The best method depends entirely on your research question. Data structure probably follows your survey, and repeated measures can be handled/modeled in SPSS. Can you elaborate on the model you aim to build?

Is there a reason that you are required to use SPSS?

And is there any reason you are imputing missing data? Epidemiological studies are full of holes but it is difficult, often impossible, to argue that the data meet the criteria for imputation.

This in addition to what your research question is ( Sandro J Martins ) and why you are stuck with SPSS ( Daniel Wright ).

Hi,

Thank you for your responses.

Yes, Sandro J Martins I aim to build a regression model on how maternal & paternal alcohol use & tobacco smoking affect child’s head circumference. In my mind, I have figured that the data has a hierarchical structure: The pregnancy i.e. cases are the level 1, and the maternal identifier level 2. What is your proposal for doing that kind of a model?

As an answer to Daniel Wright, I use SPSS, cause (for the time being) that is the software I am able to use & I have no time to learn R. I am aware of that there are better programs to be used, but that’s how it is right now.

Ronán Michael Conroy, I have used multiple imputation to account for the missing data because I know that the information in my cohort is not MCAR. So, doing complete case analysis would yield (more) biased estimates in addition of losing a lot of the power.

I hope these information answered your questions and I look forward for further advice.

I hope you got an answer! I am struggling with a big SPSS puzzle of my own, so I’ve stumbled through some similar questions. I believe the generalized linear mixed model/ linear mixed model in SPSS are not be completely interchangeable. You have to do workarounds in the software:

https://www.ibm.com/support/knowledgecenter/SSLVMB_sub/statistics_casestudies_project_ddita/components/glmm/glmm_cablesurvey_howto.html

https://www.spss.ch/upload/1126184451_Linear%20Mixed%20Effects%20Modeling%20in%20SPSS.pdf

Read it carefully because I could be totally wrong- I’m trying to figure out some things in SPSS right now too.

I haven’t gotten my hands on any papers directly addressing this that are helping me decide what to do with my data in SPSS yet! :-) But I just requested the article below. you might have direct access at your institution. it might be helpful as you proceed:

Bivariate meta-analysis of sensitivity and specificity with sparse data: a generalized linear mixed model approach. (Chu & Cole, 2006)

Best of luck!

I believe that SPSS is adequate to fit regression models in your project, either by multiple regression or ANCOVA (dependent: head circunference in cm), logistic regression (dependent: normal/abnormal), or generalized linear model (dependent: microcephaly, normal, macrocephaly). Be aware that If your dataset has multiple pregnancies from a same woman, those are not independent observations and should be either some excluded or evaluated by repeated mesures methods.

If you have access to them, the books by Heck, Thomas & Tabata on multilevel modeling via SPSS are pretty good.

• https://www.routledge.com/Multilevel-and-Longitudinal-Modeling-with-IBM-SPSS/Heck-Thomas-Tabata/p/book/9780415817110
• https://www.routledge.com/Multilevel-Modeling-of-Categorical-Outcomes-Using-IBM-SPSS/Heck-Thomas-Tabata/p/book/9781848729568

HTH.