The data are from a longitudinal study of 1,682 women in the volunteer Australian Twin Registry who were screened with a self-report health and lifestyle survey that included information about disordered eating. At the time of screening, in 1988–1989, they were between the ages of 30 and 45 years (mean=36.5, SD=4.7). These women were further interviewed over the telephone in 1992–1993 with a general semistructured psychiatric interview that included information about lifetime bulimia nervosa. The women indicating problems with eating from the first two waves of data were followed up during 1994–1995 and interviewed over the telephone more extensively about past history of bulimia nervosa, parental bonding, and temperament. Nine monozygotic twin pairs were found to be discordant for lifetime history of bulimia nervosa. The mean age of this group at the time of the third data collection was 40.3 years (SD=5.8), and their mean age at the first binge episode was 21.8 years (SD=7.3, range=16–38 years). Oral informed consent was obtained before the telephone interviews.
The wave 3 interview was performed with the Eating Disorder Examination
(5), conducted by interviewers blind to information about the co-twins. The diagnostic questions of the Eating Disorder Examination were adapted to assess the presence of bulimia nervosa during the subject’s lifetime as well as over the previous 3 months. The Eating Disorder Examination also provides scores on four subscales, measuring dietary restraint, eating concern, weight concern, and shape concern over the previous 3 months. Other wave 3 measures included the Rosenberg Self-Esteem Scale
(6), the subscale for dichotomous thinking style of the Dysfunctional Attitude Scale
(7), and the “emotional reliance on another person” dimension of the Interpersonal Dependency Inventory
(8) to measure external control. Warmth and overprotection from parents in the first 16 years of life was measured with the 25-item Parental Bonding Inventory
(9), extensively used in the assessment of parental characteristics as influences on adjustment in adulthood
(10). Scores on the Tridimensional Personality Questionnaire
(11) were collected at wave 2, and information on birth weight, birth order (first born or second born), difficulties at birth, and six medical conditions before age 14 (migraine, hay fever, asthma, eczema, dust allergy, arthritis) was collected at wave 1. All wave 1 variables, with the exception of birth weight, were categorical.
Paired t tests were used for all continuous measures, and McNemar’s exact test was used for categorical variables. Given the clear directional hypotheses from the research literature, one-tailed significance was used for all t tests (with the exception of birth weight). All analyses were conducted by using SPSS for Unix, Release 6.1 (Chicago, SPSS).