Database Reference
In-Depth Information
can sometimes yield an answer that is misleading. Among our examples in that chapter,
we considered which hypothesis we should believe about the population mean-satisfac-
tion-level based on what all potential customers would report for a particular design; in
essence, whether we should accept or reject a speciic hypothesis.
We considered whether to believe that the mean satisfaction with a modiied
design was higher than the mean satisfaction with the old design, which was 4.1 on
a scale of 1 (not at all satisied) to 5 (extremely satisied). Formally, we were testing
two hypotheses, the null hypothesis (H0) and the alternate hypothesis (H1):
•
H0: The true mean satisfaction with the modiied design is no higher than 4.1.
•
H1: The true mean satisfaction with the modiied design is greater than 4.1, that
of the original design.
As we noted in Chapter 1, the statistician would write these hypotheses more
succinctly, by deining the true mean satisfaction with the modiied design as “mu”
(the Greek letter, μ)—which, of course, we don't know and likely will never know,
and writing:
H0: μ ≤ 4.1 (the modiied design mean satisfaction is not above that of the
original design—weep!!)
H1: μ > 4.1 (indeed, the modiied design mean satisfaction is above 4.1; there
has been an increase in mean satisfaction—yeah!!)
We noted in Chapter 1 that Excel or SPSS was each able to take the raw data
gathered during a typical usability test and put together an analysis that tells us which
hypothesis (H0 or H1) to believe. To reach its conclusion, the software performed a
“one-sample t-test.”
2.4
INDEPENDENT SAMPLES
In this chapter we introduce the reader to an easy extension of the above type exam-
ple, and present the situation in which we have
two unknown means
that we wish to
compare, based on
two sets of data
. The software will still do all the work, and it is
likely not a surprise to the reader that what the software would be doing is called a
“two-sample t-test.”
In the case of Mademoiselle La La, you have to compare the means of two indepen-
dent samples, and determine the one, if either, that has the higher perception of sophis-
tication. To do so, we will create two hypotheses in an effort to decide between the two:
H0: μ1 = μ2
(The two designs do not differ with respect to mean sophistication)
H1: μ1 ≠ μ2
(The two designs do, indeed, differ with respect to mean sophistication)
The software will easily calculate which hypothesis is correct. If the second
hypothesis is determined to be true, then we can conclude that the means are
indeed different. As a consequence, we can conclude that the mean perceptions of
Search WWH ::
Custom Search