A regression approach to testing genetic influence on communication behavior: Social media use as an example

A behavior genetics perspective suggests both social and biological forces influence human behavior, including highly specialized media and communication behaviors. In this paper, I use a behavior genetics framework and twin study data from the 2013 Midlife in the United States (MIDUS III) survey to examine how both environmental and genetic factors contribute to social media use. By applying a straightforwarddand easily replicabledanalytical extension to linear regression called DeFries-Fulker (DF) regression, I demonstrate that approximately one-to two-thirds of variance in social media use is attributable to additive genetic traits; unique and shared environmental factors account for the remainder of variance. In addition to showing social media use is partially motivated by underlying genetic traits, this paper, more importantly, provides an analytical blueprint for using DF regression in future investigations of genetic influence on communication behaviors and media effects. © 2017 Elsevier Ltd. All rights reserved. Behavior genetics research suggests complex human characteristicsdfrom political ideology to cigarette smokingdare motivated by heritable biological traits in addition to the environmental forces of culture, parents, peers, and institutions (e.g., Do et al., 2015; Plomin, DeFries, & Fulker, 1988; Plomin, DeFries, Knopik, & Neiderheiser, 2013). From a behavior genetics perspective, sophisticated human beliefs, attitudes, and behaviors are byproducts of a rich interplay between genes underpinning human neurological function and exposure to the social environment, between our biological nature and the nurturewe receive from theworld around us, not one influence exclusive of the other. This theoretical logicdthat both genes and environment play an interactive role in shaping observable human activitydhas been applied to a broad spectrum of behaviors investigated throughout the social sciences (Polderman et al., 2015), including human communication and media consumption patterns (Cappella, 1991, 1996; Sherry, 2001, 2004; Shoemaker, 1996), and while a handful of empirical studies have borne out its theoretical postulates in observable detail (e.g., Kirzinger, Weber, & Johnson, 2012), the extent to which genes influence the use of newer information and communication technologies such as social network sites (e.g., Facebook) remains unclear. In the past, motivations for social media exposure have been primarily investigated through a uses and gratifications (U&G) framework that proposes individuals actively seek media content to fulfill goal-oriented psychological needs, such as the need to obtain new information and to escape from reality (Katz, 1959; Katz, Blumler, Geurovitch, 1973; Katz, Haas, & Gurevitch, 1973). U&G research on social media suggests users indeed turn to sites like Facebook and Twitter to seek news, establish and maintain social relationships, and for personal amusement, among other reasons (Chen, 2011; Krause, North, & Heritage, 2014; Park, Kee, & Valenzuela, 2009; Quan-Haase & Young, 2010; Raacke & BondsRaacke, 2008; Sheldon & Bryant, 2016; Tosun, 2012; Zhang, Tang, & Leung, 2011), although it is unclear the extent to which such consciously articulated motivations are guided by antecedent genetic variation that is foundational to neuroanatomical structure and cognitive processing, and thus, perceived psychological needs satisfied by using social media. In this article, I apply a behavior genetics framework to a novel set of online communication activities: frequency of using social media sites such as Facebook to communicate with friends and family. Using survey data from the University ofWisconsin's Midlife in the United States (MIDUS) Longitudinal Study of Health & WellBeing collected on identical and fraternal twin pairs and a simple extension of linear regression called DeFries-Fulker (DF) regression, I demonstrate that individual genetic traits explain a non-trivial amount of variance in frequency of social media use; the impact of a person's environment, including shared experiences (e.g., 2 Additionally, the authors showed that the majority of traits (69%) are best explained “with a simple and parsimonious model where the observed variation is solely due to additive genetic variation” (p. 1, italics mine). This is a striking finding, but it is important to note that the behavior genetics approach argues that genes influence rather than determine human characteristics (see Plomin et al., 2013). As Cappella (1996) suggests, “The fallacy that biological approaches to social behavior determine an individual's actions is based on naïve and long-outmoded theories” (p. 5, italics mine). Perhaps it could be argued that behavior genetics is “deterministic” in cases of physical anatomy, where a purely genetic model that omits environmental influence may best explain phenotypic variation. In twin studies of social behavior, however, such characteristics are often better explained by models possessing