Message/Author |
|
Andy Ross posted on Monday, September 24, 2012 - 9:38 am
|
|
|
Dear Prof Muthen. Can you please confirm if for type=basic (a) Mplus gives polychoric correlations for ordinal variables with 3 or more categories, and (b)Mplus bases these on the full sample taking account of any variable missingness Many thanks Andy |
|
|
If the ordinal variables are on the CATEGORICAL list, you obtain a pairwise present polychoric correlation matrix. |
|
Andy Ross posted on Wednesday, September 26, 2012 - 1:35 am
|
|
|
Many thanks for getting back to me so promptly, I had thought they would be, but just wanted to be certain. Could you also confirm whether these are for the full sample in the case of missing data at the variable level? Many thanks again Andy |
|
|
Each correlation uses all non-missing values for that pair of variables. The entire sample is not used for each correlation unless there is no missing data for that pair of variables. |
|
Andy Ross posted on Wednesday, September 26, 2012 - 7:22 am
|
|
|
Many thanks that's very much appreciated. |
|
|
Hi Prof. Muthen, When using Type=basic in Mplus for descriptive statistics, a) for nominal variables (dummy coded), are the correlations between nominal variables provided in Mplus, Person Chi-square analysis, or same results as crosstab analysis? b)is the interpretation the same as regular correlations? --------------------- Nominal variables correlation Matrix in Mplus: Expereince 1 dentity -0.038 1 Affliation -0.004 0.356 1 DEGREE 0.213 0.063 0.099 1 FIELD -0.057 0.053 0.011 -0.321 1 JobSetting 0.013 -0.108 -0.127 -0.246 0.231 1 Background -0.089 -0.026 0.136 -0.134 0.203 0.183 1 Gender 0.057 0.046 0.102 0.158 -0.021 -0.063 0.025 1 -------------------- Thanks a lot! |
|
Back to top |