Dental data

[PhilF]

Has anyone done any analysis of dental data in SPSS? I do it all the
time but I'm now running into problems of size. My most recent
database has over 700 variables per patient and if I want to follow
them over time I get a second exam set with another 700 variables. How
do I compare or track the changes in those variables efficiently? Do I
use macros? And if so, is there an easy guide to SPSS macros? I
haven't seen any intro guides to it so any of the (many) examples of
it I see on the web just baffle me.

Any help would be greatly appreciated.

[Bruce Weaver]

It sounds like you have the same 700 variables measured on another
occasion, not 700 new variables. For that situation, I'd be
inclined to have multiple rows per person, with a variable called
TIME (or DATE, perhaps). I.e.,

ID TIME V1 V2 ... V700

If the outcome variable you are tracking changes for is
continuous, you could use the MIXED procedure, which requires this
data file structure. On the other hand, if you are using a
procedure that requires one row per subject (e.g., GLM - Repeated
Measures), you could restructure the file (Data - Restructure in
the menus), keeping only the subset of variables you need for the
analysis.

--
Bruce Weaver
bwe...@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

[PhilF]

> bwea...@lakeheadu.cahttp://sites.google.com/a/lakeheadu.ca/bweaver/

> "When all else fails, RTFM."

Thanks for this! You are right - it is the same 700 variables with
either new data or the same for the second exam. The data is not
continuous.
What is the command for comparing two (-->700) variables for the first
and second exam? Is it a simple "if" loop or a macro?

--
Phil Feeley
phi...@interchange.ubc.ca

[Bruce Weaver]

I think it would help if you gave a specific example or two of the
types of things you want to track over time.

--
Bruce Weaver
bwe...@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/Home

[PhilF]

> bwea...@lakeheadu.cahttp://sites.google.com/a/lakeheadu.ca/bweaver/Home

> "When all else fails, RTFM."

Okay. The crowns of the teeth are in one condition at baseline
(filled, carious, sound, etc.). The roots are described using a colour
scale (brown, black, leathery, etc.). I'd like to track all the
surfaces and any changes from first to second exam. Of course, I'll be
getting all kinds of changes - from carious to filled, and even
carious to sound (it does happen!) - or even no change. The same with
the roots. I need to be able to find out - over the 700+ variables -
how many change for the worse and how many (however few) for the
better.

Does that make sense?

Phil

[Bruce Weaver]

I can't quite make out if all of the variables are ordinal. But if
they are, you could compute new variable that is -1 for worsening, 0
for no change, and 1 for improvement, for example. One way to do that
is by computing the Time2 - Time1 difference, and then recoding all
negative values to -1 and all positive values to +1. The 0's can be
left alone. Does this give what you want?

Given that there are 700 variables, you'll want to use some kind of
looping, such as you get with DO-REPEAT. There should be examples in
the Help files.

--
Bruce Weaver
bwe...@lakeheadu.ca

[PhilF]

Thanks for this help. I will try it out.

The variables are all categorical rather than scale.