The Gettin Out the Gunk intervention and evaluation are moving forward. Baseline survey data have been collected, entered, and cleaned. Knowledge, self-efficacy, and motivation scales have been created. And now, most recently, volunteers have been randomized to intervention and control groups, and a new variable indicating which group each volunteer is in has been added to the dataset. That variable is named TREAT, and the updated dataset, named GunkPre.sav, has been posted on the course website.
Now it is time to begin getting to know the data better by providing preliminary answers to two basic questions: (a) What patterns of association exist among the baseline variables? And (b) Do there appear to be baseline differences on baseline variables between the intervention and control groups? We do not need statistically compelling answers to these questions at this point; that is, we will not yet concern ourselves with p-values or confidence intervals. Rather, we are looking for informal impressions based upon cross-tabulations, multiple-group descriptive statistics, and correlation coefficients.
Please download the GunkPre2.sav dataset from the Practice Assignment 4 page of the Assignments section of this course.
Assignment Use SPSS to conduct analyses to answer the following questions, and write a report that describes your findings.
(1) Were females and males similar or dissimilar in their pre-intervention knowledge, selfefficacy, and motivation scale scores? And are they similar or dissimilar in their pre-intervention LDL levels? Two answer these questions, present (a) two-group boxplots of the knowledge, self-efficacy, and motivation scales, and of LDL levels, and (b) a cross tabulation of sex by high vs. not high LDL.
(2) Were race/ethnic groups (non-Hispanic Blacks, non-Hispanic Whites, non-Hispanic Asians, non-Hispanic Others, and Hispanics/Latinos) similar or dissimilar in their pre-intervention knowledge, self-efficacy, and motivation scale scores? And are they similar or dissimilar in their pre-intervention LDL levels. To answer these questions, present (a) five-group boxplots of the knowledge, self-efficacy, and motivation scales, and of LDL levels, and (b) a cross tabulation of race/ethnicity by high vs. not high LDL.
(3) This intervention is based upon the theory that knowledge, self-efficacy, and motivation all contribute to dietary behaviors that, in turn, influence cholesterol levels in the blood. Are the scale scores for knowledge, self-efficacy, and motivation correlated with measured LDL levels in ways that are consistent with this theory? To answer this question, present correlation coefficients between (a) knowledge and LDL levels, (b) self-efficacy and LDL levels, and (c) motivation and LDL levels. Briefly discuss the implications of your findings.
(4) What pattern of correlations exists among the knowledge, self-efficacy, and motivation scale scores? To answer this question, present a correlation matrix for these three variables, and describe briefly in words what you see.
(5) Ideally, randomization of volunteers into treatment and control groups should ensure that the two groups are comparable on all background variables. In our data, does it appear that volunteers from all demographic groups were equally likely to be assigned to the treatment as opposed to the control group in our study? Consider specifically: sex and race/ethnicity (NH Black, NH White, NH Asian, NH Other, and Hispanic/Latino). To answer this question, present a cross-classification of race/ethnicity with experimental group, and present a bar chart that shows the proportion of volunteers assigned to the treatment and control groups by sex. Comment briefly on what you observe. Considering that volunteers were randomized to the treatment and control groups, do these findings surprise you?
(6) Similarly, were volunteers with higher levels of LDL more or less likely than those with low LDL to be assigned to the treatment as opposed to the control group? Examine this issue in two ways. First, use the raw LDL scores. Present a table of descriptive statistics according to experimental group for this variable, as well as boxplots of LDL for the treatment and control groups. Second, use the variable that categorizes volunteers according to whether or not their pre-intervention LDL was high (i.e., above 160 mg/dL).
(7) Finally, are volunteers assigned to the treatment group similar to those assigned to the control group in terms of their knowledge, self-efficacy, and motivation scale scores? To answer this question, present two-group boxplots for each of these three scales, and comment on what you see.
As you complete this assignment, please remember that we are not yet conducting statistical hypothesis tests to determine whether observed associations or group differences are in the range of what would likely occur due solely to chance. At this point, we are simply “eyeballing” the data and giving our impressions.
Submission In the Assignments section, please upload the following items to the Practice Assignment 4 page 24 hours before live session 4: 1. Your SPSS syntax file containing commands for running all cross-tabulations, multigroup descriptive statistics, and correlation analyses, as well as all graphs, requested above. 2. Your report or memorandum presenting your results in tables and graphs, and discussing your interpretation of whatever patterns you observe in the data as a result of these analyses.