Combining two categorical variables

[Diana Cedeno]

Hello everyone!

I am trying to merge two categorical variables in SPSS (gender and marital status). My goal is to create one variable so I can look at single parent households when I run my model. I have crosstabulated these two variables and seen sone interesting numbers.

I am using PSID secondary data (national). Any advice? Thanks in anticipation!

[Bruce Weaver]

I don't understand why you need a single variable that combines gender and marital status to look at single parent households. What are the categories for the two variables in their current form? And what exactly do you wish to test in your regression model? It may be that including an interaction term and following up with some contrasts would allow you to address your research question.

HTH.

[Diana Cedeno]

Hello,

Thanks so much for your reply. I want to test social exclusion for families, and the possible impact upon child development via parental stress (mediation). Because I use secondary data, the variable of single parent households is not available. However, I do have marital status and gender (1 male, 2 female) My adviser said I should figure out a way to combine both variables, to see if there is more stress in single parent households vs two parent ones. My model depends on a scale for measuring parental stress and child behavioral outcomes.

Categories are:
gender (1 male, 2 female)

marital status
1 Married
2 Never married
3 Widowed
4 Divorced, annulled
5 Separated

I did cross-tabulation and found interesting patterns, but I need a variable that I can use to see if single parents are more prone to stress.

Thanks in anticipation for any help provided!

[Bruce Weaver]

On Wednesday, July 26, 2017 at 12:44:20 PM UTC-4, Diana Cedeno wrote:
> Hello,
>
> Thanks so much for your reply. I want to test social exclusion for families, and the possible impact upon child development via parental stress (mediation). Because I use secondary data, the variable of single parent households is not available. However, I do have marital status and gender (1 male, 2 female) My adviser said I should figure out a way to combine both variables, to see if there is more stress in single parent households vs two parent ones. My model depends on a scale for measuring parental stress and child behavioral outcomes.
>
> Categories are:
> gender (1 male, 2 female)
>
> marital status
> 1 Married
> 2 Never married
> 3 Widowed
> 4 Divorced, annulled
> 5 Separated
>
> I did cross-tabulation and found interesting patterns, but I need a variable that I can use to see if single parents are more prone to stress.
>
> Thanks in anticipation for any help provided!

Are you assuming that Married = two-parent family and all other marital status categories represent single-parent families? Does gender enter into it at all? If so, how?

I still can't work out why you need to compute a new variable. Can you not get what you need with contrasts of category 1 vs. other categories of your current marital status variable?

Maybe it would help if you showed us the 2x5 cross-tabulation of gender and marital status.

HTH.

[Rich Ulrich]

I don't think I would combind Dad-only single parent with Mom-only,
unless it was after-the-fact of seeing that (oddly) they were similar
on everything.

If you want to flatten some variables transparently, you can take
(say) COMPUTE Both= 10*var1 + var2 ... and recode the values
to create your two groups, or however many groups you want.


What does "secondary data" (without the variable) mean? - that
you are only dealing with summary data from some analysis?
[ If that is the case ....
Dis-aggregating data that are in aggregate form is something that
is wanted, for instance, from vote counts. Some proposals are crap.
I did not spend enough time with the best-regarded one to figure if
it really worked, and there wasn't (10 years ago) broad empirical
support for it. ]


--
Rich Ulrich

[Diana Cedeno]

Bruce,

Thanks so much for your reply. Yes, I am assuming that marrie couples are two parent households. All other would be single headed households. Gender does enter into it, because family composition would be good to analayze. Below is the crosstable:

SEXHEAD * HEADMARITAL Crosstabulation
Count
Total
Married Never Married Widowed Divorce Sep
Male 2240 143 13 81 33 2510
Female 14 507 65 276 189 1051

Total 2254 650 78 357 222 3561

[Diana Cedeno]

Married Separated
M 2240 33
F 14 189

Never Married
143
507

Widowed
13
65

Divorce
81
276

Separated
33
189


Total 3561

[Diana Cedeno]

Hi Rich,

Thanks for your reply. Secondary data is data that I did no collect - I am using the PSID which has been collecting data from US families since 1969. It's a national public datset.

I am still trying to create a variable on single women head of household because I believe these families will be more vulnerable towards social exclusion, and thus might create behavior problems in children.