genlinmixed confusion

[Maguin, Eugene]

I find the genlinmixed command documentation extremely confusing and I need a bit of help. What I’m doing is a mixed model (multilevel) logistic regression. At this point, I’m simply doing bivariate analyses, cycling through a list of IVs to assess magnitude of the IV coefficient and significance of the intercept variance. Here’s where I’m confused. The reference subcommand of the  target_options command is said to specify the reference category with the default being the last category the DV is sorted in ascending order. So, I say reference=0, given a 0/1 coding on my DV and customary practice for logistic regression. Further down is the build_options command. The target_category_order subcommand is said to control the designation of the reference category. Ok. But what is the difference in function and controlling priority between the reference subcommand in target_options and the target_category_order subcommand?

 

Thanks, Gene Maguin

===================== 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

[Jon Peck]

From the CSR:

TARGET_CATEGORY_ORDER. This determines the order of the target's categories for purposes of

determining the "last" category, which is used as the reference category if the REFERENCE keyword is not

used

On Wed, Sep 12, 2018 at 8:22 AM Maguin, Eugene <[hidden email]> wrote:

I find the genlinmixed command documentation extremely confusing and I need a bit of help. What I’m doing is a mixed model (multilevel) logistic regression. At this point, I’m simply doing bivariate analyses, cycling through a list of IVs to assess magnitude of the IV coefficient and significance of the intercept variance. Here’s where I’m confused. The reference subcommand of the  target_options command is said to specify the reference category with the default being the last category the DV is sorted in ascending order. So, I say reference=0, given a 0/1 coding on my DV and customary practice for logistic regression. Further down is the build_options command. The target_category_order subcommand is said to control the designation of the reference category. Ok. But what is the difference in function and controlling priority between the reference subcommand in target_options and the target_category_order subcommand?

 

Thanks, Gene Maguin

===================== 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

[Maguin, Eugene]

Jon, thank you. I now see and understand the connection. I certainly didn’t before. I understood Target_options and Build_options as separate, non-overlapping commands.

 

From: Jon Peck <[hidden email]>
Sent: Wednesday, September 12, 2018 12:32 PM
To: Maguin, Eugene <[hidden email]>
Cc: SPSS List <[hidden email]>
Subject: Re: [SPSSX-L] genlinmixed confusion

 

From the CSR:

TARGET_CATEGORY_ORDER. This determines the order of the target's categories for purposes of

determining the "last" category, which is used as the reference category if the REFERENCE keyword is not

used

 

On Wed, Sep 12, 2018 at 8:22 AM Maguin, Eugene <[hidden email]> wrote:

I find the genlinmixed command documentation extremely confusing and I need a bit of help. What I’m doing is a mixed model (multilevel) logistic regression. At this point, I’m simply doing bivariate analyses, cycling through a list of IVs to assess magnitude of the IV coefficient and significance of the intercept variance. Here’s where I’m confused. The reference subcommand of the  target_options command is said to specify the reference category with the default being the last category the DV is sorted in ascending order. So, I say reference=0, given a 0/1 coding on my DV and customary practice for logistic regression. Further down is the build_options command. The target_category_order subcommand is said to control the designation of the reference category. Ok. But what is the difference in function and controlling priority between the reference subcommand in target_options and the target_category_order subcommand?

 

Thanks, Gene Maguin

===================== 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

 

--

Jon K Peck
[hidden email]

===================== 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

[Jon Peck]

Designing syntax that is understandable, powerful, unambiguous, has good defaults, and is implementable can be a challenge.  It's nice when subcommands are orthogonal, but sometimes that clashes with other criteria.

On Wed, Sep 12, 2018 at 11:34 AM Maguin, Eugene <[hidden email]> wrote:

Jon, thank you. I now see and understand the connection. I certainly didn’t before. I understood Target_options and Build_options as separate, non-overlapping commands.

 

From: Jon Peck <[hidden email]>
Sent: Wednesday, September 12, 2018 12:32 PM
To: Maguin, Eugene <[hidden email]>
Cc: SPSS List <[hidden email]>
Subject: Re: [SPSSX-L] genlinmixed confusion

 

From the CSR:

TARGET_CATEGORY_ORDER. This determines the order of the target's categories for purposes of

determining the "last" category, which is used as the reference category if the REFERENCE keyword is not

used

 

On Wed, Sep 12, 2018 at 8:22 AM Maguin, Eugene <[hidden email]> wrote:

I find the genlinmixed command documentation extremely confusing and I need a bit of help. What I’m doing is a mixed model (multilevel) logistic regression. At this point, I’m simply doing bivariate analyses, cycling through a list of IVs to assess magnitude of the IV coefficient and significance of the intercept variance. Here’s where I’m confused. The reference subcommand of the  target_options command is said to specify the reference category with the default being the last category the DV is sorted in ascending order. So, I say reference=0, given a 0/1 coding on my DV and customary practice for logistic regression. Further down is the build_options command. The target_category_order subcommand is said to control the designation of the reference category. Ok. But what is the difference in function and controlling priority between the reference subcommand in target_options and the target_category_order subcommand?

 

Thanks, Gene Maguin

===================== 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

 

--

Jon K Peck
[hidden email]

===================== 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