Discussion
Few studies have investigated the psychological predictors of risky health behaviors at an aggregated country level. The present study addressed this gap in the literature by obtaining reliable estimates of health and psychological variables from big publicly available datasets comprising millions of individuals, and using them to test theoretical predictions.
How Do Country-Level Values and Traits Relate to Risky Health Behaviors?
Cultural values: A recent study using similar datasets has demonstrated that cultural values show a particular pattern of association with alcohol consumption across countries (Inman et al., 2017). Specifically, alcohol consumption was greater in countries that value harmony and autonomy, and lower in countries that value embeddedness and hierarchy. Given that the present study used the same data for cultural values, it was unsurprising that the partial correlations in the present study replicated those of Inman et al. Specifically, significant correlations were observed for harmony (positive) and embeddedness (negative), although the effect sizes were small.
Beyond alcohol, the present study was interested in how values would be associated with smoking and obesity. The pattern of partial correlations for these health indicators suggested some similarity with alcohol consumption. For example, the direction of the associations between alcohol, smoking, and obesity were the same for hierarchy (negative) and mastery (negative). There were, however, some clear differences. Alcohol and smoking were positively correlated with harmony, but obesity appeared to be unrelated. Inman et al. (2017) explained that harmony may be positively correlated to alcohol consumption at the country level because cultures high in harmony regulate how their members relate to the social world via an emphasis on appreciation and “fitting in” (Schwartz, 2006). Such an emphasis may promote “having a good time” as a social motive and thus encourage individuals to engage in risky social behaviors such as drinking, smoking, and illicit drug use. In contrast, obesity may be unrelated to harmony values as its associated risky behaviors are less social in nature.
A second finding was that hierarchy was negatively associated with alcohol consumption. On the other hand, obesity appeared unrelated to embeddedness, and smoking had a positive correlation. Cultures that value hierarchy emphasize responsible behavior in line with rules assigned to their respective roles (Schwartz, 2006). Alcohol has long been considered a threat to public order (Mold, 2018), and contemporary evidence links alcohol consumption to socially undesirable behaviors and health problems (World Health Organization, 2014), both of which are threatening hierarchies. Smoking and obesity, on the other hand, are not linked to disinhibited social behaviors in the same manner as alcohol (e.g., intoxicated behavior) and thus may not be considered as threatening to authorities.
Finally, egalitarianism had a significant negative association with smoking but was unrelated to alcohol, and obesity. This can be understood by considering that egalitarian cultures socialize their populations to feel concern for everyone’s welfare and to act for the benefit of others (Schwartz, 2006). Smoking may be negatively related to egalitarianism because it is a behavior that presents a risk to others, via secondhand smoke, as well as the individual. Indeed, there are a wide range of risks associated with passive smoking including heart disease, stroke and cancer (U.S. Department of Health and Human Services, 2006), and studies within individual countries, such as the U.S.A, have shown that a large proportion of people perceive secondhand smoke as being harmful (Kruger et al., 2016). In short, egalitarian populations may be less likely to smoke because their members try and act in a manner that does not risk the health of others.
Big-5 traits: The study expands on Inman et al. (2017), Mackenbach (2014) and others by also considering country-level aggregates of personality traits. Firstly, it was evident that neither alcohol consumption nor smoking was linked to aggregate personality. Partial correlations did, however, hint that countries with high aggregated scores for openness to new experiences had increased alcohol consumption (although this correlation fell short of being statistically significant, p = .052).
Unlike smoking and alcohol, obesity did present a clear pattern of associations with aggregate personality traits. Specifically, obesity had a strong positive association with extraversion, and a moderate negative association with neuroticism. These findings add to a growing body of evidence that links extraversion to obesity at a country level (McCrae & Terracciano, 2008) and concur with those of several studies that have linked extraversion to increased BMI (Armon et al., 2013; Kakizaki et al., 2008). A clear question that emerges from these results is: why is a personality trait linked to descriptors such as “outgoing” and “energetic” linked to increased obesity? Indeed, some studies at the individual level have linked extraversion to healthier eating habits (Mõttus et al., 2012). To account for this, some authors have argued that the positive mood state linked to high extraversion leads individuals to perceive themselves as less vulnerable to negative health conditions and thus to engage in behaviors, such as overeating, that can lead to obesity (Grant & Schwartz, 2011).
Interestingly, our findings only partly replicate those on an individual level. For example, we found that harmony is positively associated with alcohol consumption, whereas Rudnev & Vauclair (2018) have not found a significant relation on an individual level; we found no association between conscientiousness and obesity, whereas a range of other studies found a negative association (e.g., Allen et al., 2015; Gerlach et al., 2015; Kim, 2016; cf. Table 1); we found a positive association between obesity and extraversion which is in line with the literature (e.g., Kim, 2016; Mõttus et al., 2012); however, we also found a negative association between neuroticism and obesity, whereas other studies either found no or positive associations (Kim, 2016; Mõttus et al., 2012).
