From 2015... The Unifying Moral Dyad: Liberals and Conservatives Share the Same Harm-Based Moral Template. Chelsea Schein, Kurt Gray. Personality and Social Psychology Bulletin, Vol 41, Issue 8, June 19, 2015. https://doi.org/10.1177/0146167215591501
Abstract: Do moral disagreements regarding specific issues (e.g., patriotism, chastity) reflect deep cognitive differences (i.e., distinct cognitive mechanisms) between liberals and conservatives? Dyadic morality suggests that the answer is “no.” Despite moral diversity, we reveal that moral cognition—in both liberals and conservatives—is rooted in a harm-based template. A dyadic template suggests that harm should be central within moral cognition, an idea tested—and confirmed—through six specific hypotheses. Studies suggest that moral judgment occurs via dyadic comparison, in which counter-normative acts are compared with a prototype of harm. Dyadic comparison explains why harm is the most accessible and important of moral content, why harm organizes—and overlaps with—diverse moral content, and why harm best translates across moral content. Dyadic morality suggests that various moral content (e.g., loyalty, purity) are varieties of perceived harm and that past research has substantially exaggerated moral differences between liberals and conservatives.
Keywords: moral pluralism, values, culture, moral foundations, political psychology
---
The existence of moral disagreement across the political spectrum is uncontroversial. One need only open a news-paper to see that liberals and conservatives are divided on many issues, including abortion, capital punishment, gay rights, women’s rights, gun ownership, environmentalism, euthanasia, and the justifiability of war. What is contro-versial is whether this disagreement reflects deep differ-ences in moral cognition. Do liberals and conservatives have fundamentally different moral minds? One popular theory of moral cognition argues that liberals and conser-vatives rely on different sets of moral mechanisms (or foundations; Haidt, 2012).In contrast, we suggest that liberals and conservatives fundamentally have the same moral mind. Rather than dis-tinct and differentially activated mechanisms, we suggest that moral judgment involves a common template grounded in perceived harm (the moral dyad; Gray, Waytz, & Young, 2012). This template is not only consistent with categorization in non-moral domains but also reconciles modern moral pluralism with historic harm-centric accounts, and provides hope for bridging political differ-ences. In this article, we test six predictions of dyadic morality, which can be summarized as follows: Harm is central in moral cognition for both liberals and conservatives.
Descriptive Diversity
An important first step in science is collecting and cataloging diversity. Biology began with natural history, in which living organisms were collected from around the world and placed into taxonomies. The most famous biological taxonomy is Linnaean classification—proposed by Carl Linnaeus—which divides organisms into five different kingdoms based on their appearance.The new renaissance of moral psychology also began with collecting and taxonomizing moral diversity based on descriptive appearance, akin to Linnaean classification. Anthropological accounts of morality in rural India were divided into three content areas of autonomy, community, and divinity (Shweder, Much, Mahapatra, & Park, 1997). One later account inspired by American political disagree-ment—Moral Foundations Theory (MFT)—taxonomizes morality into the five content areas of harm, fairness, author-ity, in-group, and purity (Haidt, 2012).Beyond providing a moral taxonomy, MFT also suggests differences in morality across liberals and conservatives, with only conservatives being concerned with authority, in-group, and purity. The idea of differences between liberals and conservatives is not new, as decades of research reveal that conservatives are more tolerant of inequality, are more religious, and believe more in a just world (for review, see Jost, Glaser, Kruglanski, & Sulloway, 2003). In particular, classic research on right-wing authoritarianism (RWA) reveals that conservatives are more submissive to authority, more likely to use aggression to protect the in-group, and more conventional in terms of sexuality (Altemeyer, 1988).Given that RWA demonstrates the sensitivity of conserva-tives to authority, in-group, and sexual/religious convention-alism, it is safe for MFT to suggest the same—especially given that MFT questionnaires about authority, in-group, and purity correlate up to r = .70 with the RWA questionnaire (Graham etal., 2011, Table 7, p. 377). Moreover, as RWA fully accounts for liberal-conservative differences revealed by MFT (Kugler, Jost, & Noorbaloochi, 2014), one may wonder about the novelty of MFT claims. However, MFT does make one bold, unique claim—that these political dif-ferences arise from deep differences in moral cognition.
