4.    The effects of Marketing Capabilities on Performance Outcomes of International New Ventures: The Mediating Role of Market Information Resources

Silvia Lozano Martin, and Rajshekar Javalgi

In an emerging market context, this study examines the impact of market information resources on marketing capabilities and performance outcomes at the export venture level of international new ventures (INVs). To date, the literature on high-tech INV firms has failed to consider the importance of market information resources within the marketing capabilities interplay in approaching new markets overseas. With survey data from 260 INV firms in Mexico structural equation modeling was used to test measurement veracity and proposed relationships between the constructs constrained in the measurement model. The results indicate that market information resources play a central role in the process of attaining superior export venture performance. These findings indicate that the rapid environment that INVs find themselves requires an entrepreneurial orientation mediated with market information resources to impact on marketing capabilities from which to achieve superior export venture performance. This investigation contributes to the study of INVs by demonstrating that market information resources help INVs to focus on deploying marketing capabilities based on new product development, sales and distribution service to enhance export venture performance.

 

5.    Changes in Risk Behavior Achieved by Activating Dynamic Coupling Processes: dynamic growth modeling of a health prevention intervention

Emil Coman, Carolyn Lin, Suzanne Suggs, Eugen Iordache, Maria Coman, and Russell Barbour

We introduce the modeling of bivariate dynamical changes as a method of assessing the intervention effects of a substance use prevention program. Applying the techniques of dynamical modeling with latent change scores (LCS) to a four-wave data set, we show how the intervention has spurred a strong coupling effect between Internal locus of Control (ILC) and Intent to Use Substances (IUS) in the intervention group, as well as new mutual dynamic links – i.e., changes in ILC causing subsequent IUS changes, and vice versa.

 

6.    Performance of the MIMIC-Model Methods for DIF Detection

Pei-Hsuan Chiu, and H. Jane Rogers

Recently, a number of MIMIC-model methods have been proposed for DIF testing. These methods differ in the statistical criteria or the procedures that are used during the DIF detection process. In addition, a key difference of these methods is how the (DIF-free) items are selected to serve as anchors. The purpose of this study is to compare four different MIMIC-model methods for DIF testing. A series of Monte Carlo simulation studies were conducted using MPLUS with manipulation of test length, sample size and the percentage of DIF-items including in the test (including DIF-free condition). The Type I error rate of misclassifying DIF-free items and the power of identifying DIF items for each condition were assessed as the outcome variables.  Finally, a couple of real datasets from large-scale student assessments were analyzed to show applications of the MIMIC-model methods that were compared in the simulation.

 

7.    The Relationship between Middle Years Programme Student Performance and Diploma Programme Student Performance

Melissa Gordon, and Liz Bergeron

The International Baccalaureate (IB) offers three educational programmes for a worldwide community of schools, aiming to create a better, more peaceful world.  The IB offers Middle Years Programme (MYP) for students aged 11 to 16 and Diploma Programme (DP) for students aged 16-19.  Previous research suggests that participation in an IB preparatory program better prepares students for the rigour and challenges of the IB DP (Bland & Woodworth, 2011).  The goal of the present study is to investigate if IB Diploma Programme (DP) students benefit from previously completing the IB Middle Year Programme (MYP).  This is accomplished by conducting regression analyses, multilevel modeling (HLM), and an independent t test on the existing data. Findings suggest that students who perform better during MYP moderation tend to perform better on their DP subject exams.

 

 

8.    Housing Profiles and Changes in Low-Income Children’s Academic, Emotional, and Behavioral Functioning

Rebekah Levine Coley, Melissa Kull, Tama Leventhal, and Alicia Doyle Lynch

The present study sought to comprehensively assess the “housing bundle” of a representative sample of low-income families using a latent class analysis (LCA). Data were drawn from the longitudinal Three City Study of families and children living in low-income neighborhoods in Boston, Chicago and San Antonio. Profiles identified in the LCA were subsequently used as predictors in individual fixed effects models predicting child functioning. Results revealed that shifts between housing profiles was associated with changes in children’s academic, emotional, and behavioral functioning. This study makes a unique contribution to the literature linking families’ housing contexts and children’s functioning.

