SPSS Study Description: List the Research Question

SPSS

Study Description:

List the research question for the study. The researcher is interested in looking at whether or not an appraisal from a person's manager regarding their job performance affects the person's self-esteem as measures on a self-esteem appraisal survey (SEM). According to the vignette while the researcher has managers give either a positive rating or a negative rating; he/she has no pre-experimental hypotheses regarding of how these specific appraisals will affect self-esteem (other than appraisals may affect self-esteem) and therefore the directionality of positive or negative ratings does not appear to be important in terms of the hypothesis.

Ma = Mb (Here the researcher is interested in the change on a measure of self-esteem from pre-to post intervention). Note to Customer: I'm using M. To represent mean here; it would also be correct to use the variance of the groups).

H1: Ma ? Mb

Variables. The independent variable is the feedback from the manager (either positive or negative), an ordinal discrete variable (if we assume a direction from negative to positive, but it is treated as nominal in the analysis). The dependent variable is a score on a measure of self-esteem, particularly the change in the score on the measure of self-esteem (SEM) following the intervention. The dependent variable is a ratio level discreet variable (considering you can score 0 on it). An interesting note is that the IV has two levels (positive feedback or negative feedback, a between subjects variable), but according to the vignette the researcher is not hypothesizing that positive feedback will increase self-esteem or negative feedback will lower it. The researcher is only interested in whether or not feedback changes self-esteem from a pre-to post score.

4. There are 50 individuals in the study; 25 per group according to the description.

The overall measures of central tendency in the analysis (mean and median values) indicate relatively normal distributions of overall pre and post scores on the measures of self-esteem. The overall pre-SEM mean score is was 76.52 with a median of 74 (there were multiple modes). The range was 62-92. The standard deviation (8.12) indicates that scores were adequately distributed and the he skewness (.267; standard error .337) indicates a slight positive skew but that is insignificant (skew/standard error; Creswell, 2012). The post SEM score is a mean of 71.3, the median of 70, and a mode of 70 (basically a normal distribution) with a range of 42-98. The standard deviation is a bit larger meaning more dispersion in the distribution (13.93). There was a slight negative skew although this is not significant (skew = -.169: standard error, .337).

The pre-SEM score for the positive evaluation group indicated a mean of 78.08 and the median of 80 indicating a problem normal distribution. The range was 65 to 92. Standard deviation was 8.67 with skewness .146 (slightly positive but not significant; standard error = .464). The post-mean score for the positive group was 82.24 with a median of 83 and a standard deviation of 8.05 indicating a relatively normal distribution. The range was 69 to 98. Skewness was slightly positive (.233; standard error = .464) but not significant.

As the pre-SEM mean score for the negative group was 74.96 with the median of 74 and a standard deviation of 7.37 indicating a relatively normal distribution and could spread. The range was 62 to 90. Skewness was slightly positive (.257; standard error, .464) but not significant. The post mean for the negative group was 60.36 with the median of 61 and a standard deviation of 9.06. The minimum was 42 to 73 the skewness of -.645 (standard error, .464) indicating nonsignificant negative skew.

In summary the distributions were relatively normal for the overall pre -- and post-SEM scores and four the pre -- and post SEM scores in each specific intervention. Levene's test was not significant (Pretest, F (1, 48) = 2.73, p = .105; post test F (1, 48) = .172, p = .68), so homogeneity of variance can be assumed and no corrections to the degrees of freedom need be taken in the analyses (Jackson, 2012). Observed power was 1.0 in all analyses.

4. A mixed ANOVA was conducted, even though the researcher was not interested in the effects of positive or negative evaluations. An ANOVA is a considered a two-tailed test; however, the F. distribution is technically only one-tailed (Runyon, Coleman, & Pittenger, 2000). In the repeated measures analysis the results indicate that there is a significant effect between pre and post tests on SEM; F (1, 48) = 67.74, p < .001. The effect size ( = .585) indicates a very robust effect. The between subjects effect of appraisal type was significant F (1, 48) = 30.49, p < .001. The effect size ( = .388) indicates a very robust effect. The interaction of test by appraisal type was also significant, F (1, 48) = 218.74, p < .001. This means that scores on the SEM measure varied depending of the appraisal type (Tabachnick & Fidell, 2012). Positive appraisals were associated with more positive post SEM scores and vice versa. The effect size ( = .820) also indicates a very robust effect.

5. The researcher's hypothesis that the evaluation would affect SEM was supported by the repeated measures ANOVA. The omnibus ANOVA result tells us that there is a significant finding; however, there are many other considerations here. First, since the researchers only interested in a pre-post effect there's no reason to do post hoc tests (only two comparisons in the omnibus test; Jackson, 2012). We can look at the means pre -- and post and determine that there is a difference after we know that the omnibus test is significant. We can also look at the means of the different appraisal groups and see that there was a significant difference in the positive and negative appraisal groups on the pre-measurement of SEM. It would be interesting to see if that difference was significant. Next we can look at the means of the within group changes for the appraisal differences and see that those who had a positive appraisal scored higher and those in the negative appraisal scored lower compared to their pretest results. Analyses here determine the magnitude of this effect which is probably significant. The interaction of the appraisal and SEM indicates that at least immediately following appraisal people's score on the self-esteem measure follows the direction of the appraisal. However, we really don't know how much control in the appraisal itself occurred. Did everyone get the same positive or negative appraisal? This would be something to discuss with the researcher.