Flashcards in Statistics for Public Health Deck (39):

1

## Define Odds Ratio

###
The odds ratio is:

The probability of an event happening divided by the probability of it not happening.

2

## State the formula for Odds Ratio

### Odds = p / 1 - p

3

## What entities can you calculate Odds Ratios for?

###
Can calculate for:

1 entity eg CD4>250 response to AZT

Many groups eg compare different CD4 count groups to a reference group.

4

## Why would you convert a RR to the log scale?

### So that the range of possible positive and negative values are equivalent.

5

## How do you convert a RR result to a log scale result?

### Just take the final RR and use the ln function on the calculator.

6

## What’s a narrative difference between relative risk (RR) and odds ratio (OR)?

###
Relative risk is how frequently (likely) one thing happens compared to another. [Both can happen]

Odds ratio is how likely is one thing to happen *versus* how likely is another. [one versus the other happening]

7

## What is the direction of logRR when numerator event is more likely than denominator event? i.e numerator is bigger than denominator.

###
When numerator is bigger the LEGO tower falls to the right in logRR. ie values will be greater than 0.

When numerator is smaller than denominator the tower leans left, like stable society! ie logRR values are less than 0.

8

## What is the formulas for Incidence Rate Ratio?

### IRR = number of events / study time

9

## When would you use an IRR?

### To quantify the TTE differences from 2 samples.

10

## How would you compare IRR across 3 or more groups?

### Use one as the reference group.

11

## What is an advantage of using IRR?

### It’s a relative risk measure that enables comparison between groups that have differences in follow-up time.

12

## What does Kaplan-Meier summarise?

### Cumulative proportion event-free (survival)

13

## What’s a synonym for Relative Risk?

### Rate Ratio

14

## What is the Survival equation in the TTE topic?

###
N(t) - E(t)

S(t) = ————— - S(pEt)

N(t)

15

## Where do you take the median time from for a Kaplan-Meier curve that does not exactly equal 0.5?

### From the first time point where Survival drops below 0.5

16

## Which groups do RR and RD (ARR) relate to? Personal or population-level?

###
RD to Population

RR to individual risk

17

## Define the ‘sampling distribution’

###
The sampling distribution is:

The theoretical distribution of all possible values of a sample statistic,

From all possible random samples,

Of the same size from the same population

18

## State 3 characteristics of the Central Limit Theorem.

###
Theoretical sampling distribution of a sample statistics will:

1) be approximately normal

2) contain average of the true population level statistic being estimated

3) have variability that is a function of:

a) variation of population-level values

b) the sample size

19

## What is the formula for standard error for continuous data?

### It is:

20

## What is the SE formula for binary data?

### It is:

21

## What hypothesis test would you use for comparing two population means comprising unpaired continuous data?

###
Two-sample t-test

aka

unpaired t-test

22

## What are the 3 types of data that can be used in a Hypothesis Test?

###
Continuous (paired, unpaired)

Binary

Time to event

23

## What does the p-value report?

### The probability of getting your test result if the null hypothesis is true.

24

## What hypothesis test would you use for paired (continuous) data?

### Paired t-test

25

## What is the mnemonic for the steps involved in hypothesis testing?

###
SAC.TAR

A sac of tar

26

## What is the mnemonic and each of the steps when performing a hypothesis test?

###
SAC.TAR

S State the competing hypotheses (Ho and Ha)

A Assume Ho is True (ie no difference)

C Compute the d of your test result from the null (0) in SEs (standard errors)

T Translate the d into a p-value

A/R Accept or Reject the null hypothesis Ho

27

## What is the mnemonic for accepting the null hypothesis?

###
PGAA

Tiger Woods played in the PGA, and had to go to AA.

28

## What do each of the letters in the hypothesis test p-value result stand for?

###
PGAA

p-value

Greater than

alpha

Accept Ho

29

## What is the mnemonic for rejecting the null hypothesis and what do the letters stand for?

###
PLAR

Klingons reject that humans and Klingon’s are no different (Ho) “kPLAR Picard!

p-value

Less than

alpha

Reject Ho

30

## What hypothesis tests would you use for hypothesis testing proportions (RD, RR, OR) between two populations?

###
2-sample z-test

Chi-square test

Fisher’s Exact Test

31

## What tests is the chi-square equivalent to?

###
Two-sample z-test

And Fisher’s Exact Test

32

## What tests gives the same result as a 2-sample z-test?

###
Chi-square test

Fisher’s Exact Test

33

## When would you use Fisher’s exact test instead of chi-square or 2-sample z-test?

### When samples from both populations are small.

34

## What goes on the rows (left side) and columns (top) of 2x2 error matrix table, and what goes inside it?

###
Rows:

Decision: Reject Ho

Decision: Accept Ho

Columns:

u1 = u2

u1 != u2

T1 error (alpha). Power

Correct. T2 error

35

## What test could you use for hypothesis test comparison of means between more than 2 populations?

###
ANOVA

Analysis of Variance

36

## What test could you use for hypothesis test comparison of proportions between more than 2 populations?

### Chi-square

37

## What test is ANOVA an extension of, for more than two populations?

### 2-sample t-test

38

## What test could you use for hypothesis test comparison of TTE between more than 2 populations?

### Log-rank test

39