Measurement Scales (2 of 6)
Displays were rated on a 7 point scale where a 1 was the lowest rating and a
7 was the highest rating. This rating scale is only an ordinal scale since there
is no assurance that the difference between a rating of 1 and a rating of 2 represents
the same degree of difference in preference as the difference between a rating
of 5 and a rating of 6.
The mean rating of the color display was 5.5 and the
mean rating of the black and white display was 3.9. The first question the experimenter
would ask is how likely is it that this big a difference between means
could have occurred just because of chance factors such as which subjects
saw the black and white display and which subjects saw the color display.
Standard methods of
statistical inference can
answer this question. Assume these methods led to the conclusion that
the difference was not due to chance but represented a "real" difference
in means. Does the fact that the rating scale was ordinal instead of
interval have any implications for the validity of the statistical
conclusion that the difference between means was not due to chance?