 

9.    Model Specification for Interaction and Quadratic Effects Between Formative Latent Variables

Shu-Ping Chen, and Chung-Ping Cheng

Estimating the interaction and quadratic effects of latent variables is an important issue in the social and behavioral sciences. A variety of approaches have recently been developed for the estimation of nonlinear structural equation modeling. To our knowledge most approaches have predominantly been used to estimate latent nonlinear effects between reflective exogenous variables. While Chen and Cheng (accepted) generalize the constrained approach to a matrix form that encompasses the latent nonlinear effects between exogenous and/or endogenous variables, their framework only accommodates reflective variables. In the current research, we extend Chen and Cheng’s original framework to encompass the nonlinear effects between formative latent variables which can be exogenous and/or endogenous variables. Constraints are specified in matrix form and the matrices involved in model specification are partitioned to fit into our nonlinear model framework. The usage and validity of the procedure is demonstrated with a simulated dataset example using the OpenMx package.

 

10.  Longitudinal Influence of Marital Satisfaction on Harmonious Family Interaction: Using SEM to Estimate the APIM with Distinguishable Dyads

Yeonsoo Yoo, Rachel B. Tambling, and JoAnn L. Robinson

This poster presents a longitudinal investigation of interparental relationship, parental sensitivity, and family interaction quality. These findings suggest that maternal and paternal perceptions on marital satisfaction appear to reciprocally affect each other before but not after child age 3 years old. Only mothers’ marital satisfaction influenced their own sensitive behaviors toward their children and further, harmonious family interactions at dinner. This study has important implications for intervention with couples and families at critical time points.

 

11.  Mediational Analysis of Intervention Effects on Dietary Fat Reduction Using Parallel Process Latent Growth Modeling

Leslie Brick, Si Yang, and Lisa L. Harlow

Data from three randomized trials of Transtheoretical Model (TTM) tailored multiple behavior interventions (N=9461) were pooled to examine the meditational relationship of three TTM measures (Pros, Cons, Temptations for Dietary Fat) with treatment group and stage of change. The focus of the current study was on energy balance behavior, represented as having a diet low in fat (i.e. less than 30% daily consumption) with measures at three time points: baseline, 12-months, and 24 months. Parallel processes latent growth curve modeling was used to assess mediation of measures. In this study, mediation was supported when the treatment group significantly affected the trajectory of the TTM measure, which then affected the trajectory of stage membership. The Cons scale was the only measure with a significant mediation, suggesting that intervention treatment groups resulted in an increase in the growth rate of Cons, which decreased over time, and an increase in the growth rate of stage, which increased over time.

 

12.  Associations between Perceived Racial Discrimination, Alcohol Use, and Drinking Problems: A Structural Equation Model with Moderated Mediation

Ross E. O’Hara, Marcella H. Boynton, Denise Scott, Stephen Armeli, Howard Tennen,and Jonathan Covault

The effects of racial discrimination on alcohol use and drinking problems were examined using structural equation modeling with moderated mediation. This study extended prior knowledge by examining whether discrimination effects were mediated by discrimination-specific or generalized anger; how these mediation patterns related to drinking problems; and whether these processes differed by gender. Survey data came from 741 African-American students at a historically Black university (53% female). Discrimination predicted alcohol use only among men, and mediated fully by generalized anger. Similarly, discrimination predicted men’s drinking problems through generalized anger and depressive symptoms. For women, only a risky family environment predicted alcohol outcomes. These results replicate and extend prior research by showing that African-American men exposed to discrimination may become more prone to anger, in general, prompting increased alcohol use. Additionally, indirect effects of discrimination on drinking problems via changes in affect may help explain male African-Americans’ disproportionate risk for these outcomes.

