Multinominal hierarchical logistis regression follow up
Locked
 
|
I just discovered that if I moved my variables which I entered as covariates into the factor’s box I then get the option to use various enter methods. But my variables are not factors, many are continuous variables so I treated all of my variables as covariates. With covariates only one does not have the option to use various enter methods. Puzzled. Martin F. Sherman, Ph.D. Professor of Psychology Director of Masters Education in Psychology: Thesis Track Loyola University Maryland Department of Psychology 222 B Beatty Hall 4501 North Charles Street Baltimore, MD 21210 410-617-2417 |
|
|
Martin,
I don't know which procedure you're using. Regardless, however, you should not use step-wise methods. It's fraught with problems, one of which is that valid predictors might not end up in the final model because of they are correlated. Ryan On Tue, Aug 16, 2011 at 8:50 PM, Martin Sherman <[hidden email]> wrote: > I just discovered that if I moved my variables which I entered as covariates > into the factor’s box I then get the option to use various enter methods. > But my variables are not factors, many are continuous variables so I treated > all of my variables as covariates. With covariates only one does not have > the option to use various enter methods. Puzzled. > > > > Martin F. Sherman, Ph.D. > > Professor of Psychology > > Director of Masters Education in Psychology: Thesis Track > > > > Loyola University Maryland > > Department of Psychology > > 222 B Beatty Hall > > 4501 North Charles Street > > Baltimore, MD 21210 > > > > 410-617-2417 > > [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 |
Re: Multinominal hierarchical logistis regression follow up
Administrator
|
In reply to this post by msherman
Martin, when you say you treated all variables as covariates, does that mean you just stuck "factors" in the covariates box without doing anything to them first? For any factors that have more than two categories, you definitely do not wan to do that. For a factor with k levels (or categories), you would need to construct k-1 variables to enter as covariates (to account for the k-1 degrees of freedom). One common approach is to compute k-1 indicator variables (1=in the category, 0 = not in the category). The omitted kth category is the reference category to which the others are compared (via Wald tests) in the table of coefficients.
Re doing hierarchical regression with NOMREG, see my other post. 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: Multinominal hierarchical logistic regression follow up
|
|
And note, Bruce, that the SPSSINC CREATE
DUMMIES extension command will do all the dirty work of creating dummies
and even define a macro covering k-1 of them. Of course, it requires
the Python Essentials.
Jon Peck (no "h") Senior Software Engineer, IBM [hidden email] new phone: 720-342-5621 From: Bruce Weaver <[hidden email]> To: [hidden email] Date: 08/16/2011 08:22 PM Subject: Re: [SPSSX-L] Multinominal hierarchical logistis regression follow up Sent by: "SPSSX(r) Discussion" <[hidden email]> Martin, when you say you treated all variables as covariates, does that mean you just stuck "factors" in the covariates box without doing anything to them first? For any factors that have more than two categories, you definitely do not wan to do that. For a factor with k levels (or categories), you would need to construct k-1 variables to enter as covariates (to account for the k-1 degrees of freedom). One common approach is to compute k-1 indicator variables (1=in the category, 0 = not in the category). The omitted kth category is the reference category to which the others are compared (via Wald tests) in the table of coefficients. Re doing hierarchical regression with NOMREG, see my other post. HTH. msherman wrote: > > I just discovered that if I moved my variables which I entered as > covariates into the factor's box I then get the option to use various > enter methods. But my variables are not factors, many are continuous > variables so I treated all of my variables as covariates. With covariates > only one does not have the option to use various enter methods. Puzzled. > > Martin F. Sherman, Ph.D. > Professor of Psychology > Director of Masters Education in Psychology: Thesis Track > > Loyola University Maryland > Department of Psychology > 222 B Beatty Hall > 4501 North Charles Street > Baltimore, MD 21210 > > 410-617-2417 > [hidden email]<[hidden email]; > ----- -- Bruce Weaver [hidden email] http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." NOTE: My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Multinominal-hierarchical-logistis-regression-follow-up-tp4706318p4706527.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 |
|
|
In reply to this post by Bruce Weaver
Bruce: That makes sense if the variables are categorical but for continuous variables it would generate a huge number of cells making it basically uninterpretable. mfs
