DQ Two week five
Post hoc statistical analysis is a statistical test that is conducted to identify the specific pairs of means where the difference is statistical significant. Post hoc statistical analysis is conducted when the results of an ANOVA test indicate statistical significance. When an F-test indicates significant difference exists among several groups, the significant difference may not exist between all the groups. Therefore, further analysis is necessary. In post hoc analysis, the mean of each group is compared to the mean of all the other groups.
There are three main ways of conducting post hoc analysis. They include; least significant difference Tukey method, least significant difference Scheffe method, confidence interval of difference between means. LSD Tukey method is the most commonly used because it is the strongest post hoc analysis test in controlling type 1 error. The LSD Tukey method tests all feasible pair wise comparisons separately. It assumes that all the groups have the same size. The LSD Scheffe method is more appropriate if the groups are of different sizes. Similar to the LSD Tukey method, this method tests all feasible pair wise comparisons separately. The computation of the least significant difference for each feasible pairs makes the LSD Tukey method and the LSD Scheffe method tedious.
Confidence interval of difference between means presents the differences the groups as confidence intervals. The confidence interval is given by +/-t*SE. The t-value is obtained using the residual degrees of freedom while the standard error (SE) is computed using the sample size of the two groups that are being compared and the residual mean square. Confidence interval is increasingly being preferred over significance testing because it allows the evaluation of differences as well as equivalence.
References
Healey, J. F. (2011). Statistics: A Tool for Social Research (revised ed.). London: Cengage Learning.
Jackson, S. L. (2011). Research Methods and Statistics: A Critical Thinking Approach (4 ed.). London: Cengage Learning.