time series/repeated measure spss analysis
Locked
 
|
I have some time series/repeated measures data that I'm trying to analyze with SPSS. Each case is listed on one row with multiple repeated measures (some up to 8 times). The measures were collected on different dates for each case. Data on number and type of interventions and demographics are also identified in a case's row. My main questions are do # and type of intervention and demographics impact the repeated measures. I do have some missing data--some cases only have, for example, 4 measures on a certain assessment while others have 8. I can't seem to figure out the appropriate analysis although I've tried forecasting, multivariate analysis, etc---all I get in the output is empty boxes with no clear interpretable results. Can someone help--I really need some clear instructions. Thank you.
|
Re: time series/repeated measure spss analysis
Administrator
|
You've used "time series" and "repeated measures" to describe your data. I'm no expert on time series analysis, but I don't think those two terms are synonymous. Assuming it is really closer to "repeated measures", you might want to take a look at "Applied Longitudinal Data Analysis", by Singer & Willett (2003). There is a companion page here that may also be helpful:
http://www.ats.ucla.edu/stat/spss/examples/alda/default.htm HTH.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
I didn't get the original reply, but it is far too vague to provide any *specific* guidance using SPSS. There are circumstances in which a mixed model can very closely resemble a time-series model based on the specified residual correlation structure. I am aware of some [unpublished] work that has been done in this area showing the similarities under some circumstances). Ryan On Wed, Apr 30, 2014 at 2:08 PM, Bruce Weaver <[hidden email]> wrote: You've used "time series" and "repeated measures" to describe your data. I'm |
|
|
Thank you, Ryan and Bruce for replying to my post. I have been looking through the Singer and Willett link, Bruce, and will spend more time doing so. I am not familiar with using a lot of syntax in data analysis so the information was a little overwhelming--but I will continue to study it.
It might be helpful to say that this is a program evaluation study, a pretty big one too. The data that I've described are what we are examining for the implementation study. The program staff want to know if dosage (# of interventions) and intervention type impacts client behavior, measured through various assessments repeatedly every 4 months. Since the program participants are very diverse, they also want to know if certain populations benefit more or less. I had initially thought a multiple regression analysis would work but the more that I read, the more I thought perhaps a time series model or repeated measures analysis would work. If this explanation helps and you have more ideas, I'd appreciate it. Thanks. |
Re: time series/repeated measure spss analysis
|
|
In reply to this post by mdaugher
I do not see a specifics that are specific enough. Are there two variables of
interest as IVs, "Number" and "Type" of intervention? What sort of impact do you expect when you say, "impact the repeated measures"? Do you expect some trend across repeats, because they are across time? Or do the "repeats" reflect something else, (say) the different treatments? Could the hypothesis be well-covered by either the average or the final score of a set of repeats, if not the trend? From one statement, it sounds like there are "sometimes 4, sometimes 8 scores" in each of the sets that is repeated, to account for what is Missing; but that does not account for the statement, "(some up to 8 times)". -- Rich Ulrich > Date: Wed, 30 Apr 2014 10:49:11 -0700 > From: [hidden email] > Subject: time series/repeated measure spss analysis > To: [hidden email] > > I have some time series/repeated measures data that I'm trying to analyze > with SPSS. Each case is listed on one row with multiple repeated measures > (some up to 8 times). The measures were collected on different dates for > each case. Data on number and type of interventions and demographics are > also identified in a case's row. My main questions are do # and type of > intervention and demographics impact the repeated measures. I do have some > missing data--some cases only have, for example, 4 measures on a certain > assessment while others have 8. I can't seem to figure out the appropriate > analysis although I've tried forecasting, multivariate analysis, etc---all I > get in the output is empty boxes with no clear interpretable results. Can > someone help--I really need some clear instructions. Thank you. > |
|
|
Rich, There are actually several IV's of interest: Number and Type of treatment and several demographic variables. The main question is do measures of the repeated assessments change based on the IV's. What analysis would you recommend for this type of data?
I do have quite a bit of missing data though. For example, I was looking at one repeated measure and all cases had at least 4 times of being assessed on it, then it starts to drop with 3/4's having been assessed 5 times, then 1/2 6 times, etc. all the way to only 4 cases measured 8 times. I'm not quite sure as to how to handle this in an analysis. |
Re: time series/repeated measure spss analysis
|
|
Since you have much Missing, one of the first things you should look
at is whether the amount of Missing (or which ones) is associated with lower/higher scores, or with any of the demographic data. If it is not Missing At Random, then you have extra limitations on what you might try to infer. If data are complete through #4, then it *might* be worthwhile to focus attention on that first half; is that sensible and justifiable? - It would avoid the problems that arise from arguing around the Missing. Since you seem to have several treatments, you probably don't have much power for comparisons, in general. (The proper time to Ask a Statistician is throughout the design phase. Just saying.) So you should probably figure on how to do a convincing job of Exploratory Analysis. I raised questions before of what you expected for change - overall, linear, etc. You have not described that, nor have you mentioned what the DVs are so that we could apply our own experience in making guesses. Was the study justified by prior hypotheses? In particular, WHAT? Why are there several DVs? Are they equally important, or should they be used to create a composite outcome of Success? -- Rich Ulrich > Date: Thu, 1 May 2014 08:53:13 -0700 > From: [hidden email] > Subject: Re: time series/repeated measure spss analysis > To: [hidden email] > > Rich, There are actually several IV's of interest: Number and Type of > treatment and several demographic variables. The main question is do > measures of the repeated assessments change based on the IV's. What analysis > would you recommend for this type of data? > > I do have quite a bit of missing data though. For example, I was looking at > one repeated measure and all cases had at least 4 times of being assessed on > it, then it starts to drop with 3/4's having been assessed 5 times, then 1/2 > 6 times, etc. all the way to only 4 cases measured 8 times. I'm not quite > sure as to how to handle this in an analysis. > > > > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/time-series-repeated-measure-spss-analysis-tp5725738p5725758.html > Sent from the SPSSX Discussion mailing list archive at Nabble.com. > > ===================== > To manage your subscription to SPSSX-L, send a message to > [hidden email] (not to SPSSX-L), with no body text except the > command. To leave the list, send the command > SIGNOFF SPSSX-L > For a list of commands to manage subscriptions, send the command > INFO REFCARD |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |