T-Test with Chi-Square

noetsi

Fortran must die
#3
T test and chi square are totally different test that assume different distributions and different type of data (interval for t test and ordinal or nominal usually for Chi square). You would never use one with the other.
 

Englund

TS Contributor
#4
T test and chi square are totally different test that assume different distributions and different type of data (interval for t test and ordinal or nominal usually for Chi square). You would never use one with the other.
The Chi square distribution is used very, very often. Not only when we have ordinal or nominal data. Not sure that was what you meant though.
 

noetsi

Fortran must die
#5
I meant the chi square test of independence was used with (or anyhow assumed) nominal or ordinal data (actually nominal). I thought that was what was being raised since it is a very common test. Obviously there are many other chi square test some of which (like the SEM model fit tests) which use interval data.
 
#7
Sorry for the confusion. I really did a bad job of composing the question !!
Here's my question: When using a z-score test we're dealing with a normal curve. For the t-test with a skewed distribution.
Now the actual areas under the normal curve can be computed using an integral.
How are the tables for t-tests and chi square tests computed. Not how are they used but how are the actual
entries computed? Are they also integrals?
 

Dason

Ambassador to the humans
#8
Yes the entries in a t-table are computed using integrals of the pdf of the t-distribution and the entries in Chi-square tables are also integrals of the pdf of the chi-square distribution.