Moral Modules
Inspired by theories of basic emotions, MFT suggests that harm, fairness, in-group, authority, and purity each represent a distinct functional moral mechanism or cognitive module(Haidt, 2012). MFT defines cognitive modules as “little switches in the brains of all animals” that are “triggered” by specific moral “inputs” (Haidt, 2012, p. 123). These modules are suggested to be ultimately distinct from each other, involving fundamentally “distinct cognitive computations” (Young & Saxe, 2011, p. 203), such that violations of one content area (e.g., harm) are processed differently from those of another (e.g., purity).This “distinct cognition” prediction is best explained by the MFT analogy of moral foundations as different “taste receptors” (e.g., purity as “saltiness”), such that each moral concern triggers only one specific receptor, which gives rise to a corresponding distinct moral experience. However, recent evidence casts doubt on claims of distinct cognition, as even harm and purity—often discussed as maximally dis-tinct (Haidt, 2012)—are highly correlated (r = .87; Gray & Keeney, 2015). Moreover, the apparent cognitive differ-ences between these two content areas stem from scenario sampling bias: MFT purity violations are weirder and less severe (e.g., necrophilia) than harm violations (e.g., mur-der), and it is these general differences that give the illusion of distinct cognition (Gray & Keeney, 2015).MFT also posits that moral modules are uniquely linked to specific emotions, such that harm is specially linked to anger and purity to disgust. Recent evidence also casts doubt on this claim, as links between moral content and emotion can be explained with broader affective and conceptual con-siderations (Cameron, Lindquist, & Gray, in press; Cheng, Ottati, & Price, 2013). Studies that purport to find a unique effect of disgust on purity fail to control for other high arousal emotions, such as anger or fear, or use statistical pro-cedures that ignore substantial shared variance between anger and disgust (Cameron etal., in press).Despite the lack of evidence for cognitive distinctness, the idea of moral “foundations” is intuitively compelling because it aligns with psychological essentialism—provid-ing a deep mental/biological explanation for important polit-ical differences. However, descriptive differences between liberals and conservatives—whether in personality, music preferences, or the moralization of specific issues—need not reflect cognitive differences. Even incredible diversity can be underlain by a common process.
Common Cognition
If science begins with cataloging diversity, its next step is often developing theories that explain this diversity with a common mechanism. In biology, Linnaeus and his contem-poraries believed that each species was distinct and immu-table, uniquely created by God. However, Darwin discovered that this incredible diversity stemmed from the simple algo-rithm of evolution. Could moral diversity also be underlain by a simple common mechanism? Could liberals and conser-vatives—despite their political disagreements—ultimately have the same moral cognition?Decades of research in cognitive psychology suggest that non-moral categorization decisions (is x a member of y cat-egory?) rely upon the process of template comparison. This template (or prototype) represents the most common, salient or important features across category instances (Murphy, 2004). Categorization decisions are made by comparing potential examples with this template, with closer matches being more robustly categorized as belonging to that cate-gory. For the category of “birds,” the template includes the features of “small,” “flying,” and “seed-eating,” which explains why sparrows are judged as more bird-like than penguins (Rosch, 1978). The same process occurs in social categorization, in which people are compared with cognitive templates called stereotypes (Smith & Zarate, 1990).The principle of parsimony suggests that moral judgments (is x act a member of the category immorality?) should also be made via template comparison. Acts should be compared to a moral template—or prototype—that extracts the most common, salient, and important elements across instances of immorality. Moral psychology suggests many potential can-didates for these elements, such as concerns about intention, causation, and outcome (Alicke, 2000; Cushman, Young, & Hauser, 2006; Malle, 2006); norm and affect (Nichols, 2002); and mind perception (Gray, Young, & Waytz, 2012). The concept of harm is related to many of these elements, and we suggest it forms the basis of a cognitive moral template.
Dyadic Morality
Harm can manifest itself in different ways, but within moral contexts, it typically involves the intentional action of one person causing suffering to a second person—a perpetrator and a victim. More technically, harm involves the perception of two interacting minds, one mind (an agent) intentionally causing suffering to another mind (a patient)—what we call the moral dyad (Gray, Waytz, & Young, 2012; see also Mikhail, 2007).The complementary roles of agent and patient stem from the two-dimensional nature of mind perception (Bastian, Laham, Wilson, Haslam, & Koval, 2011; Gray, Jenkins, Heberlein, & Wegner, 2011) and the general dyadic structure of language (Brown & Fish, 1983) and action (Aristotle, BC350), in which agents act upon patients (Strickland, Fisher, & Knobe, 2012). The psychological power of a harm-based template stems not only from the presence of inten-tional harm in many canonical acts of immorality (e.g., murder, rape, assault, and abuse) but also from the affective potency of suffering victims (Blair, 1995), the hypersensitiv-ity of agency detection (Barrett, 2004), the early develop-ment of empathy and harm-based concerns (Decety & Meyer, 2008; Govrin, 2014; Hamlin, Wynn, & Bloom, 2007), and the obvious evolutionary importance of harm (DeScioli & Kurzban, 2013).It is clear that harm plays a key role in morality, helping to separate counter-normative acts into those that are immoral from those that are violations of mere social con-vention (Sousa, Holbrook, & Piazza, 2009; Turiel, Killen, & Helwig, 1987). Dyadic morality provides a mechanism for the role of harm. We suggest that a norm violation “x” leads people to automatically ask “is x immoral?”