 

13.  Using Structural Equation Modeling to Assess The Effect Affective Variables and Socio-Economic Status on Mathematics Achievement at Lower Secondary Level

Khemduth Singh Angateeah , Dr Kaviraj Sharma Sukon, and Dr Preethee Nunkoo-Gonpot

The purpose of this study was to examine the effect of socio-economic status and three school related construct – Attitude, Motivation and Perceived usefulness of Mathematics – on 8th grade (13 years old) achievement in mathematics in Mauritius. Though cognitive abilities of students of students and their family background are important predictors of achievement, there is an increasing amount of research evidence on the salient role of affective variables on mathematics achievement. A sample of 491 grade 8 students from 14 secondary schools was involved in 2011. Structural Equation models (LISREL 9.0) was used to measure the influence of affective variables and Socio-economic status on mathematics achievement. Socio-economic status and attitude had greatest influence on mathematics achievement. Perceived usefulness of mathematics has positive influence on motivation and attitude but negative direct influence on mathematics performance.

 

14.  Using the Rasch Model to Assess the Interest for Shakespeare and Illustrate Rating Scale Diagnostics

Gloria Yeomans-Maldonado

An illustration of how the Rasch Model was used to investigate the psychometric properties of a 35-item survey developed to gather student attitudes regarding Shakespeare will be presented. Similar to many surveys, it employed an ordinal level of measurement. However, one of the sections utilized a “Don’t know” [DK] response choice nested between the “Yes” and “No” responses choices. The test developer’s intention was that this option might act as a middle point. Standard practice is to set such values to missing because the scale point does not fall along the continuum of the other anchor points. The third section of the survey instrument employed a standard three-point ordinal scale. A common practice is to treat such scales as interval and to create composite scores via averaging. However, this practice is only valid if equidistance holds between adjacent scale points. Rasch measurement diagnostics (logit scores, threshold estimates, probability curves, and category fit) were used to determine both whether DK could be treated as a middle point as well as whether the equidistance assumption was met.

 

 

15.  Examination of Growth Models within Ohio Vendor Assessments

Lauren M. Porter, and Susan Mauck

Beginning in 2008, Ohio has used value-added measures as an accountability component in district and school report cards. Under the new Ohio Teacher Evaluation System (OTES) and the Ohio Principal Evaluation System (OPES) value-added measures will become an integral component of teacher and principal evaluations. Given that current statewide assessments do not cover all subjects or grades, the use of non-state assessments has been approved for evaluation purposes in select cases. This study seeks to examine the methodology behind and definitions of growth as applied to Ohio Department of Education designated Vendor Assessments. Within this study we seek to review and compare differences in the definition of growth, methodologies used to determine growth and reliability and validity measures for the 16 assessments identified in January 2013 as Vendor Assessments. Results will be used to inform policymakers, including the Ohio Department of Education, on potential changes for Vendor Assessment selection criteria.

 

16.  Learning Using Hidden Information Applied to Inference of Motor Neuron Function from Differential Gene Expression

Therese M. Smith, and Greg Johnson

Recent work by Dr. Pamela Shaw and colleagues, obtaining differential gene expression in spared and vulnerable motor neurons in Amyotrophic Lateral Sclerosis(Unravelling the enigma of selective vulnerability in neurodegeneration: motor neurons resistant to degeneration in ALS show distinct gene expression characteristics and decreased susceptibility to excitotoxicity, Acta Neuropathologica 2013) has made use of DAVID (http://david.abcc.ncifcrf.gov/) to infer the functions these differently expressed genes might support.

We undertook to review the methods of inferring functions from gene expression in the multiple tools used by DAVID and at Gene Ontology, to see whether the work of Dr. Vladimir Vapnik, described as learning using hidden information, might offer additional benefits.

