Amanda Kraha

Tools to Support Interpreting Multiple Regression in the Face of Multicollinearity

While multicollinearity may increase the difficulty of interpreting multiple regression (MR) results, it should not cause undue problems for the knowledgeable researcher. In the current paper, we argue that rather than using one technique to investigate regression results, researchers should consider multiple indices to understand the contributions that predictors make not only to a regression model, but to each other as well. Some of the techniques to interpret MR effects include, but are not limited to, correlation coefficients, beta weights, structure coefficients, all possible subsets regression, commonality coefficients, dominance weights, and relative importance weights. This article will review a set of techniques to interpret MR effects, identify the elements of the data on which the methods focus, and identify statistical software to support such analyses.

Coping with Stressful or Traumatic Events: What Aspects of Trauma Reactions are Associated with Health Outcomes?

The presence of posttraumatic stress disorder (PTSD) symptoms has been shown to be related to a number of health outcomes. In the current study, we explored which specific aspects of PTSD are most related to health measures. The associations between the specific DSM-IV-TR PTSD criteria (criteria A-F) and five indicators of health and well-being--physical health symptoms, quality of life, mental health, depression and negative affect--were examined. The sample consisted of 711 undergraduates. A non-clinical sample was recruited so there would be variability in the various criteria for PTSD. Multiple regression analyses revealed that the hyperarousal (criterion D) was the most consistent and strongest predictor of outcomes. However, the F criterion (causes significant impairment) predicted additional variance in quality of life, depression and negative affect. These results suggest that it is not just the mere frequency of trauma symptoms that affect well-being but also the disruptive capability of these symptoms. In addition, follow-up analyses indicated that hyperarousal mediated the association between the A2 criterion (traumatic response) and all five outcome measures. These results underscore the importance of the hyperarousal criterion, while also suggesting the need for increased attention to the F criterion when considering the impact of stressful events on health and well-being.

Parents and Vehicle Purchases for Their Children: A Surprising Source of Weight Bias

According to current estimates, 68% of the population in United States is considered either overweight or obese. Despite its relative frequency, prejudice and discrimination against overweight and obese Americans is a common occurrence. This study seeks to build on previous findings that overweight individuals are more likely than their skinnier peers to rely on themselves, rather than family, to fund their education. We examined whether this trend continued in car‐buying practices. Results suggest that overweight and obese individuals differentially rely on their own sources of income to finance a car, even after controlling for socioeconomic differences. Possible explanations and implications are discussed.

Why so negative? Positive flashbulb memories for a personal event.

Flashbulb memories are highly vivid and long-lasting memories for events that are emotionally significant and personally important. Research in this area typically focuses on negative events such as the assassination of John F. Kennedy, the events of 11 September 2001, and the start of Operation Desert Storm. However, the current study examines flashbulb memory formation to the positive event of receiving an invitation to join a university-wide social organisation (Greek Life). Results suggest that this is an event capable of producing vivid flashbulb memories with high ratings of belief in accuracy and high consistency over time.

Reliability Generalization of the Motivated Strategies for Learning Questionnaire: A Meta-Analytic View of Reliability Estimates

A reliability generalization meta-analysis was performed to explore the relationship between study factors and levels of alpha reliability for the 15 subscales of the Motivated Strategies for Learning Questionnaire (MSLQ). The MSLQ has been widely adapted over the past 25 years to investigate the role of motivation and strategies in learning, primarily at the postsecondary level. A literature search from the years 1991 to 2015 yielded 295 peer-reviewed journal articles and 1,369 alpha reliability coefficients. Articles were coded for six potential moderator study variables. A novel varying coefficient (VC) model was adopted to determine average reliabilities across studies for each subscale and to perform multiple regression analyses to identify study variables that may moderate alpha reliability estimates. Commonality analyses were used to aid in interpretation of regression results. Meta-analyzed alpha reliabilities were lower than values published in the test manual for all but three of the subscales. Ability of specific moderators to predict score reliability varied across subscales; however, studies in North America, in English, or using a 7-point response scale generally corresponded to increased reliability estimates. Knowledge of expected levels of score reliability under varying sample and study conditions may provide useful information for researchers planning future use of the MSLQ.

Student Perceptions of and Attitudes toward Peer Review

ABSTRACT The current study seeks to analyze university students’ perceptions of online peer assessment (PA) with regard to its effectiveness as a learning tool and as a contributing factor to motivation and interaction. Participants also responded to questions about self-perceived writing strength, attitudes toward peer review, and experiences with peer review. The survey was administered via link online for students who report having completed peer review in online courses. Results from participants revealed that students generally believe peer review to be effective, helpful to learning, increasing communication skills, and as contributing to a stronger written end product. Students also agreed that peer review increases interaction, participation, and a sense of community. Attitudes toward peer review indicate that female students are more intimidated by giving peer reviews than their male classmates, but they felt less intimidated online as compared to face-to-face. More experienced students and students farther along in their college careers are less likely to give peer reviews that are affected by the ones they receive. The results indicate that peer review is an important teaching and learning tool that merits more research with particular attention to gender and other demographic data.

How did you hear the news? The role of traditional media, social media, and personal communication in flashbulb memory

Flashbulb memories are at the intersection of public and private event memory. We investigated whether the source of news (traditional media, social media, or another person) shaped how people remembered learning of an event. Individuals were asked how they learned of Osama bin Laden’s assassination immediately after the event and 7, 42, 224, or 365 days later. Initial memory reports from those who learned from traditional media showed enhanced phenomenological features (i.e., a sense of recollection and vividness) relative to those who learned from social media or from another person. Both phenomenological and metacognitive (i.e., belief in the memory’s accuracy) features of memory reports decreased over time; however, there were no differences as a function of source. Consistency of the memory reports did not differ as a function of time or source. Although sources differed as a function of social group salience, these differences did not seem to influence memory.

All-Possible-Subsets for MANOVA and Factorial MANOVAs: Less than a Weekend Project

Multivariate techniques are increasingly popular as researchers attempt to accurately model a complex world. MANOVA is a multivariate technique used to investigate the dimensions along which groups differ, and how these dimensions may be used to predict group membership. A concern in a MANOVA analysis is to determine if a smaller subset of variables may be used in the classification functions without any loss of explanatory power when precision of parameter estimates or parsimony needs to be addressed cf. Huberty, 1984; Huberty & Olejnik, 2006. One way to address these concerns is through the use of all possible subsets. However, not all common statistical packages easily facilitate this analysis, and the analysis can be a weekend project Huberty & Olejnik, 2006. As such, the purpose of the current paper is to examine and demonstrate R and SPSS solutions to conduct an all-possible-subsets MANOVA, including all-possible-subsets factorial MANOVA.