both genetic and environmental components. It is worth noting again, in other words, that the behavior genetics paradigm does not suggest that explicit, observable behaviors such as cigarette smoking are literally and directly passed C. York / Computers in Human Behavior 73 (2017) 100e109 101 parent-child socialization) and unique individual experiences (e.g., peer influence) account for the remainder. Taken together, the results of this study suggest that social media use is at least partially influenced by genetic traits, which is a result consistent with previous research linking genetic traits to communication and media consumption patterns (Kirzinger et al., 2012). In the Discussion section, I examine the implications of the study for U&G theory as it pertains to social media use. I also discuss the application of the behavior genetics perspective and DF regression to other areas of communication and media effect research. As I later suggest, DF regression offers communication researchers a straightforward approach to analyzing the relative impact of genes and environment on communication behaviors compared to the highly specialized structural equation and Bayesian approaches that are largely taught in research methods courses outside the social sciences. 1. Literature review 1.1. The quantitative behavior genetics approach Quantitative behavior genetics has foundations in the work of R.A. Fisher (1919) and Sewell Wright (1921), both of whom highlighted a basic principle onwhich the behavior genetics perspective rests, which is that “If genetic factors affect a quantitative trait, phenotypic [observable] resemblance of relatives [on that trait] should increase with increasing degrees of genetic relatedness” (Plomin et al., 2013, p. 34). Put simply, as the amount of genetic variation shared in common with kin increases (from first-degree cousins, to half siblings, to full siblings, and so on) the greater the amount of behavioral similarity can be attributed to genetics as opposed to the environment (socialization, enculturation, etc.). Because the degree of genetic relatedness between fraternal twins is already known to be 0.5 and identical twins 1.0, researchers can leverage survey data collected on twin pairs and established degrees of genetic affiliation to “assess the relative contributions of nature [genetic relatedness set at 0.5 or 1.0] and nurture” (p. 85). Specifically, by using a variance components approach to twin survey data, behavior geneticists attempt to decompose or partition the variance of an observed behavioral, perceptual, or attitudinal trait possessed by fraternal and identical twin respondents “into the constituent parts of genetic and environmental sources of variation” (p. 377). The variance components approach commonly used in behavior genetics twin study research assumes an observable human trait or phenotype (P), such as news-reading or voting, is comprised of just three components: genetic relatedness or “heritability” (notated A), exposure to the shared environment (C), and exposure to the unique environment (E). Mathematically, then, any observed trait, including any empirically assessed communication behaviors, can be explained by the simple equation: P 1⁄4 (A þ C þ E). Put another way, this logic suggests the degree to which a twin and their cotwin vary on an observed communication variable (P) is a function of genetic resemblance, exposure to the same shared environment (e.g., parent socialization of both twins in the same household), and an array of unique events that are experienced by one twin and not the other (having a different first-grade teacher, 1 It should be noted that the behavior genetics perspective does not assume that observed variation on a given trait analyzed in a twin study directly affects the behavior. There is no “news gene” that can be directly assessed in a twin study, for example. Behavior genetics research assumes, rather, that the impact of genetic variation captured in a twin survey study underpins neuroanatomy, cognition, and personality differences that more explicitly influence the observed perceptual, attitudinal, and behavioral variables (Sherry, 2004, p. 98; Weber et al., 2008). being involved in different sports, interacting with distinct peer groups, etc.). The overarching goal of behavior genetics research is thus to assess the extent to which individual beliefs, attitudes, and behaviors are attributable to aspects of the individual's environment and to genetic traits.1 For almost a century, the behavior genetics approach has yielded analytically consistent results across the social sciences. Using adoption, sibling, and twin study survey designs, quantitative behavioral geneticists have shown that genes influence a wide range of complex human traits, including strength of political ideology, political affiliation, political participation, risk-taking, schizophrenia, depression, cigarette smoking, psychological temperament, cognitive ability, cognitive function, IQ, Machiavellianism, empathy, social attitudes, anxiety, neuroticism, height, weight, and anatomical