Procedure
The data for this study was collected in an online survey created with the Unipark software (Unipark, E.F.S. Survey, version 7). The aim was to recruit a sample with a large variance on the study variables, including maltreatment and revictimization, without focusing primarily on a clinical population. For this purpose, the link for participation was published in numerous private Facebookaccounts and in self-help groups on the Internet which are focused on the topic of traumatic life experiences. Those self-help groups were located by a Google search. First the moderators of active online self-help groups were contacted, the contents of the study were presented and a publication of the link for participation was requested. Of 19 online self-help groups 14 agreed to the publication of the link.The link´s publication was accompanied by brief information on the nature of the questions and the need for potential study participants to be of legal age. There was no incentive to participate.On the first page of the survey, the participants were informed about the contents and risks of the study, about the voluntary and anonymous nature of the survey, and the possibility to desist at any timewithout penalty. This was followed by a declaration of consent to participate in the study. The programming of the survey prevented the participants from seeing the contents of the study if they were underage.
First, demographic data (age, gender, education) were collected. Afterwards, the experience of different types of violence in childhood was investigated. In order to avoid closure effects, the pursue of various types of violence in adolescence and adulthood was only then examined. Subsequently, psychopathology was recorded. The average time taken to complete the survey was 23 minutes. The ethical considerations of this study were reviewed and approved by the Ethics Committee of the Department of Psychology at Bielefeld University. This study was part of a larger survey, that also included additional psychological variables that are not part of this analysis.
Participants
In total, N= 1062 participants commenced the study and of those n= 155 finished it, leaving the completion rate at 14.6%. Most drop-outs took place on the start page (n=789, 74.29 %). Under 3% (n= 26) ofdrop-outs occurred at the declaration of consent and n = 20, 1.88 % drop-outs occurred during the indication of socio-demographic data. For all other questions, there was no noticeably higher number of drop-outs. The study was active for a total of 182 days. The average number of participants per day was 7.08.
Of the 155 participants who completed the study, only female and adult participants were considered. Therefore, eight underage participants and twelve men were excluded. The final sample thus consisted of N=135 participants aged 19 to 67 years (M = 33.4; SD = 11.12). They had an average of 14 years (SD = 3.27) of formal education (primary school, secondary school, university).
Measures
Child abuse experiences.
The Childhood Trauma Questionnaire (CTQ[15-16])is a self-assessment tool for the retrospective assessment of abuse and neglect in childhood. Due to the specific nature of the study, participants were instructed to click on items only if they were applicable before the age of 14. With 28 items, the questionnaire covers the subscales of sexual abuse (e.g., “when I grew up, someone tried to touch me sexually, or get me to touch him/her sexually”), emotional abuse (e.g., “when I grew up, people from my family said hurtful/offending things to me”), physical abuse (e.g., “when I grew up people from my family hit me so hard that I was bruised or scarred”), emotional neglect (e.g., when I grew up, I thought my parents wished I had never been born”) and physical neglect (e.g., “When I grew up, I had enough to eat”). The items were answered on a five-point Likertscalerated from 1(not a bit) to 5(very often). The CTQ showed good internal consistency for all scales in validation studies [17] besides the scale of physical neglect. Due to the low internal consistency of this scale and its high intercorrelation with the other scales [17]it was not included foranalysis in the present study. In the present study, good internal consistency for the four scales used was confirmed (Cronbach’s alpha coefficients: sexual abuse .97, physical abuse .93, emotional abuse .92, emotional neglect .92).The decision on the existence of the different types of abuse was made on the basis of the cut-off values for the summed item scores of Walker et al. [18]. However, the frequencies thus obtained were used only for the descriptive information. In all further analyses, the summed item scores of the individual scales were used, independent of the cutoff values.
Sexual Victimization in Adolescence and Adulthood.
The Potsdam scales for recording sexual aggression and victimization (SEX_AGG_VIC)[19] are a self-report instrument for the assessment of sexual aggression and victimization in adolescence and adulthood (after the age of 14). For the purpose of this study only the sexual victimization subscale was used. The Potsdam scales record sexual victimization with three questions(e.g., “Since the age of 14, has anyone brought (or tried to bring) you into sexual contact by physically threatening or hurting you?”; “Since the age of 14, has anyone brought you (or tried to bring you) to sexual contact by pressuring you with words”). These questions were asked separately based on different potential offenders,which included the type of pre-relationship between offender and victim ((former) partners, acquaintances or strangers). In order to harmonize thisinstrument with the other assessments, the category "colleagues or supervisors at the workplace" was added here. Through the three questions combined with four possible perpetrators each, the instrument consists of 16 items. The items were answered on a four-point Likert scale regarding the frequency of occurrencehree different strategies for exerting pressure were asked (use or threat of physical violence, exploitation of inability to resist, and verbal pressure). A more precise differentiation of forced sexual acts (sexual contact, attempted sexual intercourse, sexual intercourse and other sexual acts, e.g., oral sex) is mentioned in the explanatory text. This more precise differentiation is removed from the item query, as it is not relevant for the purpose of this study. As this is a newly developed instrument, no validity or reliability criteria were available. In the present study the scale was found to demonstrate a sufficiently good internal consistency (Cronbach’s alpha coefficient .71). People are considered victimized if they have been exposed to sexual aggression on at least one occasion. For the present evaluation, the total scores of all items assessing victimization represented the experience of sexual violence in adolescence and adulthood.
