SRT 605 - Topic (8) Chi Square

SRT 605: CHI SQUARE
Dr Tan Chee Hian

Profesor Madya
Sports Management Department
FSR, UiTM Shah Alam
03-55442949
A4046
[email protected]
Chi Square Test
It is the basic technique for examining:
- the relationship between two categorical
(nominal or ordinal) variables, possibly
- controlling for additional layering variables.
- measures of association and agreement for nominal and ordinal
data.
- using cross tabulation (contingency) table
Calculation of Crosstabulation
Group A Total
Yes No
Yes Observed a b a+b
Expected (a+c) X (a+b) (b+d) X (a+b)
Group a+b+c+d a+b+c+d
B
No Observed c d C+d
Expected (a+c) X (c+d) (b+d) X (c+d)
a+b+c+d a+b+c+d
a+c b+d a+b+c+d

Expected Count = Total of row X Total of Column
Total
% of Cells with EC < 5 = No of cells with EC < 5 X 100
Total no of Cells
Hands- on
Group A
Yes No Total
Yes Observed 25 16 41
Expected ?? ??
Group B
No Observed 4 20 24
Expected ?? ??
29 36 65
Calculate Expected Count (EC) for each cell

Calculate Percentage of Cells if EC less than 5
Con’t - Steps
Step 1: What hypotheses?
Variables 2 or more proportions to explore the relationship between
normally just 2 categorical variables.
Will also tell if there is a significant association between the 2 variables.
Example:
Comparing the proportion of obesity between male and female.
- Using the example above, then
Ho: P(obesity)male = P(obesity)female
Step 2
The assumptions:
1. Random samples.
2. Observations are independent – by study design or methods.
3. Number of cells with Expected - Count (EC) less than 5, must be
less than 20% of the total number of cells. –Calculate expected count
for each cell (software will do it).
Generating Chi Square Test- Example
To generate cross tabulation table and Chi-Squared Test
1. Open Data
2. Select Analyze => Descriptive Statistics => Crosstabs
3. In Crosstabs dialogue box, select risk factor into
Row(s) and achieve_goal.3 in Column(s).
Con’t
4. Then click on the Statistics and select Chi-
square. Click Continue.
5. Then click on Cell. Tick at Observed and
Expected in the Counts box. Click Continue.
6. Click OK.
SPSS - Computer
SPSS OUTPUT
Expected Count
Number of cell with expected count less than 5 & the smallest EC is
5.4,
All the assumption for Chi-square is fulfill.
Hypothesis: Ho: P1=P2=P3

There is no association between risk factor and achievement.
Con’t
Con’t
Used Pearson’s Chi-square P-value
Interpretation: P-value = 0.003<0.05. Reject Ho.
Conclusion - There is an association between risk factor and

achievement.
If
your Number of cells with Expected Count (EC) less than 5, more than 20%???
HOW?
Can’t use p value for Pearson’s Chi-Square
Used FISHER’S EXACT

Fisher’s Exact - Example
Example 2: In Crosstabs dialogue box, select risk factor into Row(s) and
achieve_goal.4 in Column(s). Generate the cross tabulation table.
Con’t
Percentage of cells with EC < 5 = No of cells with EC < 5 x100
Total no of cells
Percentage of cells with EC < 5 = 2/6*100 =33.33%

Extra Steps
But FISHER’S EXACT only used if cross tabulation was 2 x 2

so
need to collapse the table become 2 x 2 table
SPSS
Con’t
2 x 2 table (recoding)
New cross tabulation table for new coding: 2 x 2 table
risk coding * achive_goal.4 Crosstabulation
Percentage of cells with EC < 5 = No of cells with EC < 5

Total no of cells
Percentage of cells with EC < 5 = 1/4*100 =25%
Con’t
Assumption 3 for Chi-square is not fulfill.

So we look at the P-value for Fisher’s exact test.
Hypothesis: Ho: P1=P2=P3
Example -Conclusion - There is an association
between risk factor and achievement
Chi-Square Tests
Value df Asymp. Sig (2 tailed) Exact Sig (2- tailed) Exact sig (1-T)
Pearson Chi-Square 5.386a 1 0.021

Continuity Corrections b 3.386 1 0.051
Likelihood Ratio 8.457 1 0.004
Fisher's Exact Test 0.28 0.015
Linear-by-Linear Association 5.279 1 0.022
N of Valid Cases 68
a. 1 cells (25.0%) have expected count less than 5. The minimum expected count is 3.25.
b. Computed only for a 2x2 table

SRT 605 - Topic (8) Chi Square

Uploaded by

Copyright:

Available Formats

SRT 605 - Topic (8) Chi Square

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SRT 605 - Topic (8) Chi Square

Uploaded by

Copyright:

Available Formats

SRT 605: CHI SQUARE

Dr Tan Chee Hian

a+c b+d a+b+c+d

Calculate Expected Count (EC) for each cell

Hypothesis: Ho: P1=P2=P3

Interpretation: P-value = 0.003<0.05. Reject Ho.

Conclusion - There is an association between risk factor and

Can’t use p value for Pearson’s Chi-Square

Used FISHER’S EXACT

Percentage of cells with EC < 5 = 2/6*100 =33.33%

But FISHER’S EXACT only used if cross tabulation was 2 x 2

Percentage of cells with EC < 5 = No of cells with EC < 5

Assumption 3 for Chi-square is not fulfill.

Pearson Chi-Square 5.386a 1 0.021

You might also like