variation, among many others (Bouchard, Lykken, McGue, Segal, & Tellegen, 1990; Do et al., 2015; Ferguson, Munoz, Winegard, & Winegard, 2012; Hatemi, Smith, Alford, Martin, & Hibbing, 2015; Hatemi et al., 2010; Hibbing, Smith, & Alford, 2013; Polderman et al., 2015; Smith & Hatemi, 2013). In fact, Polderman et al. (2015) meta-analysis of over 17,000 human traits studied across over 2700 twin studies estimated that the average heritability of any observed human trait is 49%, leaving environmental factors such as parent socialization an enculturation to explain the remainder of variation in individual characteristics.2 Results such as these suggest that while childhood socialization and conscious, goal-oriented decision-making play key roles in guiding human behavior, unarticulated, latent genetic influence is just as crucial to neuroanatomical composition, personality, and specific cognitions that guide performance of complex behaviors such as consuming media content. From a behavior genetics perspective, then, it is likely that the satisfaction of psychological needs through media use is at least partially rooted in the antecedent variables of genetic variation and neuroanatomy, although it is not within the scope of this study to directly explore such associations. Here I simply rely on the basic theoretical premise of quantitative behavior genetics: Genes, in addition to the environment, guide individual thoughts and actions, and there are analytical means by which the relative contribution of each can be assessed. Such an approach does not infringe on the utility or function of traditional media selection theories such as uses and gratifications (U&G); instead, as I argue, behavior genetics approach can contextualize U&G in a broader model of media choice and effects. through genes from parent to child; an individual does not, for instance, inherit the specific behavior of cigarette smoking from their parents. Rather, parents contribute genes to their children that form the neuroanatomical foundations of certain psychological and physiological characteristics (e.g., a propensity for addiction) that, combined with environmental stimuli (access to cigarettes and peer pressure to smoke), result in a measurable behavior like frequency of cigarette smoking (Plomin et al., 2013; Weber et al., 2008). Genes may, in this sense, be drivers of neuroanatomy, cognition, personality, temperament, and even specific psychological needs that individuals report seeking to satisfy through interpersonal communication and exposure to media content. C. York / Computers in Human Behavior 73 (2017) 100e109 102 1.2. Uses and gratifications of social media use Uses and gratifications (U&G) is an approach to media behavior that investigates reasons why individuals select media content. Unlike the “direct” or “hypodermic needle” media effects theories of the early 20th century that suggested individuals had little control over media's powerful persuasive influence, the U&G framework states that “people bend the media to their needs more readily than the media overpower them” (Katz, Haas, et al., 1973, pp. 164e165). Rather than acting as passive vessels through which media exerts persuasive influence, the U&G paradigm hypothesizes that individuals actively and consciously select media contentdincluding newspaper, radio, television and internet contentdwith expectations that these materials will satisfy specific psychological needs, such as needs for information, entertainment and escape, and socializing with peers (see Katz, 1959; Katz, Blumler, et al., 1973). A variety of more recent U&G studies have examined psychological needs satisfied by using social media sites such as Facebook. Using self-reported survey data typically collected on conveniently sampled college students, these studies find users seek social media content to satisfy peer interaction and relationship needs, including establishing and maintaining romantic associations (Tosun, 2012). Similar studies have found that users turn to social media sites “to thank people,” “to show others encouragement,” “because [they] need to talk about [their] problems sometimes,” and “to make friends of the opposite sex,” among other needs such as entertainment and escape (Quan-Haase & Young, 2010, p. 356; see also; Raacke & Bonds-Raacke, 2008). Additionally, individuals have reported using social media sites to engage in unobserved “social surveillance” of friends' activities (Zhang et al., 2011), document their lives through photographs (Sheldon & Bryant, 2016), listen to music via platform applications (Krause et al., 2014), experience “camaraderie” among weak social ties (Chen, 2011), and obtain information about real-world events (Park et al., 2009). Drawing on the U&G literature, McAndrew and Jeong (2012) suggested that patterns of social media use may also be influenced by an evolutionary psychology mechanism involving the unconscious desire to transmit one's genes to the next generation. Specifically, McAndrew and Jeong (2012) hypothesized that individuals might frequently, if unconsciously, use social media