COMPARE <Sample_1>,<Sample_2>,...,<Sample_s>,L
performs statisticals tests for samples from various Survo data sets.
L (optional) is the first line for the results in the edit field.
Each sample is referred to by <name_of_data>(<name_of_variable>). If the
data set consists of one variable, the name of variable may be omitted.
Typically, COMPARE is intended for small sample comparisons.
The CASES, IND and SELECT specifications are not valid in COMPARE operation.
Use FILE COPY, for example, to form samples from larger data sets.
Example:
COMPARE FILE1(Test3),FILE2(Test4),X,FILE4(Test7),11
COMPARE computes and displays relevant statistical summaries
and gives the test statistics usually with critical levels (P values).
In many cases also randomization tests are performed by simulating
the sampling process under the null hypothesis. Then the true critical
level will appear as a limit for the proportion of those cases where
the null hypotheses is rejected more clearly than in the case observed.
Also the standard error of the critical level is displayed. The user
can interrupt simulation whenever by a single keystroke.
COMPARE:
The testing procedures are selected according to the number of the
samples s and an optional TEST specification.
1 = Comparison of 2 or more independent samples
2 = Comparison of 2 paired samples, rank correlations
3 = Tests for normality
4 = Quantile-Quantile plots
5 = Specifications for randomization
S = More information on statistical tests
An interactive procedure for comparison of 2 samples is provided by the
/COMPARE sucro. It selects the tests automatically and "interprets"
the results. Usage: /COMPARE <sample1>,<sample2>
Help System (web edition)