1. What is the difference between simple (also known as univariate and bivariate) regression and multivariate regression?
A) Simple regression includes one independent variable, and multivariate regression includes more than one independent variable.
B) Simple regression includes only one dependent variable, and multivariate regression includes more than one dependent variable.
C) Simple regression uses a chi-square statistic, and multivariate regression uses maximum likelihood.
D) Simple regression adjusts for confounding, and multivariate regression does not.
2. What type of regression is used when the dependent variable is dichotomous?
A) Linear
B) Logistic
C) Poisson
D) Multinomial
3. What measure(s) of the strength and direction of association can we get from a logistic regression model? (Select all that apply.)
A) Regression coefficient (beta)
B) Risk ratio
C) Odds ratio
D) F-statistic
E) Chi-square
F) One-way ANOVA
4. Researchers used simple logistic regression to model the association between physical therapy and ability to get up from a chair unassisted among a group of men with Parkinson's disease. The beta was 0.32. What is the odds ratio?
A) 0.32
B) 32%
C) 1.38
D) - 1.14
5. One of the important benefits of multivariate regression over many other statistical tests is that it
A) allows you to adjust for potentially confounding variables.
B) allows you to prevent measurement error.
C) ascertains temporal order that is necessary for causal inference.
D) ensures that you are fulfilling all the required assumptions for statistical testing.
6. Linear regression is appropriate for what type of dependent variable?
A) Dichotomous
B) Ratio scale and normally distributed
C) Polytomous
D) Count
7. What do we mean by "regression toward the mean?"
A) Outliers in the model should always be excluded when using linear regression.
B) The linear regression equation can be used to identify the average value of each variable in the model.
C) The phenomenon that if a variable is extreme on its first measurement, it will tend to be closer to the average on a second measurement.
D) Linear regression normalizes the scale of the variables so they have a mean of zero and standard deviation of 1.
8. A strong correlation (>0.85) between the independent variables that you want to include in one multivariate linear regression model is an indication that which assumption has been violated?
A) Homoscedasticity
B) Multicolinearity
C) Normal distribution of the dependent variable
D) Residual analysis
9. What is the difference between stepwise methods versus the standard approach and hierarchical methods of selecting independent variables to include in a linear regression model?
A) Stepwise uses statistical criteria for selecting the variables to be included, whereas the other approaches use theoretical rational.
B) Stepwise uses theoretical rational, whereas the other approaches use statistical criteria for selecting the variables for inclusion.
C) Stepwise is better for hypothesis testing than the other approaches because it is less influenced by chance.
D) Stepwise methods allow you to include more variables without increasing the probability of making a type I error.
For the following research problems, indicate which statistical analysis would be BEST for answering the question. Provide a very brief (1-sentence) explanation as to why the test you chose is best. Also state what the independent and dependent variables are for each problem.
1. A researcher is interested in examining the difference between teenage males and females and their level of self-efficacy.
2. A researcher is interested in examining if there is a significant relationship between exercise and level of depression.
3. A researcher wants to examine if the prevalence of eating disorders is what would be expected when compared to the general population for a group of white/non-Hispanic, Black, and Asian women on a college campus.
4. A researcher wants to examine the effects of a smoking awareness campaign for a group of smokers. The researcher measures awareness prior to the campaign and following the campaign for a single group of participants that have participated in the campaign.
BONUS - the following is NOT a required question but merely a 5 point BONUS. You will NOT lose any points for attempting this question.
Refer to this ANOVA table:
|
Source
|
SS
|
df
|
MS
|
F
|
|
Total
|
42
|
11
|
|
|
|
Between
|
|
3
|
7
|
|
|
Within
|
|
|
|
|
1. What will be the value of SSbet?
2. What are the within degrees of freedom?
3. What's the value of the MSwith?
4. What is the value of F for this problem?