ERIC Number: ED330707
Record Type: Non-Journal
Publication Date: 1991-Apr
Reference Count: N/A
Applying the Generalized Sequential Bonferroni Method to a Set of Planned and Post Hoc Comparisons.
Edwards, Lynne K.
This study analytically examined the multiple comparison procedures for testing a combined set of planned and post hoc comparisons. To establish a practical guideline, the following issues were investigated: (1) when the generalized sequentially rejective Bonferroni method (GSRB) is and is not applicable; (2) the required alpha levels for the GSRB compared with six alternatives (classical Bonferroni (CB)-CB, CB, sequentially rejective Bonferroni (SRB)-SRB, SRB, planned comparison-SRB, and a procedure proposed by R. Rosenthal and D. B. Rubin (1984); and (3) the optimal weighting system for the GSRB. When a combined set of planned and post hoc comparisons is tested in the same study, the GSRB can be used with W, which represents the weight for each planned comparison set to n, the number of post hoc comparisons (i.e., W=n if n is greater than or equal to 2, otherwise W=2 and W=1 for the post hoc comparisons). If the planned comparisons are orthogonal and each is tested at the per comparison level, the GSRB is not recommended. Instead, the SRB can be applied to the post hoc portion of the comparisons. A numerical example with data for four groups was used to illustrate seven multiple comparison procedures. Although the procedures of controlling for the inflated Type I error arising from the multiplicity of tests remain controversial, the GSRB seems promising in conducting a combined set of planned and post hoc comparisons. Four tables contain illustrative data. (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Sponsor: Minnesota Supercomputer Inst.
Authoring Institution: N/A