The MIDUS is an interdisciplinary study of behavioral, psychological, and social factors involved in midlife health and well-being. The baseline data were collected in 1995–1996 (MIDUS I) and a follow-up of the original sample (MIDUS II) was conducted in 2004–2006, with data collection in MIDUS II largely repeating baseline assessments. Respondents (aged 25–74 years) in 1995 were drawn from a nationally representative multistage probability sample of community-based English-speaking adults in the coterminous United States. In the first stage, participating households were selected from working telephone banks via random digit dialing. In the second stage, individual respondents were selected using disproportionate stratified sampling. Elderly individuals (ages 65–74) and men were oversampled along with oversampling in five metropolitan areas (Atlanta, Boston, Chicago, Phoenix, and San Francisco). Those quarried in the survey participated in an initial telephone interview and completed a self-administered mail questionnaire. The response rate for the telephone interview in MIDUS I was 70%; among the telephone respondents, 86.8% completed the mail questionnaire, so that the overall survey response rate was 60.8% (17).
Our study uses data from the random core sample of the MIDUS in both waves (the Main Data) (MIDUS I is from the 2006 Release (2006/03/30). The most recent version of the MIDUS I data was released on 2007/04/16. MIDUS II is from the original and most recent release on 2007/03/22). The baseline 1995–1996 MIDUS study included data from three subsample data sets: the Main Data (N = 4,242 with 3,485 national random core sample participants and 757 metropolitan oversamples), the Twins Data (N = 1,996 twins), and the Siblings Data (N = 1,614 pairs with 951 participants drawn from the Main Data). Among the original 7,190 MIDUS I participants, 842 participants refused to participate in a 2004–2006 follow-up, and 1,334 could not be successfully contacted (including 421 confirmed deceased). Thus, 70% of the original MIDUS sample participated in the second wave of 2004–2006 (ages 35–86). The survey assessed physical and psychological health throughout the respondent's adult life, substance abuse, well-being, personal beliefs, socioeconomic status, social support, and various forms of perceived discrimination. Further details on the MIDUS data and methodology are available elsewhere (17,18).
Our study uses data from the MIDUS national random sample of the participants who completed at least the telephone survey. We limited our analyses to adults aged 35–74 in both waves to make data comparable across waves. We therefore excluded respondents who were 25–34 years old in the first MIDUS wave (1995–1996) and who were 75 years and older in the second wave (2004–2006). Further exclusion criteria included individuals with incomplete or inconsistent data on discrimination-related questions (e.g., reporting a cause for discrimination but no discriminatory experiences, citing experiences of discrimination but giving no reason for discrimination). As a result of all exclusions, our analytic sample for the 1995–1996 wave included 1,826 individuals, and the 2004–2006 sample included 1,136 respondents.
Our study objectives aimed to (i) determine how the prevalence of weight/height discrimination among US adults changed between the two time periods of 1995–1996 and 2004–2006 (while also examining contributors to the observed trends in weight/height discrimination), and (ii) compare changes over time across different forms of discrimination. We treated the data from the two MIDUS waves as two cross-sections. We did not examine discrimination exposure among the same respondents 10 years apart, because our aim was to capture patterns of discrimination over time for an average representative American adult. Limiting the trend analysis to participants who experienced discrimination in the first wave of data collection and tracking their discriminatory experiences over a decade would provide only a partial snapshot of trends in weight discrimination for adults in the United States.
Measure of discrimination
The MIDUS survey evaluates self-perceived discriminatory experiences by asking participants to report occurrences of discrimination over their lifetime and on a day-to-day basis. The survey asks about the primary reason for discrimination allowing participants to report multiple reasons if applicable. Specifically, the survey asks “What was the main reason for the discrimination you experienced? If more than one reason, circle all that apply” with response choices including “Your Age, Gender, Race, Height or weight, Ethnicity or nationality, Physical disability, Some aspect of appearance other than weight or height, Sexual orientation, Religion, and Other reason.”
