Solving the many-variables problem in MICE with principal component regression.

Multiple Imputation (MI) is one of the most popular approaches to addressing missing values in questionnaires and surveys. MI with multivariate imputation by chained equations (MICE) allows flexible imputation of many types of data. In MICE, for each variable under imputation, the imputer needs to specify which variables should act as predictors in the imputation model. The selection of these predictors is a difficult, but fundamental, step in the MI procedure, especially when there are many variables in a data set. In this project, we explore the use of principal component regression (PCR) as a univariate imputation method in the MICE algorithm to automatically address the many-variables problem that arises when imputing large social science data. We compare different implementations of PCR-based MICE with a correlation-thresholding strategy through two Monte Carlo simulation studies and a case study. We find the use of PCR on a variable-by-variable basis to perform best and that it can perform closely to expertly designed imputation procedures.

Middle School Effects of the Dating Matters® Comprehensive Teen Dating Violence Prevention Model on Physical Violence, Bullying, and Cyberbullying: a Cluster-Randomized Controlled Trial.

Few comprehensive primary prevention approaches for youth have been evaluated for effects on multiple types of violence. Dating Matters®: Strategies to Promote Healthy Teen Relationships (Dating Matters) is a comprehensive teen dating violence (TDV) prevention model designed by the Centers for Disease Control and Prevention and evaluated using a longitudinal stratified cluster-randomized controlled trial to determine effectiveness for preventing TDV and promoting healthy relationship behaviors among middle school students. In this study, we examine the prevention effects on secondary outcomes, including victimization and perpetration of physical violence, bullying, and cyberbullying. This study examined the effectiveness of Dating Matters compared to a standard-of-care TDV prevention program in 46 middle schools in four high-risk urban communities across the USA. The analytic sample (N = 3301; 53% female; 50% Black, non-Hispanic; and 31% Hispanic) consisted of 6th-8th grade students who had an opportunity for exposure to Dating Matters in all three grades or the standard-of-care in 8th grade only. Results demonstrated that both male and female students attending schools implementing Dating Matters reported 11% less bullying perpetration and 11% less physical violence perpetration than students in comparison schools. Female Dating Matters students reported 9% less cyberbullying victimization and 10% less cyberbullying perpetration relative to the standard-of-care. When compared to an existing evidence-based intervention for TDV, Dating Matters demonstrated protective effects on physical violence, bullying, and cyberbullying for most groups of students. The Dating Matters comprehensive prevention model holds promise for reducing multiple forms of violence among middle school-aged youth. ClinicalTrials.gov Identifier: NCT01672541.

Effects of the Dating Matters® Comprehensive Prevention Model on Health- and Delinquency-Related Risk Behaviors in Middle School Youth: a Cluster-Randomized Controlled Trial.

Teen dating violence (TDV) is associated with a variety of delinquent behaviors, such as theft, and health- and delinquency-related risk behaviors, including alcohol use, substance abuse, and weapon carrying. These behaviors may co-occur due to shared risk factors. Thus, comprehensive TDV-focused prevention programs may also impact these other risk behaviors. This study examined the effectiveness of CDC's Dating Matters®: Strategies to Promote Healthy Teen Relationships (Dating Matters) comprehensive TDV prevention model compared to a standard-of-care condition on health- and delinquency-related risk behaviors among middle school students. Students (N = 3301; 53% female; 50% black, non-Hispanic; and 31% Hispanic) in 46 middle schools in four sites across the USA were surveyed twice yearly in 6th, 7th, and 8th grades. A structural equation modeling framework with multiple imputation to account for missing data was utilized. On average over time, students receiving Dating Matters scored 9% lower on a measure of weapon carrying, 9% lower on a measure of alcohol and substance abuse, and 8% lower on a measure of delinquency by the end of middle school than students receiving an evidence-based standard-of-care TDV prevention program. Dating Matters demonstrated protective effects for most groups of students through the end of middle school. These results suggest that this comprehensive model is successful at preventing risk behaviors associated with TDV. clinicaltrials.gov Identifier: NCT01672541.

Effects of Dating Matters® on Sexual Violence and Sexual Harassment Outcomes among Middle School Youth: a Cluster-Randomized Controlled Trial.

