Gbolahan O. Olanubi and A. Janet Tomiyama
Psychology Department, University of California Los Angeles, Los Angeles, CA, USA
When attempting to understand factors that influence diet, or food choice, people are quick to assume that people eat when they are hungry and stop when they are full. Hunger, however, turns out to be one of the least important factors in eating. This is because there are a multitude of other social, economic, and psychological factors at play that shape our eating choices, which can over time affect weight and body mass index (BMI).
The purpose of this encyclopedia entry is to describe the social factors that are involved in eating, dietary choices, and obesity. We organize our discussion of social factors roughly from the macro‐ to microlevel. The literatures we cover are enormous, but space and reference constraints allow only for highlights of each topic. Throughout the entry, we note when there are seminal review papers in each area. Ultimately, our goal is to highlight the role of social factors in diet and obesity and to position psychology in its rightful place as a key player in diet and obesity research.
Culture has been shaping food choice for millennia. Rozin and Vollmecke (1986) noted that if you want to know as much as possible about a person's food preference, simply ask, “What is your culture or ethnic group?” Culture, a way of life of a society or an amalgam of shared practices and meanings among a group that guides interactions and is shaped by experience, plays a major role in diet and food choice. Culture influences attitudes toward food, our preferences, and perceptions of what is healthy compared with what is unhealthy.
Everyone needs to eat, but not everyone needs to be happy about eating. Among Americans there is an attitude not only that food is “nutrient,” a source of nourishment, but also that it may be dangerous—almost as dangerous as abstaining from food (Rozin, 1996). Americans also endorse attitudes that even trace amounts of salt and fat may be harmful (Rozin, 1996). In contrast to American attitudes, the French have a more positive attitude toward food, viewing it as both nutrient and a source of pleasure. Considering these cultural differences in attitudes, there are also cultural differences in consumption. The French, compared with Americans, eat more high‐fat foods (yet tend to be healthier—a phenomenon known as the French paradox; Ferrières, 2004). Although Americans have more negative attitudes than the French about food, Americans eat larger servings of food.
Expanding on this research Rozin, Fischler, Imada, Sarubin, and Wrzesniewski (1999) examined cultural differences in food attitudes and behavior in the United States, France, Belgium, and Japan. Congruent with previous research, they found that Americans had the highest concern about healthiness of food choice, worried about the fattening effects of food as opposed to the savoring of food, and were more likely to make nutritional, as opposed to culinary, associations with food. Americans also had the lowest food‐positive attitudes, considering food less important and drawing little pleasure from food. Finally, despite the worry, concern, and attitudes toward food, Americans also reported viewing themselves as the least healthy eaters compared with the other countries.
Food choice is predicated upon access to foods. Socioeconomic status, comprising income, education, and occupation, plays a focal role in determining purchasing power and food choice. Cost has been one of the most important factors in food choice for low‐income individuals. Ironically, the cost of food is actually more expensive in low‐income neighborhoods compared with high‐income neighborhoods because of a dearth of low‐cost supermarkets and presence of higher‐priced convenience stores (see Morland, Wing, Roux, & Poole, 2002).
If low‐income families have a restricted budget but more expensive foods, what are they eating? Low‐income families may be cornered into low‐cost energy dense diets (Darmon, Ferguson, & Briend, 2003). Energy density refers to calories per edible weight of a food item. Energy dense diets consist of eating foods high in energy density (cereals, meats, sweets) more so than foods low in energy density (fruits, vegetables, whole grains). A low‐cost energy dense diet would essentially attempt to maximize calories per dollar while still maintaining sufficient nutrients.
Supporting this perspective, Darmon et al. (2003) used a computer simulation to model the food choices of individuals under economic constraint. Under increasing economic constraint the model selected for less fruits and vegetables (FV), meats, fish, and cheese and higher contents of milk, sweets, and starchy foods compared with a normal diet.
Fortunately, this barrier appears surmountable. Researchers have intervened to remove financial barriers to increase selection and consumption of FV for low‐income individuals and found that subsidizing FV costs increases consumption for low‐income individuals (Anderson et al., 2001).
Norms surrounding food consumption are powerful drivers of eating behavior (see Herman, Roth, & Polivy, 2003). In general, an inhibitory social norm surrounding eating exists in Western countries, based on widespread weight stigma and negative stereotypes of those who overeat (Vartanian, Herman, & Polivy, 2007). This inhibitory norm appears to be most pronounced in eating among strangers, due to impression management concerns, and true particularly for women, who are seen as more feminine and socially appealing when they eat less (Vartanian et al., 2007. Inhibitory norms are highest with a non‐eating observer, according to a meta‐analysis of studies that experimentally manipulated awareness of one's eating being observed (Robinson, Hardman, Halford, & Jones, 2015).
In contrast, eating in the presence of close others appears to trigger social facilitation of eating (de Castro, 2002). In one illuminating study, Howland, Hunger, and Mann (2012) experimentally manipulated eating norms in existing friend groups. Two members of a friend trio were enlisted to set an eating norm, and the third member's eating was examined. Participants followed whatever eating norm was set, and these eating norms carried over into situations where participants were then eating alone.
