The Convergence of Retrospective and Prospective Assessments of Childhood Abuse: Prospective Longitudinal Evidence from an At-risk sample A DISSERTATION SUBMITTED TO THE FACULTY OF THE UNIVERSITY OF MINNESOTA BY Marissa D. Nivison IN PARTIAL FULFILLMENT OF THE REQUIERMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Advisor: Dr. Glenn Roisman 2023 © Marissa D. Nivison, 2023 Marissa D. Nivison’s work on this dissertation was supported by the National Institute of Mental Health of the National Institutes of Health via a training grant [T32 MH015755] supporting pre- doctoral research by Marissa D. Nivison at the University of Minnesota. This analysis was also supported in part by a National Institute on Aging grant (R01 AG039453) to J.A. Simpson and NIA R01 AG070138 to E.A. Carlson and K.M. Thomas. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Institutes of Health i ACKNOWLEDGMENTS "No one has measured, not even poets, how much the heart can hold" - Zelda Fitzgerald I have always found comfort in books and learning—which is why I am still a student at 27 years old—and will be for the rest of my life. I have used academics as a way to navigate the world, but I never anticipated all the important relationships I would make along the way. To everyone who has supported me, I would not be who I am without all of you. To Dr. Brandi Dunlap Stupica, who taught my very first college class on attachment theory, who taught me how to be a researcher, and who inspired and set the trajectory of my entire career. She invited me into her home and her family and showed me what true security is—my very own Ms. Honey. To Ms. Sarah Gornicki Pancost, who was the first person to believe in me, who taught me that learning was not about grades but about curiosity, passion, and exploration. To everyone who has scaffolded and supported my career in any way. To Dr. Michelle Englund, who helped me for countless hours with the MLSRA dataset. To Dr. Henriette Warren, for showing me what it means to be an excellent instructor. To Dr. Chryle Elieff, for being an encouraging source of support and kindness. To Dr. Debby Jacobvitz, for lending me her coding system and being willing to train us. To Lindsey Jendraszak, who has been such an unequivocal supporter and fierce advocate for all graduate students, especially for me. To anyone who has worked on the Minnesota Longitudinal Study of Risk and Adaptation (MLSRA) over the last 40 years—this study has been an incredible resource and has laid the foundation for my entire research program. To the participants of the MLSRA for sharing their entire lives with us in the name of scientific discovery. ii To Jasmine Ernst, who was my very first friend in graduate school. I am truly thankful that we have been on this journey together. To Lauren Eales, for being a constant source of support, a great friend, and for literally housing me when my world was on fire. To Cara Lucke, who always brought such joy and warmth to my life. And, of course, thanks to the rest of my BABs: Mariann Howland, Meriah DeJoseph, Emmy Reilly, Danruo Zhong. Truly the best cohort to walk the halls of ICD. To my lab mates and collaborators. To Sophie Magro, I am so thankful that I got to share a lab and an advisor with you throughout graduate school. To Clarissa Filetti, for not only dedicating months of her life to code AAIs with me but for doing such a spectacular job that I truly believe we must share one brain. To Katie DeWitt, who will always be my favorite research assistant. I am so grateful for all of your help over the past five years. To Annika Stensland, who has been one of the most supportive and loving friends I have ever had. Thank you for always checking in on me, for sharing the love of baking with me, and for loving Rooney as much as I do. To all my friends from home, thank you for always supporting me, laughing at my jokes, and listening to my research even if you did not understand it. To my best friends from college who have always pushed me to be my best self while still causing a ruckus: Dr. Kendall Kamp, Dr. Amanda McKeith, and Madam Diplomat (pretty sure that is not your real title, but it sounds cool) Marianna Smith-Mutabchi. To Erin, Frankie, and Magda Stupica, who welcomed me into their family and provided continual love and support. To my best friend of 25 years, Maddy Pasche—you are my very best friend and my person—and I simply would not be where I am without you. To my siblings, whom I love more than anyone else on this planet—goonies never iii say die. Of course, to my cat, Rooney, whose love and affection have always made me feel better. Thank you to all the faculty members serving on my dissertation committee. To Dr. Jeff Simpson, I have been very fortunate to work with him throughout graduate school. Working with Jeff has been extremely rewarding, not only because of his vast expertise but also because he has always been so supportive and kind. To Dr. Dante Cicchetti, for all his support and for all he has done for the field of developmental psychopathology. My dissertation would not have been possible if it were not for the foundation Dante laid within the field of child maltreatment. To Dr. Ann Masten, I am so very grateful to have her on my committee as her work in the field of resilience has given hope to many children and adults who have experienced the kind of trauma investigated in the present dissertation. And, of course, to my advisor, Dr. Glenn Roisman. I was once asked what my favorite part of graduate school was, and my answer came to me easily: working with Dr. Glenn Roisman. I feel incredibly fortunate that I have been able to spend the past five years working with Glenn. I don't think I have had a working relationship as positive and successful as this one has been. He has always been incredibly supportive while also pushing me to be my absolute best. I am grateful for all the mentorship and experience I received prior to graduate school, but I am the scholar I am today because of Glenn’s influence. He has taught me what it means to conduct high-quality science, how to navigate academia, and he has supported my entire career. I am incredibly grateful for all he has done for me professionally— and I know he would say he is just doing his job—but like everything he does, he goes above and beyond what is required of him. Not only am I thankful to have Glenn as a mentor, but I am genuinely lucky to know him at all. Glenn is a kind, funny, and relatable person whom I can always count on. I am so very appreciative of all the flexibility Glenn has given me over the iv years when life events have come up and work cannot be a priority. I am so very grateful to have an advisor who creates an environment where I feel comfortable being myself, joking around, and remembering that the world does not revolve around academia. Truly, thank you, Glenn, for everything you have done for me. It is a pleasure just to know you. v ABSTRACT Scholars across multiple disciplines, both within and beyond developmental psychology, have been long interested in the effects of childhood abuse. Many studies have investigated the legacy of experiences of childhood abuse. However, most of these studies take a retrospective approach to studying early life experiences. Recent meta-analytic evidence (Baldwin et al., 2019) has demonstrated that retrospective measures of child abuse are not a valid proxy for prospective longitudinal data. The present report builds on these meta-analytic findings by investigating the convergence of prospectively documented child abuse from birth to age 17.5 in the Minnesota Longitudinal Study of Risk and Adaptation (MLSRA) and retrospective assessments of physical and sexual abuse newly coded in the context of the Adult Attachment Interview (AAI). Surprisingly, the convergence between prospective and retrospective assessments of childhood abuse was notably higher (k = .71) than previously seen in meta-analytic findings (k = .19). Convergence between retrospective and prospective measures were similar across dichotomous and continuous measures of abuse as well as type of abuse (physical and sexual). Convergence was lowest for abuse perpetrated by non-caregivers and for abuse perpetrated in infancy. The implications for future research investigating childhood experiences of physical and sexual abuse are discussed. vi TABLE OF CONTENTS ABSTRACT ............................................................................................................................................. v i. INTRODUCTION ................................................................................................................................ 1 The Present Dissertation ....................................................................................................................................... 4 ii. METHOD........................................................................................................................................... 10 Participants ......................................................................................................................................................... 10 Measures ............................................................................................................................................................. 11 iv. DISCUSSION ................................................................................................................................... 27 Strengths, Limitations, and Future Directions .................................................................................................... 35 Conclusion .......................................................................................................................................................... 39 v. REFERENCES .................................................................................................................................. 41 vi. LIST OF TABLES ........................................................................................................................... 50 Table 1. ............................................................................................................................................................... 51 Table 2. ............................................................................................................................................................... 52 Table 3. ............................................................................................................................................................... 53 Table 4. ............................................................................................................................................................... 54 Table 5. ............................................................................................................................................................... 55 Table 6. ............................................................................................................................................................... 56 Table 7. ............................................................................................................................................................... 57 Table 8. ............................................................................................................................................................... 58 Table 9. ............................................................................................................................................................... 59 Table 10. ............................................................................................................................................................. 60 Table 11. ............................................................................................................................................................. 61 Table 12. ............................................................................................................................................................. 62 Table 13. ............................................................................................................................................................. 63 Table 14. ............................................................................................................................................................. 64 Table 15. ............................................................................................................................................................. 65 Table 16. ............................................................................................................................................................. 66 Table 17. ............................................................................................................................................................. 67 vii. LIST OF FIGURES ........................................................................................................................ 68 Figure 1 ............................................................................................................................................................... 69 Figure 2 ............................................................................................................................................................... 70 Figure 3 ............................................................................................................................................................... 71 Figure 4 ............................................................................................................................................................... 72 Figure 5 ............................................................................................................................................................... 73 Figure 6 ............................................................................................................................................................... 74 Figure 7 ............................................................................................................................................................... 75 viii. SUPPLEMENTARY MATERIALS ............................................................................................ 76 Neglect dropped analyses ................................................................................................................................... 78 Suspect abuse analyses ....................................................................................................................................... 86 vii ix. APPENDIX: SPSS SYNTAX .......................................................................................................... 94 Re-computing the prospective abuse data to exclude neglect cases ................................................................... 94 SPSS syntax for transforming the retrospective abuse data for analytic variables ............................................ 96 Analysis Syntax .................................................................................................................................................. 99 Attrition analyses Syntax .................................................................................................................................. 