NCS Statistical Analysis Flashcards
what are common incorrect assumptions in statistics
you can use popular comparison tests aka t tests
groups are equal at baseline
groups are normally distributed
what are the 4 types of data
nominal, ordinal, interval, ratio
male or female is an example of what kind of data
nominal data
why is nominal data not great for clinical research
it is less sensitive and more difficult to demonstrate significant change
pain NRS or MMT grades are examples of what kind of data
ordinal data
mile markers or temperature in C or F are examples of what kind of data
interval data
weight, temperature in K, 0-100% scoring on a test are examples of what kind of data
ratio data
what type of data is the easiest and most sensitive
ratio
Bob collects data for his study, this is an example of what kind of data
primary data
Bob collects data for his study. A few years later, he then uses that same data for another study. this is an example of what kind of data
secondary data
Bob is looking to find out if there is a trend between patient admissions and adequate staffing in the ED over the past 5 years using EMR. This is an example of what kind of data
secondary data
what is continuous data
ratio or interval data
what is categorical data
ordinal or nominal data
what are the parametric assumptions
observations are independent
the data is normally distributed
variance is the same within groups
can only be used for ratio or interval data
true or false: it is easy to demonstrate meaningful effect sizes in nonparametric data
false - it is easier in parametric data
Tyler falls into the 75th percentile of students in his cohort. this is an example of parametric or nonparametric data
nonparametric (ranks/proportions)
according to central limit theorem, a sample distribution becomes more normal or abnormal as the sample size increases?
abnormal
according to central limit theorem, at what sample size/n can you not assume normal distribution
n > 30
you are looking to determine if there is a significant difference between data sets, what tests can yo u use
kolmogrov-smirnov or shapiro-wilks
Bob is conducting a study for a new intervention. His new intervention demonstrates clinical improvement but he does not recruit enough participants to reject his hypothesis that it is better than the standard of care. This is an example of what kind of error
type 1 error
Bob is conducting a study for a new intervention. His new intervention is not better than the standard of care but in his analysis he chose the wrong test and it produced a statistically significant finding. This is an example of what kind of error
type 2 error
Bob is analyzing nonparametric data using a test of proportions. what test should he use?
Chi squared
Bob is analyzing outcomes for an intervention group and control group that was not matched, what should he use if the data is normally distributed
parametric t test
Bob is analyzing outcomes for an intervention group and control group that was not matched, what should he use if the data is not normally distributed
nonparametric mann whitney u test
Bob is analyzing outcomes for an intervention group and control group that was matched, what should he use if the data is normally distributed
parametric paired T test
Bob is analyzing outcomes for an intervention group and control group that was matched, what should he use if the data is not normally distributed
non parametric wlcoxon test
Bob is analyzing outcomes for an intervention group, a standard of care group, and placebo control group, what should he use if the data is normally distributed
parametric 1 way ANOVA
Bob is analyzing outcomes for an intervention group, a standard of care group, and placebo control group, what should he use if the data is not normally distributed
nonparametric kruskal wallis test
Bob is analyzing outcomes for matched intervention group, standard of care group, and placebo control group in which there was a experimental period, a wash out period, and a second experimental period, what should he use if the data is normally distributed
parametric repeated measures ANOVA
Bob is analyzing outcomes for matched intervention group, standard of care group, and placebo control group in which there was a experimental period, a wash out period, and a second experimental period, what should he use if the data is not normally distributed
non parametric freidmans test
you are looking to find if there is an association between exercise time and age, what type of regression analysis will you do
linear regression
you are looking to find if there is an association between age, geographic location, and meeting physical activity guidelines, what type of regression analysis will you do
logistic regression
the odds ratio between smoking and hypertension is 3.5. which of the following are true:
a. smokers are 3.5x more likely to develop hypertension over non smokers
b. smokers are 250% increased odds of developing hypertension over non smokers
c. smokers are 2.5x more likely to develop hypertension over non smokers
d. a and b
e. b and c
d. a and b
the odds ratio between exercising and obesity is -0.75. which of the following are true:
a. exercisers are 0.75x less likely to develop obesity than non exercisers
b. exercisers are 1.33x more likely to not develop obesity than non exercisers
c. exercisers are 0.25x less likely to develop obesity than non exercisers
d. a and b
e. b and c
d. a and b
Bob is analyzing data from a control and experimental group to determine associations, assuming the data is normally distributed what type of test should he use
parametric pearson product
Bob is analyzing data from a control and experimental group to determine associations, assuming the data is not normally distributed what type of test should he use
nonparametric kendall tau
Bob is analyzing data from a waitlist, standard of care, and experimental group to determine associations, assuming the data is normally distributed what type of test should he use
parametric pearson product
Bob is analyzing data from a waitlist, standard of care, and experimental group to determine associations, assuming the data is not normally distributed what type of test should he use
non parametric kendall tau
Bob is analyzing the outcomes/associations of therapeutic exercise on neck pain vs waitlist control. assuming the data is normally distributed, what test should he use
parametric linear regression
Bob is analyzing the outcomes/associations of steriods, modalities, and therapeutic exercise on neck pain. assuming the data is not normally distributed, what test should he use
nonparametric logistic regression
Bob is analyzing the outcomes/associations of steriods, modalities, and therapeutic exercise on neck pain and NDI score. assuming the data is not normally distributed, what test should he use
discriminate analysis
Bob is researching different interventions for his patient. What statistic should Bob look at to determine if a particular intervention will be clinically meaningful to his patient?
a. p value of <0.05
b. p value of <0.001
c. effect size
c. effect size
Bob is researching different interventions for his patient. In reading an article about a particular intervention, Bob sees that the confidence interval crosses zero. how should bob interpret this
there is no difference between groups so the intervention may not be effective
Bob is researching different interventions for his patient. In reading an article about a particular intervention, Bob sees that the confidence interval is very wide. how should bob interpret this
a. this intervention is helpful and precise
b. this intervention is helpful but not precise
c. this intervention will not differ from the control
b. this intervention is helpful but not precise
Bob is researching different interventions for his patient. In reading an article about a particular intervention, Bob sees that the confidence interval is very narrow. how should bob interpret this
a. this intervention is helpful and precise
b. this intervention is helpful but not precise
c. this intervention will not differ from the control
a. this intervention is helpful and precise
the AIMS is an example of an outcome measures that uses what kind of data
a. nominal
b. ordinal
c. interval
d. ratio
a. nominal
the FIM is an example of an outcome measures that uses what kind of data
a. nominal
b. ordinal
c. interval
d. ratio
b. ordinal
the Stroke impact scale is an example of an outcome measures that uses what kind of data
a. nominal
b. ordinal
c. interval
d. ratio
c. interval
distance on the 6MWT is an example of an outcome measures that uses what kind of data
a. nominal
b. ordinal
c. interval
d. ratio
d. ratio
