Exam 2 Flashcards
T tests
When there are only 2 levels in a test with 1 independent variable
ANOVA
When you have more than 1 independent variable and more than 3 levels
What type of stats are T-tests and ANOVA?
Parametric inferential stats
What are the basic assumptions for all parametric tests?
Samples are randomly drawn from populations
Normal distributions
Homogeneity of variance within groups
Continuous data (ratio and interval)
Factor
Variable manipulated by experimenter (independent variable)
Level
Groups/conditions being compared
Repeated measures factor
Within subjects
Non repeated factors
Between subjects
One way ANOVA
Compare means of 3+ groups with 1 independent variable with 3+ levels
Two way ANOVA
Compare means of 3+ groups with 2 independent variables with 3+ levels
Repeated measures ANOVA
Repeats measures overtime accounting for successive measurement in the same individual (within subjects)
What has more statistical power than one way ANOVA?
Repeated measures ANOVA
Mixed design ANOVA
Both between and within subjects (two way ANOVA with one repeated measure)
What are mixed design ANOVA most commonly used in?
RCT
Type 1 error
When you say there is a change but you should not have
False positive
Rejecting the null when you should not have
Liar
Type 2 error
When you say there was not a change but there was
False negative
Fail to reject the null when you should have
P value
Score you get that you compare to alpha (if less than alpha than statistically significant, if more than not statistically significant)
When should you perform a post hoc?
After ANOVA
What does a post hoc test?
Every difference so it is exploratory (most common)
Bonferroni calculation
Divide alpha by number of planned comparisons
What is the goal of the bonferroni calculations?
Decrease family wise error rate
What must the bonferroni calculation balance?
Type 1 and type 2 error
Between subjects
Two groups doing two separate tests (vertical jump one with shoes one without)
Within subjects
Same group doing two different tests (same group does vertical with shoes and then without)
Disordinal interaction
Lines cross
Cannot interpret significant main effects
Ordinal interaction
Lines don’t cross
Can interpret significant main effects
What does lines parallel mean?
No interaction
What does lines not parallel mean?
Interaction
What are the assumptions for a repeated measures ANOVA?
Sphericity- homogeneity of variance of differences
Test with mauchlys test of sphericity
Non significant finding means no difference in variance
If fail assumption use correction/ adjusted p value
What is the type of statistical analysis used in a RCT?
2x2 ANOVA- mixed design
What is an ANCOVA example?
1 independent variable with 2 levels
Dependent variable is continuous
Covariant
When you cannot control a variable through research design and sampling
Covariate
Can effect the outcome of the study but is not of interest in a study
What are the basic assumptions of ANCOVA?
The usual ones plus
Linear relationships between covariant and dependent variable (as one variable changes so does the other)
Homogeneity of slopes (relationship between covariate and dependent variable is the same for both groups)
What are the rules of the covariate in ANCOVA?
Must be associated with the dependent variable and relate to each group in the same way and should be measured/assessed prior to the start of the intervention
What are all covariates highly correlated to?
A dependent variable and not correlated with one another
What is more powerful ANOVA or MANOVA?
MANOVA
Most randomized controlled trials include which statistical analysis?
Mixed model ANOVA
What is an example of a one way ANOVA?
Measuring stress during meditation of DPT students and they are all tested in 3 positions
When is bonferroni used?
When an ANOVA is statistically significant
What is the F statistic in ANOVA?
Effect size (larger F stat= larger effect size)
Which is most powerful tests?
Paired t test
Repeated measures ANOVA
What does sphericity apply to?
Repeated measures ANOVA (all other assumptions are the same in both ANOVAs and T tests)
If different between groups exists before data collection what should be considered?
ANCOVA
When should you use a MANOVA?
Where there is multiple dependent variables (outcomes)
When is an ANOVA more likely to be statistically significant?
When F is bigger
When between group variance is bigger than within group variance what does this mean?
It is more likely to be significantly significant
What do repeated measures ANOVA and paired t test have in common?
Within group
Why would you use an ANCOVA?
When there is differences between groups at baseline
If a question says data is skewed or bimodel that means what?
Nonparametric
What is more powerful between parametric tests and nonparametric tests?
Parametric testing
What is used for nonparametric testing?
Median
What is used for parametric testing?
Mean
What is the non parametric test or an unpaired t test (independent) (between subjects)?
Mann Whitney U test
What is the non parametric test or a paired t test (dependent) (within subjects)?
Wilcoxon signed ranks test
What is the non parametric test or a one way ANOVA?
Kruskal Wallis one way ANOVA
What is the non parametric test or a repeated measures ANOVA?
Friedmans