T-Test Flashcards
3 forms
one sample, paired samples, indep samples
steps for sig testing
- 1.Set alpha (p value cutoff).
- 2.State the null and alternative hypotheses.
- 3.Calculate the test statistic.
- 4.Find the critical value (cutoff) of the statistic.
- 5.State the decision.
- 6.State and interpret the conclusion.
what types of study for one sample t-test
cross-sectional
does population mean differ from a particular value?
is mean BP of patients 140 mmHg?
paired sampled t-test type of study
before-after
matched pair (tumor from patient and nontomorous sample from same patient)
RCT
example of questions that answer paired sample t-test
is there change/difference in means before and after?
independent sample t-test what kind of studies
cross-sectional - compare two groups
case/control
Vitamin D concentration in men and women
independent sample t-test answers
is there change/diff in group means?
one sided t-test
one direction of effect. mean greater than some value
two sided t-test
difference can go either way
Assumption of unpaired t-test
Equal variance, normality of data
variance of vitamin D concentration in men is same as variance in women
why want equal variance
when high variability need a lot of data to conclude means are different
if assume equal variance between groups use_____ ttest
POOLED
otherwise saccterwire
for all t-tests formally assume
underlying population is normal
t-test if normality assumption violated
t-test not appropriate for smaller sample sizes
may still use for larger (CLT)
possible transformation
use non-parametric Mann-Whitney U test (for unpaired)
how to check normality in SAS
univariate