The key variable in our study is perceived discrimination due to the respondent's weight or height. Because the survey used one category for body size discrimination combining height and weight, we refer to this variable as weight/height discrimination throughout the paper. We compared average body weight, height, and BMI (defined as weight in kilogram relative to height in meters squared) between the participants reporting weight/height discrimination and the rest of the sample to test whether weight was more likely to be a source of discrimination than height. On average, body weight and BMI were significantly higher among people reporting weight/height discrimination relative to other participants in both samples (e.g., BMI of 34 vs. 26, P < 0.01 in 1995–1996, and BMI of 35 vs. 28 in 2004–2006, P < 0.01). At the same time, there was no difference in both samples in the average height of women by weight/height discrimination, so that a short body stature was unlikely to be a source of weight/height discrimination. Body height among men was on average slightly higher in the group reporting weight/height discrimination (P < 0.10), so that a short stature is an unlikely cause of weight/height discrimination among men. We therefore have reasonable evidence to believe that the reported experiences of discrimination due to weight or height are reflective of higher body weight and obesity rather than height.
We examined lifetime experiences of discrimination in major settings such as employment, medical care, and education as well as interpersonal discrimination on a day-to-day basis. Lifetime experiences were self-reported in the question: “How many times in your life have you been discriminated against in each of the following ways because of such things as your race, ethnicity, gender, age, religion, physical appearance, sexual orientation, or other characteristics? (If the experience happened to you, but for some reason other than discrimination, enter “0”).” Eleven ways of lifetime discrimination were evaluated, including: “discouraged by a teacher or advisor from seeking higher education,” “denied a scholarship,”“not hired for a job,” “not given a job promotion,” “fired,” “prevented from renting or buying a home in the neighborhood you wanted,” “prevented from remaining in a neighborhood because neighbors made life uncomfortable,” “hassled by the police,” “denied a bank loan,” “denied or provided inferior medical care,” and “denied or provided inferior service by a plumber, car mechanic, or another service provider.” Discrimination in interpersonal experiences on a day-to-day basis was evaluated with the question: “How often on a day-to-day basis do you experience each of the following types of discrimination ?” with nine response items including: “you are treated with less courtesy than other people,” “you are treated with less respect than other people,” “you receive poorer service than other people at restaurants or stores,” “people act as if they are afraid of you,” “people act as if they think you are dishonest,” “people act as if they think you are not as good as they are,” “you are called names or insulted,” and “you are threatened or harassed.” Participants indicated how often they had experienced these situations using these categories: “Often,” “Sometimes,” “Rarely,” “Never.”
In both survey waves, we constructed a dichotomous variable of perceived discrimination indicating whether an individual reported occurrences of any types of discrimination (e.g., work-related discrimination, personal insults). Multiple and single occurrences of discrimination were weighted equally so that a person reporting a single occurrence of discrimination (e.g., in the form of “denied or provided inferior medical care”) would be treated equally in our discrimination measure as someone who reported several discriminatory experiences (e.g., by “not given a job promotion”). We coded responses “Often” or “Sometimes” in the question about daily discrimination as an indicator of discrimination. We also constructed measures of lifetime discrimination exposure and daily interpersonal discrimination indicating any occurrences of lifetime discrimination or discrimination in personal relationships, respectively.
We conducted comparative analyses of the prevalence of different forms of discrimination and forms of discrimination across population groups in both waves of data. Stratifying by age, race, education, marital status, weight status, and occupation, we performed a t -test for each characteristic to evaluate the hypothesis that rates of discrimination in the group remained unchanged between the two waves (reporting P values). Similarly, we used a t -test to compare rates of discrimination between men and women across groups in each survey wave. We also compared sociodemographic characteristics and weight distribution of the samples in both waves and tested for differences in these attributes between the samples. To account for the complex sampling design and to obtain nationally representative estimates, we used individual sample weights in presenting sample statistics.