procedure to analyze correlation / predict
Locked
 
|
im looking for a procedure in spss, other than regression, to analyze the correlation between two variables and to see which variables provide a good predictive model for my dependent variable. can anyone recommend anhy of the other statistical procedures in spss for this? thanks!
|
|
|
i understand that neural networks would work well but i dont have the newest neural networks addition to spss. is there anything else that you guys recommend? thanks
|
Re: procedure to analyze correlation / predict
|
|
In reply to this post by jimjohn
I think it would be wise for you to at least read something of the SPSS
documentation before you ask such questions. In addition, it might be wise to consult a statistics book or take a course. I don't mean to be rude, but questions on this website often sound like students asking website members to do their homework -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of jimjohn Sent: Tuesday, May 27, 2008 3:04 PM To: [hidden email] Subject: procedure to analyze correlation / predict im looking for a procedure in spss, other than regression, to analyze the correlation between two variables and to see which variables provide a good predictive model for my dependent variable. can anyone recommend anhy of the other statistical procedures in spss for this? thanks! -- View this message in context: http://www.nabble.com/procedure-to-analyze-correlation---predict-tp17497 715p17497715.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 |
|
|
thanks, i had briefly gone through the spss documentation and each topic is very detailed. i would be very grateful if someone could just provide me the name of a procedure they recommend and after that, i can go ahead and do my own research on how to use that procedure in spss.
|
Re: procedure to analyze correlation / predict
|
|
In reply to this post by Burleson,Joseph A.
By and large, I'm in agreement with Joseph (as are others, apparently - this issue has been brought up previously in response to similar questions).
Is there any way that posts originating from certain domains (e.g., nabble.com) may be banned? This would seem to reduce a sizable chunk of such questions. Best, Matt "Burleson,Joseph A." <[hidden email]> wrote: I think it would be wise for you to at least read something of the SPSS documentation before you ask such questions. In addition, it might be wise to consult a statistics book or take a course. I don't mean to be rude, but questions on this website often sound like students asking website members to do their homework -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of jimjohn Sent: Tuesday, May 27, 2008 3:04 PM To: [hidden email] Subject: procedure to analyze correlation / predict im looking for a procedure in spss, other than regression, to analyze the correlation between two variables and to see which variables provide a good predictive model for my dependent variable. can anyone recommend anhy of the other statistical procedures in spss for this? thanks! -- View this message in context: http://www.nabble.com/procedure-to-analyze-correlation---predict-tp17497 715p17497715.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: procedure to analyze correlation / predict
|
|
In reply to this post by jimjohn
Why isn't linear regression satisfactory?
Scott R Millis, PhD, MEd, ABPP (CN,CL,RP), CStat Professor & Director of Research Dept of Physical Medicine & Rehabilitation Wayne State University School of Medicine 261 Mack Blvd Detroit, MI 48201 Email: [hidden email] Tel: 313-993-8085 Fax: 313-966-7682 --- On Tue, 5/27/08, jimjohn <[hidden email]> wrote: > From: jimjohn <[hidden email]> > Subject: procedure to analyze correlation / predict > To: [hidden email] > Date: Tuesday, May 27, 2008, 3:03 PM > im looking for a procedure in spss, other than regression, > to analyze the > correlation between two variables and to see which > variables provide a good > predictive model for my dependent variable. can anyone > recommend anhy of the > other statistical procedures in spss for this? thanks! > -- > View this message in context: > http://www.nabble.com/procedure-to-analyze-correlation---predict-tp17497715p17497715.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 |
|
|
Hi scott, linear regression works fine for me. I just wanted to see if
there were other predictor models/procedures that would work well too, so maybe I could compare my results to the regression results. Thx. Quoting SR Millis <[hidden email]>: > Why isn't linear regression satisfactory? > > > Scott R Millis, PhD, MEd, ABPP (CN,CL,RP), CStat > Professor & Director of Research > Dept of Physical Medicine & Rehabilitation > Wayne State University School of Medicine > 261 Mack Blvd > Detroit, MI 48201 > Email: [hidden email] > Tel: 313-993-8085 > Fax: 313-966-7682 > > > --- On Tue, 5/27/08, jimjohn <[hidden email]> wrote: > >> From: jimjohn <[hidden email]> >> Subject: procedure to analyze correlation / predict >> To: [hidden email] >> Date: Tuesday, May 27, 2008, 3:03 PM >> im looking for a procedure in spss, other than regression, >> to analyze the >> correlation between two variables and to see which >> variables provide a good >> predictive model for my dependent variable. can anyone >> recommend anhy of the >> other statistical procedures in spss for this? thanks! >> -- >> View this message in context: >> http://www.nabble.com/procedure-to-analyze-correlation---predict-tp17497715p17497715.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: procedure to analyze correlation / predict
|
|
Linear regression can be a good starting point. After fitting the model, examine the residuals. Is there evidence of heteroscedasticity? If so, the Box-Cox transformation might be considered.
