x*******i 发帖数: 1237 | 1 To compare k independent samples, i.e. to test the association between two
ordinal variables) e.g. 5-point likert scale variable vs 4-level group
variable
I used NPAR1WAY to provide a Kruskal-Wallis test and specified the SCORES=
DATA option to use the input data observations as scores.
Is there any option in NPAR1WAY that I can request a rank-ordered Kruskal-
Wallis,,in order to take advantage of the rank of the groups? |
|
i**f 发帖数: 1195 | 2 ??
npar1way is able to come up with ttest output with anova option but i don't
think it is appropriate to say 'if the data don't follow normality, you can
do the t-test using npar1way'. In that case, people will go for wilcoxon.
you can still use proc ttest even data is not normal, but is it a correct
way to do that? |
|
|
y***6 发帖数: 46 | 4 If the data don't follow normality, you can do the t-test using non-
parametric technique. In SAS, try proc npar1way for more details. |
|
g******n 发帖数: 339 | 5 You did not specify your problem clearly.
What is your goal? Comparing two groups by conducting a hypothesis testing?
If this is the case, I would think about nonparametric test
such as Kruskal-Wallis test(SAS procedure NPAR1WAY).
is |
|
o******6 发帖数: 538 | 6 You need investigate the normal distribution assumption of these two
variables.If the assumption is satisfied ,you can use paired comparison.
PROC NPAR1WAY should be used if it is not satisfied.
significance |
|
z**********i 发帖数: 12276 | 7 Wilcoxon-Mann-Whitney test?
proc npar1way data = wilcoxon;
class x;
var count;
run;
d x count
1 1 10
2 1 11
...
1 2 9
2 2 9
...
The results suggest that there is a statistically significant difference between the underlying distributions of the count of x1 and the count of x2 (z = ***, p = ***). |
|
|
T*******I 发帖数: 5138 | 9 NPAR1WAY检验的是秩和间的差异,而非中位数间的差异。 |
|
h*********n 发帖数: 278 | 10 PROC LOGISTIC does not provide the KS test. However, you can obtain this
test by using the EDF option in PROC NPAR1WAY to analyze the predicted
values in the OUT= data set from PROC LOGISTIC. This is discussed and
illustrated in this usage note. |
|
|