Post Hoc or Multiple Comparison Testing To determine which group means are statistically different Ideally should test just what groups you want to compare particularly if small sample size Select type of post hoc testing ID: 770307
Download Presentation The PPT/PDF document "Post Hoc or Multiple Comparison Testing" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Post Hoc or Multiple Comparison Testing To determine which group means are statistically differentIdeally, should test just what groups you want to compare – particularly if small sample sizeSelect type of post hoc testing (eg, Tukey, Bonferroni, Scheffe) based on variance assumptions and how many comparisons are needed ** cannot just do multiple t-tests because it increases the risk of Type I error **
Examples: One-Way ANOVA Research Question: Is there a difference in confidence scores among those who are rarely, sometimes or often in a depressed state of mind?
Interpretation There is a statistically significant difference in overall confidence score between depression groups F(3, 676) = 111.46, p <.001. The post hoc testing indicated that there was a significant difference in confidence scores between those who were rarely depressed and all of the other groups. The group that is rarely depressed rated their confidence score (mean = 68.89) significantly higher than the sometimes (mean = 58.22) and often (mean = 46.23) groups. Depression Level N Mean (SD) Confidence Score F-Ratio and p-valueMultiple Comparisons Rarely SometimesRarely35668.89(10.05)F = 111.46 (p=.001) Sometimes26658.22(10.83)p = .001 Often4746.23(12.50)p = .001p = .001
Which of the following conclusions can be made based on the above data? A. There is a not a significant difference between the depression levels. B. The sometimes group reported a significantly higher mean confidence score as compared to the often group. C. The sometimes group reported a significantly lower mean confidence score as compared to the often group. D. There is not a significant difference between the mean confidence scores of the rarely and sometimes groups. Depression Level N Mean (SD) Confidence Score F-Ratio and p-valueMultiple Comparisons Rarely SometimesRarely35668.89(10.05)F = 111.46 (p=.001) Sometimes26658.22(10.83)p = .001 Often4746.23(12.50)p = .001p = .001
Nursing Staff N Mean (SD) Job Satisfaction Score F-Ratio and p-value MultipleComparisons RN LPNRN2045.0(3.95)F = 13.23 (p=.001) LPN2041.7(5.79)p = .258 UAP2036.2(6.37)p = .024p = .004 Levene’s Test = .5261.We see that the mean job satisfaction score for the RN group is 45.0 (SD 3.95), the LPN group is 41.7 (SD 5.79) and the UAP group is 36.2 (SD 6.37). 2. Levene’s test result=.526 which is not significant because p>.05 so we can assume that there is no significant difference between the variances of the two groups.3.The next value is the F-ratio which is 13.23 and the p-value corresponding to the F-ratio is .001. Since the p-value is <.05 (p=.001) we can reject the null hypothesis.4. The next step is to determine where the differences exist among the groups (i.e., RN vs. LPN, RN vs. UAP, LPN vs. UAP). RNs to LPNs – the value for p=.258 which is greater than .05, which means it is not significant, and you conclude that there is no difference in the mean job satisfaction scores of RNs vs. LPNs (45.0 and 41.7 respectively). RNs to UAPs – the value for p=.024 whichis less than .05, which means it is significant, and you conclude that there is a statistically significant difference in the mean job satisfaction scores of RNs vs. UAPs (45.0 and 36.2)LPNs to UAPs – the value for p=.004 which is less than .05, which means it is significantIn conclusion, the UAPs are less satisfied than both the RNs and LPNs.