Sexual violence (SV), including sexual harassment (SH), is a significant public health problem affecting adolescent health and well-being. This study extends prior research by evaluating the effectiveness of a comprehensive teen dating violence prevention model, Dating Matters, on SV and SH perpetration and victimization, inclusive of any victim-perpetrator relationship, among middle school students. Dating Matters includes classroom-delivered programs for youth in 6th, 7th, and 8th grades; community-based programs for parents; a youth communications program; training for educators; and community-level activities. Middle schools in four urban areas in the USA were randomly assigned to receive Dating Matters (DM, N = 22) or a standard-of-care intervention (SC, N = 24) over four consecutive school years (2012-2016). The analytic sample included two cohorts who entered the study in 6th grade and completed 8th grade by the end of the study allowing for full exposure to Dating Matters (DM: N = 1662; SC: N = 1639; 53% female; 50% black, non-Hispanic; 6 waves of data collection for each cohort). Structural equation modeling was employed with multiple imputation to account for missing data. Dating Matters was associated with significant reductions in SV and SH perpetration and victimization scores in most-but not all-sex/cohort groups by the end of 8th grade relative to an evidence-based TDV prevention program. On average, students receiving Dating Matters scored 6% lower on SV perpetration, 3% lower on SV victimization, 4% lower on SH perpetration, and 8% lower on SH victimization by the end of middle school than students receiving an evidence-based violence prevention program. Overall, Dating Matters shows promise for reducing SV and SH, occurring both within and outside dating relationships, through middle school. Clinicaltrials.gov Identifier: NCT01672541.

Principled Missing Data Treatments.

We review a number of issues regarding missing data treatments for intervention and prevention researchers. Many of the common missing data practices in prevention research are still, unfortunately, ill-advised (e.g., use of listwise and pairwise deletion, insufficient use of auxiliary variables). Our goal is to promote better practice in the handling of missing data. We review the current state of missing data methodology and recent missing data reporting in prevention research. We describe antiquated, ad hoc missing data treatments and discuss their limitations. We discuss two modern, principled missing data treatments: multiple imputation and full information maximum likelihood, and we offer practical tips on how to best employ these methods in prevention research. The principled missing data treatments that we discuss are couched in terms of how they improve causal and statistical inference in the prevention sciences. Our recommendations are firmly grounded in missing data theory and well-validated statistical principles for handling the missing data issues that are ubiquitous in biosocial and prevention research. We augment our broad survey of missing data analysis with references to more exhaustive resources.

A Comparison of Methods for Creating Multiple Imputations of Nominal Variables.

Many variables that are analyzed by social scientists are nominal in nature. When missing data occur on these variables, optimal recovery of the analysis model's parameters is a challenging endeavor. One of the most popular methods to deal with missing nominal data is multiple imputation (MI). This study evaluated the capabilities of five MI methods that can be used to treat incomplete nominal variables: multiple imputation with chained equations (MICE) using polytomous regression as the elementary imputation method; MICE based on classification and regression trees (CART); MICE based on nested logistic regressions; the ranking procedure described by Allison ( 2002 ); and a joint modeling approach based on the general location model. We first motivate our inquiry with an applied example and then present the results of a Monte Carlo simulation study that compared the performance of the five imputation methods under conditions of varying sample size, percentage of missing data, and number of nominal response categories. We found that MICE with polytomous regression was the strongest performer while the Allison ( 2002 ) ranking procedure and MICE with CART performed poorly in most conditions.

On the joys of missing data.

We provide conceptual introductions to missingness mechanisms--missing completely at random, missing at random, and missing not at random--and state-of-the-art methods of handling missing data--full-information maximum likelihood and multiple imputation--followed by a discussion of planned missing designs: Multiform questionnaire protocols, 2-method measurement models, and wave-missing longitudinal designs. We reviewed 80 articles of empirical studies published in the 2012 issues of the Journal of Pediatric Psychology to present a picture of how adequately missing data are currently handled in this field. To illustrate the benefits of using multiple imputation or full-information maximum likelihood and incorporating planned missingness into study designs, we provide example analyses of empirical data gathered using a 3-form planned missing design.

Children's Utilization of Toys is Moderated by Age-Appropriateness, Toy Category, and Child Age.

Play during childhood is essential to growth and learning. Little is known about whether categories of toys moderate play behaviors at different ages, or how children interact with toys that are simple, appropriate, or complex for their developmental level. Two hundred and forty-three children between the ages of 1 and 8 years, divided into four age groups, played with toys that were targeted to their age group as well as toys aimed at one age group younger and older. Toys fell into nine different categories. Whether children fully utilized each toy was evaluated. Analyses examined how children's utilization of toys was affected by the age-appropriateness of the toy, the category of toy, and the child's age. Considering all age groups and toys, children were less likely to fully utilize toys targeted toward older children than age-appropriate toys, but this effect was moderated by the toy category and the child's age.

A novel item-allocation procedure for the three-form planned missing data design.