—per-haps to the degree that x induces negative affect (Nichols, 2002)—which then activates the dyadic harm-based tem-plate. The more an act is inherently dyadic (i.e., harmful), the better the template matches, and the more robustly it is judged as immoral, explaining why murder is judged as more immoral than masturbation. Importantly, this process of dyadic comparison is intuitive and need not rely on effortful reason, like moral judgment in general (Haidt, 2001).Although a dyadic template should be reliably present during moral judgments, we acknowledge the influence of other domain-general psychological factors, such as misin-terpreting affective arousal (Wheatley & Haidt, 2005) or rote learning (i.e., “the bible says abortion is wrong”). However, a dyadic template suggests that such misinterpretation and rote learning are easier with more harmful actions. Just as it is easier to rote-learn that a jerboa (a little desert kangaroo) is a mammal than a platypus (which non-prototypically lays eggs and has a bill), it should also be easier to rote-learn that abortion is immoral than to rote-learn that dropping the Torah is immoral.Importantly, we are not suggesting that moral cognition consists of only one moral module (i.e., a foundation) of harm. Dyadic morality, with its roots in psychological con-structionism (Cameron etal., in press), denies the very exis-tence of moral modules. This template is not an on-or-off “switch” but is instead a domain-general process that allows for gradations of harm. It is also activated no matter the con-tent of the norm violation—that is, even when an act initially seems harmless (Gray, Schein, & Ward, 2014). Because harm represents the essence of immorality, it serves as a con-stant backdrop in moral cognition—one that exerts a power-ful cognitive gravity (Schein & Gray, 2014).The Pluralism of Perceived HarmModular theories such as MFT have long argued against such a common template because of the ostensible existence of harmless wrongs. For example, scenarios of consensual incest carefully constructed to be “objectively harmless” are still rated as immoral by participants (Haidt, 2001). However, we argue against the very idea of “objective” harm. Harm, like morality, is in the eye of the beholder. In fact, both harm and morality are rooted in the ambiguous perceptions of other minds. Judgments of immorality require seeing a mind capable of doing evil, and judgments of harm require seeing a mind capable of suffering (i.e., an agent and a patient; Gray & Schein, 2012).The subjective nature of harm means that bizarre “harm-less” scenarios concocted by liberal researchers (e.g., mas-turbating with a dead chicken) may not seem harmless to their more conservative participants. Indeed, many studies document the perception of harm in “harmless” cases of reli-gious blasphemy, anti-patriotism, and aberrant sexuality (DeScioli, Gilbert, & Kurzban, 2012; Kahan, 2007; for a full treatment, see Gray etal., 2014).Consider a case described by anthropologist Richard Shweder (2012): Oriya Hindu Brahmans believe it is extremely immoral for the eldest son to eat chicken immedi-ately after his father’s death. Westerners fail to see this action as wrong—or harmful—viewing it as a mere matter of reli-gious protocol, whereas Hindus consider it the eldest son’s duty to process the father’s “death pollution” through a veg-etarian diet. When the son eats chicken, he “places the father’s spiritual transmigration in deep jeopardy” (Shweder, 2012, p. 96). By understanding the perceived harm in these actions, even Western liberals can understand its perceived immorality.1 Who can deny the immorality of condemning your father to eternal suffering? MFT interprets such perceived harm as mistaken, but dyadic morality sees these perceptions as legitimate. In the language of social anthropology, dyadic morality advocates for not only moral pluralism (accepting the legitimacy of dif-ferent perceptions of morality) but also harm pluralism (accepting the legitimacy of different perceptions of harm). Harm pluralism suggests that different moral content such as purity and loyalty are (less prototypical) varieties of per-ceived harm. In contrast, MFT endorses harm monism, rejecting the legitimacy of harm in anything but direct physi-cal or emotional suffering.Indeed, the very act of separating harm into a specific modular “foundation” denies its perceived existence in moral issues such as treason or sexual impropriety. The harm monism of MFT discounts the harm that conservatives see in matters of religious and sexual propriety (Gray etal., 2014). We suggest that this harm monism stems from the liberal bias in social psychology (Inbar & Lammers, 2012), which also once long denied the legitimacy of moral pluralism. Echoing the cries of moral anthropologists, we suggest that understanding harm requires cultural sensitivity (Shweder, 2012); moral psychol-ogy should prioritize the harm pluralist perceptions of partici-pants over the harm monist theories of researchers.
