SIMPLE TWO GROUP TESTS Prof Peter T Donnan Prof Peter T Donnan.

SIMPLE TWO GROUP SIMPLE TWO GROUP TESTSTESTS

Prof Peter T DonnanProf Peter T Donnan

1. Chi-squared test (2x2)2. t-test3. Logrank test

1. Chi-squared test (2x2)2. t-test3. Logrank test

CHI SQUARED TESTCHI SQUARED TESTUsed to compare the proportions of observations in different categories.

Example:

Are gender and achieving LDL target related?

Null hypothesis: There is no association of gender with achieving LDL target (or no difference between men and women)

Achieved LDL target

TotalNo Yes

GenderMale 140 563 703

Female 149 531 680Total 289 1094 1383

SIGNIFICANCE OF THE TEST SIGNIFICANCE OF THE TEST STATISTICSSTATISTICS

The value 2 = 0.834 is not significant (p = 0.361)

The null hypothesis can not be rejected.

Gender and achieving LDL target are not related

% Males meeting target: 80.1%

% Females meeting target: 78.1%

CONDITIONSCONDITIONS

1. Data are categorical.

2. If you tried to do Crosstabs with continuous

variables you will get one column or row for

each unique value!

T-TEST – COMPARE TWO T-TEST – COMPARE TWO MEANSMEANS

1. Parametric test since comparing means

2. Paired samples t-test – the mean difference

between two linked groups

3. Independent samples t-test – the mean

difference between two independent groups.

PAIRED SAMPLE T-PAIRED SAMPLE T-TESTTEST

One variable measured in: 2 different groups who are matched or same group at 2 different times (e.g before / after)

Example:

Is there a difference in LDL level before and after treatment in the total sample?

Is this paired or unpaired?

INDEPENDENT SAMPLES T-INDEPENDENT SAMPLES T-TESTTEST

Example: Is there a difference between the population mean ages of the males and females?

Assumptions: - In the population of interest the variable is normally distributed.- The variances of the 2 groups are the same

LOGRANK TESTLOGRANK TEST

Example: Survival curves for women with glioma by diagnosis.

Bland J M , Altman D G BMJ 2004;328:1073

The most popular method of comparing the survival between groups, which takes the whole follow up

period into account.

LOGRANK TESTLOGRANK TEST

Log rank test involves calculating Chi-squared

(2) statistic for difference in median survival

between two groups.

Example: Do colorectal cancer patients with

hypertension have worse survival than patients

without hypertension?

The Null hypothesis for the logrank test:

Hazard Rate group A = Hazard Rate for group B

= HR = OA / EA = 1 OB / EB

The Null hypothesis for the logrank test:

Hazard Rate group A = Hazard Rate for group B

= HR = OA / EA = 1 OB / EB

THANK YOU FOR THANK YOU FOR LISTENINGLISTENING

Explore 2-group tests further with datasets: LDL Data.sav colorectal.sav