We propose a new method of constructing questionnaire forms in the three-form planned missing data design (PMDD). The random item allocation (RIA) procedure that we propose promises to dramatically simplify the process of implementing three-form PMDDs without compromising statistical performance. Our method is a stochastic approximation to the currently recommended approach of deterministically spreading a scale's items across the X-, A-, B-, and C-blocks when allocating the items in a three-form design. Direct empirical support for the performance of our method is only available for scales containing at least 12 items, so we also propose a modified approach for use with scales containing fewer than 12 items. We also discuss the limitations of our procedure and several nuances for researchers to consider when implementing three-form PMDDs using our method. ● The RIA procedure allows researchers to implement statistically sound three-form planned missing data designs without the need for expert knowledge or results from prior statistical modeling. ● The RIA procedure can be used to construct both "paper-and-pencil" questionnaires and questionnaires administered through online survey software. ● The RIA procedure is a simple framework to aid in designing three-form PMDDs; implementing the RIA method does not require any specialized software or technical expertise.

An RCT of Dating Matters: Effects on Teen Dating Violence and Relationship Behaviors.

INTRODUCTION: Teen dating violence is a serious public health problem with few effective prevention strategies. This study examines whether the Dating Matters comprehensive prevention model, compared with a standard of care intervention, prevented negative relationship behaviors and promoted positive relationship behaviors. STUDY DESIGN: This longitudinal, cluster-RCT compared the effectiveness of Dating Matters with standard of care across middle school. Standard of care was an evidence-based teen dating violence prevention curriculum (Safe Dates) implemented in eighth grade. SETTING/PARTICIPANTS: Forty-six middle schools in high-risk urban neighborhoods in four U.S. cities were randomized. Schools lost to follow-up were replaced with new schools, which were independently randomized (71% school retention). Students were surveyed in fall and spring of sixth, seventh, and eighth grades (2012-2016). The analysis sample includes students from schools implementing Dating Matters or standard of care for >2 years who started sixth grade in the fall of 2012 or 2013 and had dated (N=2,349 students, mean age 12 years, 49% female, and 55% black, non-Hispanic, 28% Hispanic, 17% other). INTERVENTION: Dating Matters is a comprehensive, multicomponent prevention model including classroom-delivered programs for sixth to eighth graders, training for parents of sixth to eighth graders, educator training, a youth communications program, and local health department activities to assess capacity and track teen dating violence-related policy and data. MAIN OUTCOME MEASURES: Self-reported teen dating violence perpetration and victimization, use of negative conflict resolution strategies, and positive relationship skills were examined as outcomes. Imputation and analyses were conducted in 2017. RESULTS: Latent panel models demonstrated significant program effects for three of four outcomes; Dating Matters students reported 8.43% lower teen dating violence perpetration, 9.78% lower teen dating violence victimization, and 5.52% lower use of negative conflict resolution strategies, on average across time points and cohorts, than standard of care students. There were no significant effects on positive relationship behaviors. CONCLUSIONS: Dating Matters demonstrates comparative effectiveness, through middle school, for reducing unhealthy relationship behaviors, such as teen dating violence and use of negative conflict resolution strategies, relative to the standard of care intervention. TRIAL REGISTRATION: This study is registered at www.clinicaltrials.gov NCT01672541.

The Support Needs of Children with Intellectual Disability and Autism: Implications for Supports Planning and Subgroup Classification.

The Supports Intensity Scale-Children's version (SIS-C) was developed to provide a standardized measure of support needs of children with intellectual disability. Over half of the norming sample had a secondary diagnosis of autism. Using this subset of the sample, we engaged in exploratory analysis to examine the degree to which latent clusters were present in the data, and after identifying these clusters, the degree to which they mapped on the SIS-C standard scores. A four latent class solution provided the best fit to the data. When mapped on SIS-C standard scores, specific patterns of differences were found in life activity domain scores and overall support needs scores. Implications for future research and practice are discussed.

Creating a SIS-A Annual Review Protocol to Determine the Need for Reassessment.

The Supports Intensity Scale - Adult Version (SIS-A) has been widely adopted throughout North America and the world since its publication a little over a decade ago. Many organizations and jurisdictions operate under regulations that require an annual assessment of people who receive services and supports that are financed through public funds. The time and energy devoted to an annual SIS-A reassessment has become a concern in cases where the resulting information is largely redundant with information from a prior assessment. This article presents findings from an investigation of two approaches to creating a protocol to assist SIS-A users in distinguishing situations where there is a high likelihood that support needs have not changed in meaningful ways from situations where there is a reasonable possibility that support needs have changed. The SIS-A Annual Review Protocol was created based on these analyses as well as consideration of conceptual issues associated with support needs assessment. Ways in which this protocol might be used, as well as data that need to be collected to evaluate its usefulness, are discussed.