The Centrality of Harm for Liberals and Conservatives
The diversity of harm provides the possibility for a unify-ing moral template in both liberals and conservatives. Rather than distinct moral mechanisms for each kind of moral content, dyadic morality suggests that immoral acts—even those of “authority” or “purity”—will activate a prototype of harm. Of course, some acts are more harmful than others, and dyadic morality predicts that increased harm (i.e., better template matches) will result in more severe judgments of immorality. Consistent with past research on RWA, we acknowledge political differences between liberals and conservative—and the possibility that these differences may translate to some differences in moral judgment. However, we predict that moral differences between liberals and conservatives have been greatly exag-gerated by MFT (a prediction consistent with Frimer, Biesanz, Walker, & MacKinlay, 2013; Janoff-Bulman & Carnes, 2013; Skitka & Bauman, 2008; Skitka, Morgan, & Wisneski, in press).We suggest that liberals and conservatives share the same dyadic template, rather than categorically different moral minds. A harm-based moral template predicts that harm should be cen-tral in moral cognition across both moral diversity (i.e., many different moral acts) and political orientation (i.e., for both lib-erals and conservatives). Because centrality is a relatively broad concept, we operationalize it through six specific hypotheses:Hypothesis 1 (H1: Accessibility): Harm is most cogni-tively accessible across moral diversity and political ori-entation (Study 1).Hypothesis2(H2: Importance): Harm is most impor-tant across moral diversity and political orientation (Studies 2 and 3).Hypothesis3(H3: Organization): Harm organizes judg-ments of immorality across moral diversity and political orientation (Study 4).Hypothesis4(H4: Overlap): Harm overlaps substan-tially with other moral concerns across political orienta-tion (Study 5).Hypothesis5(H5: Translation): Harm is the best lingua franca for translating across moral diversity and political orientation (Study 6).Hypothesis6 (H6: Association): Harm is more implic-itly associated with moral diversity than descriptively similar concerns, across political orientation (Study 7).
Bipartisan Alliance, a Society for the Study of the US Constitution, and of Human Nature, where Republicans and Democrats meet.
Monday, April 1, 2019
Demonstrate the potential for population structure to create spurious results, especially when using methods that rely on the accumulation of large numbers of small effects
Population Genetics: Why structure matters. Nick Barton, Joachim Hermisson, Magnus Nordborg. eLIFE Sciences, Mar 21 2019. https://elifesciences.org/articles/45380
Abstract: Great care is needed when interpreting claims about the genetic basis of human variation based on data from genome-wide association studies.
---
As the name suggests, GWAS scan the genome for variants – typically single nucleotide polymorphisms (SNPs) – that are associated with a particular condition or trait (phenotype). The first GWAS for height found a small number of SNPs that jointly explained only a tiny fraction of the variation. Because this was in contrast with the high heritability seen in twin studies, it was dubbed ‘the missing heritability problem’ (reviewed in Yang et al., 2010). It was suggested that the problem was simply due to a lack of statistical power to detect polymorphisms of small effect. Subsequent studies with larger sample sizes have supported this explanation: more and more loci have been identified although most of the variation remains ‘unmappable’, presumably because sample sizes on the order of a million are still not large enough (Yengo et al., 2018).
One way in which the unmappable component of genetic variation can be included in a statistical measure is via so-called polygenic scores. These scores sum the estimated contributions to the trait across many SNPs, including those whose effects, on their own, are not statistically significant. Polygenic scores thus represent a shift from the goal of identifying major genes to predicting phenotype from genotype. Originally designed for plant and animal breeding purposes, polygenic scores can, in principle, also be used to study the genetic basis of differences between individuals and groups.
This, however, requires accurate and unbiased estimation of the effects of all SNPs included in the score, which is difficult in a structured (non-homogeneous) population when environmental differences cannot be controlled. To see why this is a problem, consider the classic example of chopstick-eating skills (Lander and Schork, 1994). While there surely are genetic variants affecting our ability to handle chopsticks, most of the variation for this trait across the globe is due to environmental differences (cultural background), and a GWAS would mostly identify variants that had nothing to do with chopstick skills, but simply happened to differ in frequency between East Asia and the rest of the world.
Several methods for dealing with this problem have been proposed. When a GWAS is carried out to identify major genes, it is relatively simple to avoid false positives by eliminating associations outside major loci regardless of whether they are due to population structure confounding or an unmappable polygenic background (Vilhjálmsson and Nordborg, 2013). However, if the goal is to make predictions, or to understand differences among populations (such as the latitudinal cline in height), we need accurate and unbiased estimates for all SNPs. Accomplishing this is extremely challenging, and it is also difficult to know whether one has succeeded.
One possibility is to compare the population estimates with estimates taken from sibling data, which should be relatively unbiased by environmental differences. In one of many examples of this, Robinson et al. used data from the GIANT Consortium (Wood et al., 2014) together with sibling data to estimate that genetic variation contributes significantly to height variation across Europe (Robinson et al., 2015). They also argued that selection must have occurred, because the differences were too large to have arisen by chance. Using estimated effect sizes provided by Robinson et al., a more sophisticated analysis by Field et al. found extremely strong evidence for selection for height across Europe (p=10−74; Field et al., 2016). Several other studies reached the same conclusion based on the GIANT data (reviewed in Berg et al., 2019; Sohail et al., 2019).