 

17.  Under Detection Problem of Cytokines: an AFT Model Application

Yu-Bo Wang, James Grady, Rong Wu, and Zhao (Helen) Wu

The roll of proinflammatory cytokines as a biomarker for human diseases has become increasingly investigated in clinical studies. Among them, IL-6, IL-10, IL-1ra, and TNF are of most interest. However, cytokine data present two typical analysis challenges including 1) left censoring due to in- struments that cannot detect very low levels of cytokines, 2) right-skewed distribution. To identify better methods to appropriately analyze this type of data, we utilize cytokine data from an NIH sponsored longitudinal study between 2006-2011, which examines cytokines as predictors of stress and drug use in a cohort of women. We will compare several approaches that have been suggested in the literature, including log transformation, mixture models and the Tobit model to a parametric accelerated failure time model borrowed from the survival analysis field. Simulation results will be presented to compare the performances of those methods.

 

18.  Using HLM to Understand Alliances in High Conflict Divorce Treatment

Scott C Huff, Shayne R. Anderson, and Rachel B. Tambling

This poster will present an example of using Hierarchical Linear Modeling with longitudinal dyadic data. Our study compared the nature of the therapeutic alliance between high conflict divorce couples and traditional couples. Using data from 140 clinical couples, we modeled growth curves on several measures of the therapeutic alliance comparing divorced couples to traditional couples. Analysis showed that on each measure the two groups had different trajectories. Implications will be discussed, especially the value of HLM for marriage and family therapy research. Our study is an example of how HLM answers many of the common critiques of typical marriage and family therapy research.

 

 

19.  Methods for Intuitive Presentations of Hierarchical Categorical Models: An Example Predicting Futile Treatment in Critical Care

Joshua F. Wiley, Thanh N. Huynh, MD, and Neil S. Wenger, MD

Categorical outcomes and hierarchical data structures are common in medical research. Generalized linear mixed models can be difficult to interpret; however, predicted probabilities and marginal effects on the probability metric can be used to provide meaningful, interpretable results for researchers and clinicians.

In this poster, we demonstrate an application of the method with a study of predictors of receiving futile treatment in critical care units using a cross classified random effects Bayesian ordered probit model and the newly developed R package, postMCMCglmm to assess the accuracy of the model as number correctly classified, understand random effects in terms of standard deviations of probabilities, compute average marginal predicted probabilities to assess overall effects of predictors, and create graphs of predicted probabilities with 95% highest posterior density intervals.

 

20.  Day-to-Day Stress-Pain Relationships among Fibromyalgia Patients: Does Social Support Moderation Change the Daily Relationship over Time?

Deborah A. Forrest

Participants in the study were a sub-sample of the national sample of women diagnosed with FM referred by Rheumatology practices in the United States.  Participants (N = 210) completed a telephone interview and then completed mailed daily diaries in a Measurement Burst design with seven daily diary measurements in three bursts across three years on their daily pain, stressors and negative emotions.  Social support was assessed from a baseline interview as it was found to be unchanging over the course of the study.   The results indicated that individuals with FM experienced increased pain on days they experienced higher than average levels of daily stressors, and this relationship increased in magnitude over time.  In addition, there was support for the notion that perceived availability of social support moderated the relationship over time.  For those with low social support, the relationship between stress and pain intensified over time but in those with high social support, this relationship did not change over time.  These findings have implications for FM patient education and cognitive behavioral therapies capitalizing on reduced stress, particularly for those with low social support.

 

21.  Investigating the Directionality and Pattern of Mutual Changes of Health Outcomes: Adding Dynamic Perspectives to Static Longitudinal Analyses

Emil Coman, Marco Bardus, Suzanne Suggs, Eugen Iordache, Maria Coman,and Holly Blake

We walk the audience through stepwise analytical procedure meant to analyze complex mutual dynamic relationships between health outcomes. We introduce first simple difference scores for a set of 6 outcomes measured every six months over five time points and describe their relations, then explain the latent (true) change scores (LCS) specification. We detail alternative specifications of the LCS scores, and how they might impact the fit and the estimates of the models (e.g. latent change scores defined as a 1 indicator factor, by the prior X1 or the subsequent wave variable X2). We report on models linking the key variables from the Theory of Planned Behavior (TPB) based health intervention aimed at improving physical activity (PA), examine combinations of sets of outcomes and their mutual changes, and conclude with implications for methodologists and practitioners of such dynamic change modeling.

