Summary of Prior Findings on Intergenerational Mental Health Correlations
Study | Country | Data | Survey Data | Maximum Sample Size | Health Outcomes | Main Results |
---|---|---|---|---|---|---|
(1) | (2) | (3) | (4) | (5) | (6) | (7) |
Johnston, Schurer, Shields (2013) | UK | 1970 British Cohort Study | Yes | 8,194 | (1) Mother and children (two generations): 9-question subset of the 24-item Malaise Inventory (2) Mother and grandmother (three generations): 9-question subset of the 24-item Malaise Inventory (3) Children (three generations): SDQ questionnaire | (2) Two generations: mother–child mental health IC of 0.13* to 0.19* (SD) (2) Three generations: mother–child mental health IC of 0.31* (SD) and grandmother–child (conditional on mother) of 0.03 (SD) |
Hancock et al. (2013) | Australia | Growing Up in Australia: The Longitudinal Study of Australian Children | Yes | 4,069 | (1) Children: SDQ questionnaire (2) Parents: Kessler K6 scale of nonspecific psychological distress (3) Grandparents: reported by parents based on binary question “Did your father/mother suffer from nervous or emotional trouble or depression?” | (1) Additional 1.9*, 0.9*, and 1.2* SDQ points if mother, father, or both parents had mental health problems, respectively (2) Additional 0.4* (maternal), 0.2 (paternal), 0.5* (maternal), and 0.4 (paternal) SDQ points if grandmothers and grandfathers had mental health problems, respectively |
Knight, Menard, Simmons (2014) | USA | National Youth Survey Family Study | Yes | 1,725 | Annual substance use frequency (alcohol, marijuana, other drugs) of both parents and children | Elasticities ranging from −0.03 (for other drugs use at ages 12–17) to 0.23* (for alcohol use at ages 18–24) |
Eley et al. (2015) | Sweden | Twin and Offspring Study of Sweden | Yes | 876 | (1) Parental anxiety: 20 items from the Karolinska Scales of Personality (2) Children anxiety: items from Child Behavior Checklist (3) Neuroticism based on Eysenck Personality Questionnaire for both parents and children | (1) Anxiety ICCs of 0.02–0.20* (2) Neuroticism ICCs of 0.03–0.21* |
Akbulut-Yuksel and Kugler (2016) | USA | NLSY79 | Yes | 19,165 | Indicator of self-reporting being depressed (sometimes, a moderate amount of this or most of the time during past week) for both mother and children | Mother’s depression increases the likelihood of child being depressed by 9* and 0.3 percentage points for native-born and immigrant children, respectively |
Eyal and Burns (2019) | South Africa | National Income Dynamics Survey | Yes | 3,111 | Center for Epidemiological Studies Short Depression Scale (CES-D 10) for both parents and children | Parental depression increases likelihood of adolescent depression by 31* to 35* percentage points |
Bencsik, Halliday, Mazumder (2021) | UK | British Household Panel Survey and UK Household Longitudinal Survey | Yes | 5,292 | Mental health index based on 5 questions in the Short Form 12 Survey | Parents–children IC of 0.22* |
Vera-Toscano and Brown (2021) | Australia | Household, Income and Labor Dynamics in Australia | Yes | 1,960 | Mental health index based on five questions in the Short Form 36 Health Survey | Parents–children ICs of 0.18* to 0.21* |
This paper | Norway | Administrative health and social security data | No | 370,498 | Medically diagnosed mental health conditions based on ICPC-2 classification (code P) | (1) Parent–child mental health IC of 0.05*. Extensive margin: 10* percentage points increase in P(diagnosed) (2) Extended family–child mental health IC of 0.09*. Extensive margin: 17* percentage point increase in P (diagnosed) |
Notes: This table summarizes findings from prior research on intergenerational correlations in mental health. Column 1 provides study reference, Column 2 gives the country of origin of the data, Column 3 gives the specific data sets used, Column 4 indicates if the data sets are survey-based, Column 5 provides the maximum sample size used in the paper, Column 6 describes the mental health outcomes used, and Column 7 describes the main findings. ∗ implies that the result is statistically significant at least at the 10 percent level. The lack of asterisk implies that the result is not statistically significant at conventional levels.