T-Test Flashcards
(28 cards)
when can a t-test become applicable
if comparing two treatment variables’ effect on the dependent variable
treatment A vs treatment B on desired outcome
treatment variables are also known as
independent
main differences observed for a t-test
variables are compared between response or difference over time
how do multiple treatment variables affect the alternative hypothesis
turns the statement into a comparison of the treatment variables
ie… normal = difference in walking when doing x treatment
new = difference in walking when doing x compared to those doing y
between group variance
difference between patients of two groups after they receive assigned treatment
within group variance
difference within same group before and after treatment
what is the basis of a t-test
difference in the mean of dependent variable (outcome variable) between the groups
ie - comparing walking differences between those who did treatment x vs treatment y
if the t-test determines that it is significantly different than zero difference, one should
accept alternate hypothesis
ie - outcome difference between treatments was significant, therefore one treatment was better
if the t-test determines that it is not significantly different than zero difference, one should
accept the null hypothesis
ie - the difference between outcomes was not significant, there was not a better performing variable of the two
how to find degrees of freedom
- equation
[(n1-1)+(n2-1)]
explain the degrees of freedom equation
n1 = # of pts in first group
n2 = # of pts in second group
one tailed t-test
specifies the direction of the difference of interest
ie - tells if one independent variable (treatment) is better than the other(s) being tested
two-tailed t-test
test of any difference between groups, regardless of the direction of difference
compare one and two-tailed t-tests
one = see if one variable is better than the other
two = see if the variables have a (+ or -) effect
define paired t-tests
comparison of mean scores from the same group of people being assessed at two different time points
how else can paired t-tests be used
if there is a comparison made for samples of individuals that are paired in some way
(siblings/daugters or sons)
define independent sample t-test
difference between mean scores of two different populations
explain difference between paired t-test and independent sample test
paired = mean of a treatment prior and post in a specific population
indep = treatment group vs control group
equal variance
aka pooled variance
same number of people in both groups
unequal variance
aka separate variance
unequal number of people between the groups (statistically different variance between two groups)
explain confidence intervals
we are ____% confident that the mean for the outcome (dependent) variable for x treatment variable (independent) is within _______ range.
explain sample size and statistical difference
small sample
= people in the sample exhibit more significant variability
–> may not yield statistical difference
large
= small differences are shown as statistically significant
short-comings of statistical difference
fails to provide interpretation about magnitude of the difference’s importance
–> does not provide interpretation of results on clinical significance
relationship between statistical and clinical importance
square rectangle idea
need statistical difference to establish clinical significance
– can’t be the other way around