Physical and Emotional Victimization in Adolescence and Adulthood.
A recently created screening instrument was used to record and quantify the experience of physical and emotional violence in adolescence and adulthood. It was developed for one of the main German epidemiological studies on health, the "Study on Adult Health in Germany” (DEGS1) of the Robert Koch Institute (RKI) [20]. It explores whether there have been physical or emotional experiences of violence, both from the victim's and the perpetrator's perspective. The preliminary relationship between perpetrator and victim is also recorded ((former) partners, acquaintances, work colleagues/supervisors at the workplace or strangers). In order to avoid confusion with experiences of intrafamily childhood violence, the category of "one person from the family" was omitted. For the same reason, participants were instructed to select the items only if the experience had taken place at or after the age of 14. An additional question on the burden of the respective experience of violence was removed as it was not relevant for the purpose of this study. In total the instrument consisted of eightitems, four concerning physical violence and four concerning emotional violence (e.g., “Has an (ex-)partner physically attacked you from the age of 14 onwards (e.g., hit you, slapped you, pulled your hair, kicked you, threatened you with a gun or an object”; “Has a friend or an acquaintance devalued you from the age of 14 onwards (in terms of your appearance, the way you dress, the way you think, act or work, or possible disability? Or has a friend or acquaintance insulted, threatened, harassed or pressured you?”). The items were answered on a four-point Likert scale regarding the frequency of occurrence rated from As this study only considered the victim`s perspective, eight items about the perpetrator`s perspective were left out. Due to the recent development of this instrument, no reliability or validity criteria was yet available for use. The internal consistencies identified in this study were considered sufficient in view of the low number of items (four per scale; Cronbach’s alpha coefficients: physical violence .59, emotional violence .71). Participants were considered victimized if they had been exposed to physical or emotional aggression on at least one occasion. However, this information was used only for descriptive purposes. For the present evaluation, the total scores of all items concerning victimization represented the experience of physical or emotional violence in adolescence or adulthood.
PTSD Symptoms.
The Primary Care PTSD Screen (PC-PTSD)[21] is a screening instrument for the detection of post-traumatic stress disorder (PTSD). The scale consists of four items in dichotomous response format (Yes/No). It asks whether a person has experienced four symptoms typical of PTSD in the last month: Re-experience, numbness, avoidance, and hyperarousal. The PC-PTSD has optimum efficiency in terms of the best possible combination of sensitivity and specificity at a cut-off value of three[21]. The cut-off value was only used for descriptive purposes in this study. The summed item scores were used in the evaluation, representing a value for the exposure to symptoms of PTSD. The PC-PTSD has good retest reliability and correlates highly with the standard instrument for the detection of PTSD, the Clinician Administered PTSD Scale (CAPS)[21].
Symptoms of Depression.
The health questionnaire for patients (PHQ-9) [22-23] is a screening instrument used for the detection of a depressive disorder. The self-report questionnaire contains nine items that assess whether typical symptoms of depression (based on the DSM-IV criteria) have occurred in the last two weeks. The items were answered on a four-point scale regarding the frequency of occurrence rated from . The PHQ-9 has previously demonstrated good validity and retest reliability [23]. In addition, good values for sensitivity and specificity were confirmed for the stated cut-off values for the severity of depressive symptoms [24].These cut-off values were only used for descriptive purposes in this study. The summed item scores were used in the evaluation representing a value for the burden of symptoms of depression.
Statistical Analyses
The statistical analyses of the study were carried out with the statistical program IBM SPSS Statistics, version 21. All procedures refer to the significance level ∝ = .05. To describe the sample characteristics, the variables of sexual, physical, emotional abuse, emotional neglect in childhood and sexual, physical, and emotional experiences of violence in adulthood were treated as dichotomous variables using the cut-off values described above. They were used as continuous variables in all further analyses. The scales of emotional abuse and emotional neglect were combined due to their high correlation (sum of the scores of the individual scales). The scale of emotional abuse reflected this composite value.
In order to test the predictors for the experience of violence in adulthood, linear regressions were calculated. To carry out the regression analyses, the normal distribution of the residuals was checked by visual inspection first. According to the question of what specific contribution different types of childhood abuse experiences make to the prediction of a subsequent revictimization, all potential predictors were simultaneously included in the regression model. In a first linear regression, the variable sexual experiences of violence in adulthood served as a dependent variable. Independent variables were sexual abuse, emotional abuse, and physical abuse in childhood. In the following two linear regressions, the same independent variables were used. The dependent variables were experiences of physical violence in adulthood and experiences of emotional violence in adulthood.