Advanced Quantitative Methods Assessed Coursework. Answer both sections. Section A. Posttraumatic stress disorder (PTSD) is a psychological phenomenon characterised by involuntary intrusive thoughts that can affect memory, impair concentration and attentional focus, and interfere with day to day activities (see Buckley et al., 2000, for a review). Interestingly, people with a history of PTSD are strongly biased towards stimuli (e.g., words or pictures) that convey the nature of the experienced trauma. This bias has been shown in the magnitude of the startle blink response, when participants are exposed to traumatic information in pictures and/or words. The magnitude of the startle blink response (in microvolts) has been used to index a selective attentional bias towards a particular stimulus, with greater blink responses indicative of a greater bias. You are interested in examining how PTSD predicts and explains the startle blink response towards trauma related cues in soldiers whove undergone extensive exposure traumatic experience during war. You assess the startle blink magnitude (in microvolts) while participants were exposed to trauma-related scenes (., battle-field) as well as neutral information (., a chair). The magnitude of the blink response in vulnerable populations has also been shown to be affected by other variables such depression, anxiety, and levels of inhibitory control. So, in your experiment you assess anxiety vulnerability using a well known anxiety questionnaire (scores range between 20 and 80 with higher scores indicating greater anxiety) and depression (scores range between 0 and 30 with higher scores indicating higher depression). You also assess participants level of attentional control using an antisaccade task where faster antisaccade latencies (in msec) are indicative of better inhibitory control. PTSD was assessed using a PTSD scale (1 10, with higher scores indicative of greater PTSD experiences). The startle blink response was measured in microvolts and indicates a relative effect of blink magnitude to threatening scenes vs neutral scenes. The data for this question is in the file: Now attempt the following questions 1. Provide a correlation matrix of the variables under investigation and comment on the patterns of correlations you see amongst the variables as well as in relation to the predicted variable of interest. Would you be concerned about multicolinearity? How and when does multicolinearity bias a regression model? 2. Perform a standard regression analysis with the IVs predicting startle blink response. Report on the effects observed, and comment on the contribution of the variables in predicting startle blink response. 3. Perform a stepwise regression analysis and compare your results with that of 2 above. How has the stepwise method altered the contribution of the independent variables? How can the regression analysis check for multicolinearity? And are there cases for concern? 2 4. Assume that you are interested in assessing the contribution of PTSD first, and Antisaccade (inhibitory control) on the last step. Comment on the contribution of each model and the Fchange and R2change on every step. How do the results of this model compare with the Stepwise model? 5. Write a brief summary on the role of PTSD in explaining startle blink response in soldiers when exposed to upsetting scenes. What did the analysis reveal as the most important contributor? 6. Using the regression equation of the final model of the hierarchical method (as in 4 above), if a soldier scored 8 on PTSD, 46 on Anxiety, 25 on Depression, and had an antisaccade latency of 323msec, what would the model predict for his startle blink response? Section B Major depression is a clinical disorder that usually follows in response to a traumatic event, or following a history of dealing with chronic stressors in life. Recent research suggests that cognitive control is a major predictor of depression relapse such that higher levels of inhibitory control are associated with lower rates of relapse in depressed individuals (Koster, & Derakshan, 2017). In a recent breakthrough (Owens, Koster & Derakshan, 2013, see Koster et al 2017 for a review) it was shown that by training working memory capacity it is possible to enhance inhibitory control in depression with effects generalised to other untrained tasks. You are interested in examining how improvement brought about by cognitive training predicts the likelihood of relapse during a six month period in a sample of 30 subclinically depressed participants, some of whom underwent a course of four weeks of adaptive cognitive training aimed at enhancing working memory capacity. Those participants who did not undergo adaptive cognitive training participated in a non-adaptive control training. At the end of the training course all participants were assessed on behavioural (WMC: working memory capacity) and neural (FE: filtering efficiency) measures of inhibitory control. Participants were also assessed on depression severity before the course of training (with higher scores on this assessment indicating more severe depressive symptomatology). The data for this question is in the file: Now attempt the following: Perform a standard logistic regression analysis to assess the predictive utility of the variables under investigation in predicting the likelihood of depression relapse in your depressed sample. 1. Comment on the classification table(s) of interest. What percentage of the cases is correctly and incorrectly classified by the models attempt at each stage of the analysis? 3 2. Comment on the significance of the model(s) tested, the amount of variation explained by the model(s), and the -2LLs of interest. 3. Which variable(s) predict the likelihood of relapse significantly? Provide an interpretation of the significance of Exp (B) of the significant variable(s). 4. Using the maximum likelihood method estimate the likelihood of no relapse for someone with a WMC score of . Note that e (the base of the natural logarithms) is .
Hide