Berg et al. (who are based at Columbia University, Stanford University, UC Davis and the University of Copenhagen) and Sohail et al. (who are based at Harvard Medical School, the Broad Institute, and other institutes in the US, Finland and Sweden) now re-examine these conclusions using the recently released data from the UK Biobank (Sudlow et al., 2015). Estimating effect sizes from these data allows possible biases due to population structure confounding to be investigated, because the UK Biobank data comes from a (supposedly) more homogenous population than the GIANT data.
Using these new estimates, Berg et al. and Sohail et al. independently found that evidence for selection vanishes – along with evidence for a genetic cline in height across Europe. Instead, they show that the previously published results were due to the cumulative effects of slight biases in the effect-size estimates in the GIANT data. Surprisingly, they also found evidence for confounding in the sibling data used as a control by Robinson et al. and Field et al. This turned out to be due to a technical error in the data distributed by Robinson et al. after they published their paper.
This means we still do not know whether genetics and selection are responsible for the pattern of height differences seen across Europe. That genetics plays a major role in height differences between individuals is not in doubt, and it is also clear that the signal from GWAS is mostly real. The issue is that there is no perfect way to control for complex population structure and environmental heterogeneity. Biases at individual loci may be tiny, but they become highly significant when summed across thousands of loci – as is done in polygenic scores. Standard methods to control for these biases, such as principal component analysis, may work well in simulations but are often insufficient when confronted with real data. Importantly, no natural population is unstructured: indeed, even the data in the UK Biobank seems to contain significant structure (Haworth et al., 2019).
Berg et al. and Sohail et al. demonstrate the potential for population structure to create spurious results, especially when using methods that rely on large numbers of small effects, such as polygenic scores. Caution is clearly needed when interpreting and using the results of such studies. For clinical predictions, risks must be weighed against benefits (Rosenberg et al., 2019). In some cases, such as recommendations for more frequent medical checkups for patients found at higher ‘genetic’ risk of a condition, it may not matter greatly whether predictors are confounded as long as they work. By contrast, the results of behavioral studies of traits such as IQ and educational attainment (Plomin and von Stumm, 2018) must be presented carefully, because while the benefits are far from obvious, the risks of such results being misinterpreted and misused are quite clear. The problem is worsened by the tendency of popular media to ignore caveats and uncertainties of estimates.
Finally, although quantitative genetics has proved highly successful in plant and animal breeding, it should be remembered that this success has been based on large pedigrees, well-controlled environments, and short-term prediction. When these methods have been applied to natural populations, even the most basic predictions fail, in large part due to poorly understood environmental factors (Charmantier et al., 2014). Natural populations are never homogeneous, and it is therefore misleading to imply there is a qualitative difference between ‘within-population’ and ‘between-population’ comparisons – as was recently done in connection with James Watson’s statements about race and IQ (Harmon, 2019). With respect to confounding by population structure, the key qualitative difference is between controlling the environment experimentally, and not doing so. Once we leave an experimental setting, we are effectively skating on thin ice, and whether the ice will hold depends on how far out we skate.
Abstract: Great care is needed when interpreting claims about the genetic basis of human variation based on data from genome-wide association studies.
---
As the name suggests, GWAS scan the genome for variants – typically single nucleotide polymorphisms (SNPs) – that are associated with a particular condition or trait (phenotype). The first GWAS for height found a small number of SNPs that jointly explained only a tiny fraction of the variation. Because this was in contrast with the high heritability seen in twin studies, it was dubbed ‘the missing heritability problem’ (reviewed in Yang et al., 2010). It was suggested that the problem was simply due to a lack of statistical power to detect polymorphisms of small effect. Subsequent studies with larger sample sizes have supported this explanation: more and more loci have been identified although most of the variation remains ‘unmappable’, presumably because sample sizes on the order of a million are still not large enough (Yengo et al., 2018).
One way in which the unmappable component of genetic variation can be included in a statistical measure is via so-called polygenic scores. These scores sum the estimated contributions to the trait across many SNPs, including those whose effects, on their own, are not statistically significant. Polygenic scores thus represent a shift from the goal of identifying major genes to predicting phenotype from genotype. Originally designed for plant and animal breeding purposes, polygenic scores can, in principle, also be used to study the genetic basis of differences between individuals and groups.
