Running MIXED with time as a factor
Locked
 
|
Dear subscribers,
I would like to get the parameter estimation on ALL the categories (time 1, time 2, time 3) in my three-way interaction analysis. As of now, SPSS only estimates time 1 and 2 (as it often does with categorical data). How may I obtain this information? Please find present syntax below: mixed Exlog with OPce Posce pssce pcsce Gender Age Sector Respons Tenure Hour_w BY time /CRITERIA=MXITER (5000) /print=solution /method=reml /fixed = Gender Age Sector Respons Tenure Hour_w OPce time OPce*time OPce*POSce OPce*PsSce OPce*PcSce PoSce pssce pcsce pssce*time pcsce*time Posce*time POSce*OPce*time PsSce*OPce*time PcSce*OPce*time /random intercept time | subject(id) covtype(VC). Best regards, Kine |
Re: Running MIXED with time as a factor
|
|
I think you may be a bit confused on how to interpret the coefficients. The 3rd category is in the model, it's just a redundant parameter. All the information for getting the time 3 coefficient is in the model. Let's look at it a bit more simply first. If you had a model that only had time and nothing else. Then time 3 would be the intercept value. Now let's add in a gender variable. If 1 is Boy, then the intercept is now the time 3 coefficient for girls. If you believe growth is different for boys and girls, you would include an interaction for gender by time. So how do you know the time 3 coefficient for boys. It's the intercept plus the different for gender, plus the time coefficient plus the difference for gender of that. Where is that coefficient, it's the interaction coefficient.
I know a lot of people will calculate all the coefficients in something like excel to make this easier, and that is certainly something you could do. Matthew J Poes Research Data Specialist Center for Prevention Research and Development University of Illinois 510 Devonshire Dr. Champaign, IL 61820 Phone: 217-265-4576 email: [hidden email] -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of kine Sent: Thursday, June 07, 2012 9:00 AM To: [hidden email] Subject: Running MIXED with time as a factor Dear subscribers, I would like to get the parameter estimation on ALL the categories (time 1, time 2, time 3) in my three-way interaction analysis. As of now, SPSS only estimates time 1 and 2 (as it often does with categorical data). How may I obtain this information? Please find present syntax below: mixed Exlog with OPce Posce pssce pcsce Gender Age Sector Respons Tenure Hour_w BY time /CRITERIA=MXITER (5000) /print=solution /method=reml /fixed = Gender Age Sector Respons Tenure Hour_w OPce time OPce*time OPce*POSce OPce*PsSce OPce*PcSce PoSce pssce pcsce pssce*time pcsce*time Posce*time POSce*OPce*time PsSce*OPce*time PcSce*OPce*time /random intercept time | subject(id) covtype(VC). Best regards, Kine -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Running-MIXED-with-time-as-a-factor-tp5713561.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 ===================== 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 |
Re: Running MIXED with time as a factor
Administrator
|
In reply to this post by kine
MIXED has an EMMEANS sub-command. You can use it to generate fitted values of Y at various combinations of the explanatory variables. Will that give you what you're after? Here's an example for a much simpler model from the fine manual.
MIXED Y BY A B WITH X /FIXED A B X /EMMEANS TABLES(A*B) WITH(X=0.23) COMPARE(A) ADJ(SIDAK) /EMMEANS TABLES(A*B) WITH(X=MEAN) COMPARE(A) REFCAT(LAST) ADJ(LSD). MIXED also has a TEST sub-command, which can be used for crafting any comparisons you wish to make. See several posts to this list by Ryan for nice examples of how to use it. 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/). |
Re: Running MIXED with time as a factor
|
|
In reply to this post by kine
Kine,
You have a complex analysis in that it has one factor and lots of covariates and interactions among covariates and with the factor plus a random intercept and time effects. But you have duplicated terms (see below). In the common meaning of 'three way interaction', you can't have one because you have one factor, time. Yes, you interactions involving time and two covariates but covariates are typically thought of as control variables and in the EMMEANS statements that Bruce just referred to you'll specify that you want the means of the DV at the levels of the time factor with the covariates fixed at specific points. You say ' ... get the parameter estimation on ALL the categories (time 1, time 2, time 3) in my three-way interaction analysis. ... SPSS only estimates time 1 and 2 (as it often does with categorical data). Is it possible that missing data on covariates and DV eliminate all the time=3 cases? fixed = Gender Age Sector Respons Tenure Hour_w OPce time OPce*time OPce*POSce OPce*PsSce OPce*PcSce PoSce pssce pcsce pssce*time pcsce*time Posce*time POSce*OPce*time PsSce*OPce*time PcSce*OPce*time Gene Maguin -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of kine Sent: Thursday, June 07, 2012 10:00 AM To: [hidden email] Subject: Running MIXED with time as a factor Dear subscribers, I would like to get the parameter estimation on ALL the categories (time 1, time 2, time 3) in my three-way interaction analysis. As of now, SPSS only estimates time 1 and 2 (as it often does with categorical data). How may I obtain this information? Please find present syntax below: mixed Exlog with OPce Posce pssce pcsce Gender Age Sector Respons Tenure Hour_w BY time /CRITERIA=MXITER (5000) /print=solution /method=reml /fixed = Gender Age Sector Respons Tenure Hour_w OPce time OPce*time OPce*POSce OPce*PsSce OPce*PcSce PoSce pssce pcsce pssce*time pcsce*time Posce*time POSce*OPce*time PsSce*OPce*time PcSce*OPce*time /random intercept time | subject(id) covtype(VC). Best regards, Kine -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Running-MIXED-with-time-as-a-factor-tp5713561.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 ===================== 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 |
|
|
In reply to this post by Bruce Weaver
For some reason I never received the original post in my inbox. At any rate, I have concerns about both the complexity of the fixed effects component and type of the random effects covariance structure selected.
Since the OP has not provided any information regarding the actual study, variables, data, research questions, etc., it's not possible for me to provide any general advice about the model in question.
Finally, given the way the specific question was stated, it's likely the OP's understanding of the fixed effects coefficients is incorrect. Ryan
On Thu, Jun 7, 2012 at 10:31 AM, Bruce Weaver <[hidden email]> wrote: MIXED has an EMMEANS sub-command. You can use it to generate fitted values |
|
|
In reply to this post by Poes, Matthew Joseph
Wow, thank you all for the ellaborate and quick replies! |
| Free forum by Nabble | Edit this page |