106 1 i. INTRODUCTION Across all developmental psychology, and indeed across all psychological science, scholars have shown an interest in the degree to which childhood caregiving experiences shape mental and physical health in adulthood. For example, a common measure from a high impact paper was created to assess Adverse Childhood Experiences (ACEs; Felitti et al., 1998). ACEs includes a simple tally of adverse childhood events such as childhood verbal, physical, sexual abuse or having a parent go to prison in childhood. This tally of adverse experiences was associated with alcoholism and drug abuse, depression, smoking, poor self-rated health, sexually transmitted diseases, and severe obesity—along with the diagnosis of serious adult illnesses and injuries including heart disease, cancer, lung disease, skeletal fractures, and liver disease. The impact of childhood adversity has been examined in nearly every discipline of psychology including biological psychology (Herzog et al., 2020; Carpenter et al., 2009; van Harmelen et al., 2010), clinical psychology (Vallati et. al., 2020), and social/personality psychology (Neel et al., 2016). However, a critical issue with studies such as these is that they take a retrospective, self- report approach to measuring childhood caregiving quality. As such, the findings of these studies hinge crucially on the ability to measure childhood experiences accurately and validly from a retrospective perspective. Despite the assumption that retrospective, self-report measures of early caregiving might be used as a proxy for prospectively documented evidence of caregiving in childhood, few measures have been validated with prospective longitudinal evidence. One benefit of retrospective measures is that they are quite simple, easy, cost-effective, and efficient to administer as they typically require only one assessment time point and are often self-report in nature, which can be particularly beneficial considering the significant resources that support 2 prospective, longitudinal work. More specifically, prospective, longitudinal studies interested in studying the legacy of early experiences need to follow children often starting at birth, or even prenatally across the life course. This kind of work takes decades and often a scholar’s entire career, if not more. Longitudinal work can be quite resource intensive requiring extensive funding which is likely not guaranteed for more than 5-6 years at a time. Given the difficult nature of prospective longitudinal work it is no wonder that scholars have often used a one- assessment retrospective measure. However, despite the extensive resources that prospective longitudinal work requires, it is most advantageous as it often allows scholars to study the events of childhood as they are unfolding. Scholars can examine prospectively, if resources allow, childhood caregiving through multiple methods (e.g., observed, parent-report, child-report etc.) by multiple informants (e.g., parents, children, teachers etc.) across development. Such rich data is unattainable in principle via one retrospective self-report measure of childhood caregiving. However, on the one hand, in the case of childhood maltreatment, retrospective measures may give insight into an individual’s perception of experiences which is especially beneficial given the difficulty of prospectively measuring childhood maltreatment (e.g., the hidden nature of maltreatment; Barnett, Manly, & Cicchetti, 1993; Danese, 2020). On the other hand, given that retrospective measures ask individuals to recall past information (that sometimes occurred decades before), these measures may be subject to a number of biases such as recall bias and depressogenic bias (Raphael, 1987; Brewin, Andrews, & Gotlib, 1993; Danese, 2020). Fortunately, the extent to which retrospective and prospective reports of childhood caregiving converge has been addressed in the developmental literature with one of the first empirical investigations reported nearly 90 years ago (i.e., Pyles, Stolz, & MacFarlane, 1935). Subsequent work investigating the validity of retrospective reports has resurfaced periodically 3 every few decades since the 1930s (e.g., Yarrow et al., 1970; Henry et al., 1994; Hardt & Rutter, 2004). It was not until recently, however, that this work had been investigated on a larger scale— mostly with a focus on the convergence of retrospective and prospective reports of childhood maltreatment (e.g., Reuben et al., 2016; Newbury et al., 2018; Baldwin et al., 2019; Danese, 2020), though there has still been limited work investigating the retrospective reporting of normative caregiving experiences (i.e., parental sensitivity; Nivison et al., 2021a). Of note, most work investigating the convergence of retrospective and prospective measures of childhood maltreatment have come from two large longitudinal studies: the Dunedin Multidisciplinary Health and Development Study and the Environment Risk (E-Risk) Longitudinal Twin Study (but also see Tajima et al., 2004; Shaffer et al., 2008; Naicker et al., 2017). The Dunedin study examined childhood adversity prospectively from ages 3-15 years of age and retrospectively at age 38. Reuben and Colleagues (2016) found that retrospective and prospective measures of ACEs converged moderately (k = .31). The E-risk study examined prospective childhood maltreatment was collected at ages 5, 7, 10, and 12 and were reported by caregivers, researchers, and clinicians. Newbury and colleagues (2018) reported that retrospective measures of childhood maltreatment were assessed when participants were 18 years old. There was low agreement between retrospective and prospective measures on overall maltreatment (k = .19) though when examining maltreatment type, agreement was highest on sexual abuse (k = .31). Authors concluded that retrospective and prospective measures capture largely non-overlapping groups of individuals. Moreover, a recent meta-analysis has demonstrated exactly that. Baldwin and colleagues (2019) meta-analyzed data from 16 unique studies and found that agreement between retrospective and prospective measures of childhood maltreatment converged poorly (k = .19). 4 Agreement was higher when retrospective measures were based in interviews rather than self- reports. Agreement was not moderated by prospective measure type, age at retrospective reporting, or biological sex. Furthermore, more than half of individuals with prospective records of child maltreatment did not retrospectively report maltreatment. Likewise, more than half of individuals who retrospectively reported childhood maltreatment did not have concordant prospective record. Baldwin and colleagues concluded that retrospective and prospective different groups of people and cannot be used interchangeably. Despite the evidence from these existing studies, they have often been constrained to samples where: (1) prospective measures of abuse do not cover the entirety of childhood, (2) retrospectively assessed maltreatment is assessed via a self-report measure (e.g., rather than a structured interview), and (3) only one retrospective assessment of maltreatment was administered. The Present Dissertation The present dissertation extends the literature on the convergence of retrospective and prospective reports of childhood maltreatment by drawing on the Minnesota Longitudinal Study of Risk and Adaptation (MLSRA; Sroufe et al., 2005). The MLSRA is a prospective longitudinal study that has followed mothers and their children from three months prior to their birth to their mid-40s. The MLSRA contains extensive childhood abuse and neglect data from birth to 17.5 years of age including information on type, perpetrator, and developmental timing. Abuse and neglect data were assessed through multiple methods and via multiple informants, including parent-child observations, caregiver interviews, reviews of available child protection and medical records, adolescent reports, and teacher interviews. At ages 19 and 26 years of age, participants were administered the Adult Attachment Interview (AAI; Main et al., 1985). Although the AAI 5 was not designed to be a measure of retrospective maltreatment the interview often contains information related to childhood maltreatment, and specifically asks interviewees if they had experienced childhood physical and sexual abuse. For the present report, the age 19 and 26-year AAIs were recoded for retrospective abuse using a coding system developed by Leon, Jacobvitz, and Hazen (2004; see also Jacobvitz, Leon, & Hazen, 2006). The retrospective coding system focuses on experiences of harsh physical punishment such as spanking or repeated hitting and experiences of sexual abuse. The retrospective coding system does not, however, account for physical neglect. More specifically, the retrospective system focuses on abusive acts of commission (i.e., physical, and sexual abuse) rather than omission (i.e., physical neglect) largely because the AAI does not explicitly ask if participants have experienced physical neglect, whereas it does explicitly ask about physical and sexual abuse. It is important to note that the 19-year AAIs (but not the 26-year AAIs) in the MLSRA have been previously coded for retrospective abuse (Shaffer et al., 2008), however, they used a different coding system that had been developed based on the criteria outlined by Barnett and colleagues (1993). Although this coding scheme is very rigorous and detailed, it was not designed specifically for use in the AAI. Therefore, instead of using the existing data and coding system, for this dissertation were recoded both the 19 and 26-year AAIs by using a system that was developed specifically for the AAI (i.e., Leon et al., 2004; Jacobvitz et al., 2006). This is particularly important because the AAI is not an interview explicitly focused on experiences of childhood abuse and the coding systems developed by Jacobvitz and colleagues were based on the original coding schemes developed for the AAI. The present dissertation examined the convergence between prospectively assessed abuse from birth to age 17.5 years and an overall measure of retrospectively reported abuse at ages 19 6 and 26 years. Additionally, the current project was one of the first studies to leverage two retrospective assessments of childhood abuse (i.e., at ages 19 and 26). This dissertation was uniquely positioned to examine whether the magnitude of associations between prospective abuse and retrospective abuse varied at age 19 compared to 26 years. Given little-to-no prior evidence this question was largely exploratory. On the one hand, it was possible that convergence of abuse may be higher at age 19 as it is the most proximal to childhood. On the other hand, age 26 convergence with the prospective data could have been higher given the time since childhood to process events and individuals may disclose more in the AAI. It was also possible that convergence would be similar across both assessments. A recent examination of the MLSRA (Köber et al., 2019) found that autobiographical memories from the 19- and 26-year AAIs were moderately stable (~22-28% adjectives consistent from 19 and 26 AAIs), though this study primarily found that the adjectives individuals used to describe their primary caregivers were moderately stable and did not focus on specific experiences of abuse. In light of this preliminary evidence, the stability of retrospective measures of abuse was examined from the 19 to 26-year AAIs—this was largely exploratory, though moderate stability was expected given the evidence from Köber and colleagues (2019). Additionally, given the wealth of the prospective data in the MLSRA, this dissertation examined the convergence of an overall assessment of childhood abuse as well as examined specific parameterizations of abuse including the subtype of abuse (i.e., physical and sexual abuse), the developmental period in which abuse occurred (i.e., infancy, early childhood, middle childhood, adolescence), as well as perpetrator of abuse (i.e., mother-figure, father-figure, or non-parent). Although examining by specific parameterizations of abuse was largely exploratory, it is important to examine convergence beyond just a simple dichotomous variable of whether 7 abuse is present given calls from within the maltreatment literature to examine specific parameterizations of maltreatment (e.g., Cicchetti, 2013). Previous research has examined convergence by abuse type (e.g., Baldwin et al., 2019) but age at which abuse occurred as well as perpetrator of abuse had not previously been examined in relation to convergence and was exploratory in nature. It was possible that abuse that occurred in infancy and early childhood (i.e., before the age of five) may be less likely to be recalled in the context of the AAI given infantile amnesia (Hardt & Rutter, 2004). Abuse occurring in middle childhood and adolescence (i.e., age 6 to 17.5 years of age) may be better recalled compared to early life given the temporal proximity to the retrospective measures. These specific parameterizations of abuse are consistent with previous MLSRA studies (i.e., Raby et al., 2017; Labella et al., 2018; Nivison et al., 2021b; VanMeter et al., 2021). Overall, the MLSRA was very well positioned to examine questions regarding the convergence of retrospective and prospective measures of childhood abuse and had the following aims: Aim 1: Examined the extent to which a composite measure of retrospectively recalled abuse assessed at ages 19 and 26 years converged with prospectively assessed childhood abuse from birth to 17.5 years of age operationalized as a binary presence of abuse variable. Hypothesis: Consistent with the meta-analytic association between retrospective and prospective measures (i.e., k = .19; Baldwin et al., 2019), I expected agreement between dichotomous measures of retrospective and prospective abuse to be a kappa of approximately .20. Convergence between retrospective and prospective measures of abuse were interpreted according to the benchmarks outlined in the interrater reliability literature (i.e., McHugh, 2012). According to this literature, kappa values between 0-.20 are considered to have no agreement; kappa values between .21-.39 are considered to 8 have minimal agreement; .40-.59 = weak agreement; .60-.79 = moderate agreement, .80- .90 = strong agreement; and above .90 = almost perfect agreement. Aim 2: Compared the magnitude of the associations between prospectively assessed abuse and each measure of retrospective abuse assessed at 19 and 26 years of age, respectively. Hypothesis: This aim was exploratory in nature. On the one hand, it was possible that there would be a larger association between the prospective assessment and the 19-year retrospective assessment given that the 19-year assessment was closer in proximity to childhood and individuals may thus be able to be better recall their childhood events. On the other hand, convergence could have been larger with the 26-year retrospective assessment as individuals will have had more time to process the events of their childhood and may disclose more experiences during the AAI. Aim 2a: Examined the stability of retrospectively recalled abuse from age 19 to age 26 years. Hypothesis: This aim was largely exploratory in nature. Although the test-retest reliability of this specific scale had yet to be investigated, the test-retest reliability of the AAI itself is moderately high (k = .63, Bakermans-Kranenburg & van IJzendoorn, 1993) and other measures of retrospective adversity were also moderately high (r = .61, Pinto, Correia, & Maia, 2014). Additionally, within the MLSRA, recent evidence that has found moderate stability of adjectives from the 19 to 26-year AAI (Köber et al., 2019) given this evidence I expected that retrospectively recalled abuse would be somewhat moderately stable from the 19 to 26-year AAIs. 9 Aim 3: Examined the magnitude of the associations outlined in Aims 1-2 using continuous measures of both prospective and retrospective abuse. (Note: Aim 4 below could not be run on a continuous basis as the parameterizations of abuse are dichotomous in nature). Hypothesis: This aim was largely exploratory. Though it was possible that a continuous severity variable may have captured more variation in experiences of childhood abuse than the binary variable assessing whether abuse ever occurred in childhood. Aim 4: Examined whether the convergence of retrospective and prospective measures of abuse varied by specific parameterizations of prospectively documented abuse. Aim 4a: Compared the magnitude of associations of retrospective and prospective measures by prospectively documented abuse type (i.e., physical and sexual abuse) Aim 4b: Compared the magnitude of associations of retrospective and prospective measures by prospectively documented abuse developmental period in which the abuse occurred. Aim 4c: Compared the magnitude of associations of retrospective and prospective measures by prospectively documented perpetrator of abuse. Hypothesis: Aim 4 was largely exploratory in nature though a recent meta-analysis did find that convergence of retrospective and prospective measures varied by abuse subtype (i.e., sexual abuse: k = .16; physical abuse: k = .17; neglect: k = .09; Baldwin et al., 2019). As such, it was possible that magnitude of convergence may have varied between physical and sexual abuse (Aim 4a). Whether convergence varies depending on the age in which retrospective abuse occurred had not been directly tested in the literature previously, however some (e.g., Hardt & Rutter, 2004) have suggested that infantile amnesia may explain the low convergence between retrospective and prospective 10 measures. Though it was also possible that abuse occurring in middle childhood or adolescence may converge higher given that it is most recent to the retrospective report (Aim 4b). Finally, no study to date had examined whether convergence varies across different perpetrators of abuse (Aim 4c). ii. METHOD Participants Participants were drawn from the Minnesota Longitudinal Study of Risk and Adaptation (MLSRA; Sroufe et al., 2005), a prospective longitudinal study following mothers and their children from 3 months before the child’s birth to around 40 years of age. Between 1975 and 1977, 267 pregnant mothers seeking free prenatal care in Minneapolis, Minnesota were recruited to participate. Individuals were recruited if they were living at or below the poverty line at the time of their child’s birth. Forty-eight percent of participating were adolescents, 65% were single, and 42% had not completed a high school education. The analysis focused on participants who have completed at least one codable AAI at either 19 or 26 years of age who also had complete prospective abuse data. The present subsample was comprised of 162 participants (48% female, 68% white, non-Hispanic) overall, but varies by the parameterizations of abuse. The subsample (N = 162) did not differ from the original sample (N = 267) on socioeconomic status, ethnicity/race, or biological sex. However, the present subsample (n = 162, M = 12.30, SD = 1.59) when compared to those excluded (n = 104, M = 11.73, SD = 2.01) had significantly higher maternal education (t [183.45] = −2.41, p=.02, equal variances not assumed). Despite this, average levels of maternal education in the current sample were still equal to or less than a high school education, consistent with an at-risk cohort. 11 Measures Prospective abuse. The MLSRA uses the rubric childhood experiences of adverse caregiving as an umbrella term to refer to a variety of atypical parent-child experiences that were prospectively measured in the MLSRA cohort and are believed to be harmful to children’s development. The present study focused exclusively on information collected about MLSRA participants’ adverse caregiving experiences of physical abuse, sexual abuse, and neglect. This information was re-coded to apply contemporaneous definitions of abuse and neglect, to identify the specific perpetrator and ages of the abuse and neglect experiences, and to assess the reliability of those coding decisions. Coding criteria were based on definitions developed by the Centers for Disease Control and Prevention (CDC) in order to “promote consistent terminology and data collection related to child maltreatment” (Leeb et al., 2008, p. 4). The coding included: 1) neglect of a child’s basic physical or cognitive needs, defined as a caregiver’s failure to provide adequate hygiene, shelter, clothing, medical care, supervision, or education, 2) physical abuse, defined as a caregiver’s “intentional use of physical force against a child that results, or has the potential to result in, physical injury” (Leeb at al., 2008, p. 14), 3) sexual abuse, defined as sexual contact (e.g., molestation, rape) or noncontact exploitation (e.g., intentional exposure of child to pornography) by a custodial caregiver or by a perpetrator five or more years older than the target child. Although the CDC criteria only addresses sexual abuse perpetrated by a caregiver, the inclusion of non-caregiving perpetrators and the use of a five-year cutoff is consistent with other research in this area (e.g., Stoltenborgh et al., 2011). These CDC definitions were supplemented by a set of more specific coding guidelines that distinguished clear indicators of physical abuse, sexual abuse, and physical/cognitive neglect from ambiguous indicators that were not sufficient for classification in isolation of other 12 evidence. These additional guidelines were developed in consultation with MLSRA senior researchers, Minnesota state law, and available research literature (e.g., Barnett et al., 1993) and are available from the first author upon request. However, the classifications of childhood experiences of abuse or neglect do not necessarily reflect criteria for maltreatment used by child protective services, which vary from state to state. As such, our scoring of abuse and neglect does not necessarily mean that these children or their families were involved with child protective services. Although emotional unavailability or lack of caregiver responsiveness has proven to be an important dimension of adverse caregiving (especially for young children), with pernicious developmental consequences (National Scientific Council on the Developing Child, 2012; Sroufe et al., 2005), this dimension was not included in the current coding criteria due to insufficient information across developmental periods. Similarly, exposure to violence between caregivers and other forms of environmental violence were not included in the current set of codes. Exposure to violence between caregivers is captured by a separate variable in the MLSRA dataset (e.g., Narayan, Englund, & Egeland, 2013), and insufficient information was available to code adequately exposure to other forms of environmental violence. Judgments regarding abuse and neglect experiences were made for participants whose records had been previously flagged as potentially ever abused or neglected (n = 139, 52% of the original sample). For these cases, all available data collected from birth to 17.5 years (up to 25 assessments) were reviewed for information regarding caregiving quality, physical discipline, supervision, home environment, physical and sexual assault, child protective service involvement, and foster care history. Information was obtained from parent-child observations, caregiver interviews, reviews of available child protection and medical records, adolescent 13 reports, and teacher interviews. Disclosures of childhood physical or sexual abuse during the Adult Attachment Interview (AAI; George, Kaplan, & Main, 1985), a retrospective interview regarding early caregiving experiences administered at 19 years of age, were not included in the present set of codes except in situations in which an experience of abuse was initially identified based on records through age 17.5 years, but there was insufficient detail to code the specific developmental period or perpetrator (e.g., an adolescent disclosed a history of sexual assault without specifying whether the perpetrator was a peer). In these cases, available AAIs were consulted only for clarifying information about the previously identified incident (i.e., prospective judgements on whether abuse had occurred was not made based on the AAI). Coding focused on the presence or absence of physical abuse, sexual abuse, and/or neglect in each of four developmental periods (Infancy: birth to 24 months; Early Childhood: 25 months to five years; Middle Childhood: 6-12 years; and Adolescence: 13-17.5 years). For incidents of physical and sexual abuse, coders additionally specified the perpetrator. Perpetrators included maternal caregivers (biological mothers, stepmothers, grandmothers), paternal or father figures (biological fathers, stepfathers, adoptive fathers, and mothers’ live-in boyfriends), and non-parental figures (relatives, neighbors, babysitters, and family friends). Two coders reviewed each case and demonstrated good to excellent reliability for all parameters: kappas were between .80 and .98 for presence or absence of physical abuse, sexual abuse, and/or neglect, .80 and .84 for presence or absence of each type during each development period; and .80 and .98 for incidents of physical or sexual abuse by each category of perpetrator. All discrepancies were resolved by consensus. Within the full sample of MLSRA participants (N = 267), 102 individuals were classified as having ever experienced physical abuse, sexual abuse, and/or neglect; 81 were coded as not 14 having experienced abuse or neglect; and the status of 84 was deemed unclear due to missing data (see below). By developmental period, 47 individuals were classified as being abused and/or neglected in infancy (of the 211 with sufficient data to allow for confident classifications of abuse and/or neglect during this developmental period), 66 in early childhood (of the 185 with sufficient data during this developmental period), 66 in middle childhood (of the 190 with sufficient data during this developmental period), and 21 in adolescence (of the 179 with sufficient data during this developmental period). Within the current sample of participants who had at least one AAI in young adulthood (N = 178), 69 individuals were classified as having ever experienced physical or sexual abuse. Among participants with histories of abuse, 46% had experienced sexual abuse and 74% had experienced physical abuse (not mutually exclusive). Within the abused group, 14% experienced abuse in infancy, 46% during early childhood, 65 % during middle childhood, and 28% during adolescence (not mutually exclusive). In terms of chronicity, 54% of this group experienced abuse during one developmental period, 25% during two periods, 12% during three periods, and 1% during all four developmental periods; 8% had insufficient data to determine the number of developmental periods during which abuse occurred. Among participants with histories of abuse, 73% experienced one type of abuse, 20% experienced two types; 7% had insufficient data to determine the number of abuse types experienced. With respect to perpetrator, 51% of participants who experienced abusive acts of commission were abused by a maternal perpetrator, 49% by a paternal perpetrator, and 32% by a non-parental perpetrator (not mutually exclusive). In order to separate participants who had not experienced abuse and/or neglect from those with missing data, the abuse and neglect variables were coded as missing if: (a) the participant was not coded as having been abused based on the available information, and (b) the participant 15 was missing two or more full assessments within any given developmental period. Within the current sample, 15 participants were classified as having missing information related to abuse. The remaining 94 individuals comprised the non-abused group; the number of missing assessments for this group did not differ from the group of individuals who were classified as having experienced abuse (t [108.61] = -1.32, p = .19). Although the MLSRA has sufficient prospective physical neglect data these data were not included in the present analysis given that the retrospective coding system does not code for physical neglect as it focuses on acts of commission (i.e., physical and sexual abuse) rather than omission (i.e., physical neglect). However, the present analyses examined prospective abuse in two predominate ways: 1) examining indicators of overall presence of abuse and 2) examining specific parameterizations of abuse. To examine abuse overall, two variables were created. The first variable is a binary variable indicating whether physical or sexual abuse occurred any time between birth and 17.5 years of age. The second variable is a quasi-continuous scale of total experiences of abuse. This variable was calculated by summing the number of types of abuse (i.e., physical and sexual abuse) in each developmental period (i.e., infancy, early childhood, middle childhood, and adolescence). Given that each of these subtypes were coded on a dichotomous basis for each developmental period, the total experiences of abuse scale has a theoretical minimum of zero (i.e., the participant did not experience any abuse in any developmental period) to a theoretical maximum of 8 (i.e., the participant experienced both physical and sexual abuse in every developmental period). Additionally, further parameterizations of abuse were examined including 1) type of abuse (i.e., physical and sexual abuse) 2) developmental period in which abuse occurred (i.e., infancy, early childhood, middle childhood, and adolescence) and 3) the perpetrator of abuse (i.e., mother/mother-figure, 16 father/father-figure, and non-parental figure). All parameterizations of abuse were coded on a dichotomous basis of whether the event occurred (i.e., yes was abused by mother figure versus no was not abuse by mother figure). Prospective analytic variables. The prospective variables used in the present analyses were based on variables previously used in the MLSRA (e.g., Nivison et al., 2021b; Raby et al., 2017). However, the present analysis focused on physical and sexual abuse, but not neglect as it was not possible to code neglect retrospectively in the context of the AAI. Therefore, the existing prospective variables were adjusted to remove neglect. An overall severity variable was originally created in Nivison and colleagues (2021b) which created a quasi-continuous variable by summing each type of abuse/neglect (i.e., physical, or sexual abuse, neglect) that occurred in each developmental period (i.e., infancy, early childhood, middle childhood, adolescence). This scale had a theoretical minimum of zero and a theoretical maximum of 12. Given that the present analyses did not include prospective neglect the scale has a theoretical minimum of zero and a theoretical maximum of eight. See Table 1 for the prospective variable guide. Retrospective abuse. Experiences of overall retrospective abuse were coded in the context of the AAI at ages 19 and 26 year. The AAI is a twenty-question semi-structured interview that asks individuals to recall their experiences with their primary caregivers in childhood. The AAI also inquires about experiences of loss/grief, rejection, separation, and trauma/abuse. Although the AAI was not originally developed to be a retrospective measure of abuse experiences, Leon and colleagues (2004) developed a scale to assess physical and sexual abuse in the context of the AAI adapted from the original abuse scales in the Main and Goldwyn coding system (see also, Jacobvitz et al., 2006). AAIs were coded for experiences of physical and sexual abuse on a 9- point severity of abuse scale: physical and sexual abuse were coded separately, resulting in two 17 9-point severity abuse scales. Individuals who did not indicate any experiences of abuse in the AAI received a score of one. Low scoring individuals (i.e., a score of 2-3) described occasional spankings, frequent, but not harsh spankings or comments with sexual connotation. Those scoring a middle to higher score (i.e., 4-7) described experiences of harsh physical contact (e.g., spanking) that did not reach the threshold of abuse in the original AAI coding systems, extreme threats, or experiences that the individual was told about but does not explicitly recall happening to them. Finally, individuals with a high score (i.e., 8-9) described incidents of severe physical and sexual abuse, which also meet the criteria of the original AAI coding manual (Main et al., 2003-2008). Including things such repeated hitting of the child in the face, physical contact that leaves a mark, severe hitting that results in the child experiencing extreme fear of the parent, any sexual contact between the parent and the child. A dichotomous variable was created based on presence of abuse meeting legal criteria (consistent with Leon et al., 2004; Jacobvitz et al., 2006) for both the physical and sexual abuse scales. Participants who received high scores (8-9) on the either the physical or sexual abuse scales were placed in the abuse category consistent with Leon and colleagues (2004). All AAIs were coded by an expert coder with 33% of cases double coded for reliability. Coders demonstrated excellent interrater reliability for both the 19-year AAIs (physical abuse: ICC = .98; sexual abuse: ICC = .94, n = 52) and 26-year AAIs (physical abuse: ICC = .98; sexual abuse: ICC = .95, n = 52). The individuals who coded the 19- and 26-year AAIs for retrospective abuse were not involved in any other coding of the MLSRA AAIs (e.g., traditional Main & Goldwyn coding or secure base script coding). Retrospective analytic variables. Adult Attachment Interviews (AAIs) were administered at ages 19 and 26. Each AAI was coded for recalled experiences of retrospective physical and sexual abuse. AAIs were coded on a continuous 1-9 scale with 1 representing no abuse, and 9 18 representing severe abuse. Consistent with prior work (Leon et al., 2004; Jacobvitz et al., 2006), the scales were transformed from the continuous scales into a dichotomous variable representing whether the participant retrospectively reported abuse— individuals scoring 1-7 on the scale were coded as zero indicating that they had not recalled severe abuse and individuals scoring an 8 or 9 were coded as a one indicating that they had recalled severe abuse. This cutoff threshold was previously established by the original creator of the scale (Leon, Jacobvitz, & Hazen, 2004; Jacobvitz, Leon, & Hazen, 2006). This resulted in the following dichotomous variables: “recalled physical abuse age 19,” “recalled sexual abuse age 19,” “recalled physical abuse age 26,” and “recalled sexual abuse age 26,” wherein (0 = non-abused, 1 = abused). In an effort of further aggregation, an overall abuse variable was created within each assessment (i.e., 19 and 26 year)—if an individual had reported neither physical or sexual abuse they were coded as a 0 in the “non-abuse” group- if they had experienced at least one or both types of abuse they were coded as a 1 in the “abuse” group. This resulted in the following dichotomous variables: “recalled abuse age 19” and “recalled abuse age 26.” Finally, the recalled abuse variables at age 19 and 26 were averaged to create an overall “omnibus retrospective abuse” variable—this resulted in three possible outcomes: 0 = no recalled abuse at either the 19 or 26 assessments, .5 = recalled abuse at either the 19 or 26 assessment, 1= abuse recalled at both the 19- and 26-year assessments. The omnibus variable is the primary retrospective variable used in the following analyses. Additionally, retrospective abuse was examined using a continuous scale. The AAIs were originally coded on a 1-9 scale, but to be consistent with the prospective data, the AAIs were transformed to be on a 0-8 scale. This resulted in the continuous age 19 physical abuse, continuous age 19 sexual abuse, continuous age 26 physical abuse, and continuous age 26 sexual abuse variables. Like the dichotomous variables, the physical and sexual abuse scales were 19 averaged within assessment to produce the overall continuous recalled abuse age 19 and continuous recalled abuse age 26 variables. Again, I averaged both assessments together to create an “continuous omnibus retrospective abuse scale.” See Table 2 for the retrospective variable guide. Planned Analysis To address Aim one, raw agreement between the ever abused prospective and omnibus retrospective abuse was assessed using a Cohen’s kappa test. Agreement between these measures of retrospective and prospective abuse was expected to be a kappa of approximately .20 consistent with meta-analytic findings (Baldwin et al., 2019). Convergence between retrospective and prospective measures of childhood abuse were interpreted based on the benchmarks outlined in the interrater reliability literature (i.e., McHugh, 2012). Values between 0-.20 are considered to have very little agreement; values between .21-.59 are considered to have weak-to-minimal agreement; .60-.79 = moderate agreement; and .80-1.00 = strong-to-perfect agreement. To address Aim two Cohen’s kappas were estimated between the prospective dichotomous ever abused variable and each retrospective measure of abuse and 19 and 26 years. This aim was exploratory in nature. However, if the difference between the kappa values were considerable, then I would have concluded that retrospective assessments of childhood abuse vary in their convergence with prospective measures depending on the age at which the retrospective measure is administered. However, if the difference between the kappa values were negligible then I would have concluded that retrospective reports of abuse in our sample do not vary by the time at which they are reported. To address Aim 2a bivariate correlations were run between the 19 and 26-year recalled experiences of abuse. This aim was largely exploratory as 20 the test-retest reliability of this scale had yet to be investigated but given that the test-retest reliability of the traditional scales of the AAI are moderately high (k = .63, Bakermans- Kranenburg & van IJzendoorn, 1993) and other measures of retrospective adversity were also moderately high (r = .61, Pinto, Correia, & Maia, 2014)— I expected the stability to be moderately high. To address Aim three, Aims one and two were repeated using the quasi-continuous measure of prospective abuse experiences as well as a continuous measure of retrospective abuse. The prospective variable was operationalized as the sum of each type of abuse (i.e., physical/sexual abuse) in each development period (infancy, early childhood, middle childhood, adolescence) this scale had a theoretical minimum of zero (i.e., no experiences of abuse in any developmental period) and a theoretical maximum of eight (i.e., both types of abuse in every developmental period). The retrospective variable was a severity of abuse score on a 0-8 scale (i.e., higher scores reflect more severe experiences of abuse). For the omnibus analysis, the 19- and 26-year retrospective abuse scores were averaged. The logic outlined in Aims 1 and 2 was repeated, however, to account for the nature of the continuous variables intraclass correlations (ICC) were run in place of kappas. This aim was largely exploratory though it was possible that convergence would be higher when using continuous measures of abuse as continuous measures may capture more variation in abusive experiences than a dichotomous cut off variable. To address Aim four, Cohen’s kappas were run between the overall omnibus retrospective measures of abuse and three different parameterizations of prospective abuse. To address Aim 4a two variables of prospective abuse type were examined in relation to the overall omnibus retrospective dichotomous variable: prospective presence of physical abuse (did any physical abuse occur anytime between birth and 17.5 years; 0 = no physical abuse, 1 = yes, 21 physical abuse) and prospective presence of sexual abuse (did any sexual abuse occur anytime between birth and 17.5 years; 0 = no sexual abuse, 1= yes, sexual abuse). Two kappas were run examining the extent to which 1) physical abuse and 2) sexual abuse converge with the overall omnibus dichotomous variable of retrospective abuse. The same process and logic were repeated with both developmental period (i.e., did abuse occur in any of the four developmental periods; Aim 4b) and perpetrator type (i.e., ever abused by mother-figure, father-figure, or non-parent; Aim 4c). Sensitivity Analyses. The scope of this project was to examine the extent to which physical and sexual abuse can be accurately recalled in a retrospective report and therefore, individuals who were neglected (prospectively documented), but not physically or sexually abused fall into the “non-abused” group when these analyses were run. Although neglect is not specifically abuse, but is a form of maltreatment, it is possible that convergence may have varied by operationalizing neglected individuals in the “non-abuse” category. However, because the retrospective report did not measure neglect, I cannot compare the neglect group on their own. Therefore, to fully make accurate conclusions about the extent to which physical and sexual abuse can be retrospectively recalled, Aims 1-4 were rerun excluding participants with prospectively documented cases of neglect only (i.e., did not experience physical or sexual abuse). In the current sample, 15 participants had experienced only neglect, but not physical or sexual abuse and were excluded in these sensitivity analyses. Because this was a relatively low number of participants, and I did not expect convergence to vary greatly once these participants were excluded. Additionally, there were a few cases (~ 5 participants) where the individual either refused to answer the abuse question or did not provide enough information (or was not properly probed by the interviewer) and therefore coders could reasonably discern whether or which type 22 of abuse had occurred. In these cases, the individuals were not included in main analyses, but because abuse was suspected to have occurred, these individuals were included in sensitivity analyses in an effort to examine the full sample of available AAIs. iii. RESULTS Descriptive statistics for all prospective and retrospective abuse variables are presented in Tables 3 and 4, respectively. The distribution of overall abuse documented prospectively, and overall abuse recalled retrospectively are presented in Figures 1 and 2, respectively (the distribution of the continuous scales are presented in Supplementary Figures 1 and 2). The following results are presented in two sections: the first examines the descriptive breakdown of the retrospective and prospective reports of abuse overall, by abuse type, and by assessment. The second section focuses on the convergence of retrospective and prospective reports of abuse. Overall abuse. Within the current analytic frame, 162 participants had an AAI at either or both the 19- and 26-year assessment and complete prospective abuse data. Of those 162 participants, 68 participants recalled at least one experience of physical or sexual abuse at either the 19- or 26-year assessment. Of those who had recalled an experience, 48 participants also had prospective evidence of abuse, resulting in 14 individuals who had recalled abuse, but did not have prospectively documented evidence of abuse occurring in childhood. Finally, 20 participants did have prospectively documented abuse in childhood, but did not report having experienced any abuse during the AAI at either age 19 or 26. This information is outlined in both Table 5 and Figure 3. Physical abuse. Within the 162 participants who had an AAI at either or both the 19- and 26-year assessment and complete prospective abuse data, 50 participants had reported 23 experiencing physical abuse in childhood during the 19- or 26-year AAI. Of those, 31 participants also had prospective evidence of physical abuse and 16 participants recalled abuse but did not have prospective evidence. 19 participants who had documented prospective abuse did not retrospectively recall abuse. This information is outlined in both Table 6 and Figure 4. Sexual abuse. Of the 157 participants who had an AAI at either or both the 19- and 26- year assessment and complete prospective abuse data, 26 participants had reported experiencing sexual abuse in childhood during the 19- or 26-year AAI. Of those, 20 participants also had prospective evidence of sexual abuse and 6 participants recalled sexual abuse but did not have prospective evidence. 11 participants who had documented prospective abuse did not retrospectively recall abuse. This information is outlined in both Table 7 and Figure 5. 19-year recall. Of the 153 participants who had an AAI at age 19 and complete prospective data, 42 participants recalled at least one experience of physical or sexual abuse at the 19-year assessment. Of those who had recalled an experience, 34 participants also had prospective evidence of abuse, resulting in 8 individuals who had recalled abuse, but did not have prospectively documented evidence of abuse occurring in childhood. Finally, 30 participants did have prospectively documented abuse in childhood, but did not report having experienced any abuse during the AAI at age 19. This information is outlined in both Table 8 and Figure 6. 26-year recall. Of the 146 participants who had an AAI at age 26 and complete prospective data, 52 participants recalled at least one experience of physical or sexual abuse at the 26-year assessment. Of those who had recalled an experience, 40 participants also had prospective evidence of abuse, resulting in 12 individuals who had recalled abuse, but did not have prospectively documented evidence of abuse occurring in childhood. Finally, 22 24 participants did have prospectively documented abuse in childhood, but did not report having experienced any abuse during the AAI at age 26. This information is outlined in both Table 9 and Figure 7. For the following analyses, convergence between retrospective and prospective measures of abuse were interpreted according to the benchmarks outlined in the interrater reliability literature (i.e., McHugh, 2012) values between 0-.20 are considered to have no agreement; values between .21-.39 are considered to have minimal agreement; .40-.59 = weak agreement; .60-.79 = moderate agreement, .80-.90 = strong agreement; and above .90 = almost perfect agreement. For the stability analyses, correlations will be interpreted according to the benchmarks outlined by Cohen (1994): associations 0-.10 = weak, ~.24 = moderate, and .37 = strong. To what extent does retrospectively recalled abuse in young adulthood converge with prospectively documented abuse from birth to 17.5 years? To address Aim 1, a Cohen’s Kappa was run between the omnibus retrospective abuse variable and the prospective ever abused variable. Retrospectively recalled abuse and prospectively documented abuse demonstrated moderate agreement (k = .71). Results are outlined in Table 10. The results did not change materially when individuals who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental Table 1) or when those who were suspected of having been abused during the retrospective assessment were included in analyses (outlined in Supplemental Table 9). To address Aim 3, the above analyses were re-run using the continuous measures of retrospective and prospective abuse. Consistent with the dichotomous variables, retrospectively recalled abuse and with prospectively documented abuse demonstrated moderate agreement (ICC = .65). Results presented in Table 11. The results did not change materially when individuals 25 who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental Table 2). Note that the continuous analyses could not be done on the suspected abuse cases as they were coded as dichotomous in nature. To what extent does prospectively documented abuse converge with each measure of retrospectively recalled abuse assessed at 19 and 26 years of age, respectively? To address Aim 2, Cohen’s Kappa’s were run between the prospective ever abused and each individual retrospective abuse at ages 19 and 26. Consistent with the omnibus retrospective variable, the convergence between prospective abuse and retrospective abuse demonstrated moderate agreement at age 19 years (k = .65) and age 26 years (k = .68). Results outlined in Table 12. The results did not change materially when individuals who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental Table 3) or when those who were suspected of having been abused during the retrospective assessment were included in analyses (outlined in Supplemental Table 10). To address Aim 3, the above analyses were re-run using the continuous measures of retrospective and prospective abuse. Prospectively documented abuse and 19-year retrospective abuse demonstrated moderate convergence (ICC = .61). Prospective abuse and 26-year retrospective abuse demonstrated weak convergence (ICC = .56). Results outlined in Table 11. The results did not change materially when individuals who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental Table 2). To what extent is retrospectively recalled abuse stable from age 19 to age 26? To address Aim 2a, bivariate correlations were run between the retrospective abuse at age 19 and age 26. According to Cohen’s (1994) criteria retrospectively recalled abuse from age 19 to 26 was largely stable (r = .62, p < .05). Additionally, recalled physical abuse was largely 26 stable (r = .65, p < .05) from age 19 to 26 as was sexual abuse (r = .60, p < .05) from age 19 to 26. Results outlined in Table 13. The results did not change materially when individuals who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental Table 4) or when those who were suspected of having been abused during the retrospective assessment were included in analyses (outlined in Supplemental Table 11). To address Aim 3, the correlations between the 19- and 26-year assessments were run with the continuous variables—the pattern of results held. See Table 14 for more information. The results did not change materially when individuals who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental Table 5). Does the convergence of retrospective and prospective measures of abuse vary by specific parameterizations of prospectively documented abuse? Consistent with prior work in the MLSRA (e.g., Nivison et al., 2021b), prospectively documented abuse was broken down into various parameterizations: type of abuse, timing of abuse, and perpetrator of abuse. To address Aim 4a, Cohen’s Kappas were run between omnibus retrospective abuse and both prospective ever physically abused and ever sexually abused. Convergence was moderately weak for both prospective physical (k = .59) and sexual abuse (k = .59). To further examine the convergence of abuse type, retrospective physical abuse and retrospective sexual abuse were compared directly with the prospective physical and sexual abuse scales. Retrospective and prospective physical abuse demonstrated moderate agreement (k = .65). Retrospective and prospective sexual abuse also demonstrated moderate agreement (k = .78). Results outlined in Table 15. The results did not change materially when individuals who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental 27 Table 6) or when those who were suspected of having been abused during the retrospective assessment were included in analyses (outlined in Supplemental Table 12). To address Aim 4b, Cohen’s Kappas were run between omnibus retrospective abuse and whether abuse had occurred in each developmental period, documented prospectively. Convergence was moderate in middle childhood (k = .65), weak in early childhood (k = .48) and adolescence (k = .56), and minimal in infancy (k = .36). Results outlined in Table 16. The results did not change materially when individuals who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental Table 7) or when those who were suspected of having been abused during the retrospective assessment were included in analyses (outlined in Supplemental Table 13). To address Aim 4c, Cohen’s Kappas were run between omnibus retrospective abuse and whether abuse had been perpetrated by mother-figure, father-figure, or non-caregiver, documented prospectively. Convergence was moderate for abuse perpetrated by father-figures (k = .64), but weak for mother-figures (k = .48), and minimal for non-caregivers (k = .38). Results outlined in Table 17. The results did not change materially when individuals who had experienced only neglect documented prospectively were dropped from the analyses (outlined in Supplemental Table 8) or when those who were suspected of having been abused during the retrospective assessment were included in analyses (outlined in Supplemental Table 14). iv. DISCUSSION The present dissertation sought to reproduce previous findings that the convergence between prospective and retrospective reports of childhood abuse was quite low (e.g., k = ~ .19, Baldwin et al., 2019). In strong contrast, I found that in the MLSRA convergence between prospectively documented childhood abuse and retrospectively recalled abuse as measured in the 28 AAI was much higher than expected (k = .71), especially given the temporal lag between the retrospective and prospective assessments. These findings suggest that the AAI may provide a better context to evaluate retrospective abuse, as compared to other commonly used retrospective measures of abuse, which are often brief self-report measures (e.g., ACEs). Although prospectively documented experiences of childhood abuse provide more precise measurement of abusive experiences (e.g., timing, perpetrator, chronicity of abuse; Widom et al., 2004), these findings suggest that the AAI can be used in future research to assess retrospective childhood abuse more validly than previous measures (e.g., ACEs, Childhood Trauma Questionnaire) have demonstrated. Furthermore, results were consistent when examining abuse on a continuous scale (ICC = .65) and when broken down by retrospective assessment (19-year, k = .65; 26-year, k = .68). Convergence did vary slightly when broken down by the specific parameterization of abuse. Retrospective sexual abuse converged more highly with prospective sexual abuse (k = .78) than did retrospective and prospective physical abuse (k = .65). When broken down by prospective developmental period, retrospective abuse converged minimally with abuse occurring in infancy (k = .31), weakly in early childhood and adolescence (k = .48, and .55, respectively), and moderately in middle childhood (k = .65). Finally, when broken down by perpetrator, convergence was weak for abuse perpetrated by mother-figures and non-caregivers (k = .49, and .38, respectively), and moderate for father-figures (k = .64). Retrospectively recalled abuse was also moderately stable from age 19 to age 26 (r = .62). In the MLSRA, prospectively documented abuse also includes codes for physical neglect— in the present report, neglect was not included as it was not possible to code for retrospective neglect in the AAI. However, there were some cases (i.e., 15 participants) where a 29 participant had prospective evidence of physical neglect (but not physical or sexual abuse) and therefore, to be maximally conservative (i.e., rather than be included in the “non-abused group”) these participants were excluded in sensitivity analyses. The overall pattern of results held when these participants were excluded. Additionally, there were a few cases where a participant had refused to answer the abuse question during the AAI, or there was not enough information to decisively code whether the abuse had occurred or not. These cases were not included in the main analyses, but I conducted sensitivity analyses wherein these participants were included if I had suspected they had experienced abuse. Overall, the pattern of results remained largely the same. I had hypothesized that the convergence of retrospective and prospective reports of abuse would be around .20, consistent with meta-analytic findings (i.e., Baldwin, et al., 2019). The present findings were notably higher than expected, and indeed, much higher than most of the current literature investigating the convergence of retrospective and prospective reports of childhood abuse. Meta-analytic findings did reveal that convergence was higher in interview- based retrospective measures rather than self-report measures (for example, the Adverse Childhood Experiences Scale). It is possible that in our study our high convergence is due to the interview-based methodology— the current interview based coding system does not rely on the participant to self-identify as having been abused the way most retrospective measures do. Although the AAI does ask the participant if they had ever experienced abuse, this is not until much later in the interview after the participant has had the chance to spontaneously produce narratives surrounding their childhood relationship with their caregivers. Indeed, in our coding of the AAIs we had participants spontaneously produced explicit content providing evidence that they had been abused, but when asked directly if they had experienced abuse, they said no. If the 30 AAI coding system relied only on participants to self-identify, they would not have endorsed an item saying they had been abused. The interview-based methodology of the AAI provides a context for individuals to report on their potential abusive experiences, sometimes without even realizing they are doing so. After all, the AAI was explicitly constructed by Dr. Mary Main and colleagues (1985; George et al. 1996, p. 3) to “surprise the unconscious.” That is, Main believed that the protocol itself helped scaffold participants' ability to recall and discuss even difficult childhood experiences, which may explain why convergence in the present sample was much higher than expected. Furthermore, the prospective data in our study was collected from multiple sources such as Child Protective Services (CPS) reports, and reports from parents, grandparents, teachers, and participants (if additionally corroborated by another source)—most studies investigating convergence often rely on only one source of prospective data (usually parent report or CPS records). Perhaps it was the combination of the interview based retrospective measure, but also the very rigorous coding system applied to the prospective data that explains why convergence was much higher than expected in the present sample. Moreover, convergence in the present study may be higher than other studies due not only to the interview-based assessment, but to the very specific coding scale used. The 19-year MLSRA AAIs were previously coded for retrospective abuse (Shaffer et al., 2008) using a coding scale based on criteria outlined by Barnett and colleagues (1993). Retrospective abuse in that sample only converged with prospectively documented abuse weakly (k = .36). Although the criteria outlined by Barnett and colleagues (1993) is very rigorous, it was not developed specifically for use in the AAI, whereas the current coding system developed by Jacobvitz and colleagues (2006) was developed based on the existing AAI abuse scales from the traditional 31 AAI coding system. However, this is not fully conclusive as two reports cannot directly be compared despite being from the same sample, as the MLSRA team revised the prospective coding system in recent years to create a standard scale of abusive experiences across the entirety of childhood. Interestingly enough, convergence did not vary much between the 19 and 26-year assessment suggesting that a one-time assessment sometime in young adulthood may be sufficient in assessing retrospective abuse, using the current coding system, in the current sample. However, I would heavily caution this conclusion until this finding can be replicated in other, larger, more representative samples and across the lifespan. This finding is not especially surprising given that the stability of retrospective reporting between the 19 and 26-year assessments was moderately high (r = .62), consistent with evidence from the AAI and retrospective abuse literatures (e.g., Bakermans-Kranenburg & van IJzendoorn, 1993; Pinto et al., 2014, respectively). Convergence was similar when examining abuse on a dichotomous and a continuous basis. Although this aim was largely exploratory, I had hypothesized that it was possible that agreement would be somewhat higher on a continuous basis as it would allow for more variation in abuse experiences (i.e., not only evaluating abuse on the most extreme basis). It is possible that convergence did not vary much given that the continuous scales for the retrospective and prospective assessments were not exactly homogenous. The prospective scale was a sum of the types of abuse that occurred in each developmental period (a quasi-continuous scale, but more so just a sum of dichotomous variables), whereas the retrospective scale was more about the actual severity of the incidents reported (e.g., spanking with a hand versus with an object, whether it left marks or not). An analogous retrospective scale might involve rating each individual incident 32 reported during the AAI and then averaging the severity of all those incidents. The present coding system instead focuses on the highest degree of abuse rather than scoring individual instances (although they are all considered when scoring). Finally, I examined convergence by specific parameters of prospective abuse: abuse type, prospective abuse timing, and prospective perpetrator of abuse. Both retrospective physical and sexual abuse converged with prospective physical and sexual abuse moderate-to-strongly, though sexual abuse had slightly higher agreement (k = .78) than physical abuse (k = .65). This aim was largely exploratory but was inconsistent with meta-analytic findings which found weak convergence between retrospective and prospective abuse types, as well as minimal variation between abuse types (physical abuse, k = .16; sexual abuse, k = .17; Baldwin et al., 2019). It is possible that agreement between retrospective and prospective sexual abuse is higher than physical abuse because sexual abuse may be inherently more salient when recalling childhood experiences—or at the least, more definitive to code. For example, there may be ambiguity in retrospective reports of physical abuse on the severity— specifically on whether the physical abuse left a mark (if the participant did not disclose, or if the interviewer did not probe fully). Furthermore, societal interpretations of abuse may impact the reporting and recall of abuse—for example, physical abuse may be more difficult to define in a societal context—spanking, hitting, etc. is more normalized and less likely to be perceived as abuse, whereas sexual abuse is more apparent, and typically collectively agreed to be abuse at a societal level (Barnett et al., 1993). This finding is particularly interesting given that the base rate of sexual abuse (n = 37) was almost half that of physical abuse (n = 66) in our sample. Though this aim was largely exploratory, convergence did vary slightly by prospective abuse timing, consistent with my hypothesis. Convergence was weakest for abuse occurring in 33 infancy (k = .31), which is consistent with claims that individuals would not remember abuse occurring in childhood due to infantile amnesia (Hardt & Rutter, 2004). In the present study, an individual had reported that they had heard from one parent that the other parent had physically abused them in infancy but did not recall experiencing it themselves—in the current coding scheme this would be rated as a six which is not high enough to be qualified as having been abused, and therefore this participant was not coded as having retrospectively recalling abuse. Agreement was weak in early childhood and adolescence (k = .48, and .55, respectively), and moderately in middle childhood (k = .65). Although it is not particularly clear why the findings present the way they do, it is possible that agreement was highest for middle childhood because abuse was most prevalent during that developmental period (i.e., of those who had experienced abuse, 65% had experienced abuse in middle childhood, whereas only 46% experienced abuse in early childhood, 28% during adolescence, and 14% during infancy, not mutually exclusive). Prior evidence in the MLSRA has also found that middle childhood may be a uniquely important developmental period in our sample (Nivison et al., 2021b; Young et al., 2021). Lastly, I examined whether abuse varied by perpetrator. This was entirely exploratory in nature. Convergence was weak for abuse perpetrated by non-caregivers (k = .38) and mother- figures (k = .49), and moderate for father-figures (k = .64). It is possible that because the prospective assessments were conducted predominately with mothers rather than fathers the mothers were more willing to disclose abuse perpetrated by fathers than abuse they had perpetrated. Additionally, given the hidden nature of abuse (Barnett et al., 1993) it is possible that abuse perpetrated by non-caregivers was not disclosed by the children themselves and it is possible that parents were not aware that this abuse had occurred or it is simply because the AAI is an interview designed to assess the relationship an individual has with their caregivers and 34 non-caregivers are not often brought up, nor necessarily probed for in the context of abuse, unless spontaneously produced by the participant. Although, more research is needed to make any definitive conclusions about the convergence of specific parameterizations of abuse. Overall, the present report provides some evidence that abuse recalled in the AAI using the coding system developed by Jacobvitz and colleagues (2006) may provide a more valid assessment of retrospective experiences of physical and sexual abuse than other retrospective measures previously evaluated in the literature (see Baldwin et al., 2019 for more information). Further studies are needed to test whether this finding is reproducible in a variety of contexts (e.g., older adults, a more representative sample, etc.). Despite finding evidence that this measure may validly assess childhood abuse it is critical to note that the current measure only produces information on whether abuse occurred. It does not provide information on when the abuse occurred, how long the abuse occurred, who perpetrated the abuse, the cumulation of abusive events, or other types of abuse and neglect beyond physical and sexual abuse. These factors are critically important when trying to understand not only how to develop interventions for children who have experienced child abuse, but critical for prevention and policy efforts. The current coding system does not provide enough information to fully understand the etiology of child abuse. In its current form, the coding system allows us to conduct basic research to investigate the association between child abuse and specific factors such as symptoms of psychopathology, it does not, however, allow us to investigate how retrospectively recalled abuse occurring in middle childhood, specifically, is related to adult attachment quality, the way prospective evidence allows (see Nivison et al., 2021b). Widom, Raphael, and Dumont (2004) have long suggested that retrospective reports of child abuse do not allow for the complex study of child abuse, of the etiology or the eventual causality. As such, although this report provides promising 35 evidence of a more valid assessment of retrospectively recalled child abuse, it is not enough to solve the inherent issues related with retrospective reporting. Strengths, Limitations, and Future Directions There are a number of strengths associated with the present dissertation. There are the obvious strengths to this report including the prospective longitudinal design that has data on experiences of child abuse from birth to 17.5 years and retrospective assessments at two points in young adulthood. Additionally, the prospectively identified data in the MLSRA do not rely on only one source of information (e.g., only Child Protective Services reports), but includes information from parents, grandparents, teachers, CPS records, and participants (if corroborated with another source). A particular strength of this study is that the retrospective measures are interview based and do not rely on the participant to self-define whether they had experienced abuse—instead, coders made objective decisions using the present coding scheme to determine whether the shared experiences qualified as abusive. This is particularly important given that retrospective reporting is subject to a number of biases. Memory is a reconstructive process in that when individuals try to recall the past, new information is assimilated within the framework of existing knowledge—individuals may dismiss facts that do not fit their expectations (Loftus, 1982). Additionally, information learned after an event occurred can reshape how an individual recalls those events. For example, in the present study one individual disclosed during the AAI that they had been told by a family member that they had experienced abuse and had been shown the Child Protective Services report, but they denied the events and would not believe that it ever occurred. Of course, it is possible that the prospective evidence was not accurate—it is not goal of the present report to undermine an individual’s experiences or prove whether experiences 36 occurred, only to highlight the difficulties that surround retrospective reporting when individuals are asked to self-define. Of course, there are cases where there is not prospective evidence, but there are retrospective reports. The most obvious explanation is the hidden nature of abuse and the extreme difficulty of defining, let alone measuring abusive experiences in childhood (Barnett et al., 1993). Within our own sample, we had many participants disclose experiences of abuse and furthermore, explicitly say that they did not tell anyone about it while in childhood because they were too scared to disclose the information to anyone. There were even participants who had reported during the AAI that they had experienced abuse and told their caregivers, but their family members had dismissed them and told them it was not an issue. Although having an interview based retrospective measure does not solve the issue of under/over reporting, it does provide us, as researchers, insight into why, historically, the convergence between retrospective and prospective reports has been low. Despite the advantages of the AAI, given that it was not explicitly designed to be a measure of retrospective abuse, it naturally has its limitations. I was only able to examine physical and sexual abuse in the context of the AAI, but not physical neglect, or other types of abuse (though a frightening experiences scale does exist that is related to emotional abuse, see Jacobvitz et al., 2006). Additionally, I was constrained by the way in which the prospective data were operationalized. For example, witnessing domestic violence in the MLSRA was not included in the prospective abuse codes, and therefore we did not code for it in the context of the AAI. We also did not code for frightening experiences, despite the available scale for the AAI, because there is not a congruent variable in the prospective data. 37 Finally, the present coding system only focuses on whether abuse had been recalled— it does not provide the same detailed and rigorous information that prospective data can—such as timing, perpetrator, chronicity and more. It is reasonable to try and code for such detailed information, as Shaffer and colleagues (2008) originally set out to do. However, they ultimately concluded that there was not enough available information in the AAI to extract this information and they ultimately just focused on presence of abuse. Other retrospective measures do aim to extract this information (see the ICAST-R, Runyan, Dunne, & Zolotor, 2009). However, they do rely on participants to self-identify as having experienced abuse. Despite these limitations, the AAI may provide a somewhat valid context to assess experiences of childhood physical and sexual abuse though further studies are needed in a variety of populations to examine the validity of the measure. Perhaps the development of a new interview-based maltreatment interview would allow researchers to extract more detailed information using the advantages of the AAI, but without having the participants explicitly self-identify their experiences. Additional research on the retrospective abuse scale in the AAI is needed in samples with prospective abuse data as well to further examine whether the retrospective measure does indeed provide a somewhat valid assessment of experiences of child abuse. Further research in the MLSRA could examine whether self-report measures (such as ACEs or the CTQ) converge as highly with the prospective data as the AAI did. Furthermore, we could test whether the participants would be able to recall the specific details of abuse (e.g., timing, perpetrator etc.) via measures such as the ICAST-R. It is also possible that the age at which the retrospective measure was administered influences the validity of the measure. Although I did not find a substantial difference between the age 19 and age 26 AAIs in the present sample, these measures are confined only to young 38 adulthood. This could be empirically tested in the present sample if the AAIs were re- administered now that the participants are in their mid-40s. Other studies investigating the convergence of retrospective and prospective reports have examined potential factors that may bias retrospective reporting such as recall bias and depressogenic bias (e.g., Sheikh, 2018). Depressogenic bias has been previously experimentally manipulated in the context of the AAI itself. Roisman and colleagues (2014) found that secure individuals who were experimentally manipulated into a sad (versus happy) mood recalled more negative early childhood events during an interview about their early childhood experiences. It is possible, given this evidence, that depressogenic biases may also influence the retrospective recall of childhood abuse. Future work could also experimentally manipulate individuals in either a sad or happy mood (consistent with Roisman et al., 2014) and ask them to complete retrospective measures of early experiences, including the AAI or other self-report measures of abuse. This would be particularly advantageous in populations with existing prospective data. This experiment may provide insight into the extent to which current mood can bias retrospective reporting. Beyond experimentation, future work could investigate the association between retrospective recall of abuse and symptoms of depression. Given the wealth of data in the MLSRA, we could also control for childhood symptoms of depression given that childhood abuse has been shown to be associated with depressive symptoms in adulthood (VanMeter et al., 2021). Previous research has also suggested that it is important to consider both the informant as well as the time in which measures are collected when examining reports of childhood abuse. Specifically, whether reports are not only collected retrospectively or prospectively, but also if the source of the information is objective (e.g., CPS records) or subjective (e.g., parent-report) as 39 this may have implications for our understanding of the sequalae of early life experiences (Danese, 2020).Within the MLSRA, because our prospective data is a cumulation of multiple sources—if we were to disaggregate these data— would we find varying results by the type of prospective report (i.e., official records versus subjective report)? There has been a wealth of studies published using the prospective abuse data in the MLSRA (e.g., Nivison et al., 2021b; Raby et al., 2017; Labella et al., 2018, VanMeter et al., 2021). What is unclear is if comparable results might be found if the retrospective measure was used in place for the prospective data, In other samples, retrospective measures of adversity were more associated with subjective measures of life-outcomes (e.g., self-reported physical health) than objective measure of life-outcomes (e.g., objective measures of physical health) whereas prospective measures of childhood adversity were more associated with objective measures of health-outcomes than self-reported measures (Reuben et al., 2016). In that sample, the convergence of retrospective and prospective measures was also much lower (k = .31), whereas in the present sample convergence was much higher (k = .71) and therefore it is unclear whether the retrospective and prospective measures in the MLSRA would demonstrate differential prediction to various outcome variables. Conclusion In the present dissertation, convergence between retrospective abuse as measured in the AAI and prospectively documented abuse was much higher than hypothesized, and indeed much higher than meta-analytic evidence. The AAI may provide a more valid context in which to assess childhood experiences of physical and sexual abuse. However, more studies are needed to test the validity of this scale. Additionally, although convergence was higher in the present sample, it only provides information on whether abuse occurred, but does not provide detailed 40 enough information that is needed to inform intervention, prevention, or policy work. Although this report provides promising evidence of a more valid assessment of retrospectively recalled child physical and sexual abuse, it is not enough to solve the inherent issues related with retrospective reporting—prospective longitudinal evidence is still critically important when investigating the etiology and consequences of childhood experiences of abuse. 41 v. REFERENCES Bakermans-Kranenburg, M. J. & van IJzendoorn, M. H. (1993). A psychometric study of the Adult Attachment Interview reliability and discriminant validity. Developmental Psychology, 29, 870-879. Baldwin, J. R., Caspi, A., Meehan, A. J., Ambler, A., Arseneault, L., Fisher, H. L., Harrington, H., Matthews, T., Odgers, C. L., Poulton, R., Ramrakha, S., Moffitt, T. E., & Danese, A. (2021). Population vs Individual Prediction of Poor Health From Results of Adverse Childhood Experiences Screening. JAMA Pediatrics, 175(4), 385. https://doi.org/10.1001/jamapediatrics.2020.5602 Baldwin, J. R., & Esposti, M. D. (2021). Triangulating evidence on the role of perceived versus objective experiences of childhood adversity in psychopathology. JCPP Advances. https://doi.org/10.1111/jcv2.12010 Baldwin, J. R., Reuben, A., Newbury, J. B., Danese, A. (2019). Agreement between prospective and retrospective measures of childhood maltreatment: A systematic review and meta- analysis. JAMA Psychiatry, 76, 584–593. Barnett, D., Manly, J. T., & Cicchetti, D. (1993). Defining child maltreatment: The interface between policy and research. In D. Cicchetti & S. L. Toth (Eds.), Child Abuse, Child Development, and Social Policy (pp. 7-74). Norwood, NJ: Ablex. Bellis, M. A., Lowey, H., Leckenby, N., Hughes, K., Harrison, D. (2014). Adverse childhood experiences: retrospective study to determine their impact on adult health behaviours and health outcomes in a UK population, Journal of Public Health, 36, 81–91. https://doi.org/10.1093/pubmed/fdt038 42 Berg, M. T., Lei, M.-K., Beach, S. R., Simons, R. L., & Simons, L. G. (2020). Childhood Adversities as Determinants of Cardiovascular Disease Risk and Perceived Illness Burden in Adulthood: Comparing Retrospective and Prospective Self-Report Measures in a Longitudinal Sample of African Americans. Journal of Youth and Adolescence, 49(6), 1292–1308. https://doi.org/10.1007/s10964-020-01207-z Bouman, T., van der Meulen, M., Goossens, F. A., Olthof, T., Vermande, M. M., & Aleva, E. A. (2012). Peer and self‐reports of victimization and bullying: Their differential association with internalizing problems and social adjustment. Journal of School Psychology, 50(6), 759–774. Brewin, C. R., Andrews, B., & Gotlib, I. H. (1993). Psychopathology and early experience: A reappraisal of retrospective reports. Psychological Bulletin, 113, 82-98. Burt, D. B., Zembar, M. J., & Niederehe, G. (1995). Depression and memory impairment: A meta-analysis of the association, its pattern, and specificity. Psychological Bulletin, 117, 285-305. doi:10.1037/0033-2909.117.2.285 Carpenter, L. L., Tyrka, A. R., Ross, N. S., Khoury, L., Anderson, G.M., & Price, L. H. (2000). Effect of emotional abuse and age on cortisol responsivity in adulthood. Biological Psychiatry, 66, 69-75. doi: 10.1016/j.biopsych.2009.02.030. Cicchetti, D. (2013). Resilient functioning in maltreated children: Past, present, and future perspectives. Journal of Child Psychology and Psychiatry, 54, 402-422. Cohen, J. (1994). The earth is round (p<.05). American Psychologist, 49, 997- 1003. Cohen, P., Brown, J., & Smaile, E. (2001). Child abuse and neglect and the development of mental disorders in the general population. Development and Psychopathology, 13, 981– 999. 43 Danese, A. (2020). Annual Research Review: Rethinking childhood trauma—new research directions for measurement, study design and analytical strategies. Journal of Child Psychology and Psychiatry 61, 236–250. Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., Koss, M. P., & Marks, J. S. (1998) Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. The Adverse Childhood Experiences (ACE) Study. American Journal of Preventive Medicine; 14, 245 – 58. George, C., Kaplan, N. & Main, M. (1985). Adult Attachment Interview. (Unpublished manuscript, University of California, Department of Psychology, Berkeley). George, C., Kaplan, N., & Main, M. (1996). Adult Attachment Interview protocol (3rd ed. Unpublished manuscript, University of California at Berkeley. Hardt, J., & Rutter, M. (2004). Validity of adult retrospective reports of adverse childhood experiences: review of the evidence. Journal of Child Psychology and Psychiatry, 45, 260-273. Henry, B., Moffitt, T.E., Caspi, A., Langley, J., & Silva, P.A. (1994). On the "remembrance of things past": A longitudinal evaluation of the retrospective method. Psychological Assessment, 6, 92-101. Herzog, J. I., Thome, J., Demirakca, T., Koppe, G., Ende, G., Lis, S., Rausch, S., Priebe, K., Müller-Engelmann, M., Steil, R., Bohus, M., & Schmahl, C. (2020). Influence of severity of type and timing of retrospectively reported childhood maltreatment on female amygdala and hippocampal volume. Scientific Reports, 10, 1903. Jacobvitz, D., Leon, K., & Hazen, N. (2006). Does expectant mothers’ unresolved trauma predict frightened/frightening maternal behavior? Risk and protective factors. Development and 44 Psychopathology, 18, 363-379. Jaffee, S. R. & Christian, C. W. (2014). The biological embedding of child abuse and neglect: Implications for policy and practice. Social Policy Report, 28. Köber, C., Facompré, C. R., Waters, T. E. A., and Simpson, J. A. (2019). Autobiographical memory stability in the context of the Adult Attachment Interview. Cognition, 191, 103980. Labella, M. H., Johnson, W. F., Martin, J., Ruiz, S. K., Shankman, J. L., Egeland, B., Englund, M. M., Collins, W. A., & Roisman, G. I., & Simpson, J. A. (2018). Multiple dimensions of childhood abuse and neglect prospectively predict poorer romantic functioning in adulthood. Personality and Social Psychology Bulletin, 44, 238-251. doi: 10.1177/0146167217736049 Leeb, R. T., Paulozzi, L., Melanson, C., Simon, T., & Arias, I. (2008). Child maltreatment surveillance: Uniform definitions for public health and recommended data elements (Ver. 1.0). Atlanta, GA: Centers for Disease Control and Prevention, National Center for Injury Prevention and Control. Leon, K., Jacobvitz, D., & Hazen, N. (2004). Maternal resolution of loss and abuse: Association with adjustment to the transition to parenthood. Infant Mental Health, 25, 130–148. Loftus, E. F. (1982). Memory and its distortions. In A. G. Kraut (Ed.), The G. Stanley Hall lecture series. The G. Stanley Hall lecture series, Vol. 2 (pp. 123–154). American Psychological Association. Main, M., & Goldwyn, R. (1998). Adult attachment scoring and classification systems, Version 6.2. Unpublished manuscript, University of California, Berkeley. Main, M., Goldwyn, R., & Hesse, E. (2003-2008). Adult attachment scoring and classification 45 system. Unpublished manuscript, University of California at Berkeley. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. Monographs of the Society for Research in Child Development, 50(1-2), 66-104. McHugh (2012). Interrater reliability: the kappa statistic. Biochemia Medica, 3, 276-282. Naicker, S. N., Norris, S. A., Mabaso, M., & Richter, L. M. (2017). An analysis of retrospective and repeat prospective reports of adverse childhood experiences from the South African Birth to Twenty Plus cohort. PLOS ONE, 12(7), e0181522. https://doi.org/10.1371/journal.pone.0181522 Narayan, A. J., Englund, M. M., & Egeland, B. (2013). Developmental timing and continuity of exposure to interparental violence and externalizing behavior as prospective predictors of dating violence. Development and Psychopathology, 25, 973-990. National Scientific Council on the Developing Child. (2012). The science of neglect: The persistent absence of responsive care disrupts the developing brain: Working Paper 12. Retrieved from http://www.developing- child.harvard.edu Neel, R., Kenrick, D. T., White, A. E., & Neuberg, S. L. (2016). Individual differences in fundamental social motives. Journal of Personality and Social Psychology, 110, 887– 907. Newbury, J. B., Arseneault, L., Moffitt, T. E., Caspi, A., Danese, A., Baldwin, J. R., & Fisher, H. L. (2018). Measuring childhood maltreatment to predict early-adult psychopathology: Comparison of prospective informant-reports and retrospective self-reports. Journal of Psychiatric Research, 96, 57-64. 46 Nivison, M. D., Vandell, D. L., Booth-LaForce, C., & Roisman, G. I. (2021a). Convergent and Discriminant Validity of Retrospective Assessments of the Quality of Childhood Parenting: Prospective Evidence from Infancy to Age 26 Years. Psychological Science, 32, 721-734. DOI: 10.1177/0956797620975775 Nivison, M. D., Facompré, C. R., Raby, K. L., Simpson, J. A., Roisman, G. I., & Waters, T. E. A. (2021b). Childhood abuse and neglect are prospectively associated with scripted attachment representations in young adulthood. Development and Psychopathology, 33(4), 1143-1155. doi:10.1017/S0954579420000528 Pinto, R., Correia, L., & Maia, A. (2014). Assessing the reliability of retrospective reports of adverse childhood experiences among adolescents with documented childhood maltreatment. Journal of Family Violence, 29, 431-438. DOI 10.1007/s10896-014-9602- 9 Pyles, M. K., Stolz, H. R., & MacFarlane, J. W. (1935). The accuracy of mothers’ reports on birth and developmental data. Child Development, 6, 165-176. Raby, K. L., Labella, M. H., Martin, J., Carlson, E. A., & Roisman, G. I. (2017). Childhood abuse and neglect and insecure attachment states of mind in adulthood: Prospective, longitudinal evidence from a high-risk sample. Development and Psychopathology, 29, 347–363. Raphael, K. (1987). Recall bias: a proposal for assessment and control. International Journal of Epidemiology, 16, 167–170. Reuben, A., Moffitt, T.E., Caspi, A., Belsky, D.W., Harrington, H., Schroeder, F., Hogan, S., Ramrakha, S., Poulton, R., Danese, A. (2016). Lest we forget: comparing retrospective 47 and prospective assessments of adverse childhood experiences in the prediction of adult health. Journal of Child Psychology and. Psychiatry, 57, 1103-1112. Roisman, G. I., Haltigan, J. D., Haydon, K. C., & Booth-LaForce, C. (2014). Earned-security in retrospect: Depressive symptoms, family stress, and maternal and paternal sensitivity from early childhood to mid-adolescence. Monographs of the Society for Research in Child Development, 79, 85-107. Runyan, D. K., Dunne, M. P., Zolotor, A. J., Madrid, B., Jain, D., Gerbaka, B., Menick, D. M., Andreva-Miller, I., Kasim, M. S., Choo, W. Y., Isaeva, O., Macfarlane, B., Ramirez, C., Volkova, E., & Youssef, R. M. (2009). The development of the ISPCAN Child Abuse Screening Tools: The ICAST-P (Parent Version). Child Abuse & Neglect, 33, 826–832. Shaffer, A., Huston, L., & Egeland, B. (2008). Identification of child maltreatment using prospective and self-report methodologies: A comparison of maltreatment incidence and relation to later psychopathology. Child Abuse and Neglect, 32, 682–692. Sheikh, M. A. (2018). Coloring of the past via respondent's current psychological state, mediation, and the association between childhood disadvantage and morbidity in adulthood. Journal of Psychiatric Research, 103, 173-181. Stoltenborgh, M., van IJzendoorn, M. H., Euser, E. M., & Bakermans-Kranenburg, M. J. (2011). A global perspective on child sexual abuse: Meta-analysis of prevalence around the world. Child Maltreatment, 16, 79-101. Sroufe, L. A., Egeland, B., Carlson, E., & Collins, W. A. (2005). Placing early attachment experiences in developmental context. In K. E. Grossmann, K. Grossmann, & E. Waters (Eds.), The power of longitudinal attachment research: From infancy and childhood to adulthood (pp. 48–70). New York, NY: Guilford. 48 Tajima, E. A., Herrenkohl, T. I., Huang, B., & Whitney, S. D. (2004). Measuring childhood maltreatment: A comparison of prospective parent reports and retrospective adolescent reports. American Journal of Orthopsychiatry, 74, 424-435. Vallati, M., Cunningham, S., Mazurka, R., Stewart, J. G., Larocque, C., Milev, R. V., Bagby, R. M., Kennedy, S. H., & Harkness, K. L. (2020). Childhood maltreatment and the clinical characteristics of major depressive disorder in adolescence and adulthood. Journal of Abnormal Psychology. Advance online publication. VanMeter, F., Nivison, M. D., Englund, M. M., Carlson, E. A., & Roisman, G. I. (2021). Childhood abuse and neglect and self-reported symptoms of psychopathology through midlife. Developmental Psychology, 57(5), 824–836. https://doi.org/10.1037/dev0001169 van Harmelen, A., van Tol, M., van der Wee, N. J., Veltman, D. J., Aleman, A., Spinhoven, P., van Buchem, M. A., Zitman, F. G., Penninx, B. W. J. H., Elzinga, B. M., (2010). Reduced Medial Prefrontal Cortex Volume in Adults Reporting Childhood Emotional Maltreatment. Biological Psychiatry, 68, 832-838. doi: 10.1016/j.biopsych.2010.06.011. Widom, C. S., Raphael, K. G., & DuMont, K. A. (2004). The case for prospective longitudinal studies in child maltreatment research: Commentary on Dube, Williamson, Thompson, Felitti, and Anda (2004). Child abuse & neglect, 28(7), 715-722.h Yarrow, M. R., Campbell, J. D., & Burton, R. V. (1970). Recollections of childhood: A study of the retrospective method. Monographs of the Society for Research in Child Development, 35, 1-83. Young, E. A., Doom, J. R., Farrell, A. K., Carlson, E. A., Englund, M. M., Miller, G. E., Gunnar, M. R., Roisman, G. I., Simpson, J. A. (2021). Life stress and cortisol reactivity: An 49 exploratory analysis of the effects of stress exposure across life on HPA-axis functioning. Development and Psychopathology, 33(1), 301-312. doi:10.1017/S0954579419001779 50 vi. LIST OF TABLES 51 Table 1. Prospective analytical variable guide Based on dichotomous data Theoretical values Label Ever abused (prospective) 0 or 1 Ever physically or sexually abused from 0-17.5 years, documented prospectively Ever physically abused (prospective) 0 or 1 Ever physically abused from 0-17.5 years, documented prospectively Ever sexually abused (prospective) 0 or 1 Ever sexually abused from 0-17.5 years, documented prospectively Ever abused by mother figure (prospective) 0 or 1 Ever abused by mother figure from 0- 17.5 years, documented prospectively Ever abused by father figure (prospective) 0 or 1 Ever abused by father figure from 0- 17.5 years, documented prospectively Ever abused by non-caregiver (prospective) 0 or 1 Ever abused by non-caregiver from 0- 17.5 years, documented prospectively Ever abuse in infancy (prospective) 0 or 1 Ever abused in infancy, documented prospectively Ever abused in early childhood (prospective) 0 or 1 Ever abused in early childhood, documented prospectively Ever abused in middle childhood (prospective) 0 or 1 Ever abused in middle childhood, documented prospectively Ever abused in adolescence (prospective) 0 or 1 Ever abused in adolescence, documented prospectively Based on quasi-continuous data Theoretical values Label Severity of abuse (prospective) 0-8 Sum of total experiences of physical and sexual abuse across all developmental periods 0-17.5 years, documented prospectively 52 Table 2. Retrospective analytic variable guide Based on dichotomous data Theoretical values Label Omnibus retrospective abuse 0, .5, 1 Any recalled abuse at either age 19 or 26 assessments Recalled abuse age 19 0 or 1 Any recalled abuse at the age 19 assessment Recalled abuse age 26 0 or 1 Any recalled abuse at the age 26 assessment Recalled physical abuse age 19 0 or 1 Any recalled physical abuse at the age 19 assessment Recalled sexual abuse age 19 0 or 1 Any recalled sexual abuse at the age 19 assessment Recalled physical abuse age 26 0 or 1 Any recalled physical abuse at the age 26 assessment Recalled sexual abuse age 26 0 or 1 Any recalled sexual abuse at the age 26 assessment Based on continuous data Theoretical values Label Continuous omnibus retrospective abuse 0-8 Any recalled abuse at either age 19 or 26 assessments Continuous recalled abuse age 19 0-8 Any recalled abuse at the age 19 assessment Continuous recalled abuse age 26 0-8 Any recalled abuse at the age 26 assessment Continuous age 19 physical abuse 0-8 Any recalled physical abuse at the age 19 assessment Continuous age 19 sexual abuse 0-8 Any recalled sexual abuse at the age 19 assessment Continuous age 26 physical abuse 0-8 Any recalled physical abuse at the age 26 assessment Continuous age 26 sexual abuse 0-8 Any recalled sexual abuse at the age 26 assessment 53 Table 3. Descriptive statistics for prospective analytic variables Based on dichotomous data N Min Max M SD Ever abused (prospective) 162 0 1 0.42 0.50 Ever physically abused (prospective) 162 0 1 0.31 0.46 Ever sexually abused (prospective) 157 0 1 0.20 0.40 Ever abused by mother figure (prospective) 161 0 1 0.22 0.41 Ever abused by father figure (prospective) 160 0 1 0.21 0.41 Ever abused by non-caregiver (prospective) 156 0 1 0.14 0.35 Ever abuse in infancy (prospective) 158 0 1 0.06 0.24 Ever abused in early childhood (prospective) 160 0 1 0.19 0.40 Ever abused in middle childhood (prospective) 162 0 1 0.27 0.45 Ever abused in adolescence (prospective) 161 0 1 0.12 0.32 Based on quasi-continuous data N Min Max M SD Severity of abuse (prospective) 157 0 5 0.69 1.07 Note. Min = minimum observed scale value, Max = maximum observed scale value 54 Table 4. Descriptive statistics for retrospective analytic variables Based on dichotomous data N Min Max M SD Omnibus retrospective abuse 162 0 1 0.31 0.42 Recalled abuse age 19 153 0 1 0.27 0.45 Recalled abuse age 26 146 0 1 0.36 0.48 Recalled physical abuse age 19 162 0 1 0.29 0.46 Recalled sexual abuse age 19 162 0 1 0.16 0.37 Recalled physical abuse age 26 152 0 1 0.20 0.40 Recalled sexual abuse age 26 152 0 1 0.10 0.30 Based on continuous data N Min Max M SD Continuous omnibus retrospective abuse 162 0 8 2.25 2.12 Continuous recalled abuse age 19 153 0 8 2.05 2.21 Continuous recalled abuse age 26 146 0 8 2.49 2.38 Continuous age 19 physical abuse 152 0 8 3.08 3.07 Continuous age 19 sexual abuse 152 0 8 0.95 2.48 Continuous age 26 physical abuse 146 0 8 3.51 3.28 Continuous age 26 sexual abuse 146 0 8 1.48 2.89 Note. Min = minimum observed scale value, Max = maximum observed scale value 55 Table 5. Frequency of the overall reports of abuse Abuse documented only prospectively Abuse only reported at the retrospective assessment Abuse both documented prospectively and reported retrospectively No reported abuse at either prospective or retrospective assessment 20 14 48 80 Note. n = 162 56 Table 6. Frequency of physical abuse only Physical abuse documented only prospectively Physical abuse only reported at the retrospective assessment Physical abuse both documented prospectively and reported retrospectively No reported physical abuse at either prospective or retrospective assessment 19 16 31 96 Note. n = 162 57 Table 7. Frequency of sexual abuse only Sexual abuse documented only prospectively Sexual abuse only reported at the retrospective assessment Sexual abuse both documented prospectively and reported retrospectively No reported sexual abuse at either prospective or retrospective assessment 11 6 20 120 Note. n = 157 (1 retrospective report was refused, 5 prospective reports did not have enough information to code for SA) 58 Table 8. Frequency of only the 19-year retrospective assessment (including both physical and sexual abuse) Abuse documented only prospectively Abuse only reported at the 19-year retrospective assessment Abuse both documented prospectively and reported retrospectively at the 19-year assessment No reported abuse at either prospective or 19-year retrospective assessment 30 8 34 81 Note. N = 153 59 Table 9. Frequency of only the 26-year retrospective assessment (including both physical and sexual abuse) Abuse documented only prospectively Abuse only reported at the 26- year retrospective assessment Abuse both documented prospectively and reported retrospectively at the 26-year assessment No reported abuse at either prospective or 26-year retrospective assessment 22 12 40 72 Note. N = 146 60 Table 10. Comparison of overall dichotomous retrospective and prospective ever abused Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective ever abused .71 .56* 162 Note. *p < .05 61 Table 11. Comparison of overall continuous retrospective and prospective ever abused Comparison groups Intraclass Correlation Pearson’s Correlation N 1. Continuous retrospective abuse & continuous prospective abuse .65 .60* 157 2. Continuous age 19 retrospective abuse & continuous prospective abuse .61 .57* 148 3. Continuous age 26 retrospective abuse & continuous prospective abuse .56 .53* 141 Note. *p < .05 62 Table 12. Comparison of overall dichotomous retrospective by assessment and prospective ever abused Comparison groups Cohen’s Kappa Pearson’s Correlation N Age 19 retrospective abuse & prospective ever abused .65 .49* 153 Age 26 retrospective abuse & prospective ever abused .68 .52* 146 Note. *p < .05 63 Table 13. Bivariate associations outlining the stability between age 19 and 26 retrospective abuse (dichotomous variables) 1 2 3 4 5 6 1. Age 19 overall abuse – 2. Age 26 overall abuse .62* – 3. Age 19 physical abuse .82* .57* – 4. Age 26 physical abuse .53* .85* .64* – 5. Age 19 sexual abuse .54* .29* .02 .06 – 6. Age 26 sexual abuse .38* .54* .10 .19* .60* – N 137 146 136 146 136 146 Note. *p < .05 64 Table 14. Bivariate associations outlining the stability between age 19 and 26 retrospective abuse (continuous variables) 1 2 3 4 5 6 1. Age 19 overall abuse (cont) – 2. Age 26 overall abuse (cont) .67* – 3. Age 19 physical abuse (cont) .83* .60* – 4. Age 26 physical abuse (cont) .53* .80* .68* – 5. Age 19 sexual abuse (cont) .72* .39* .18* .06 – 6. Age 26 sexual abuse (cont) .53* .74* .22* .19* .61* – N 137 146 136 146 136 146 Note. Cont = continuous variable on a 0-8 scale *p < .05 65 Table 15. Comparison of retrospective abuse and type of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective physical abuse .59 .42* 162 Omnibus retrospective abuse & prospective sexual abuse .59 .42* 157 Retrospective physical abuse & prospective physical abuse .65 .49* 162 Retrospective sexual abuse & prospective sexual abuse .78 .64* 157 Note. *p < .05 66 Table 16. Comparison of retrospective abuse and developmental period of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective abuse in infancy .31 .21* 158 Omnibus retrospective abuse & prospective abuse in early childhood .48 .32* 160 Omnibus retrospective abuse & prospective abuse in middle childhood .65 .49* 162 Omnibus retrospective abuse & prospective abuse in adolescence .55 .40* 161 Note. *p < .05 67 Table 17. Comparison of retrospective abuse and perpetrator of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective abuse by mother-figure .49 .32* 161 Omnibus retrospective abuse & prospective abuse by father-figure .64 .47* 160 Omnibus retrospective abuse & prospective abuse by non-caregiver .38 .24* 156 Note. *p < .05 68 vii. LIST OF FIGURES 69 Figure 1 94 68 0 10 20 30 40 50 60 70 80 90 100 Fr eq u en cy Frequency of overall abuse prospecitvely documented from birth to 17.5 years 0 1 70 Figure 2 100 23 39 0 20 40 60 80 100 120 Fr e q u e n cy Frequency of overall abuse retrospecitvely recalled retrospective at ages 19 and 26 0 0.5 1 71 Figure 3 72 Figure 4 73 Figure 5 74 Figure 6 75 Figure 7 76 viii. SUPPLEMENTARY MATERIALS Supplemental Figure 1 Note. This variable is the sum of the number of types of abuse (i.e., physical and sexual abuse) in each developmental period (i.e., infancy, early childhood, middle childhood, and adolescence). 0 reflects no instances of abuse in any developmental period. 94 36 14 8 4 1 0 10 20 30 40 50 60 70 80 90 100 0 1 2 3 4 5 Fr eq u en cy Distribution of overall abuse prospectively documented from birth to 17.5 years (quasi-continuous scale) 77 Supplemental Figure 2 Note. The above variable is the average of the 19 and 26-year physical and sexual abuse scales that were coded on a 0-8 basis. 0 indicates no recalled experiences of any type of abuse during either assessment. 36 8 4 5 15 3 11 1 11 7 1 8 3 3 3 20 1 1 3 4 1 3 2 1 1 2 1 3 0 5 10 15 20 25 30 35 40 0 0. 2 5 0. 5 0. 7 5 1 1. 2 5 1. 5 1. 7 5 2 2. 5 2. 7 5 3 3. 2 5 3. 5 3. 7 5 4 4. 2 5 4. 7 5 5 5. 5 5. 7 5 6 6. 2 5 6. 5 6. 7 5 7 7. 7 5 8 Fr eq u en cy average of the 19 and 26 year continouous on 0-8 scale Distribution of overall abuse retrospecitvely recalled at ages 19 and 26 (continuous scale) 78 Neglect dropped analyses Supplemental Table 1. Comparison of overall dichotomous retrospective and prospective ever abused Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective ever abused .71 .56* 147 Note. *p < .05 79 Supplemental Table 2. Comparison of overall continuous retrospective and prospective ever abused Comparison groups Intraclass Correlation Pearson’s Correlation N 4. Continuous retrospective abuse & continuous prospective abuse .67 .62* 142 5. Continuous age 19 retrospective abuse & continuous prospective abuse .63 .59* 135 6. Continuous age 26 retrospective abuse & continuous prospective abuse .58 .54* 127 Note. *p < .05 80 Supplemental Table 3. Comparison of overall dichotomous retrospective by assessment and prospective ever abused Comparison groups Cohen’s Kappa Pearson’s Correlation N Age 19 retrospective abuse & prospective ever abused .65 .48* 140 Age 26 retrospective abuse & prospective ever abused .68 .52* 132 Note. *p < .05 81 Supplemental Table 4. Bivariate associations outlining the stability between age 19 and 26 retrospective abuse (dichotomous variables) 1 2 3 4 5 6 1. Age 19 overall abuse – 2. Age 26 overall abuse .60* – 3. Age 19 physical abuse .81* .54* – 4. Age 26 physical abuse .50* .86* .62* – 5. Age 19 sexual abuse .53* .26* -.04 .01 – 6. Age 26 sexual abuse .35* .51* .05 .16 .57* – N 125 132 124 132 124 132 Note. *p < .05 82 Supplemental Table 5. Bivariate associations outlining the stability between age 19 and 26 retrospective abuse (continuous variables) 1 2 3 4 5 6 1. Age 19 overall abuse (cont) – 2. Age 26 overall abuse (cont) .65* – 3. Age 19 physical abuse (cont) .82* .59* – 4. Age 26 physical abuse (cont) .52* .80* .68* – 5. Age 19 sexual abuse (cont) .71* .35* .16 .02 – 6. Age 26 sexual abuse (cont) .50* .73* .19* .16 .59* – N 125 132 124 132 124 132 Note. Cont = continuous variable on a 0-8 scale *p < .05 83 Supplemental Table 6. Comparison of retrospective abuse and type of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective physical abuse .59 .42* 147 Omnibus retrospective abuse & prospective sexual abuse .60 .42* 142 Retrospective physical abuse & prospective physical abuse .69 .52* 124 Retrospective sexual abuse & prospective sexual abuse .77 .63* 119 Note. *p < .05 84 Supplemental Table 7. Comparison of retrospective abuse and developmental period of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective abuse in infancy .31 .21* 143 Omnibus retrospective abuse & prospective abuse in early childhood .48 .31* 145 Omnibus retrospective abuse & prospective abuse in middle childhood .65 .49* 147 Omnibus retrospective abuse & prospective abuse in adolescence .56 .40* 146 Note. *p < .05 85 Supplemental Table 8. Comparison of retrospective abuse and perpetrator of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective abuse by mother-figure .48 .32* 146 Omnibus retrospective abuse & prospective abuse by father-figure .64 .48* 145 Omnibus retrospective abuse & prospective abuse by non-caregiver .38 .23* 141 Note. *p < .05 86 Suspect abuse analyses Supplemental Table 9. Comparison of overall dichotomous retrospective and prospective ever abused Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective ever abused .74 .59* 163 Note. *p < .05 87 Supplemental Table 10. Comparison of overall dichotomous retrospective by assessment and prospective ever abused Comparison groups Cohen’s Kappa Pearson’s Correlation N Age 19 retrospective abuse & prospective ever abused .66 .49* 157 Age 26 retrospective abuse & prospective ever abused .70 .53* 149 Note. *p < .05 88 Supplemental Table 11. Pearson Correlation between age 19 and age 26 overall abuse Age 26 overall abuse 1. Age 19 overall abuse .62* Note. N = 143 *p < .05 89 Supplemental Table 12. Comparison of retrospective abuse and type of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective physical abuse .61 .44* 163 Omnibus retrospective abuse & prospective sexual abuse .65 .49* 158 Note. *p < .05 90 Supplemental Table 13. Comparison of retrospective abuse and developmental period of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective abuse in infancy .36 .26* 159 Omnibus retrospective abuse & prospective abuse in early childhood .48 .31* 161 Omnibus retrospective abuse & prospective abuse in middle childhood .70 .53* 163 Omnibus retrospective abuse & prospective abuse in adolescence .53 .37* 162 Note. *p < .05 91 Supplemental Table 14. Comparison of retrospective abuse and perpetrator of prospective abuse Comparison groups Cohen’s Kappa Pearson’s Correlation N Omnibus retrospective abuse & prospective abuse by mother-figure .46 .30* 162 Omnibus retrospective abuse & prospective abuse by father-figure .66 .50* 161 Omnibus retrospective abuse & prospective abuse by non-caregiver .46 .30* 157 Note. *p < .05 92 Supplemental Table 15. Descriptive table of refusals and suspected abuse at both the 19- and 26-year assessments Refusal at age 19 Suspected abuse at age 19 Refusal age 26 Suspected abuse at age 26 3 3 3 3 Note. The three individuals who had refused at age 19 were not the same individuals who refused at age 26—so in total, across both assessments, six individuals had refused to disclose information. 93 Supplemental Table 16. Descriptive table of cases with not enough information and suspected abuse at both the 19- and 26-year assessments Not enough information at age 19 Suspected abuse at age 19 Not enough information age 26 Suspected abuse at age 26 Both scales: 1 Physical abuse scale only: 1 Sexual abuse scale only: 1 3 0 0 Note. Although for two individuals there was not enough information to code both scales, these individuals did score high enough on the scale they did have to be qualified as having overall experienced abuse and therefore these two were included in the overall main analyses. 94 ix. APPENDIX: SPSS SYNTAX Re-computing the prospective abuse data to exclude neglect cases ## re-computing propsective data to exclude the neglect cases ### #First creating a summary of Physical abuse (PA) and Sexual abuse (SA) which will be used to recompute the overall ombnibus variable DATASET ACTIVATE DataSet1. COMPUTE Sum_abuse=Pabuse_ever+Sabuse_ever. EXECUTE. ## Now recoding the sum_abuse variable to re-create our dichotomous ever variable. If sum abuse = 0, then 0, if =1 then 1, if =2, then 1 RECODE Sum_abuse (0=0) (1=1) (2=1) INTO Abuse_Ever. VARIABLE LABELS Abuse_Ever 'PA or SA ever from birth to 17.5'. EXECUTE. ## note I did manually enter the codes for the cases where PA was 1 and SA was 999 ## note that the ever PA or ever SA and ever by perp dont need to be recalculated as they already exclude neglect originally ## re-computing the ever in each developmental period ## Infancy ## starting with summing the ever PA or SA in infancy, no need to manually enter as none are one case 999 COMPUTE sum_infancy=SUM(Pabuse_infancy+Sabuse_infancy). EXECUTE. ## creating our ever abuse in infancy variable RECODE sum_infancy (0=0) (1=1) (2=1) INTO ever_infancy. VARIABLE LABELS ever_infancy 'ever PA or SA in infancy'. EXECUTE. ## Early childhood ## summary variable of PA and SA in EC COMPUTE sum_earlyChildhood=SUM(Pabuse_earlychild+Sabuse_earlychild). EXECUTE. 95 ## creating ever abused variable in early childhood variable RECODE sum_earlyChildhood (0=0) (1=1) (2=1) INTO Ever_EarlyChildhood. VARIABLE LABELS Ever_EarlyChildhood 'Ever PA or SA in early childhood '. EXECUTE. ## middle childhood COMPUTE sum_middleChildhood=SUM(Pabuse_middlechild+Sabuse_middlechild). EXECUTE. RECODE sum_middleChildhood (0=0) (1=1) (2=1) INTO Ever_MiddleChildhood. VARIABLE LABELS Ever_MiddleChildhood 'ever PA or SA in MC_MDN'. EXECUTE. ## adolescence COMPUTE sum_adol=SUM(Pabuse_adol+Sabuse_adol). EXECUTE. RECODE sum_adol (0=0) (1=1) (2=1) INTO Ever_Adolescence. VARIABLE LABELS Ever_Adolescence 'ever pa or sa in adolescence_MDN'. EXECUTE. ## Creating overall continuous severity variable by summing ever PA or SA in each developmental period ### note: excluding cases with any missing assessment info! Coded as a 333 in the dataset COMPUTE Severity=SUM(Pabuse_infancy+Sabuse_infancy+Pabuse_earlychild+Sabuse_earlychild+ Pabuse_middlechild+Sabuse_middlechild+Pabuse_adol+Sabuse_adol). EXECUTE. ##Recomputing the chronicity of abuse and total types of abuse variables to exclude neglect— these variables will only be used in reporting out the rates of prospective abuse in the methods section## ## recomputing number of types of abuse without neglect DATASET ACTIVATE DataSet7. COMPUTE Types_abuse_ever_MDN=SUM(Pabuse_ever+Sabuse_ever). EXECUTE. 96 ## recomputing chronicity without neglect, starting by summing each type within developmental period COMPUTE sum_chron_infancy=SUM(Pabuse_infancy,Sabuse_infancy). EXECUTE. COMPUTE sum_chron_EC=SUM(Pabuse_earlychild,Sabuse_earlychild). EXECUTE. COMPUTE sum_chron_MC=SUM(Pabuse_middlechild,Sabuse_middlechild). EXECUTE. COMPUTE sum_chron_adol=SUM(Pabuse_adol,Sabuse_adol). EXECUTE. #### now re-coding so that the above summary variables are now 0 to 1 ## note only needed to do this for early and middle childhood since infancy and adolescence did not go up to 2 RECODE sum_chron_EC (0=0) (1=1) (2=1) INTO recode_sum_chron_EC. VARIABLE LABELS recode_sum_chron_infancy 'recoded to be on a 0 to 1 scale'. EXECUTE. RECODE sum_chron_MC (0=0) (1=1) (2=1) INTO recode_sum_chron_MC. VARIABLE LABELS recode_sum_chron_infancy 'recoded to be on a 0 to 1 scale'. EXECUTE. ## now computing the overall chronicity variable only based on physical and sexual abuse, no neglect COMPUTE Chronicity_any_MDN=SUM(sum_chron_infancy,recode_sum_chron_EC,recode_sum_chron_M C, sum_chron_adol). EXECUTE. SPSS syntax for transforming the retrospective abuse data for analytic variables ### Transforming retrospective AAI coding data into all the variables needed for analyses ## ## making retrospective physical abuse (PA) at 26 into a dichotomous variable: 1-7 -> 0; 8-9 -> 1 DATASET ACTIVATE DataSet3. RECODE Final_Retro_PA_26 (1=0) (2=0) (3=0) (4=0) (5=0) (6=0) (7=0) (8=1) (9=1) INTO Ever_Retro_PA_26. 97 VARIABLE LABELS Ever_Retro_PA_26 'Dichotomous PA cut off '. EXECUTE. ## making retrospective sexual abuse (SA) at 26 into a dichotomous variable: 1-7 -> 0; 8-9 -> 1 RECODE Final_Retro_SA_26 (1=0) (2=0) (3=0) (4=0) (5=0) (6=0) (7=0) (8=1) (9=1) INTO Ever_Retro_SA_26. VARIABLE LABELS Ever_Retro_SA_26 'Dichotomous SA cut of 26'. EXECUTE. ## 19 year ## making retrospective physical abuse (PA) at 19 into a dichotomous variable: 1-7 -> 0; 8-9 -> 1 RECODE Final_Retro_PA_19 (1=0) (2=0) (3=0) (4=0) (5=0) (6=0) (7=0) (8=1) (9=1) INTO Ever_Retro_PA_19. VARIABLE LABELS Ever_Retro_PA_19 'Dichotmous PA cutoff 19'. EXECUTE. ## making retrospective sexual abuse (SA) at 19 into a dichotomous variable: 1-7 -> 0; 8-9 -> 1 RECODE Final_Retro_SA_19 (1=0) (2=0) (3=0) (4=0) (5=0) (6=0) (7=0) (8=1) (9=1) INTO Ever_Retro_SA_19. VARIABLE LABELS Ever_Retro_SA_19 'Dichotomous SA cut off 19'. EXECUTE. ## creating overall composite variables within each assessment ## sum of PA and SA at 26 in order to make dichotomous variable COMPUTE Sum_26=Ever_Retro_PA_26+Ever_Retro_SA_26. EXECUTE. ## recoding sum variable to create the dichotmous ever reported 26 year AAI variable RECODE Sum_26 (0=0) (1=1) (2=1) INTO Ever_26. VARIABLE LABELS Ever_26 'Recalled any abuse of any kind during the 26 year AAI '. EXECUTE. ## 19 year sum of PA and SA COMPUTE Sum_19=Ever_Retro_PA_19+Ever_Retro_SA_19. EXECUTE. 98 ## recoding sum variable to create the dichotomous ever reported 19 year AAI variable RECODE Sum_19 (0=0) (1=1) (2=1) INTO Ever_19. VARIABLE LABELS Ever_19 'Reported any time of abuse at the 19 year assessment '. EXECUTE. ##### Creating overall omnibus variable of 19 and 26 by averaging the dichotmous variables at each assessment COMPUTE Ombinus_Retro_Abuse=MEAN(Ever_26,Ever_19). EXECUTE. ## rescaling the retro continuous scales to be on a 0-8 scale to match prosp severity scale ## 26 PA RECODE Final_Retro_PA_26 (1=0) (2=1) (3=2) (4=3) (5=4) (6=5) (7=6) (8=7) (9=8) INTO rescaled_retro_PA_26. VARIABLE LABELS rescaled_retro_PA_26 'retro 26 PA on a 0-8 scale '. EXECUTE. ## 26 SA RECODE Final_Retro_SA_26 (1=0) (2=1) (3=2) (4=3) (5=4) (6=5) (7=6) (8=7) (9=8) INTO rescaled_retro_SA_26. VARIABLE LABELS rescaled_retro_SA_26 'retro 26 SA on a 0-8 scale'. EXECUTE. ## 19 PA RECODE Final_Retro_PA_19 (1=0) (2=1) (3=2) (4=3) (5=4) (6=5) (7=6) (8=7) (9=8) INTO rescaled_retro_PA_19. VARIABLE LABELS rescaled_retro_PA_19 'retro 19 PA on a 0-8 scale'. EXECUTE. # 19 SA RECODE Final_Retro_SA_19 (1=0) (2=1) (3=2) (4=3) (5=4) (6=5) (7=6) (8=7) (9=8) INTO rescaled_retro_SA_19. VARIABLE LABELS rescaled_retro_SA_19 'retro 19 SA on a 0-8 scale'. EXECUTE. #### average of the continuous 0-8 scale of PA and SA at 19 COMPUTE avg_19_continuous=MEAN(rescaled_retro_PA_19,rescaled_retro_SA_19). VARIABLE LABELS avg_19_continuous 'average of the continuous 0-8 scale of PA and SA at 19'. 99 EXECUTE. #### average of the continuous 0-8 scale of PA and SA at 26 COMPUTE avg_26_continuous=MEAN(rescaled_retro_PA_26,rescaled_retro_SA_26). VARIABLE LABELS avg_26_continuous 'average of the continuous 0-8 scale of PA and SA at 26'. EXECUTE. #### compositing the 19 and 26 0-8 scores to create one omnibus continouous retro abuse scale COMPUTE cont_omnibus_retro_abuse=MEAN(avg_19_continuous,avg_26_continuous). VARIABLE LABELS cont_omnibus_retro_abuse 'average of the 19 and 26 year continouous on 0-8 scale '. EXECUTE. ## Making an ever physically abused (PA) or sexually abused (SA) ever (across 19 and 26) ## Ever physically abused retrospective ever between 19 and 26 # first summing 19 year PA and 26 year PA to use that to make our ever variable COMPUTE Sum_PA_ever=Ever_Retro_PA_26+Ever_Retro_PA_19. EXECUTE. ## now doing the ever PA variable retro RECODE Sum_PA_ever (0=0) (1=1) (2=1) INTO Retro_PA_ever. VARIABLE LABELS Retro_PA_ever 'Dichotomous y/n whether PA reported at either 19 or 26'. EXECUTE. # first summing 19 year SA and 26 year SA to use that to make our ever variable COMPUTE Sum_SA_ever=Ever_Retro_SA_26+Ever_Retro_SA_19. EXECUTE. ## now making the SA dichotomous variables RECODE Sum_SA_ever (0=0) (1=1) (2=1) INTO Retro_SA_ever. EXECUTE. Analysis Syntax * Encoding: UTF-8. ## Dissertation analyses January 30, 2023 #### 100 ## Aim one RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Abuse_Ever /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ## Aim two ## 19 year RELIABILITY /VARIABLES=Abuse_Ever Ever_19 /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ## 26 year RELIABILITY /VARIABLES=Abuse_Ever Ever_26 /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ### Aim 3 ## redoing Aim 1 with continuous overall variables RELIABILITY /VARIABLES=cont_omnibus_retro_abuse Severity /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA /ICC=MODEL(MIXED) TYPE(CONSISTENCY) CIN=95 TESTVAL=0. ## redoing Aim 2 19 RELIABILITY /VARIABLES=Severity avg_19_continuous /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA /ICC=MODEL(MIXED) TYPE(CONSISTENCY) CIN=95 TESTVAL=0. ## redoing Aim 2 26 RELIABILITY /VARIABLES=Severity avg_26_continuous /SCALE('ALL VARIABLES') ALL 101 /MODEL=ALPHA /ICC=MODEL(MIXED) TYPE(CONSISTENCY) CIN=95 TESTVAL=0. ## Aim 4a, omnibus retrospective PA RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Pabuse_ever /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. #SA RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Sabuse_ever /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ## Doing Aim 4a by specific parameters of retro too, not in propsectus #19 PA RELIABILITY /VARIABLES=Pabuse_ever Ever_Retro_PA_19 /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. #26 PA RELIABILITY /VARIABLES=Ever_Retro_PA_26 Pabuse_ever /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. #19 SA RELIABILITY /VARIABLES=Sabuse_ever Ever_Retro_SA_19 /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. # 26 SA RELIABILITY /VARIABLES=Sabuse_ever Ever_Retro_SA_26 102 /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ## doing this just by PA and SA retro ever not further broken down, added 02.21.23 ## PA ever DATASET ACTIVATE DataSet1. RELIABILITY /VARIABLES=Retro_PA_ever Pabuse_ever /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ## SA ever RELIABILITY /VARIABLES=Retro_SA_ever Sabuse_ever /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. #### Aim 4b #infancy RELIABILITY /VARIABLES=Ombinus_Retro_Abuse ever_infancy /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. # early childhood RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Ever_EarlyChildhood /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ## middle childhood RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Ever_MiddleChildhood /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ## adolescence 103 RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Ever_Adolescence /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ### Aim 4c #mother RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Maltx_mother /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. #father RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Maltx_father /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. ##non caregiver RELIABILITY /VARIABLES=Ombinus_Retro_Abuse Maltx_nonparent /SCALE('ALL VARIABLES') ALL /MODEL=ALPHA. #### stability analyses for 19 to 26 CORRELATIONS /VARIABLES=Ever_19 Ever_26 Ever_Retro_PA_19 Ever_Retro_PA_26 Ever_Retro_SA_19 Ever_Retro_SA_26 /PRINT=TWOTAIL NOSIG FULL /MISSING=PAIRWISE. ### correlation of all study variables added 02.15.23 ## just dichotomous variables first since that is a separate aim CORRELATIONS /VARIABLES=Ombinus_Retro_Abuse Ever_19 Ever_26 Retro_PA_ever Retro_SA_ever Ever_Retro_PA_19 104 Ever_Retro_SA_19 Ever_Retro_PA_26 Ever_Retro_SA_26 Abuse_Ever Pabuse_ever Sabuse_ever Maltx_father Maltx_mother Maltx_nonparent ever_infancy Ever_EarlyChildhood Ever_MiddleChildhood Ever_Adolescence /PRINT=TWOTAIL NOSIG FULL /STATISTICS DESCRIPTIVES /MISSING=PAIRWISE. ## in order to run the sensitivity analyses to exclude those who were purely neglected I just deleted those cases and created a new dataset ## I then just re-ran the above analyses in the "Neglect Sensitivity Analyses_Analytic Frame_Dissertation Dataset_v2_02.08.23.SAV" file ## only 15 partiicpants who were purely neglected ## in order to run the sensitivity analyses for the supect abuse I just created a new data file called "Suspect Abuse_Analytic Frame_Dissertation Dataset_v2_02.08.23.SAV" ## can't do Aim 3 and can't look at retro PA and SA disaggregated for these analyses. But just ran the above anlayses again ## all I did to change the supsect file was to change the Ever_19 and Ever_26 and Omnibus retrospective variables based on the supsect abuse variable ##### Now we are just running the descriptive statistics to look at the breakdown of our results 02.13.23 #### ## just examining the groups of prosp only, retro only, or both no, or both yes DATASET ACTIVATE DataSet1. CROSSTABS /TABLES=Abuse_Ever BY Ombinus_Retro_Abuse /FORMAT=AVALUE TABLES /CELLS=COUNT /COUNT ROUND CELL. ## now breaking this down by abuse type ## PA first CROSSTABS /TABLES=Pabuse_ever BY Retro_PA_ever /FORMAT=AVALUE TABLES /CELLS=COUNT /COUNT ROUND CELL. ## now SA 105 DATASET ACTIVATE DataSet1. CROSSTABS /TABLES=Sabuse_ever BY Retro_SA_ever /FORMAT=AVALUE TABLES /CELLS=COUNT /COUNT ROUND CELL. ## now breaking it down by the 19 and the 26 year assessment ## 19 first CROSSTABS /TABLES=Abuse_Ever BY Ever_19 /FORMAT=AVALUE TABLES /CELLS=COUNT /COUNT ROUND CELL. ## 26 CROSSTABS /TABLES=Abuse_Ever BY Ever_26 /FORMAT=AVALUE TABLES /CELLS=COUNT /COUNT ROUND CELL. ## adding in all of the correlations to the tables 02.21.23 ## aim one CORRELATIONS /VARIABLES=Ombinus_Retro_Abuse Abuse_Ever /PRINT=TWOTAIL NOSIG FULL /MISSING=PAIRWISE. ## Aim two CORRELATIONS /VARIABLES=Ever_19 Ever_26 Abuse_Ever /PRINT=TWOTAIL NOSIG FULL /MISSING=PAIRWISE. ## aim three CORRELATIONS /VARIABLES=cont_omnibus_retro_abuse Severity avg_19_continuous avg_26_continuous 106 /PRINT=TWOTAIL NOSIG FULL /MISSING=PAIRWISE. ## doing the stability correlations with the continuous variables too CORRELATIONS /VARIABLES=avg_19_continuous avg_26_continuous rescaled_retro_PA_19 rescaled_retro_PA_26 rescaled_retro_SA_19 rescaled_retro_SA_26 /PRINT=TWOTAIL NOSIG FULL /MISSING=PAIRWISE. ##aim 4a CORRELATIONS /VARIABLES=Ombinus_Retro_Abuse Pabuse_ever Sabuse_ever Retro_PA_ever Retro_SA_ever /PRINT=TWOTAIL NOSIG FULL /MISSING=PAIRWISE. ## aim 4b CORRELATIONS /VARIABLES=Ombinus_Retro_Abuse ever_infancy Ever_EarlyChildhood Ever_MiddleChildhood Ever_Adolescence /PRINT=TWOTAIL NOSIG FULL /MISSING=PAIRWISE. ## aim 4c CORRELATIONS /VARIABLES=Ombinus_Retro_Abuse Maltx_mother Maltx_father Maltx_nonparent /PRINT=TWOTAIL NOSIG FULL /MISSING=PAIRWISE. Attrition analyses Syntax ## first comparing whether the present subsample (n = 162) differed from those excluded (n = 105) on key demographics (i.e., biological sex, ethnicity/race, socioeconomic status, and maternal education) ## comparing analytic to attrited on sex DATASET ACTIVATE DataSet2. 107 T-TEST GROUPS=analyticframe(0 1) /MISSING=ANALYSIS /VARIABLES=Female_M /CRITERIA=CI(.95). ## comparing analytic to attrited on ethnicity T-TEST GROUPS=analyticframe(0 1) /MISSING=ANALYSIS /VARIABLES=babyrace_Recoded /CRITERIA=CI(.95). ## comparing analytic to attrited on SES T-TEST GROUPS=analyticframe(0 1) /MISSING=ANALYSIS /VARIABLES=TSEIComposite42mto16y /CRITERIA=CI(.95). ## comparing analytic to attrited on maternal education T-TEST GROUPS=analyticframe(0 1) /MISSING=ANALYSIS /VARIABLES=Comp_Mat_Ed /CRITERIA=CI(.95). ## Now comparing in the present subsample whether those who were not abused (prospectively) differed on the number of missing assessments than those who were abused (prospectively) ## create the actual summary variable of the total number of assessments missing DATASET ACTIVATE DataSet1. COMPUTE Total_Number_Missing=SUM(Number_missing_infancy,Number_missing_early_childhood, number_missing_middle_childhood,number_missing_adolescence). EXECUTE. ## then within your analytic frame conduct a t-test DATASET ACTIVATE DataSet4. 108 T-TEST GROUPS=Abuse_Ever(0 1) /MISSING=ANALYSIS /VARIABLES=Total_Number_Missing /ES DISPLAY(TRUE) /CRITERIA=CI(.95).