to “gossip” about same-sex and same-age friends because of a need to “keep tabs on ... competitors for status and mates” who are typically “those in our own age and sex cohorts” (p. 2360). The authors found that women, younger individuals, and those not currently in a committed relationship were among the most active users surveyed about their social media habits, which are findings consistent with public opinion data on the demographics of social media use (Perrin, 2015) and suggest that differences in perceived needs to use social media could have as much to do with the basic biological drives to maximize inclusive fitnessdthe spread of one's genes via reproduction (Williams, 1966)das they do with sexand genderbased differences in parent-child socialization of social media behavior, for example. Genetically motivated and unarticulated needs, in other words, may be among those crucial needs that influence howdand how frequentlydindividuals use social media. 1.3. Genetic and environmental foundations of social media use The notion that genetic traits influence communication behavior is not new, and has in fact been fleshed out theoretically in several essays. For instance, Shoemaker (1996) suggested that genetic traits may play a key role in the modern production and consumption of newsmedia content because individuals have been naturally selected to pay close attention to deviant events in their surroundings that may “pose potential threats” to human survival and reproduction, whether these events are observed in the immediate surroundings or, now, via news media (p. 32). Likewise, Cappella (1991) argued automated patterns of interpersonal interaction are embedded in biological traits; specifically, Cappella (1991) suggested that two types of automated patterns in interpersonal communicationdstimulation regulation and emotional responsivenessdare rooted in increasing “inclusive fitness of the species” (p. 18); in other words, genetic motivations are central to factors underpinning interpersonal communication. Similar work by John Sherry, Michael Beatty and colleagues has shown that individual personality traits, especially psychological temperament, are shaped by genetic variation underlying differences in neuroanatomy, which in turn motivate a wide range of communication and media behaviors, including verbal aggressiveness (Beatty & McCroskey, 1997), communication apprehension (Beatty, McCroskey, & Heisel, 1998), communication adaptability (Beatty, Marshall, & Rudd, 2001), interpersonal affiliation (Beatty, Heisel, Hall, Levine, & France, 2002), tolerance for disagreement (McCroskey, Heisel, & Richmond, 2001), communicator style (Horvath,1995),willingness to communicate (Hazel,Wongprasert,& Ayres, 2006), television use (Sherry, 2001) and a broad assortment of other communication and mass media consumption behaviors (Sherry, 2004). Summarizing this work, McCroskey and Beatty (2000) suggest, “The view is that while nurture certainly has some effects (via cultural influences, formal education, experience, etc.) nature has set forth in one's genetic code most of what one will become and do ...” with “... inborn, neurobiological structures [being] responsible for communication behavior and associated processes” (p. 2; see also Weber, Sherry, & Mathiak, 2008). As was mentioned above, a common methodological and analytical approach researchers use to examine the degree towhich communication and media behaviors are indeed “inborn” or heritable involves the use of twin and adoption surveys. One such study examining how frequently adoptive and nonadoptive sibling pairs used television found that “that 34% of the variance in television viewing at [age] 4 and 30% of the variance at 5 is due to genetic influences” (Plomin, Corley, DeFries, & Fulker, 1990, p. 375). Other twin studies of media and communication behavior have found that genetics influence individual levels of computer self-efficacy (Deryakulu, Mcilroy, Ursavaş, & Çalışkan, 2016), problematic internet behavior (Deryakulu& Ursavaş, 2014), frequency of talking and texting on mobile phones (Miller, Zhu, Wright, Hansell, & Martin, 2012), the effects of television use on aggressive behavior in adolescents (Rowe & Herstand, 1986), and the relationship between TV viewing and antisocial behaviors (Schwartz & Beaver, 2015). In one other twin study, the authors demonstrated that genes, in addition to environment, were primarily responsible for how frequently respondents engaged in interpersonal discussion and how frequently they used computers, video games, television, the internet, and news media (Kirzinger et al., 2012). Across all behaviors studied, the authors found that “one fifth to one third of the variance in media consumption and communication behaviors [were] explained by additive genetic factors” (p. 159). Additionally, for many of the communication behaviors studied, a model including only genetic (A) and unique environmental (E) factors provided more parsimonious explanation of variance than did models also including an indicator of shared environmental influence (C) (pp. 156e158).