ability measures may assess some unique variance in addition to the basic cognitive abilities that are relevant to context specific criteria. Thus, it remains possible that even if health literacy as a construct is not unique, the measures used do assess some unique cognitive skill that is distinct from general cognitive abilities. As such, we conducted an empirical examination of the convergent and discriminant validity of the three health literacy measures to better understand their empirical distinctiveness from basic abilities.

Section snippets

Participants

A total of 169 adults participated. Everyone enrolled in an introductory psychology course at a mid-size urban (largely commuter) university located in the southeastern U.S. were recruited through the university's undergraduate student research pool. Two participants did not complete the demographic background information. The remaining 167 participants (117 females and 50 male) were between the ages of 18 and 53 (M = 21.31, SD = 5.61), and mostly self-identified as white (62% white, 25.3% as

Descriptive analyses

Descriptive statistics and uncorrected zero-order correlations among the ability and health literacy measures are shown in Table 1. Compared to the published norms for a general population of college students, the current sample scored, on average, at the 30th percentile on both the verbal comprehension and reasoning tests, and at the 15th percentile on both numerical tests. Though low, these scores are consistent with the SAT scores of the local population of psychology majors typically

Discussion

The purpose of this study was to evaluate the uniqueness of the health literacy construct and the validity of health literacy measures relative to cognitive abilities. Based on the totality of the evidence presented here, it is difficult to conclude that “health literacy” reflects a unique construct or that so-called health-literacy tests are psychometrically adequate measures. This conclusion is based on several lines of evidence. First, our conceptual analysis of the definition of health

Conclusion

The current topic provides a critical demonstration of the need for all areas of social science to better heed the principles of measurement. As we noted earlier, posting new constructs similar to existing ones (the “old wine in new wineskins” phenomenon) can be viewed as a major failure to adhere to the canon of parsimony in science. If the mandate of parsimony were ignored, it would seem unlikely that we will achieve the broad scientific goal of uncovering the fundamental causes that underlie

References (67)

  • K.S. McGrew

    CHC theory and the human cognitive abilities project: Standing on the shoulders of the giants of psychometric research

    Intelligence

    (2009)
  • C. Murray et al.

    The association between cognitive ability across the lifespan and health literacy in old age: The Lothian Birth Cohort 1936

    Intelligence

    (2011)
  • C.L. Reeve et al.

    Average state IQ, state wealth and racial composition as predictors of state health statistics: Partial support for ‘g’ as a fundamental cause of health disparities

    Intelligence

    (2010)
  • C.L. Reeve et al.

    On the myth and the reality of the temporal validity degradation of general mental ability test scores

    Intelligence

    (2011)
  • C.L. Reeve et al.

    Survey of opinions on the primacy of g and social consequences of ability testing: A comparison of expert and non-expert views

    Intelligence

    (2008)
  • C.L. Reeve et al.

    The psychometric paradox of practice effects due to retesting: Measurement invariance and stable ability estimates in the face of observed score changes

    Intelligence

    (2005)
  • W.M. Williams et al.

    Practical intelligence for school: Developing metacognitive sources of achievement in adolescence

    Developmental Review

    (2002)
  • D.W. Baker

    The meaning and the measure of Health Literacy

    Journal of General Internal Medicine

    (2006)
  • D.W. Baker et al.

    Health literacy and mortality among elderly persons

    Archives of Internal Medicine

    (2007)
  • G. Batty et al.

    Premorbid (early life) IQ and later mortality risk: Systematic review

    Annals of Epidemiology

    (2008)
  • G.D. Batty et al.

    Mental ability across childhood in relation to risk factors for premature mortality in adult life: the 1970 British cohort study

    Journal of Epidemiology and Community Health

    (2007)
  • N.D. Berkman et al.

