Which of the following condition is FALSE from the validity point of view with respect to Chi-square as a test for 'goodness of fit'?
1
The sample observations should be independent
2
Constraints on cell frequencies, if any, should be linear (i.e. sum of observed frequencies should be equal to expected frequencies)
3
N, the total frequency should be large (>50)
4
Some of the cell frequencies can be less than 5.