We do not perceive a significant finding to be conflicting with a non-significant one because effect sizes can vary across studies: Assuming a statistical power of <1 (in many cases, the power is clearly below <.80; Brysbaert, 2019), a mix of significant and non-significant findings is expected (Lakens & Etz, 2017). Thus, our findings were mostly in line with the literature. However, the negative association between neuroticism and obesity requires further explanation because some previous research found positive associations between obesity and neuroticism (whereas some others found no association, cf. Table 1). This might indicate to an ecological fallacy: The pattern of association is reversed on the aggregated level as opposed to an individual level. Alternatively, neuroticism might have a different meaning on a country level than on an individual level. The effects of a neurotic person might be different than those of a neurotic large group of people (here: country). Future research is needed to explore this possibility. Importantly, since our findings are overall consistent with the literature, it is unlikely that the ecological fallacy is an issue.
Which Is a Better Predictor of Country-Level Risky Health Behaviors: Values or Traits?
The present study tested theoretical predictions regarding the relative importance of value and traits for understanding risky health behaviors. Our finding was that values and traits were differentially important for different behaviors. In line with our predictions, cultural values were a better predictor of alcohol consumption than traits. Alcohol consumption has always been rooted in a social context and in many societies (excluding “abstinent” societies that expressly forbid alcohol consumption for religious reasons; Room & Mäkelä, 2000) alcohol consumption is a common occurrence at social events (Galea et al., 2004). In other words, drinking alcohol is a social behavior. Because values are more important for guiding social interactions than traits (Boer et al., 2011), they were also the better predictors.
Also in line with our predictions, B5-traits were a better predictor of obesity than cultural values. Obesity can be a result of by poor eating habits and a lack of physical exercise (Prentice & Jebb, 1995). In other words, it is caused by behavior that lies in the past. Thus, obesity is conceptually closer to traits, which measure how a person is in the present (Saucier, 1994), in contrast to values, which are better able to capture future behavior (Eyal et al., 2009).
Finally, the results did not support our third hypothesis that values explain variance beyond traits in smoking. Initially, we predicted that values would be better at explaining smoking than traits because values are more relevant for social interactions (Boer et al., 2011). However, in retrospect, we believe that smoking might have a weaker social component than alcohol consumption. Indeed, the number of cigarettes daily smokers consume on average per hour varies little between 10 a.m. and midnight (Shiffman et al., 2014). If smoking was mostly social, it would be higher in the late afternoons and evenings when people are more likely to be in social situations. Note however that our data does not allow us to distinguish between social smokers and “full-time” smokers. We assume that values are better predictors for social smokers, a prediction that can be investigated by future research.
Limitations
One potential limitation pertains to differences in sample types across the measures. While smoking, obesity, and alcohol consumption were based on official representative data, not all of our traits and values data was based on representative samples. However, the value importance and structure remains mainly the same across sample types such as representative, student, or teacher samples (Hanel et al., 2018; Schwartz & Bardi, 2001), indicating that the sample type barely matters.
Another possible limitation pertains to the sample size of only 50 countries. However, the term “sample size” is misleading here: The numerical value of each country for each variable consists of hundreds to thousands of participants (traits and values) or even millions (smoking, obesity, and alcohol consumption), making our analysis very robust. Moreover, a sample of 50 countries is more representative of the total “population” of around 200 countries than, for example, a sample of even 1,000,000 Americans to the total US population of around 300, 000, 000. Finally, with 50 countries, we included more countries than many previous studies which often included 20–30 countries (see Introduction, for example). It is important to acknowledge, however, that countries from geographical regions other than Europe, and particularly the East Mediterranean and Africa regions, were underrepresented in our sample.
Finally, the trait data we used were measured with two different questionnaires, the NEO-PI-R (McCrae, 2001) and the BFI (Schmitt et al., 2007), which show on a country level relatively low convergence, rs ≤ .45 (Schmitt et al., 2007). However, both trait measures are reliable and valid, which makes it unlikely that the combination of trait measures impacted the results.
Implications
A current UN goal for sustainable development by 2030 includes ensuring healthy lives and well-being for all at all ages (United Nations General, 2015). This includes specifically, reducing by one-third premature mortality from noncommunicable diseases through prevention and treatment and promoting well-being. This study adds to a growing body of research that suggests information about cultural-level psychological characteristics, including dimensions and profiles of cultural values and aggregate personality, may be useful for informing policies and interventions aimed at meeting these goals. Campaigns promoting healthy behaviors, for example, may be differentially effective across populations with distinct cultural values or personality characteristics. Future research is, however, required to further understand the causal mechanisms between country-level psychological characteristics and risky health behaviors.
Many publicly available datasets, and particularly those including health-related variables, have data for most countries. Unfortunately, datasets including national indices of psychological variables typically have smaller samples. Moreover, while national health indicators are abundant and easily accessible, it is currently more difficult to find and obtain national indices of psychological variables. The present study demonstrates that public datasets can be useful to researchers who wish to test theoretical predictions about country-level psychological characteristics and their relations to indicators of health or well-being. Given that such research findings may have important implications for meeting international goals for sustainable development (e.g., the 2030 Agenda for Sustainable Development), a clear implication of our study is that an effort is required to construct public datasets that outline the psychological characteristics of all countries globally.
Posts
