STATS MAKE ME CRY

Hi,

So i have a very large dataset of over 16000 cases and over 500 variables..and i am planning to do a monte carlo simulation for multiple imputaion. before that i have to test for missing completely at rondom little's test using EM algorithm..for that SPSS asks for the distribution kind which is either 1) nomral (default) 2) mixed normal or 3) student's t..i would assume that i have a normal dis. coz i have a very large dataset using CLT rule of thumb but i am not sure if i am right or not? and generally how am i going to know the dis.? should i test each variable?
plz help me out here..thx

I have tried little's test with 100 iterations and it didn't converge, then with 1000 iterations and also didn't converge. Do you have any idea what would be the right number of iterations?