Hello Goh YEE Shan,

In SPSS, the default option for the program is to omit a case if it lacks a score on any of the variables called for in the desired analysis; that is called listwise deletion.

So, it appears that one of your 222 cases was lacking a score for at least one of the variables you specified (IVs and DV) in your hierarchical regression. In the regression program, you can ask to save the predicted scores for each case. If you find one case without a predicted score in your data worksheet, you should be able to figure out what's causing that case to be omitted from the analysis.

Good luck with your work.

Suppose you have a unique identifier called ID, that Y = the DV, and x1 to x5 are the explanatory variables. You can do this to list any observations with incomplete data.

COMPUTE CompleteCase = NMISS(Y,x1,x2,x3,x4,x5) EQ 0.

FORMATS CompleteCase (F1).

FREQUENCIES CompleteCase.

TEMPORARY.

SELECT IF NOT CompleteCase.

LIST ID Y x1 x2 x3 x4 x5.

HTH.

Bruce Weaver , I would not use SELECT IF, but FILTER the cases, otherwise I would need to redo all my previous commands to return to my data set as is. With USE ALL, I restore my original data set.

Maybe I also would use COUNT MISVAL = varY VarX1 to VarXn (MISSING).

To have complete cases for the items used in the analysis is very relevant to compare models. One can compare Deviances/ChiSq to test if the complexity of the model is significant. In R one can also test if a multilevel model is neeeded by comparing the OLS-null model with the GLMM-null model. This can be done as the models are nested.

The R syntax could look like:

Ana1_M0.GLS

Hello Peter Moorer. Perhaps you missed the TEMPORARY command I included before SELECT IF? It means that the SELECT IF is applied to only the first following command, i.e., the LIST command in my code.

TEMPORARY.

SELECT IF NOT CompleteCase.

LIST ID Y x1 x2 x3 x4 x5.

See the Command Syntax Reference manual for more info and examples.

• https://www.ibm.com/docs/en/spss-statistics/27.0.0?topic=temporary-overview-command

HTH.

Why don't you look at your data before you start conducting any analyses! With a small sample like this not know this until you get the n out of the stats output is very bad. You should get to know your data better before conducting statistical procedures.