*2.3. Data Analysis*

In order to obtain estimates of adverse childhood experiences in the adult noninstitutionalized population of Hungary, statistical weights were applied to ensure that estimates reflect the general adult Hungarian population gender, age group, education, settlement type, and region. The sample defined as survey sample was analyzed using Stata/IC 13.1. Single-stage design was used stratifying the sample based on the sampling units, that is, regions of the country. The Taylor method was used to estimate sampling errors; primary sampling units were sampled without replacement [48]. Analysis of variance was computed to examine the prevalence of child maltreatment and household dysfunctions by total ACE core and by type of ACE stratified by gender. The sociodemographic characteristics of respondents reporting no history (ACE score = 0) or any history of adversity (ACE score >0) were described defining the ACE score as a discrete interval variable. The number of categories of demographic variables such as education and employment type was combined to reduce the number of categories and to simplify interpretation. Considering the weighted estimates, all prevalence data were rounded.

Logistic regression was carried out by backward stepwise regression to identify the independent variables of childhood adversity. The ACE score was defined as the binary outcome variable as described above (no childhood adversity vs any history of adversity). One binary ACE score was also created considering only the five types of childhood maltreatment, and another by including only the five types of family dysfunction. Age (in years), sex (female, male), place of residence (capital, city, village), education (higher education vs less), type of work (nonmanual vs manual), and marital status (single, divorced, married, cohabiting, widowed) were tested as categorical explanatory variables. In terms of marital status, two models were tested. The first model compared those in an ongoing relationship (married or cohabiting) to all other marital categories (single, divorced, widowed) including all respondents. The second model compared those in an ongoing relationship (married or cohabiting) to those who ended their relationship (divorced) including only those with a (supposedly) living present or past partner. Post-test analysis was carried out by the adjusted Wald test.
