Download the documentation - True BASIC

T-tests and Confidence Intervals 47
The test returns:
t
p
mean
se
df
the T-statistic
significance probability of t
mean value of data()
standard error of the mean
degrees of freedom
Ttest2 (a(), b(), null, alt, t, p, mean, se, df)
Two sample T-test with assumption of equal variance for a() and b(). You must supply
the datasets in a() and b(), and the null and alternate tests in null and alt. The null
hypothesis is that the true mean difference of the populations equals whatever you
passed in null.
The alt parameter lets you construct one- or two-sided tests: a negative number means the
alternative is that the true mean is less than null; 0 means that it’s not equal; and positive
means that it’s greater. Thus the test is two-sided if alt = 0; otherwise it’s one-sided.
Example: Ttest2 with null = 17 and alt = –1 gives a one-sided T-test. The null hypothesis
is that the difference of population means is 17. The alternative hypothesis is that
the mean is less than 17.
The test returns:
t
p
mean
se
df
the T-statistic
significance probability of t
mean of a() minus mean of b()
standard error
degrees of freedom
Use Ftest, described at the end of this section, to check that two normally-distributed
datasets come from populations with equal variances. If your datasets don’t have equal
variances, use Ttest2Var instead of Ttest2.
Ttest2Var (a(), b(), null, alt, t, p, mean, se, df)
Ttest2Var gives a two sample T-test without an assumption of equal variance for a()
and b(). You must supply the datasets in a() and b(), and the null and alternate tests in
null and alt. The null hypothesis: the true mean difference of the populations equals
whatever you passed in null.
The alt parameter lets you construct one- or two-sided tests: a negative number means
the alternative is that the true mean is less than null; 0 means that it’s not equal; and
positive means that it’s greater. Thus the test is two-sided if alt = 0; otherwise it’s one-
01/01