It may be a very silly question, but I have trouble conceptualizing these two terms. I had this contingency table with 2 rows and colums. Is expected value always 50%? I remember otherwise in dice or coin game. Could you kindly explain chi-square (I know you discussed this elsewhere, but) addressing this expected values? Thank you.
The expected values are the ones we would expect if the two variables are independent*. If the departure from the expectancy is "large enough," the chi-square statistic will be significant.
* Independent means that the probability of one variable occurring or taking on a particular value is not affected by the row or column of the other variable.
So, if there is no association between the two variables in a contingency table, the observed % in a row should be the same as % in a column, am I right??? Not necessarily 50%. I have trouble with understanding why observed % has anything to do with expected %. Thank you.
Read through the section on chi-square and independence at this link:
http://davidmlane.com/hyperstat/B143466.html
If %'s in rows are not contingent on columns (or vice versa), then the two variables are independent.
Posting Permissions
|
|
Reply With Quote