Another alternative is to plan for non-linearity prior to fitting the model. First, examine the magnitude of the Spearman r^2 between each predictor variables and the response variable. Next, use restricted cubic splines in your regression model with the number of knots in each spline determined by the magnitude of the Spearman r^2: <=.2: no knots .02-.08: 3 knots .08-.15: 4 knots >.15: 5 knots Scott R Millis, PhD, MEd, ABPP (CN,CL,RP), CStat Professor & Director of Research Dept of Physical Medicine & Rehabilitation Wayne State University School of Medicine 261 Mack Blvd Detroit, MI 48201 Email: [hidden email] Tel: 313-993-8085 Fax: 313-966-7682 --- On Tue, 5/27/08, [hidden email] <[hidden email]> wrote: > From: [hidden email] <[hidden email]> > Subject: Re: procedure to analyze correlation / predict > To: [hidden email] > Date: Tuesday, May 27, 2008, 9:40 PM > Hi scott, linear regression works fine for me. I just wanted > to see if > there were other predictor models/procedures that would > work well too, > so maybe I could compare my results to the regression > results. Thx. > > > Quoting SR Millis <[hidden email]>: > > > Why isn't linear regression satisfactory? ===================== 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 |
|
|
Thanks!
Quoting SR Millis <[hidden email]>: > Linear regression can be a good starting point. After fitting the > model, examine the residuals. Is there evidence of > heteroscedasticity? If so, the Box-Cox transformation might be > considered. > > Another alternative is to plan for non-linearity prior to fitting > the model. First, examine the magnitude of the Spearman r^2 between > each predictor variables and the response variable. Next, use > restricted cubic splines in your regression model with the number of > knots in each spline determined by the magnitude of the Spearman r^2: > > <=.2: no knots > .02-.08: 3 knots > .08-.15: 4 knots >> .15: 5 knots > > > Scott R Millis, PhD, MEd, ABPP (CN,CL,RP), CStat > Professor & Director of Research > Dept of Physical Medicine & Rehabilitation > Wayne State University School of Medicine > 261 Mack Blvd > Detroit, MI 48201 > Email: [hidden email] > Tel: 313-993-8085 > Fax: 313-966-7682 > > > --- On Tue, 5/27/08, [hidden email] <[hidden email]> wrote: > >> From: [hidden email] <[hidden email]> >> Subject: Re: procedure to analyze correlation / predict >> To: [hidden email] >> Date: Tuesday, May 27, 2008, 9:40 PM >> Hi scott, linear regression works fine for me. I just wanted >> to see if >> there were other predictor models/procedures that would >> work well too, >> so maybe I could compare my results to the regression >> results. Thx. >> >> >> Quoting SR Millis <[hidden email]>: >> >> > Why isn't linear regression satisfactory? > > > > ===================== 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 |