Modules 3-4 and 5-8: Descriptive and Comparative Statistics

Question 1

Identify the inferential statistical test:

2 or more groups;
independent measures;
comparing means;
normally distributed

Answer 1

Independent-measures ANOVA (or ANOVA)

Question 2

Identify the inferential statistical test:

2 or more groups;
independent measures;
ordinal data;
NOT normally distributed

Answer 2

Kruskal-Wallis test

Question 3

Identify the inferential statistical test:

2 or more groups;
Nominal data w/categories;
Given proportion in each category

Answer 3

Chi-square test

Question 4

Identify the inferential statistical test needed to be used:

“There is no relationship between gender and willingness to use mental health services.”

Answer 4

Chi-square test
(2+ groups - male/female;
nominal data w/categories;
given proportion in each category)

Question 5

Identify the inferential statistical test needed to be used:

“There is no difference in the subject’s attitude about the medical intervention, as measured by in-house survey, before and after the intervention.”

Answer 5

Repeated measures t-test

Question 6

Is this a significant result?

t(8) = 6.00, p<0.01

E. F(3,12) = 24. 88, p<0.01, r^2 = 0.862

Answer 6

yes

Question 7

Is this a significant result?

t(18) = 4.00, p = 0.8

Answer 7

no

Question 8

Is this a significant result?

t(18) = 4.00, p<0.001,
95 CI [3.798, 12.202]

Answer 8

yes

Question 9

Is this a significant result?

F(3,16) = 8.33, p<0.05,
r^2 = 0.61

Answer 9

yes

Question 10

Is this a significant result?

F(3,12) = 24. 88, p<0.01, 
r^2 = 0.862

Answer 10

yes

Question 11

Does an r of 0.42 indicate a weak, moderate, or strong correlation?

Answer 11

Moderate (0.5-0.3)

Question 12

What is “sensitivity”?

Answer 12

The ability of a test to identify correctly who have the disease (the proportion of diseased patients who are correctly identified as positive).

Question 13

What is “specificity”?

Answer 13

The ability of a test to identify who does not have the disease - the proportion of diseased patients who are correctly identified as negative.

Question 14

What is a positive predictive value?

Answer 14

tells the probability that a patient has the disease

Question 15

What is a negative predictive value?

Answer 15

tells the probability that a patient does not have the disease

Question 16

What are problems with false positives?

Answer 16

1. Further testing increases healthcare costs;
2. patient anxiety;
3. diagnostic results follow patients

Question 17

What are problems with false negatives?

Answer 17

Patient w/disease may need treatment quickly but will not receive it if the test is negative

Question 18

Why are tests with high specificity important?

Answer 18

To rule out disease - especially if course of therapy is intensive or expensive, and requires an operation later.

Question 19

How is the likelihood ratio calculated?

Answer 19

First - need to know pre and post test odds.

Odds = Probability/(1-probability)
LR = post-test odds/pre-test odds

Question 20

How is sensitivity calculated (if given odds, and specificity)?

Answer 20

LR+ = Post-test odds/pre-test odds
LR+ = Sensitivity/(1-specificity)

Question 21

Name the statistical test:
Non-parametric distribution; continuous outcome
Categorical independent variable (1)
Independent measurements

Answer 21

Mann Whitney Rank Sum

Question 22

Name the statistical test:
Parametric distribution; continuous outcome
Categorical independent variable (1)
Independent measurements

Answer 22

t-test

Question 23

Name the statistical test:
Non-parametric distribution; continuous outcome
Categorical independent variable (1)
NO independent measurements - matched cases or before/after testing

Answer 23

Wilcoxon Signed Rank

Question 24

What is the non-parametric counterpart to the Chi-square test?

Answer 24

Fisher exact: categorical outcome, categorical independent variable (1), independent measurements