How do I weight my cases if using Dental and Insurance Variables over 2008-2012? I have reviewed user notes IHIS_

I am using the following variables over the years 2008-2012: AGE, SEX, MARST, RACENEW, EDUC, INCFAM97ON2, HIPUBCOV, HIPRIVATE, HINONE, SDENTALE, HIPDENTCOVR, DENTINT. I am trying to weight my cases in SPSS. I have read the User Notes and I am still confused. I know I need to use PERWEIGHT, but I also need to use The SUPPxWT series weight. Which SUPP series do I use? What would be my syntax for SPSS to create my weight variable?


Compute dentweight = perweight if year = 2008; dentweight = supp2wt if year 2008

Is this right?

For the variables you listed, you do not need to use any of the SUPPxWT weights. You can check the weight associated with each variable on the ‘Weights’ tab of the variable page, e.g. MARST or DENTINT. All of the variables in your analysis use PERWEIGHT, except for the variable DENTINT. It uses SAMPWEIGHT during the years in your extract.

DENTINT has a different weight, because that question was asked of one random person (adult or child) within a sample household. The other variables you are using were eligible to be asked of all persons within a household. When using DENTINT in your analysis, this will restrict your sample from ‘all respondents’ to ‘one respondent per household’ and you should use SAMPWEIGHT. For any analysis with these variables that does not include DENTINT, you can work with the full IHIS sample, which means you should use PERWEIGHT. Therefore, you actually do not have to create a new weight variable.

Hope this helps.


My dependent variable is DENTINT, so i will be using it during all of my tests. What do I do then with combining the weights? Or do I only use the SAMPWEIGHT when I am going to control for year?

If you will be using DENTINT for all tests, then you should always use SAMPWEIGHT as your weight. This is the weight applicable to DENTINT during the years of your analysis. In other words, by choosing DENTINT as your dependent variable, you will be using a subsample of the full sample; thus, SAMPWEIGHT should be used instead of PERWEIGHT.

Hope this helps.