tions. (h) false: The SE goes down to 0, but the
SD stays about the same.
4.(a) A paired ttest: the same authors are used in each
of the two samples.
(b) Assumptions are:
linearity of the relationship,
and equal variance. The linearity assumption is
clearly violated: there appears to be a curved re
lationship. In light of this, looking at the homoge
neous assumption is a little bit irrelevant. But the
variability of the points around the
curve
seems
to be quite homogeneous.
(c) i.
An independent sample ttest (forcing or not
forcing equal variances, it shouldn’t matter).
ii.
Try to transform the data and use a ttest
on the transformed data, if a transformation can
be found to make the samples look normally dis
tributed. If the transformed data have very dif
ferent variances, use the ttest that does not force
equal variances. Use the MannWhitney test if no
transformation can make both samples look nor
mally distributed.
iii.
Use a ttest.
The rather large sample size
makes it okay even if the data are not normally
distributed.
5(a)
¯
KMN
+
¯
KS
2
= 2
.
4, then 2
.
40

¯
KW1 = 0
.
165.
The standard error for this contrast is
√
.
567
*
1
15
+
4
15
+
4
15
=
.
2381. We get a tvalue of
t
=
0
.
165
0
.
2381
=
.
693 on df= 126.
Using 100 df in the
table, we get that the pvalue is
> .
20 so we fail
to reject the null hypothesis that the true average
pulse rate of crickets at KW1 is equal to the mean
This is the end of the preview.
Sign up
to
access the rest of the document.
 Fall '08
 Staff
 Normal Distribution, Variance, Student's ttest, Bonferroni, 26.92 cm

Click to edit the document details