 

22.  An Investigation of a Multidimensional Model of School Readiness 

Jill Pentimonti, Kimberly Murphy, Laura Justice, Jessica Logan,and Joan Kaderavek

The present study aimed to examine the construct of school readiness and the extent to which it captures three dimensions of readiness skills (academic, social, and behavioral), as well as how these dimensions relate to children’s end-of-kindergarten literacy skills. Moreover, this study addresses these issues among children with language impairment (LI), a population of children known to be at risk for reading difficulties. Participants were 136 preschool-aged children with LI drawn from 83 special education classrooms. Children were assessed on measures of academic, social, and behavioral skills. Confirmatory Factor Analyses indicated that school readiness for this sample of children is best characterized as two separate dimensions (academic, socioemotional). Of these components, academic readiness was found to be predictive of children’s later performance on measures of literacy. The results of the study further our theoretical understanding of the dimensions of school readiness by empirically testing the nature of school readiness models.

23.  Leveraging Both Between- and Within-Person Variability in Predicting Alcohol Use: Theory and Application of Centering Approaches in a Multilevel Modeling Context

Marcella H Boynton, Ross E. O’Hara, and Howard Tennen

Multilevel modeling (MLM) is increasingly becoming the analysis of choice when examining data with a nested structure. In MLM variable centering greatly facilitates model convergence and interpretation of the variances of the intercept and slope. Despite these benefits, many users of MLM are still hesitant to employ centering, in part because there are multiple centering strategies with distinct statistical and theoretical implications. The focus of this talk is to provide an overview of centering strategies within MLM as well as a discussion of how these strategies impact interpretation of MLM results. Results from two daily diary studies of alcohol use will be used to demonstrate the application and theoretical interpretation of different MLM centering strategies. The statistical arguments both in favor of and opposed to different centering techniques, in general, and within daily diary study analyses, specifically, will be considered.

 

24.  Predicting Condom Use Using the Information-Motivation-Behavioral Skills (IMB) Model: A Multivariate Latent Growth Curve Analysis

Jennifer L. Walsh, Theresa E. Senn, Lori A. J. Scott-Sheldon, Peter A. Vanable, and Michael P. Carey

The Information-Motivation-Behavioral Skills (IMB) model often guides sexual risk reduction programs even though no studies have examined covariation in the theory’s constructs in a dynamic fashion with longitudinal data. Using new developments in latent growth modeling, we explore how changes in information, motivation, and behavioral skills over 9 months relate to changes in condom use among STD clinic patients. Participants (N=1281, 50% female, 66% African American) completed measures of IMB constructs at three time points. We used parallel process latent growth modeling to examine associations among intercepts and slopes of IMB constructs. Initial levels of motivation, behavioral skills, and condom use were positively associated, with behavioral skills partially mediating associations between motivation and condom use. Changes over time in behavioral skills positively related to changes in condom use. Results support the key role of behavioral skills in sexual risk reduction, suggesting skills should be targeted in HIV prevention interventions.

 

 

25.  Empathic Accuracy in Daily Interactions between Romantic Partners: The Effect of Communal Behavior on the Perception of The Partner’s Negative Affect

Gentiana Sadikaj, Debbie S. Moskowitz, and David C. Zuroff  

The present study examined situational cues that influence the perception of the partner’s negative affect. Using an event-contingent recording methodology, partners in 93 cohabiting couples recruited from the community reported on their feeling of unhappiness, interpersonal behavior, and their perceptions of the partner’s unhappiness in interactions with each-other during a 20-day period. The truth and bias model of judgment (West & Kenny, 2011) was used as a conceptual and statistical framework for the data analysis. Findings suggest the presence of mean level bias (i.e., overestimation), tracking accuracy, and assumed similarity. They indicate that partner’s communal but not agentic behavior affected perception accuracy. Results extend prior research by demonstrating that the partner’s communal behavior serves as a cue that affects the perception of the partner’s affect. This study demonstrates the use of novel methodological advancements in statistical modeling and measurement in examining questions pertaining to social perception