A subtype of norms are informational norms, wherein information about eating behavior is communicated (e.g., through visual cues of leftover food). In a meta‐analysis of 15 studies examining experimental manipulations, Robinson, Thomas, Aveyard, and Higgs (2014) found that both high and low informational intake norms had moderate effects on eating behavior. Thus, there appear to be general social norms that affect eating behavior, but these processes are amenable to manipulation, pointing to potential intervention targets.
An interesting offshoot related to norms is research that investigates the weight of others in determining an individual's food intake. Some studies have found that non‐dieters consume more calories when their food server is overweight compared with when she is normal weight (e.g., McFerran, Dahl, Fitzsimons, & Morales, 2010).
Social support, characterized as instrumental, emotional, or informational, is related to eating behavior and diet quality (for a review, see Vesnaver & Keller, 2011). Social support may be particularly relevant in the context of healthy, as opposed to unhealthy, eating behavior. For example, the national representative National Health and Nutrition Examination Survey found that greater frequency of contact with family, friends, neighbors, coworkers, and others was related to eating at least five servings of FV per day (Ford, Ahluwalia, & Galuska, 2000). A systematic review of environmental determinants of FV intake (Kamphuis et al., 2006) identified social support as a contributor to more intake of FV. One rigorous study included in this review was a brief intervention in low‐income participants, where FV consumption was confirmed via plasma/urine biomarkers (Steptoe, Perkins‐Porras, Rink, Hilton, & Cappuccio, 2004). Baseline social support specifically for dietary change was related to 12‐month increases in FV intake, independent of experimental group or other controls, suggesting the effects of social support are robust to a variety of settings and populations (Kamphuis et al., 2006).
Social support is a component of Bandura's social cognitive theory (SCT; Bandura, 1997), and in a study testing multiple components of SCT, social support was an important correlate of FV intake (Anderson, Winett, & Wojcik, 2007). In the context of an online SCT‐based lifestyle intervention (Anderson‐Bill, Winett, & Wojcik, 2011), perceived social support (from friends and family) was a predictor of better nutrition.
The relationship between social support and obesity is best tested in longitudinal studies, and such studies find mixed results. Moreover, positive and negative qualities of social relationships may affect BMI differently. In 4,724 Dutch adults, low positive social support was related concurrently to FV intake, and high negative social support was related concurrently to overweight status, but neither was related to future weight change (Croezen et al., 2012). Negative relationships, however, were related to future risk of gains in BMI and waist circumference in another longitudinal study (Kouvonen et al., 2011). To inform this conflicted literature, Kershaw et al. (2014) examined a time frame when obesity tends to develop using 10‐year longitudinal data in 3,074 participants. They found that those with continuously high levels of social support had lower likelihood of gaining (by more than 10%) in BMI or waist circumference than those with continuously low levels of social support. Those who increased in frequency of negative social contacts had higher likelihood of increasing waist circumference, but there was no relationship with BMI, and increasing frequency of positive social contacts showed no effect.
One of the most famous findings in the area of modern social network analysis is that obesity spreads across networks. Christakis and Fowler (2007) demonstrated this in the 12,067 participants of the longitudinal Framingham Heart Study (FHS; 1971–2003). Individuals' risk of becoming obese was higher if their spouses, siblings, and friends previously became obese, with stronger concordance among same‐sex social ties. This phenomenon was not driven merely by similar BMI individuals forming relationships, nor was it driven by neighborhood/geographic distance effects.
Although much of the work examining social network effects on diet have been cross‐sectional, Pachucki, Jacques, and Christakis (2011) used 10‐year longitudinal data from the FHS to examine social influences (spouses, siblings, and friends) on dietary patterns. Using data classification procedures, they extracted dietary patterns based on food frequency questionnaires and found longitudinal social influence was exerted on those such as “alcohol and snacks” and “meat and soda” as well as those such as “healthier” and “caffeine avoidant.” This indicates that both healthy and unhealthy eating patterns are subject to social influence. A study in adolescents in Australia across three waves of data in 1 year found that friends appear to influence low‐nutrient, energy dense food consumption (de la Haye, Robins, Mohr, & Wilson, 2013). This effect, however, was not mediated by beliefs about unhealthy food, indicating that social network influences may be operating implicitly.
Parents, caregivers, and families exert enormous influence on eating and diet in children. Food preferences of parents affect children's eating (Patrick, Nicklas, Hughes, & Morales, 2005), due to simple availability of those foods. Attachment style is also related to eating behavior.
In their review, Ventura and Birch (2008) distinguish between parenting styles (attitudes and interaction styles) and parenting practices (behavioral patterns). Moreover, there is likely a bidirectional relationship with children shaping parent's behavior (e.g., a child rejecting broccoli, causing the parent to avoid serving it again). Authoritative feeding styles are linked to better quality eating (Patrick et al., 2005), whereas authoritarian (“clean your plate!”), permissive, and neglectful parenting styles are associated with greater likelihood of having an overweight child (see Ventura & Birch, 2008). Using longitudinal design, steeper BMI trajectories and less leveling off of BMI were observed in authoritarian and disengaged parenting styles as opposed to balanced parenting (Fuemmeler et al., 2012).