    Literacy and health outcomes. Evidence report/Technology assessment No. 87 (RTI International–University of North Carolina Evidence-based Practice Center under Contract No. 290-02-0016). AHRQ Publication No. 04-E007-2

    (2004)
  • Carnap

    Logical foundations of probability

    (1950)
  • J.B. Carroll

    Human cognitive abilities: A survey of factor-analytic studies

    (1993)
  • Centers for Disease Control and Prevention

    Behavioral Risk Factor Surveillance System Survey Questionnaire

    (2010)
  • L.J. Cronbach et al.

    Construct validity in psychological tests

    Psychological Bulletin

    (1955)
  • T. Davis et al.

    Rapid estimate of adult literacy in medicine: A shortened screening instrument

    Family Medicine

    (1993)
  • I.J. Deary et al.

    IQ at age 11 and longevity: Results from a follow up of the Scottish Mental Survey 1932

  • L.R. Derogatis et al.

    The Hopkins Symptom Checklist (HSCL): A self‐report symptom inventory

    Behavioral Science

    (2007)
  • D.A. DeWalt et al.

    Literacy and health outcomes

    Journal of General Internal Medicine

    (2004)
  • R.B. Ekstrom et al.

    Manual for kit of factor-referenced cognitive tests

    (1976)
  • D.P. Flanagan et al.

    Improvement in academic screening instruments? A concurrent validity investigation of the K-FAST, MBA, and WRAT-3

    Journal of Psychoeducational Assessment

    (1997)
  • Cited by (44)

    • An intelligent mind in a healthy body? Predicting health by cognitive ability in a large European sample

      2022, Intelligence
      Citation Excerpt :

      Health literacy has been proposed as an important factor in the intelligence-health relationship because more intelligent individuals are assumed to obtain and process relevant health information more easily than less intelligent persons. Some research even suggests that health literacy is simply a context-specific component of general intelligence (Reeve & Basalik, 2014). We had hypothesized BMI, smoking, and alcohol consumption to be negatively associated with cognitive ability.

    • A process-knowledge approach to supporting self-care among older adults

      2022, Psychology of Learning and Motivation - Advances in Research and Theory
      Citation Excerpt :

      Therefore, the construct may not be needed to explain relationships between patient abilities and health behaviors and outcomes. For example, associations between measures of health literacy and of health-related behaviors and outcomes are substantially explained by cognitive ability measures (e.g., measures of working memory capacity and processing speed), which are more important than the health literacy measures for predicting performance on health-related tasks (Deary, Hill, & Gale, 2021; Reeve & Basalik, 2014). Health literacy frameworks that include more types of patient resources (e.g., motivation) may address the redundancy between health literacy and cognitive constructs, but these frameworks in turn have been criticized as too broad.

    • The evolving relationship between premorbid intelligence and serious depression across the lifespan – A longitudinal study of 43,540 Swedish men

      2017, Journal of Affective Disorders
      Citation Excerpt :

      It was found to attenuate the effect somewhat in the present study, as it did when the association between IQ and suicide was investigated in the same cohort (Sörberg et al., 2013a). Additionally, as mentioned above, people with a higher IQ-score show better health-related behaviours (Batty et al., 2007; Hemmingsson et al., 2008), and better health literacy (Reeve and Basalik, 2014). It is conceivable that such mechanisms contribute to explaining the later occurrence of FTHD in the higher stanine groups, partly by decreasing or counteracting the cumulative effect of depressogenic stressors.

    • Intelligence in youth and health behaviours in middle age

      2018, Intelligence
      Citation Excerpt :

      Higher levels of education have been liked with better adherence to HIV treatment (Goldman & Smith, 2002). There is evidence that health literacy is a sub-component of general intelligence and education is known to correlate strongly with and may be a partial proxy for intelligence (Hagenaars et al., 2016; Herrnstein & Murray, 1994; Reeve & Basalik, 2014; Murray, 1998; Strenze, 2007). Intelligence may therefore be a factor that influences the spread and management of HIV in the population.

    • Genetic Contributions to Health Literacy

      2019, Twin Research and Human Genetics
    View all citing articles on Scopus
    View full text