How do you find the expected frequency in chi squared
E = Row Total x Column total /
Grand total
How do you find error in chi squared
( Observed - Expected ) ^2/
Expected
Assumptions of the chi squared test for association
Random sample displayed in a contingency table
Expected frequencies are atleast 5
What do you write for H0 and H1 for chi squared for association test
H0: there is no association between
H1: there is an association between
What do you write after, at the 5% level using for chi squared test for association
V = delete the last row and column and count how many cells are left
How do you find the Test Stat for chi squared test for association
X^2 = { (Observed - Expected) ^2/
Expected
In formula book page 7
What does critical region look like in chi squared test for association
Like Boltzmann distribution
Always upper tailed
Use table 6 right hand column only
How to do the analysis for contingency table
Look at error table and identify the highest error
Look in the expected and observed tables
Say why you see and what it means
For example if there was a high error such as 32 and the expected was 89 while the observed was 143
You need to say the highest source of association is in the ( say what section the high score was in ) with an error of 32
There were more than expected, with 143 observed compared with 89 expected
What if the expected frequencies are smaller than 5?
Combine rows/columns in an appropriate way to make the expected frequencies at least 5
What is the disadvantages of combing the rows/columns to make the expected frequencies at least 5
You loose degrees of freedom, making it easier to find a significant result
If you combine too much and end up at 2x2 table and the Es are still too small, the sample itself is too small and the test cannot be tested done