EBP II Exam 2
You are studying a group of 100 people randomly assigned to one of 4 treatment groups (high intensity exercise, moderate intensity exercise, low intensity exercise, no exercise). You examine lower extremity strength with a dynamometer and balance using a standardized outcome assessment. You measure strength and balance at baseline, 2 weeks, 4 weeks, and 8 weeks. NUMBER OF FACTORS
2 factors (group and time)
You are studying a group of 100 people randomly assigned to one of 4 treatment groups (high intensity exercise, moderate intensity exercise, low intensity exercise, no exercise). You examine lower extremity strength with a dynamometer and balance using a standardized outcome assessment. You measure strength and balance at baseline, 2 weeks, 4 weeks, and 8 weeks. NUMBER OF LEVELS
4 levels for each factor
If differences between groups exist before data collection, what should be considered?
ANCOVA
T test is to T statistic as ANOVA is to:
F statistic
T/F: If I have multiple dependent variables that seem to be related, I should always run a MANOVA
False
T/F: Non-parametric tests are more powerful than parametric tests.
False (parametric tests are more powerful)
T/F: An ANOVA is different than a t-test due to the fact that an ANOVA is used with 2 groups and a t-test is used when there are 3 or more groups.
False ANOVA is used with 3 or more groups and/or 2 or more IVs
Match the parametric test with the nonparametric counterpart. One-way repeated measures ANOVA
Friedman two-way analysis of variance
You are wanting to compare four different groups in a study you are designing for individuals with persistent low back pain. One group is getting mirror box training, another group is getting laterality training, the third group is getting a combination of both and the final group is the control group. The outcome of interest is tolerance to pressure at the L4 segment and measured as "hypersensitive", "normal" or "diminished". The group is randomly selected from several clinics. If you only measure at one time point after intervention, Which is most appropriate in this situation?
Kruskal-Wallis (DV is ordinal, 1 factor with 4 levels)
Match the parametric test with the nonparametric counterpart. One-way ANOVA
Kruskal-Wallis analysis of variance
In an ANOVA, the F statistic IS a measure of effect size. Which is true?
Larger F stat = Larger effect size
Same base study as above: I have examined 3 groups (exercise and good diet, exercise and good sleep, exercise and mindfulness). I am measuring pain, disability, and strength at baseline, 2 weeks, 4 weeks, and 8 weeks. In this case, I hypothesize that my outcomes of pain, disability and strength are measuring a similar construct of "function" and that each of these outcomes likely influences the other. In this case, I should consider what analysis?
MANOVA
You are measuring neck function based on your treatment with 3 groups. Neck function is measured with 3 different outcomes. What should we consider?
MANOVA (multiple outcomes for one construct)
Match the parametric test with the nonparametric counterpart. Unpaired t-test
Mann-Whitney U Test
You are wanting to see if gluteus strength improves with strengthening. There are two groups: One group is getting eccentric exercises targeting the gluteus medius muscle group and the other group is getting concentric exercises targeting the same muscle group. Subjects were randomly selected from 3 different clinics located at different geographical locations. The outcome of interest is muscle strength which will be measured as either "strong", "diminished" or "poor" via a manual muscle testing procedure. If you only measure at one time point after intervention, how would you assess the data?
Mann-Whitney U Test (DV are ordinal, two groups)
What is the most common research design in PT?
Mixed model ANOVA
if there is a stat sig difference, do i need a post hoc power analysis?
NO
Homogeneity of variance differences (sphericity) are applicable for which test?
Repeated measures ANOVA
You measure current stress during meditation with DPT students. They are all tested in three positions (sitting, lying supine, standing). What test?
Repeated measures ANOVA ( 1 IV = meditation, 3 levels = sitting, supine, standing)
Researcher uses One Way ANOVA to compare change in pain among 4 treatment groups. Alpha was .05. P is .005. What's next?
Significant finding, do post hoc test
You are reading a study that indicates that they were using a 2 x 3 ANOVA. From this, we can conclude that:
There are two levels for the first IV and there are 3 levels for the second IV
You are reading a study and notice that the authors have used a Bonferroni correction. Which of the below explain why this correction was used.
To account for the family-wise error rate where alpha is divided by the number of post-hoc comparisons
T/F: Post-hoc testing should be done when a statistically significant interaction is found after performing an ANOVA.
True
T/F: T-test is to t statistic and ANOVA is to F statistic.
True
T/F: The F-statistic is used in an ANOVA; the larger the F-ratio, the grater the differences between the group means.
True
T/F: if an ANOVA has been performed, a post hoc test would be preformed.
True
Match the parametric test with the nonparametric counterpart. Paired T-test
Wilcoxon signed-ranked test
Which of the following will tell me I should choose a MANOVA analysis? a) The relationship between my DVs is R = 0.5 b) The relationship between my DVs is R = 0.1 c) The relationship between my DVs is R = 0.8
a) The relationship between my DVs is R = 0.5 (we want DVs that are related but not too related; want R value between 0.2 and 0.7 for MANOVA)
most common post hoc test between Fishers, Tukey, and Bonferroni
bonferroni
I have examined 3 groups (exercise and good diet, exercise and good sleep, exercise and mindfulness). I am measuring pain, disability, and strength at baseline, 2 weeks, 4 weeks, and 8 weeks. I have recognized that BMI may play a role in the outcomes for my sample. If I want to use BMI as a covariate and run an ANCOVA, which of the following should be true? a) categorical data b) heterogeneity of slopes c) A linear relationship between BMI (R=.7) d) A linear relationship between BMI and amount of exercise (R=.7)
c) A linear relationship between BMI (R=.7)
Which of the follow inferential statistical analyses looks at between-group variability over within-group variability? a- ANOVA b- t-test c- t-test and ANOVA d- mean/median/mode
c- t-test and ANOVA
post hoc tests are performed when
finding from an ANOVA is statistically significant to find WHERE the difference lies
I only consider power analysis when....
if results are NOT statistically significant
If a result is NOT statistically significant and the MCID is included in a confidence interval, the study has _____________ power and we __________ do a post hoc power analysis.
inadequate power, should do post hoc power analysis
How do I know a test is statistically significant? Alpha is .01
less than 0.1
if data is clinically significant, the point estimate will:
meet or exceed the MCID
Which ANOVA design is most appropriate when you want to answer the question: Does the amount of improvement over time depend upon treatment group allocation?
mixed-model ANOVA
Bonferroni is a post hoc test that will take alpha divided by what?
number of planned comparisons
One Group we are measuring 4 times...what test?
one way repeated measures ANOVA (IV is time NOT group)
A study has 1 IV with 4 groups. The best statistical analysis would be:
one-way ANOVA
Which ARE more powerful? independent t-test paired t-test ANOVA repeated measures ANOVA
paired t -test AND repeated measures ANOVA (Within group comparisons = more power)
I want to see if there is a difference in exam scores after students experience 2 weeks of immersive labs in a DPT program. Students take an exam related to neuro content before and after lab. There is a normal distribution of exam scores and variance is homogenous. Which test should be performed?
paired t-test
You are reading a study where the investigators want to see if there is a difference in strength measured in pounds of force for 4 different test positions for the gluteus medius muscle. The study design includes a group of 25 randomly selected healthy individuals who are put into the 4 different test positions where strength is measured. The mean force (in pounds) is compared between each test position to measure differences. This is an example of a:
repeated measures ANOVA
post hoc power analysis is used when
results are NOT statistically significant
You are studying a group of 100 people randomly assigned to one of 4 treatment groups (high intensity exercise, moderate intensity exercise, low intensity exercise, no exercise). You examine lower extremity strength with a dynamometer and balance using a standardized outcome assessment. You measure strength and balance at baseline, 2 weeks, 4 weeks, and 8 weeks. WHAT IS THE DV?
strength and balance
You are studying a group of 100 people randomly assigned to one of 4 treatment groups (high intensity exercise, moderate intensity exercise, low intensity exercise, no exercise). You examine lower extremity strength with a dynamometer and balance using a standardized outcome assessment. You measure strength and balance at baseline, 2 weeks, 4 weeks, and 8 weeks. WHAT IS THE IV?
treatment group and time