This, however, requires accurate and unbiased estimation of the effects of all SNPs included in the score, which is difficult in a structured (non-homogeneous) population when environmental differences cannot be controlled. To see why this is a problem, consider the classic example of chopstick-eating skills (Lander and Schork, 1994). While there surely are genetic variants affecting our ability to handle chopsticks, most of the variation for this trait across the globe is due to environmental differences (cultural background), and a GWAS would mostly identify variants that had nothing to do with chopstick skills, but simply happened to differ in frequency between East Asia and the rest of the world.
Several methods for dealing with this problem have been proposed. When a GWAS is carried out to identify major genes, it is relatively simple to avoid false positives by eliminating associations outside major loci regardless of whether they are due to population structure confounding or an unmappable polygenic background (Vilhjálmsson and Nordborg, 2013). However, if the goal is to make predictions, or to understand differences among populations (such as the latitudinal cline in height), we need accurate and unbiased estimates for all SNPs. Accomplishing this is extremely challenging, and it is also difficult to know whether one has succeeded.
One possibility is to compare the population estimates with estimates taken from sibling data, which should be relatively unbiased by environmental differences. In one of many examples of this, Robinson et al. used data from the GIANT Consortium (Wood et al., 2014) together with sibling data to estimate that genetic variation contributes significantly to height variation across Europe (Robinson et al., 2015). They also argued that selection must have occurred, because the differences were too large to have arisen by chance. Using estimated effect sizes provided by Robinson et al., a more sophisticated analysis by Field et al. found extremely strong evidence for selection for height across Europe (p=10−74; Field et al., 2016). Several other studies reached the same conclusion based on the GIANT data (reviewed in Berg et al., 2019; Sohail et al., 2019).
Berg et al. (who are based at Columbia University, Stanford University, UC Davis and the University of Copenhagen) and Sohail et al. (who are based at Harvard Medical School, the Broad Institute, and other institutes in the US, Finland and Sweden) now re-examine these conclusions using the recently released data from the UK Biobank (Sudlow et al., 2015). Estimating effect sizes from these data allows possible biases due to population structure confounding to be investigated, because the UK Biobank data comes from a (supposedly) more homogenous population than the GIANT data.
Using these new estimates, Berg et al. and Sohail et al. independently found that evidence for selection vanishes – along with evidence for a genetic cline in height across Europe. Instead, they show that the previously published results were due to the cumulative effects of slight biases in the effect-size estimates in the GIANT data. Surprisingly, they also found evidence for confounding in the sibling data used as a control by Robinson et al. and Field et al. This turned out to be due to a technical error in the data distributed by Robinson et al. after they published their paper.
This means we still do not know whether genetics and selection are responsible for the pattern of height differences seen across Europe. That genetics plays a major role in height differences between individuals is not in doubt, and it is also clear that the signal from GWAS is mostly real. The issue is that there is no perfect way to control for complex population structure and environmental heterogeneity. Biases at individual loci may be tiny, but they become highly significant when summed across thousands of loci – as is done in polygenic scores. Standard methods to control for these biases, such as principal component analysis, may work well in simulations but are often insufficient when confronted with real data. Importantly, no natural population is unstructured: indeed, even the data in the UK Biobank seems to contain significant structure (Haworth et al., 2019).
Berg et al. and Sohail et al. demonstrate the potential for population structure to create spurious results, especially when using methods that rely on large numbers of small effects, such as polygenic scores. Caution is clearly needed when interpreting and using the results of such studies. For clinical predictions, risks must be weighed against benefits (Rosenberg et al., 2019). In some cases, such as recommendations for more frequent medical checkups for patients found at higher ‘genetic’ risk of a condition, it may not matter greatly whether predictors are confounded as long as they work. By contrast, the results of behavioral studies of traits such as IQ and educational attainment (Plomin and von Stumm, 2018) must be presented carefully, because while the benefits are far from obvious, the risks of such results being misinterpreted and misused are quite clear. The problem is worsened by the tendency of popular media to ignore caveats and uncertainties of estimates.
Finally, although quantitative genetics has proved highly successful in plant and animal breeding, it should be remembered that this success has been based on large pedigrees, well-controlled environments, and short-term prediction. When these methods have been applied to natural populations, even the most basic predictions fail, in large part due to poorly understood environmental factors (Charmantier et al., 2014). Natural populations are never homogeneous, and it is therefore misleading to imply there is a qualitative difference between ‘within-population’ and ‘between-population’ comparisons – as was recently done in connection with James Watson’s statements about race and IQ (Harmon, 2019). With respect to confounding by population structure, the key qualitative difference is between controlling the environment experimentally, and not doing so. Once we leave an experimental setting, we are effectively skating on thin ice, and whether the ice will hold depends on how far out we skate.
