lecture 8 Flashcards
what is a statistical technique that predicts an outcome?
regression
for a regression the predictors can be ____ , ___ or____
I/R , ordinal or nominal
what type of regression analyses has one predictor (any level of data) and one I/R outcome.
linear
what type of regression analyses has multiple predictor (any level of data) and one I/R outcome.
multiple linear
what type of regression analyses has One or
more predictors (any level of data) and one categorical outcome, two levels only.
logistic
what type of regression analyses has One or
more predictors (any level of data) and one categorical outcome, multiple levles
multinomial logistic
ex: does GPRE score significantly
predict PT school GPA?
what type of regression would u run
linear bc one predictor and one I/R outcome
Ex/ Does age predict days to
clearance following a concussion in
adolescent athletes?
what type of regression would u run
linear bc one predictor and one i/r data
Ex/ Does disease category predict
distance on the 6MWT?
what regression would u run
lienar bc one predictor and one I/R outcome
if we are dong a pearson correlations and we state for our scientific hypothesis that “There will be a significant relationship between time 1and time
2 at r >+.80 or r<-.80. what are we saying we believe ??? rom is being recorded
that the ROM recorded at test 1 can predict 64% of the variance at time 2
would u reject the null hypothesis if the p= < .001
yes bc since the p=<.05 (the alpha) it is saying that there is a significant difference so u reject the null
what kind of test is a dublin watson test
assumption test for linear regression
what is a concurrent validity study
testing 2 tools ( neither fro the gold standard) and u want to see if there is a relationship between them
if the relationship is strong is the effect size small or larger
larger
assumptions for regression…. data must be ____ not curvilinear (scattterplot)
linear
another assumption for regression … the data much have ____
normality and look at histograms , skewness, kurtosis , box plots and shapiro wilk test
when is the skewness/kurtosis a problem for normality
> +2 or < -2
when looking at a shapiro wilk test for normality what kind of result do we want compared to the alpha of .05
a non significant results
if the skewness is .-281 and .337 is there a problem
no bc only a problem is >2 or < -2
if the shapiro wilk test is .012 is that okay
NOOOOO bc we don’t want it to be significant bruhhhh
if the shapiro wilk test is .012 but the sample is small does it matter
no not really bc it is designed for a large sample
another assumption for regression is that the data must have ______
homoscedasticity
In relationship designs, the variance of the
outcome variable should be approximately the
same at all levels of the predictor variable… what is this called
homoscedasticity
when do u want HOV
for a independent t test
what test is used for HOV
levenes test (used when testing for differences)
do we want the levenes test to be significant
no no no no
levenes test if the variances in different groups are approximately the ___
sane
what is an example of a problem with homoscedasticity
if the spread to the data is no even around the best fit line and if it goes into t funnel shape
another assumptions for regression is that the data must be free of influential ____
outliers (cooks distance )
a point is considered an outliear if the cooks distance is what
greater then 1 - then we have to eleimaite one participant
another assumption for regression is that all data must be ____ of each other
independent ( participants cant be influenced by each other and no trends over time ) (dburbin watson test )
what is one way to check for independence of observations for regression
durbin watson test
what are the values ranges for the durbin watson test and what is it checking for
0-4 (2 is perfrect) .. checking for possible correlations between the participants which would violate our assumption
when running the regression analysis what does the R squared mean
then the % of y could be predicted from X
Ex/ Does UG GPA and GRE score significantly
predict PT school GPA?
this is an exmaple of what kind of regression
multiple linear bc multiple predictors and one I/R outcome
Ex/ Does age and sport predict days to
clearance following a concussion in adolescent
athletes?
this is an exmaple of what kind of regression
multiple linear bc multiple predictors and one I/R outcome
what is the new assumptions for multiple regression
multicolinearity
do u want to have multicolineartiy
no
for multicolinearityi for multiple regression the VIF would be ___ and the tolerance (TOL) should be what
VIF- should be <10
TOL- should be >.1
Ex/ Does TUG score predict fallers and non-
fallers?
this is an example of what type or regression
logisitc bc one or more predictors (tug) and one categorical outcome , 2 levels only (fallers and non fallers)
Ex/ Does a score on the VOM (Vestibular
Ocular Motor) Screening test predict return to
sport in the next 30 days? (yes/no)
this is na exmaple of what type of regression
logisitc bc one or more predictors (VOM) and one categorical outcome , 2 levels only (yes or no)
Ex/ Does age and type of cardiac procedure
(value repair vs bypass) predict discharge
location to home vs. other?
this is an example of what type of regression
logisitc bc one or more predictors (age and type of caradic procedure) and one categorical outcome , 2 levels only (home vs other)