-----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver Sent: Tuesday, August 16, 2011 10:20 PM To: [hidden email] Subject: Re: Multinominal hierarchical logistis regression follow up Martin, when you say you treated all variables as covariates, does that mean you just stuck "factors" in the covariates box without doing anything to them first? For any factors that have more than two categories, you definitely do not wan to do that. For a factor with k levels (or categories), you would need to construct k-1 variables to enter as covariates (to account for the k-1 degrees of freedom). One common approach is to compute k-1 indicator variables (1=in the category, 0 = not in the category). The omitted kth category is the reference category to which the others are compared (via Wald tests) in the table of coefficients. Re doing hierarchical regression with NOMREG, see my other post. HTH. msherman wrote: > > I just discovered that if I moved my variables which I entered as > covariates into the factor's box I then get the option to use various > enter methods. But my variables are not factors, many are continuous > variables so I treated all of my variables as covariates. With > covariates only one does not have the option to use various enter methods. Puzzled. > > Martin F. Sherman, Ph.D. > Professor of Psychology > Director of Masters Education in Psychology: Thesis Track > > Loyola University Maryland > Department of Psychology > 222 B Beatty Hall > 4501 North Charles Street > Baltimore, MD 21210 > > 410-617-2417 > [hidden email]<mailto:[hidden email]> > ----- -- Bruce Weaver [hidden email] http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." NOTE: My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Multinominal-hierarchical-logistis-regression-follow-up-tp4706318p4706527.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 |
|
|
Bruce was not suggesting that you dummy code continuous variables. -Ryan
On Wed, Aug 17, 2011 at 9:22 AM, Martin Sherman <[hidden email]> wrote: > Bruce: That makes sense if the variables are categorical but for continuous variables it would generate a huge number of cells making it basically uninterpretable. mfs > > -----Original Message----- > From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver > Sent: Tuesday, August 16, 2011 10:20 PM > To: [hidden email] > Subject: Re: Multinominal hierarchical logistis regression follow up > > Martin, when you say you treated all variables as covariates, does that mean you just stuck "factors" in the covariates box without doing anything to them first? For any factors that have more than two categories, you definitely do not wan to do that. For a factor with k levels (or categories), you would need to construct k-1 variables to enter as covariates (to account for the k-1 degrees of freedom). One common approach is to compute k-1 indicator variables (1=in the category, 0 = not in the category). The omitted kth category is the reference category to which the others are compared (via Wald tests) in the table of coefficients. > > Re doing hierarchical regression with NOMREG, see my other post. > > HTH. > > > > msherman wrote: >> >> I just discovered that if I moved my variables which I entered as >> covariates into the factor's box I then get the option to use various >> enter methods. But my variables are not factors, many are continuous >> variables so I treated all of my variables as covariates. With >> covariates only one does not have the option to use various enter methods. Puzzled. >> >> Martin F. Sherman, Ph.D. >> Professor of Psychology >> Director of Masters Education in Psychology: Thesis Track >> >> Loyola University Maryland >> Department of Psychology >> 222 B Beatty Hall >> 4501 North Charles Street >> Baltimore, MD 21210 >> >> 410-617-2417 >> [hidden email]<mailto:[hidden email]> >> > > > ----- > -- > Bruce Weaver > [hidden email] > http://sites.google.com/a/lakeheadu.ca/bweaver/ > > "When all else fails, RTFM." > > NOTE: My Hotmail account is not monitored regularly. > To send me an e-mail, please use the address shown above. > > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Multinominal-hierarchical-logistis-regression-follow-up-tp4706318p4706527.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 > ===================== 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: Multinominal hierarchical logistis regression follow up
|
Administrator
|
In reply to this post by msherman
You would not compute indicator (or dummy variables) for continuous variables (which are called "covariates" in SPSS lingo). They are entered (as covariates) as they are.
--
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/). |
|
|
In reply to this post by msherman
I've done this by accident (treating a continuous variable as a 'factor'). It was in SAS, but the same principle holds - a several second job wasn't done after 15 minutes or more, since the software needed to set up a huge matrix of dummy variables. Barry Date: Wed, 17 Aug 2011 07:01:19 -0700 From: Bruce Weaver <[hidden email]> Subject: Re: Multinominal hierarchical logistis regression follow up You would not compute indicator (or dummy variables) for continuous variables (which are called "covariates" in SPSS lingo). They are entered (as covariates) as they are. msherman wrote: > > Bruce: That makes sense if the variables are categorical but for > continuous variables it would generate a huge number of cells making it > basically uninterpretable. mfs |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |