Question 1(2 points)
Saved
Dr. Friedman is testing for the difference in mean pounds lost among the groups - a new weight loss drug, a placebo and the marketing leading drug. She sets the alpha level for a two-tailed test at .05. She expects the pounds lost in a month in the 3 groups to be 15, 0 and 10 with a standard deviation of 5. She will have equal numbers in each group. To determine the power of her analysis, she still needs to specify :
Question 1 options:
|
Probability of a Type I error
|
|
Total sample size
|
|
The variance
|
|
An independent t-test
|
Question 2(2 points)
Saved
Dr. Krishnan is assessing the impact of gender and severity of disease (rated on a o to 100 scale) on units of insulin prescribed per day. Two linear regression models are computed to predict units of insulin with the following results:
InsulinUnits = 55.5 + .8Gender
InsulinUnits = 68.5 + .4Gender + .05Severity
Do these results show evidence of confounding?
Question 2 options:
|
Yes, because the regression coefficient for gender has changed substantially
|
|
No, because gender is still included in the equation
|
|
Yes, because the mean is different between the two equations
|
|
Only if gender is no longer statistically significant
|
Question 3(2 points)
Saved
Computing a linear multiple regression with anxiety as a dependent variable, we obtain the following results.
Multiple regression analysis results
Independent Variable
|
Regression Coefficient
|
T
|
p-value
|
Intercept
|
5.277
|
32.52
|
.0001
|
Age
|
-.051
|
-36.38
|
.0001
|
Gender
|
.295
|
5.48
|
.0001
|
Healthy Living Index
|
.027
|
6.30
|
.0001
|
Which statement about these results is true?
Question 3 options:
|
Age has the greatest predictive value, with the largest absolute t-value
|
|
Healthy living has the greatest predictive value, with the largest t-value
|
|
Both a and c
|
|
Gender and healthy living are both more important than age as predictors, because the regression coefficients are positive
|
Question 4(2 points)
Saved
Dr. Elder finds that women who have regular mammograms and have a specific gene for breast cancer survive significantly longer than women who do not have regular mammograms. For women who donothave the gene, there is no difference in survival rates between the mammogram and no mammogram conditions. This is an example of
Question 4 options:
|
Type II error
|
|
multiple logistic regression
|
|
statistical power
|
|
statistical interaction, also known as effect modification
|
Question 5(2 points)
Saved
Give an example of two variables that would be expected to have a strong,positivecorrelation.
Question 5 options:
Question 6(2 points)
Saved
Give an example of two variables you would expect to have anegativecorrelation.
Question 6 options:
Question 7(2 points)
Saved
I have a sample of 5,000 patients and want to predict whether or not a patient develops Coronary Heart Disease with the following predictors: age, gender, weight and number of cigarettes smoked per day. The best technique to use would be:
Question 7 options:
|
Multiple logistic regression
|
|
Simple linear regression
|
|
Multiple linear regression
|
|
A nonparametric test
|
Question 8(2 points)
Saved
Computing a linear multiple regression with anxiety as a dependent variable, we obtain the following results.
Multiple regression analysis results
Independent Variable
|
Regression Coefficient
|
T
|
p-value
|
Intercept
|
5.277
|
32.52
|
.0001
|
Age
|
-.051
|
-36.38
|
.0001
|
Gender
|
.295
|
5.48
|
.0001
|
Healthy Lifestyle Index
|
.027
|
6.30
|
.0001
|
Which of these is a significant predictor of anxiety?
Question 8 options:
|
Age
|
|
Gender
|
|
Healthy Living Index
|
|
All of the above
|
Question 9(2 points)
I am conducting a study of the effectiveness of a new drug vs the current standard of care. I want my probability of a Type II error to be no higher than 10%. Therefore, the lowest power I can accept is: Please give your answer as a number.
Question 9 options:
Question 10(6 points)
The following results are from a logistic regression with diabetes as the dependent variable
Explain the relationship between these variables and diabetes. Specifically state which, if any, of the variables are statistically significant predictors. State the odds ratio for each variable, whether it is significant and which of the three variables is the best predictor of diabetes. How do you know this?
Question 10 options:
Question 11(2 points)
Saved
Non-parametric tests are used instead of parametric tests when
Question 11 options:
|
Assumptions about a normal distribution cannot be met
|
|
Sample size is small
|
|
The dependent variable of interest is time to event
|
|
Both A and B
|