The decline in adolescent substance use across Europe & North America in the early twenty-first century: Associated with declines in face-to-face contact, not with increases in use of electronic media
The decline in adolescent substance use across Europe and North America in the early twenty-first century: A result of the digital revolution? Margaretha De Looze et al. International Journal of Public Health, March 2019, Volume 64, Issue 2, pp 229–240. https://link.springer.com/article/10.1007/s00038-018-1182-7
Abstract
Objectives: Increases in electronic media communication (EMC) and decreases in face-to-face peer contact in the evening (FTF) have been thought to explain the recent decline in adolescent substance use (alcohol, tobacco, cannabis). This study addresses this hypothesis, by examining associations between (time trends in) EMC, FTF, and substance use in more than 25 mainly European countries.
Methods: Using 2002–2014 data from the international Health Behaviour in School-aged Children (HBSC) study, we ran multilevel logistic regression analyses to investigate the above associations.
Results: National declines in substance use were associated with declines in FTF, but not with increases in EMC. At the individual level, both EMC and FTF related positively to substance use. For alcohol and cannabis use, the positive association with EMC was stronger in more recent years. Associations between EMC and substance use varied across countries, but this variation could not be explained by the proportion of young people using EMC within countries.
Conclusions: Our research suggests that the decrease in FTF, but not the increase in EMC, plays a role in the recent decrease in adolescent substance use.
Keywords: Adolescence Substance use Tobacco Alcohol Cannabis Electronic media communication Internet Trends over time Europe Time spent with friends
Abstract
Objectives: Increases in electronic media communication (EMC) and decreases in face-to-face peer contact in the evening (FTF) have been thought to explain the recent decline in adolescent substance use (alcohol, tobacco, cannabis). This study addresses this hypothesis, by examining associations between (time trends in) EMC, FTF, and substance use in more than 25 mainly European countries.
Methods: Using 2002–2014 data from the international Health Behaviour in School-aged Children (HBSC) study, we ran multilevel logistic regression analyses to investigate the above associations.
Results: National declines in substance use were associated with declines in FTF, but not with increases in EMC. At the individual level, both EMC and FTF related positively to substance use. For alcohol and cannabis use, the positive association with EMC was stronger in more recent years. Associations between EMC and substance use varied across countries, but this variation could not be explained by the proportion of young people using EMC within countries.
Conclusions: Our research suggests that the decrease in FTF, but not the increase in EMC, plays a role in the recent decrease in adolescent substance use.
Keywords: Adolescence Substance use Tobacco Alcohol Cannabis Electronic media communication Internet Trends over time Europe Time spent with friends
Women in Science preferred spatial toys in childhood more than women in Arts; they scored higher in mental rotation when they had preferred spatial toys; sport practice related with mental rotation performance
Childhood preference for spatial toys. Gender differences and relationships with mental rotation in STEM and non-STEM students. Angelica Moè, Petra Jansen, Stefanie Pietsch. Learning and Individual Differences, Volume 68, December 2018, Pages 108-115. https://doi.org/10.1016/j.lindif.2018.10.003
Highlights
• Women in Science preferred spatial toys in childhood more than women in Arts.
• They scored higher in mental rotation when they had preferred spatial toys.
• Sport practice related with mental rotation performance.
Abstract: Women tend not to choose STEM degrees, for a number of reasons associated with aptitudes, motivation and experience with certain spatial tasks such as mental rotation. This study considered an unexplored experiential factor: childhood preference for spatial toys and sports. It was predicted that the higher the preference for spatial activities in childhood, the higher the mental rotation performance and intrinsic motivation, and likewise the greater the probability of choosing a STEM degree. One hundred seventy-six Italian and German students attending the first year of either a STEM (n = 90) or a no STEM (n = 86) degree filled in the Mental Rotation Test, a self-report to assess intrinsic motivation, and two questionnaires to assess their actual practice with spatial sports and their childhood preference for either spatial or non-spatial toys and sports. The results showed that women in STEM degrees preferred spatial toys more than women in non-STEM degrees and performed better in mental rotation when preferred spatial toys in childhood. The discussion focuses on the relationship between childhood toy preferences and the choice of a STEM degree.
Highlights
• Women in Science preferred spatial toys in childhood more than women in Arts.
• They scored higher in mental rotation when they had preferred spatial toys.
• Sport practice related with mental rotation performance.
Abstract: Women tend not to choose STEM degrees, for a number of reasons associated with aptitudes, motivation and experience with certain spatial tasks such as mental rotation. This study considered an unexplored experiential factor: childhood preference for spatial toys and sports. It was predicted that the higher the preference for spatial activities in childhood, the higher the mental rotation performance and intrinsic motivation, and likewise the greater the probability of choosing a STEM degree. One hundred seventy-six Italian and German students attending the first year of either a STEM (n = 90) or a no STEM (n = 86) degree filled in the Mental Rotation Test, a self-report to assess intrinsic motivation, and two questionnaires to assess their actual practice with spatial sports and their childhood preference for either spatial or non-spatial toys and sports. The results showed that women in STEM degrees preferred spatial toys more than women in non-STEM degrees and performed better in mental rotation when preferred spatial toys in childhood. The discussion focuses on the relationship between childhood toy preferences and the choice of a STEM degree.