Examples of parenting practices are using food as reward, restricting access to food, or modeling eating (Ventura & Birch, 2008; see Modeling section below). Using food as a reward was shown in experimental studies to increase consumption of the target food but decrease liking. Restricting food can backfire, where children work harder to obtain restricted snacks and eat and prefer them more.
Families create a social context around mealtimes (for a review, see Patrick et al., 2005). Families eating together is related to better diet quality. When TV watching is part of mealtime, however, diet quality is poorer, likely due to exposure to advertising.
Given the importance of parenting for child weight and diet, interventions such as the “Lifestyle Triple P” have targeted parenting. In a randomized controlled trial, this intervention positively affected intermediate outcomes such as soda consumption, but saw no differences at 12 months between intervention and control in terms of weight or body composition (Gerards et al., 2015).
Romantic relationships have effects on healthy and unhealthy eating. Women appear to play a larger role than men in terms of controlling eating behaviors of the couple (note: the literature has focused on heterosexual relationships; we know nearly nothing about eating behavior in same‐sex romantic relationships). This is likely due to cultural norms surrounding the idea that women should be the ones who cook, and marriage appears to have differential effects on the eating of men versus women. For example, when wives report poorer marital quality, they also report more unhealthy dieting behaviors, particularly if they have low self‐esteem (Markey, Markey, & Birch, 2001). In this same study, however, no marital quality variables were significantly related to healthy dieting behaviors. However, the literature on marital status and diet/nutritional quality is very mixed (reviewed in Vesnaver & Keller, 2011), indicating that it may not be marital status per se, but living alone that imparts risk for poor eating.
Marriage is associated with greater BMI, and spouses tend to have similar BMI (Jeffery & Rick, 2002). Longitudinal studies indicate that marriage is related to BMI gain, with effects most pronounced for African American women. Spouses' BMIs also predict one another's longitudinal BMI change (Jeffery & Rick, 2002). Divorce, in this study, was related to 2‐year weight loss, although other studies have found divorce and other indices of marital dissatisfaction are related to increased BMI trajectories.
Many food choices are derived from modeling—a behavioral mechanism related to social norms. Nisbett and Storms (1974) were among the first to test the effects of social influence on eating. Initially, they hypothesized that normal weight and overweight participants' eating behavior would be differentially affected by social suppression or social facilitation cues of eating compared with a no‐cue control condition. Participants completed a task with a confederate where there happened to be crackers on the table. In the suppression condition, the confederate ate one cracker; in the facilitation condition, the confederate ate 20 crackers, and there was no confederate in the control condition. Rather than a differential effect, they found that regardless of participant weight, participants typically mimicked the consumption behavior of the confederate.
This finding launched a line of research into the modeling effect and underlying mechanisms of modeling behavior in food consumption (reviewed in Herman et al., 2003). Researchers have tested if these effects were moderated by participant dieting status, participant satiation, and personality characteristics such as extraversion and self‐monitoring and found that the expected main effect of modeling occurs regardless of these moderators. Researchers have even found that participants model behavior when left alone but are given information on others' behavior (e.g., with a bogus “data sheet” left behind with information on how much the prior participant had ostensibly eaten; see also informational norms section above).
People often think of diet and obesity as matters of individual will—individuals exert control over their food choice, and the choices they make directly impact their weight. However, in this entry we covered a wide array of evidence that suggests this is not entirely the case. Culture forms our attitudes around food, socioeconomic status impacts our access to foods, norms guide behavior around food, and social connections with friends, family, and even strangers modify our behavior around food. In addition we showed that weight and obesity are also tied to social factors. Changes in BMI travel through social networks and are affected by social support and the status of our marriages.
In considering future directions, it is important to note that these factors we have discussed in isolation are deeply connected and build on one another to create our attitudes and subsequent eating behavior. Future directions should integrate these factors to build better interventions with potentially cascading impacts. For example, a future intervention aiming to change eating norms or attitudes toward food within a community should first examine the community's social network to identify particular individuals or groups who will have the strongest impact. On a larger scale, another future direction could aim to change eating norms in parents and also to explicitly test whether this also changes norms in children through modeling, with implications for the norms within both parent and child social networks. And, after some time, since culture is shaped by the people within it, these changes in norms in parents could even lead to lasting cultural changes in attitudes toward food and healthier diets.
Gbolahan O. Olanubi, C. Phil, is a doctoral student in the Department of Psychology at the University of California, Los Angeles. His research investigates concealable identities and stereotype threat, and of particular relevance to this entry, he is interested in how race affects perceptions of the healthiness and quality of food and the downstream health implications of these perceptions.
A. Janet Tomiyama, PhD, is an associate professor in the Department of Psychology at the University of California, Los Angeles. Her research investigates stress, eating, dieting, and weight stigma.