turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- hypothesis testing in threedimensional frequency t...

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

09-07-2009 05:45 AM

Hello,

I have a dataset on the level of couples: two persons, their social

status (upper and lower class), their place of living (A and B), and a

measure of their closeness (kinship level 0, 1, 2). A simple way of

analysing these data is a threedimensional (or rather two

threedimensional or one fourdimensional) frequency table, with row

percentages. The table shows e.g. that in place "A", couples in the

upper class have a tendency to be "closer" to each other (in terms of

kinship) than couples in the lower class. I need a statistical test

for this comparison of row percentages. In a statistics introductory

text, I actually found formulas for calculating confidence limits and

the like for the difference of two row percentages in twodimensional

tables. But I do not see this, particularly for more-dimensional

tables, in SAS.

Or should we try (ordered) logistic regression, with closeness

(kinship level 0, 1, 2) as the dependent variable and combinations of

the other variables (say place, class of person 1 and class of person

2) as predictors? This seems odd for theoretical reasons (social class

does not MAKE people closer in kinship terms, it's probably the other

way round).

Thanks for any suggestions!

Georg Fertig

(History department, Univ. Muenster, Germany)

I have a dataset on the level of couples: two persons, their social

status (upper and lower class), their place of living (A and B), and a

measure of their closeness (kinship level 0, 1, 2). A simple way of

analysing these data is a threedimensional (or rather two

threedimensional or one fourdimensional) frequency table, with row

percentages. The table shows e.g. that in place "A", couples in the

upper class have a tendency to be "closer" to each other (in terms of

kinship) than couples in the lower class. I need a statistical test

for this comparison of row percentages. In a statistics introductory

text, I actually found formulas for calculating confidence limits and

the like for the difference of two row percentages in twodimensional

tables. But I do not see this, particularly for more-dimensional

tables, in SAS.

Or should we try (ordered) logistic regression, with closeness

(kinship level 0, 1, 2) as the dependent variable and combinations of

the other variables (say place, class of person 1 and class of person

2) as predictors? This seems odd for theoretical reasons (social class

does not MAKE people closer in kinship terms, it's probably the other

way round).

Thanks for any suggestions!

Georg Fertig

(History department, Univ. Muenster, Germany)

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

09-09-2009 01:21 PM

Check out GENMOD. It was designed for these sorts of analyses.