Girls read more than boys: It also happens with digital devices; cause is much bigger motivation to read
Gender gap in reading digitally? Examining the role of motivation and self-concept. Nele McElvany, Franziska Schwabe. Journal for Educational Research Online, Vol 11, No 1 (2019), http://www.j-e-r-o.com/index.php/jero/article/view/882
Abstract: Reading is a core prerequisite for educational success and participation in society. However, comprehensive empirical research is needed to understand how reading may differ in a digitalized world. The current study addressed the gender gap in reading digitally. It investigated competence scores along with information on (a) reading and (b) digital motivation and self-concept in 588 elementary school students. Results revealed a gender gap in reading digitally, in reading motivation and self-concept, and in motivation and self-concept in respect to working on digital devices. Only reading motivation variables predicted reading digitally, thereby providing important information on the validity of digitally based reading tests. Reading motivation was found to fully mediate the gender effect on reading digitally. Results have important implications for research and practice.
Keywords: Reading; Motivation; Digitalization; Elementary school
.
Abstract: Reading is a core prerequisite for educational success and participation in society. However, comprehensive empirical research is needed to understand how reading may differ in a digitalized world. The current study addressed the gender gap in reading digitally. It investigated competence scores along with information on (a) reading and (b) digital motivation and self-concept in 588 elementary school students. Results revealed a gender gap in reading digitally, in reading motivation and self-concept, and in motivation and self-concept in respect to working on digital devices. Only reading motivation variables predicted reading digitally, thereby providing important information on the validity of digitally based reading tests. Reading motivation was found to fully mediate the gender effect on reading digitally. Results have important implications for research and practice.
Keywords: Reading; Motivation; Digitalization; Elementary school
.
Welfare implications of labor market power: Declining labor market concentration increase labor's share of income by 2.89 pct points 1976-2014, suggesting that labor market concentration is not the reason for a declining labor share
Labor Market Power. David W. Berger, Kyle F. Herkenhoff, Simon Mongey. NBER Working Paper No. 25719. March 2019. https://www.nber.org/papers/w25719
Abstract: What are the welfare implications of labor market power? We provide an answer to this question in two steps: (1) we develop a tractable quantitative, general equilibrium, oligopsony model of the labor market, (2) we estimate key parameters using within-firm-state, across-market differences in wage and employment responses to state corporate tax changes in U.S. Census data. We validate the model against recent evidence on productivity-wage pass-through, and new measurements of the distribution of local market concentration. The model implies welfare losses from labor market power that range from 2.9 to 8.0 percent of lifetime consumption. However, despite large contemporaneous losses, labor market power has not contributed to the declining labor share. Finally, we show that minimum wages can deliver moderate, and limited, welfare gains by reallocating workers from smaller to larger, more productive firms.
Tyler Cowen summarizing (Some implications of monopsony models, Mar 01 2019, https://marginalrevolution.com/marginalrevolution/2019/04/some-implications-of-monopsony-models.html):
Abstract: What are the welfare implications of labor market power? We provide an answer to this question in two steps: (1) we develop a tractable quantitative, general equilibrium, oligopsony model of the labor market, (2) we estimate key parameters using within-firm-state, across-market differences in wage and employment responses to state corporate tax changes in U.S. Census data. We validate the model against recent evidence on productivity-wage pass-through, and new measurements of the distribution of local market concentration. The model implies welfare losses from labor market power that range from 2.9 to 8.0 percent of lifetime consumption. However, despite large contemporaneous losses, labor market power has not contributed to the declining labor share. Finally, we show that minimum wages can deliver moderate, and limited, welfare gains by reallocating workers from smaller to larger, more productive firms.
Tyler Cowen summarizing (Some implications of monopsony models, Mar 01 2019, https://marginalrevolution.com/marginalrevolution/2019/04/some-implications-of-monopsony-models.html):
More workers ought to be in larger firms, as those firms are afraid to hire more, knowing that bids up wages for everyone. Therefore (ceteris paribus) the large firms in the economy ought to be larger.Raising the legal minimum wage also reallocates workers into larger firms, and again makes them larger.Tough stuff if you worry a lot about both monopoly and monopsony at the same time -- choose your poison!I have adapted those points from a recent paper by David Berger, Kyle Herkenhoff, and Simon Mongey, "Labor Market Power." On the empirics, they conclude: "Our theory implies that this declining labor market concentration increase labor's share of income by 2.89 percentage points between 1976 and 2014, suggesting that labor market concentration is not the reason for a declining labor share." So the paper makes no one happy (good!): monopsony is significant, but has been declining in import.