Database Reference
In-Depth Information
CHAPTER
6
Comparing more
than two means:
one factor ANOVA
with independent samples.
Multiple comparison
testing with the
Newman-Keuls test
6.1
INTRODUCTION
As the researcher on a UX team, you are probably frequently asked to compare
more than two means. You could be comparing ratings of attractiveness of a spe-
ciic design across different age brackets. Or, you may be comparing more than two
task-completion times from a usability test. Often, the most common scenario of
comparing more than two means that a UX researcher confronts is assessing the
scores from Likert scales.
A typical Likert scale is a statement to which the respondents rate their level of
agreement. Usually a ive-point scale is used, and the two ends are labeled.
SIDEBAR: ODD OR EVEN NUMBER OF OPTIONS?
There is some debate whether it is superior to have an even or odd number of choices in a Likert
type scale. This debate has been going on for what seems forever. It is the type of issue in the
marketing research world about which “reasonable people disagree.”
Having an odd number of choices (e.g., 5) gives the responder a midpoint (i.e., 3) to focus on,
while an even number gives the user no midpoint (e.g., on a 6-point scale, the midpoint is 3.5). The
issue is whether having an able-to-be-chosen midpoint is a good thing or a bad thing.
Continued
Search WWH ::
Custom Search