Help - analyzing monthly data for nursing home residents

#1
Hi there -

I am trying to analyze data from approximately 40 nursing home residents. We collected both Cohen-Mansfield and Cornell Brown monthly, and I'm trying to do a couple things with the data -

1) I want to see trends across all categories (Cohen-Mansfield has 19, for example), especially average scores over month 1 (pre) at least one month during, and post (month 14 in this case)

2) Look at each individual person over time and see if their scores have changed

3. Find out average sums of each category in Cohen-Mansfield (there are 4 over-arching ones), and whether those changed over the year.

Basically I want trends and to see if any of the changes are statistically significant. I know I know how to do this...but under pressure I am blanking and I have no one to ask.

I am using SPSS v. 22, and all data is entered.

Thank you in advance for your help. I know this shouldn't be complicated but I just can't figure out what analysis to do. :(
 
#3
They are interval, yes. Cohen Mansfield is 1-7 and Cornell-Brown is technically -2 - +2, but I have it coded as 1 - 5.

I've tried repeated-measure ANOVA but I can't figure out how to get it to work.
 
#5
Oh yeah I guess that's right, thank you so much. So how would I do it with ordinal data?

EDIT: I did change all the fields from nominal to ordinal, which will likely help but i'm still confused. It's been a while since I've done this stuff.
 
Last edited:

noetsi

Fortran must die
#6
Ordinal is more difficult for a before and after test. I do not know of any method that readily does this. Ordered logistic regression can be used for ordered data, but I have not seen it used to test change from one point to another. There are structural equation models that deal with growth, I am not sure if these can be used with categorical variables. If you have not worked with statistics recently doing structural equation growth models is not ideal in any case. the same problem occurs with time series models, they are normally interval in the dependent variable (well the ones I know of are).
 
#7
Essentially all psychological measures are ordinal though, based on likert or something similar. How do you do psychological tests over time? I know a measure of change via psychological tests is used frequently? I've read dozens of studies that have used this, and very few with nominal.