9.4 Example Data
Increasingly researchers are taking a life-course perspective to understanding how different life stages shape a variety of later in life outcomes.
In this case study, Ferraro, Schafer, and Wilkinson (2016) examine the relationship between physical health in adulthood and multiple domains of childhood disadvantage using a count regression model.
Data are drawn from the National Survey of Midlife Development in the United States (MIDUS). MIDUS contains a battery of retrospective questions concerning childhood disadvantage, as well as extensive measures of adult risks and resources. Although the authors use data from both waves of MIDUS in the paper, here we focus only on their first model of adult health outcomes, which takes into account both childhood disadvantage and the mediating effects of later life resources and risk behaviors.
9.4.1 Dependent variable
The dependent variable for this analysis is health problems at Wave 1: morbidityw1.
Here, adult health problems are measured by the self-reported occurrence of 31 diseases or health conditions. For 29 of these items respondents were asked
“In the past 12 month have you experienced or been treated for any of the following?”
For the remaining 2, cancer and heart disease, respondents were asked if they had ever been diagnosed with the disease. Finally, morbidityw1 is the sum of these 31 items, where each is coded 1 for yes, and 0 for no.
9.4.2 Explanatory Variables
9.4.2.1 Early Life Disadvantage
- ses: Childhood SES is a sum score based on standardized measures of (1) the education for the head of household, (2) financial strain and (3) receipt of welfare.
- family: Family composition is a sum score based on (1) no presence of a male in the household, (2) parental divorce, and (3) death of parent prior to age 16,
- abuse_rare: Physical or Emotional Child abuse by parents is categorized by frequency of abuse. abuse_rare indicates respondent rarely experience one or both types of abuse. The reference category is never having experienced emotional or physical abuse.
- abuse_freq1: abuse_frequency1 indicates respondents frequently (sometimes or often) experienced one type of abuse during childhood. The reference category is never having experienced emotional or physical abuse.
- abuse_freq2: abuse_frequency2 indicates respondents frequently experienced both types of abuse during childhood. The reference category is never having experienced emotional or physical abuse.
- health: Adolescent health problems are measured by self-rated physical and mental health at age 16.
9.4.2.2 Adult Characteristics
- age: Age at time of Wave 1 interview.
- nonwhite: Race (white or nonwhite).
- female: Gender (female or male).
- educate: Number of years of completed education.
- catincome: Household income adjusted by household size and recoded into five percentile categories (< 21st *percentile, 21st to 40th percentile41st to 60th percentile, 61st to 80th percentile, and > 80th percentile.).
- a1sj6: Financial strain during adulthood; responses range from 1 (no difficulty paying monthly bills) to 3 (very difficult to pay monthly bills).
- smoke_dose: Lifetime smoking is calculated from information reported by respondents: age when started smoking, year stopped (for former smokers), and average number of cigarettes smoked daily. Using a yearly metric, lifetime smoking is the product of years smoked and annual number of cigarettes, divided by 10,000 (see Footnote 11, p. 130).
- heavydr2: The measurement of heavy drinking is sex differentiated and tapped respondents’ period of greatest lifetime consumption: five or more drinks per day for men and four or more drinks for women.
- obese: Obesity, dummy variable coded 1 if body mass index [kg/m2] > 30.
- fampos: Family support as measured by four items reflecting the presence of positive relationship characteristics.
- friendpos: Friend support as measured by four items reflecting the presence of positive relationship characteristics.
- famneg: Family strain as measured by four items reflecting the presence of difficult relationship characteristics.
- friendneg: Friend strain as measured by four items reflecting the presence of difficult relationship characteristics.
- integration: Social integration as measured by three 7-item Likert–type questions.
- ever_divor: Ever divorced, a dummy variable coded 1 if the respondent reported having been divorced.
- controlw1: Average score for a 12-item index of the respondent’s feelings of personal control.