Statistics

Question 1

Paired t-test

Answer 1

Use with paired samples or individuals (matched pts or BPs before and after drug)
**Compares two samples in the same patient or in matched patients

Question 2

Unpaired t-test

Answer 2

Compares means of two independent samples/groups

Question 3

ANOVA

Answer 3

Analysis of Variance

**Different between the means of 3 or more groups

Question 4

Chi-square

Answer 4

Difference between 2 or more percentages or proportions of categorical outcomes (NOT MEANS)

Question 5

Relative Risk

Answer 5

Risk of disease in exposed/risk of disease in unexposed

**Cohort studies

(A/A+B)/(C/C+D)

Question 6

Odds ratio

Answer 6

Case-control; approximates relative risk in a rare disease

AD/BC

Question 7

Null hypothesis

Answer 7

No difference; no associated between disease and risk factor

Question 8

Alternative hypothesis

Answer 8

There is a difference; risk factor is associated with disease

Question 9

Type I/alpha error

Answer 9

Rejecting the null hypothesis when it is tru

Usually set alpha to 0.05

Question 10

Type II/beta error

Answer 10

Accepting the null hypothesis when it is false

Question 11

Power

Answer 11

Probability of rejecting the null hypothesis when it is false (saying there is a difference when there really is)

Depends on # of endpoints and differences in means

Power=1-beta

Question 12

Ordinal vs nominal variables

Answer 12

Nominal = names (sex, marital status)

Ordinal = ordered (stage, pain)

Question 13

Sensitivity

Answer 13

Chance of a positive test in a patient with disease

of true positives/# of all people with disease

A/A+C

Question 14

Specificity

Answer 14

Chance of a negative test in a negative patient

D/B+D

1- specificity = false positive rate

Question 15

Positive predictive value

Answer 15

Probability of having a condition with a positive test

of true positives divided by # who tested positive for disease

A/A+B

Question 16

Negative predictive value

Answer 16

Probability of not having a condition with a negative test

of true negatives divided by # of people who tested negative for the disease

D/C+D

Question 17

Mann-Whitney (Wilcox rank test)

Answer 17

Non-parametric analog to t-test

Question 18

Kruskal Wallis

Answer 18

Compares the means of two populations that are NOT normally distributed

Question 19

Kaplan Meier

Answer 19

Compares survival rates among groups

Question 20

Cox regression (Cox proportional hazards model)

Answer 20

Mutlivariate survival analysis - evaluates association between variable and survival; controls for confounding variable

Question 21

What is kappa?

Answer 21

Measure of reproducibility - > 0.75 = excellent; <0.4 = poor