Analysis, Interpretation, and Visual Presentation of Experimental Data
Geoffrey R. Loftus · 2002
Abstract This chapter has two main purposes: the first is to discuss serious problems with two generally accepted and widely used foundations of current statistical analysis in psychology, the linear model and null hypothesis significance testing; the second purpose is to describe some alternative valuable techniques. The alternatives are grouped into six categories: (1) use of sophisticated pictorial and graphical techniques for data display, (2) use of confidence intervals in numerous situations, (3) use of planned comparisons, emphasizing contrasts , (4) use of percent total variance accounted for, (5) representation of theoretical fits to data, and (6) use of equivalence techniques to investigate interactions. Detailed hypothetical numerical examples along with associated calculations and graphs are constructed to illustrate each of the techniques.