MANOVA SPSS RESEARCH (Statistics Project Sample)

Hilary F. Hemmatipour (BS, MS) Department of Psychology, Walden University RSCH-8260 Cameron R. John (Ph.D.) July 30th, 2023 MANOVA in SPSS Research Question: Is any difference in fear of asking for help and motivation score across stat class types? Null Hypothesis: Ho: There is no difference in fear of asking for help and motivation score across stat class type factor. The most appropriate research design that aligns with our research question is a one-way multivariate analysis of variance (MANOVA), used to determine whether there is any significant difference between the independent variable on more than one dependent variable (continuous or scale variables). This test type differs from one-way ANOVA, which measures only one continuous dependent variable. It is better to understand type 1 errors when testing for significance using type 1 errors (Statistics, 2018). This means rejecting the null hypothesis when it must not be rejected. This includes believing that you have found something significant when nothing significant has occurred. The dependent variables include; fear of asking for help and motivation score, which should be measured at the ratio or interval level. In addition, the independent variable is Stat class type, categorized as experimental, traditional, and flipped. According to the level and measurement of the independent variable in MANOVA, it is assumed that Stat class type is measured as a categorical variable (Walden University, LLC. (Producer), 2017q). The SPSS output is displayed in the appendix section of the paper. Table 2 shows the descriptive statistics section shows the mean for fear of asking for help, where flipped stat class type is slightly higher than the experimental and traditional classes. This indicates that the flipped class had better confidence in asking for statistics help than the traditional and experimental classes. Still, on the dependent variable motivation score, students in the flipped class also had the highest mean value. In contrast, the traditional class had a slightly higher mean score than the experimental stat class (Walden University, LLC. (Producer), 2017o). Consequently, Levene's test of equity of variance is used to investigate whether or not the analysis of variance between independent variables are same, also termed homogeneity of variance, which indicates equal variance between groups. On the other hand, Box’s M Test is used to understand the equality of covariance derived between the groups. It tests the significance at α = .005 since the test is measured highly sensitively (Emerson, 2018). Box's Test results show a p-value higher than the alpha level. Thus, we assume some equality of variance. For the multivariate tests in Table 4, it is always considered the independent variable Stat class type based on the Pillar's trace or Wilks' Lamba. Pillar's trace is the least sensitive to violating the assumptions of covariate measures, that is, rejecting the null hypothesis when it is not true (Stukalin & Einat, 2019). From our study here, Pillai's Trace is slightly higher above the threshold of 0.05, with an F (2.218) and P (0.066). Looking at the significance of Wilk's Lamba is slightly lower than Pillai's Trace, F (2.225) and P (0.65). Since there is no significance for Pillai's Trace and Wilk's Lamba, it can be best to look at Roy's Largest Root, F (4.403) and P (0.013), which is statistically significant (Emerson, 2018). From the multivariate test, there is some statistically significant difference to reject the null hypothesis that the Stat class types are on the same level of performance based on fear of asking for help and motivation score. There is no statistically significant difference between fear of asking for help and Stat class type, but the motivation score is significant. Moreover, partial et square (η2) displays the assumptions of covariance and is also used as the effect size for the multiple analysis of variance model (Walden University, LLC. (Producer), 2017o). As illustrated in the table, there is a partial (η2 = 0.024) between class type and motivation score. Thus, this indicates a 2% variability between class type (experimental, flipped, and traditional) and motivation score. In the pairwise comparison displayed in Table 9, there is no significance between Stat class type and fear of asking for help. Also, from these multiple comparisons, the experimental and flipped class is significant (p=0.018) for the motivation score (Emerson, 2018). This reveals a slight difference of 8.4390 between the experimental and flipped classes. Hence, there are differences in fear of asking for help and motivation scores across stat class types. To conclude, we fail to reject the null hypothesis. In summary, statistical techniques are vital in social change science. Social statistics use statistics to study and understand social environments and human behavior. Statistical methods are compulsory for every applied social research effort to attain a non-questionable conclusion. It seems that scholar depends heavily on classical statistical techniques though they could aid in using advanced and new methods. References Emerson, R. W. (2018). MANOVA (Multivariate Analysis of Variance): An Expanded Form of the ANOVA (Analysis of Variance). Journal of visual impairment & blindness, 112(1), 125-127. Statistics, L. (2018). One-way Repeated Measures MANOVA in SPSS statistics. Statistical tutorials and software guides. Stukalin, Y., & Einat, H. (2019). Analyzing test batteries in animal models of psychopathology with multivariate analysis of variance (MANOVA): one possible approach to increase external validity. Pharmacology Biochemistry and Behavior, 178, 51-55. Walden University, LLC. (Producer). (2017o). Introduction to multivariate analysis of variance [Video file]. Baltimore, MD: Author. Walden University, LLC. (Producer). (2017q). MANOVA [Video file]. Baltimore, MD: Author. Appendix Table 1: Between-Subjects Factors Between-Subjects Factors Value Label N Stats Class Type 1 Experimental 46 2 Flipped 120 3 Traditional 169 Table 2: Descriptive Statistics Descriptive Statistics Stats Class Type Mean Std. Deviation N Fear of Asking for Help Experimental .4130 .49782 46 Flipped .4583 .50035 120 Traditional .4142 .49405 169 Total .4299 .49580 335 Motivation Score Experimental 37.4994 17.48708 46 Flipped 45.9385 16.63288 120 Traditional 42.2273 18.21095 169 Total 42.9075 17.72409 335 Table 3: Box Test of Equality of Covariance Matrices Box's Test of Equality of Covariance Matrices Box's M 2.358 F .388 df1 6 df2 159952.781 Sig. .887 Tests the null hypothesis that the observed covariance matrices of the dependent variables are equal across groups. a. Design: Intercept + ClassType Table 4: Multivariate Tests Multivariate Tests Effect Value F Hypothesis df Error df Sig. Partial Eta Squared Intercept Pillai's Trace .833 825.322b 2.000 331.000 <.001 .833 Wilks' Lambda .167 825.322b 2.000 331.000 <.001 .833 Hotelling's Trace 4.987 825.322b 2.000 331.000 <.001 .833 Roy's Largest Root 4.987 825.322b 2.000 331.000 <.001 .833 ClassType Pillai's Trace .026 2.218 4.000 664.000 .066 .013 Wilks' Lambda .974 2.225b 4.000 662.000 .065 .013 Hotelling's Trace .027 2.232 4.000 660.000 .064 .013 Roy's Largest Root .027 4.403c 2.000 332.000 .013 .026 a. Design: Intercept + ClassType b. Exact statistic c. The statistic is an upper bound on F that yields a lower bound on the significance level. Table 5: Levene’s Test of Equality of Error Variance Levene's Test of Equality of Error Variances Levene Statistic df1 df2 Sig. Fear of Asking for Help Based on Mean .938 2 332 .392 Based on Median .307 2 332 .736 Based on Median and with adjusted df .307 2 331.954 .736 Based on trimmed mean .938 2 332 .392 Motivation Score Based on Mean 1.856 2 332 .158 Based on Median 1.903 ...