1
Q
How do we determine the significance of the difference between observed and expected results?
A
- Use chi-square
- x^2 = sum of (O - E)^2 / E
- O = observed; E = expected
- Chi-square statistic has positive values only, therefore only one parameter (degrees of freedom) is needed to specify any chi-squared distribution
2
Q
What is degrees of freedom?
A
Number of categories - 1
3
Q
What is used for a 1 df chi-square?
A
- A continuity correction (Yate’s correction) is recommended; not required when df > 1
- x^2 = sum of ([O - E] - 1/2)^2 / E
4
Q
What can be concluded if calculated chi square doesn’t exceed critical chi-square?
A
The difference between observed and expected values are small and are due to chance alone
5
Q
What are some assumptions for chi-square?
A
- Random sampling not essential for calculation, but may obviously impact the interpretation (bias)
- Observations must be independent
- Categories must be mutually exclusive
- Distribution of chi-square depends on # of tx being compared; this dependency is quantified in a df parameter v (equal to # of rows in the table minus 1, times # of columns in the table minus 1)
- For v = 1 contingency tables, computing chi-square using standard formula leads to p-values smaller than they ought to be (results are biased towards Ha) so Yate’s correction is used
6
Q
How do you calculate expected frequency?
A
(row total * column total) / grand total
7
Q
What conclusion can be made when calculated chi-square exceeds critical chi-square?
A
Reject null hypothesis and conclude there is an association between drug therapy and thrombus formation at alpha = 0.05
