Exam One - ttests Flashcards
What do ttests do?
estimate the probability that the mean diff between two sets of values can be attributed to chance alone
What are the two types of ttests?
independent and paired
independent t tests are between two ________- samples
different/unrelated
paired t tests are between two ____- samples
similar/related (pre/post)
if p<a, we ___ that u1 = u2 and conclude that the two samples are significantly different
reject
When do we use ttests? to compare…
- one nominal or ordinal variable
AND - one continuous (measurement) variable
1 ttest compares ____ dependent variable. if the study has 10 DVs, the _____ ttests are ran
1, 10
What kind of ttest is this example?
standing balance (s) in people with and without peripheral neuropathy?
independent ttest
What kind of ttest is this example?
standing balance (s) and functional reach (cm) in people with and without peripheral neuropathy
2 independent ttests
What kind of ttest is this example?
knee abduction excursion (degrees) during a drop landing with and without an ankle brace
paired ttest
What kind of ttest is this example?
thigh and calf circumference (cm) and among adolescents before and after a strengthening intervention
2 paired ttests
What are 2 assumptions data must have for ttest to be valid?
1 - equal variance within samples
2 - normally distributed
equal variance within groups is represented by the ______ for each group
SD
if the data is normally distributed it should not be __________
skewed left or right
ttests may be _____if the data has unequal variance or non-normal distribution
misleading
Why don’t we do tons of ttests for the study?
it increases the likelihood of type one error study-wide
if you reduce the alpha (to account for a ton of ttests) this will ______ the study power
decrease
bonferroni correction
- common but conservative method to prevent increase type one error when doing multiple ttests
- take the alpha and divide it by # of comparisons and then thats the alpha level for each comparison
Turkey and sheffe adjustments
adjust alpha to account for multiple comparisons (ttests) but are less conservative than bonferroni
- as such they are both more powerful than bonferroni
LSD (least significant difference)
- very powerful, doesn’t adjust alpha, basically says fuck it
- for the statistically liberal