Deck 14: Non-Parametric Data Analysis

Question 1

Categorical data represent the quantity of an outcome and are expressed in terms of percentages or proportions rather than means and standard deviations.

Answer 1

False

Question 2

Pretend you're creating a contingency table made up of two categorical variables. Your __________ variables would make up the rows, and the __________ variables would make up the columns.

A) predictor; exposure
B) outcome; exposure
C) exposure; outcome
D) None of these is correct.

Answer 2

C

Question 3

If your predictor variable is dichotomous and your outcome variable is not, would a simple chi-square test sufficiently explain the relationship between the two variables?

A) Yes
B) No
C) Maybe
D) The test cannot be run.

Answer 3

D

Question 4

Why would you use an effect modifier for a chi-square test?

A) To account for a confounding variable
B) To determine an extraneous effect on the predictor variable
C) To account for a confounding variable and to determine an extraneous effect on the predictor variable
D) None of these is correct.

Question 5

In the study determining the relationship between parental monitoring and teenage binge drinking, sex became an effect modifier variable; however, the association was significant in females but insignificant for males. What does this tell the investigator?

A) Sex, not parental monitoring, should be the primary variable of interest.
B) Because sex was insignificant for male binge drinking outcomes, the effect modifier should be discarded.
C) Both variables are important to the study's tests and future interventions.
D) Two separate studies should be conducted.

Question 6

The higher the chi-square value, the lower the P-value.

Question 7

Logistic regression is a form of analysis that:

A) can determine causation.
B) can have nominal, ordinal, or ratio-level independent variables.
C) is more effective than bivariate associations.
D) All of these are correct.

Question 8

An odds ratio can range from close to zero to levels exceeding 5 or 6; however, if you received an odds ratio of 1, what assumptions can you make about your results?

A) The P-value from the tests is too large.
B) The odds of the outcome are similar for people regardless of their exposure to the variable of interests.
C) You can't make an assumption without the confidence interval.
D) The P-value from the tests is too large, and the odds of the outcome are similar for people regardless of their exposure to the variable of interests.
E) The P-value from the tests is too large, and you can't make an assumption without the confidence interval.

Question 9

If you were to have an adjusted odds ratio of 0.7 with a confidence interval of 0.6-0.8, how would you interpret these results?

A) The odds of the outcome are 30% less likely for you than those exposed to the variable of interest.
B) The odds of the outcome are 70% less likely for you than those exposed to the variable of interest.
C) The positive outcome is less likely to occur.
D) Unsure

Question 10

What is a benefit of conducting a meta-analysis?

A) It could save you time and money.
B) It pools together multiple studies and allows you to conduct secondary analysis.
C) It gives you insight on more or less successful interventions.
D) All of these are correct.