Fw: Re: Logistic Regression - Threshold effect sizes?
Locked
Fw: Re: Logistic Regression - Threshold effect sizes?
 
|
Jon,
If you're uncomfortable treating your ordinal variable as though it were interval, you can always "dummy code" the ordinal variable---such that the 5-category variable become 4 separate variables---and direct comparisons can be made. This can work pretty well if you don't have a lot of ordinal vairables and if you're sample size is sufficiently large. Scott Millis > --- On Fri, 9/5/08, Justin Black > <[hidden email]> wrote: > > > From: Justin Black <[hidden email]> > > Subject: Re: Logistic Regression - Threshold effect > sizes? > > To: "SR Millis" <[hidden email]> > > Date: Friday, September 5, 2008, 6:13 PM > > Scott, thank you for following up. > > > > What's puzzling to me is the seemingly omnibus > nature > > of the effect. So, I > > could say that a 1-point increase in IV1 is associated > with > > a, e.g., 40% > > increase in the likelihood of the event occurring. > But > > that assumes that a > > 1-point increase in the IV has the same effect on the > DV > > regardless of the > > baseline level of the IV. I don't think that > > assumption is accurate in > > these particular data. > > > > I really feel like I'm missing something here, > just > > can't figure out what. > > A look back at Hosmer & Lemeshow didn't help > any. > > > > Thank you, > > > > Jon > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > <[hidden email]> wrote: > > > > > Jon, > > > > > > What is puzzling in the results? > > > > > > Have you examined the degree of collinearity > among the > > > predictors/covariates? > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > <[hidden email]> wrote: > > > > > > > From: Jon Bernard <[hidden email]> > > > > Subject: Logistic Regression - Threshold > effect > > sizes? > > > > To: [hidden email] > > > > Date: Friday, September 5, 2008, 5:42 PM > > > > Fellow SPSSers, > > > > > > > > I am struggling with a logistic regression > issue > > and > > > > thought I'd put it out > > > > to the list for some clarity. > > > > > > > > The dependent variable of interest is a > binary > > event (0 = > > > > Did not happen, 1 > > > > = Did happen). The independent variables of > > interest are > > > > ordinal > > > > attitudinal survey items, with responses on > a > > 5-point scale > > > > (1 = Strongly > > > > Disagree, 5 = Strongly Agree). I have been > using > > logistic > > > > regression for > > > > the analysis, but either I'm missing > > something in the > > > > results output or I'm > > > > using the wrong statistical technique. I > have an > > inkling > > > > that the intervals > > > > between categories of the independent > variables > > are not all > > > > equal. In other > > > > words, I think that the impact on the > dependent > > variable of > > > > an independent > > > > variable score of 3 vs. one of 2 is greater > than > > that of a > > > > score of 5 vs. > > > > one of 4. Is that clear? If so, what would > you > > recommend > > > > in order to test > > > > that hypothesis? Is there a class of > techniques > > designed > > > > particularly for > > > > this kind of test? > > > > > > > > Very many thanks in advance for your > assistance > > with this > > > > matter. > > > > > > > > Kind regards, > > > > > > > > Jon > > > > > > > > ===================== > > > > 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: Fw: Re: Logistic Regression - Threshold effect sizes?
|
|
Thank you Scott. Would you recommend any resources for a quick study on
this technique? Jon On Fri, Sep 5, 2008 at 6:18 PM, SR Millis <[hidden email]> wrote: > Jon, > > If you're uncomfortable treating your ordinal variable > as though it were interval, you can always "dummy > code" the ordinal variable---such that the 5-category > variable become 4 separate variables---and direct > comparisons can be made. This can work pretty well if you > don't have a lot of ordinal vairables and if you're > sample size is sufficiently large. > > Scott Millis > > > --- On Fri, 9/5/08, Justin Black > > <[hidden email]> wrote: > > > > > From: Justin Black <[hidden email]> > > > Subject: Re: Logistic Regression - Threshold effect > > sizes? > > > To: "SR Millis" <[hidden email]> > > > Date: Friday, September 5, 2008, 6:13 PM > > > Scott, thank you for following up. > > > > > > What's puzzling to me is the seemingly omnibus > > nature > > > of the effect. So, I > > > could say that a 1-point increase in IV1 is associated > > with > > > a, e.g., 40% > > > increase in the likelihood of the event occurring. > > But > > > that assumes that a > > > 1-point increase in the IV has the same effect on the > > DV > > > regardless of the > > > baseline level of the IV. I don't think that > > > assumption is accurate in > > > these particular data. > > > > > > I really feel like I'm missing something here, > > just > > > can't figure out what. > > > A look back at Hosmer & Lemeshow didn't help > > any. > > > > > > Thank you, > > > > > > Jon > > > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > > <[hidden email]> wrote: > > > > > > > Jon, > > > > > > > > What is puzzling in the results? > > > > > > > > Have you examined the degree of collinearity > > among the > > > > predictors/covariates? > > > > > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > > <[hidden email]> wrote: > > > > > > > > > From: Jon Bernard <[hidden email]> > > > > > Subject: Logistic Regression - Threshold > > effect > > > sizes? > > > > > To: [hidden email] > > > > > Date: Friday, September 5, 2008, 5:42 PM > > > > > Fellow SPSSers, > > > > > > > > > > I am struggling with a logistic regression > > issue > > > and > > > > > thought I'd put it out > > > > > to the list for some clarity. > > > > > > > > > > The dependent variable of interest is a > > binary > > > event (0 = > > > > > Did not happen, 1 > > > > > = Did happen). The independent variables of > > > interest are > > > > > ordinal > > > > > attitudinal survey items, with responses on > > a > > > 5-point scale > > > > > (1 = Strongly > > > > > Disagree, 5 = Strongly Agree). I have been > > using > > > logistic > > > > > regression for > > > > > the analysis, but either I'm missing > > > something in the > > > > > results output or I'm > > > > > using the wrong statistical technique. I > > have an > > > inkling > > > > > that the intervals > > > > > between categories of the independent > > variables > > > are not all > > > > > equal. In other > > > > > words, I think that the impact on the > > dependent > > > variable of > > > > > an independent > > > > > variable score of 3 vs. one of 2 is greater > > than > > > that of a > > > > > score of 5 vs. > > > > > one of 4. Is that clear? If so, what would > > you > > > recommend > > > > > in order to test > > > > > that hypothesis? Is there a class of > > techniques > > > designed > > > > > particularly for > > > > > this kind of test? > > > > > > > > > > Very many thanks in advance for your > > assistance > > > with this > > > > > matter. > > > > > > > > > > Kind regards, > > > > > > > > > > Jon > > > > > > > > > > ===================== > > > > > 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: Fw: Re: Logistic Regression - Threshold effect sizes?
|
|
Jon,
If you're using SPSS binary logistic regression, enter your ordinal variable as a covariate. Then, click on the Categorical button. This will take you to a new window where you indicate which covariates are categorical. The Contrast should be "Indicator." Then run the analysis. If you're using Stata software, you use the xi option. Scott Millis --- On Fri, 9/5/08, Jon Bernard <[hidden email]> wrote: > From: Jon Bernard <[hidden email]> > Subject: Re: Fw: Re: Logistic Regression - Threshold effect sizes? > To: "SR Millis" <[hidden email]> > Cc: [hidden email] > Date: Friday, September 5, 2008, 6:24 PM > Thank you Scott. Would you recommend any resources for a > quick study on > this technique? > > Jon > > On Fri, Sep 5, 2008 at 6:18 PM, SR Millis > <[hidden email]> wrote: > > > Jon, > > > > If you're uncomfortable treating your ordinal > variable > > as though it were interval, you can always "dummy > > code" the ordinal variable---such that the > 5-category > > variable become 4 separate variables---and direct > > comparisons can be made. This can work pretty well if > you > > don't have a lot of ordinal vairables and if > you're > > sample size is sufficiently large. > > > > Scott Millis > > > > > --- On Fri, 9/5/08, Justin Black > > > <[hidden email]> wrote: > > > > > > > From: Justin Black > <[hidden email]> > > > > Subject: Re: Logistic Regression - Threshold > effect > > > sizes? > > > > To: "SR Millis" > <[hidden email]> > > > > Date: Friday, September 5, 2008, 6:13 PM > > > > Scott, thank you for following up. > > > > > > > > What's puzzling to me is the seemingly > omnibus > > > nature > > > > of the effect. So, I > > > > could say that a 1-point increase in IV1 is > associated > > > with > > > > a, e.g., 40% > > > > increase in the likelihood of the event > occurring. > > > But > > > > that assumes that a > > > > 1-point increase in the IV has the same > effect on the > > > DV > > > > regardless of the > > > > baseline level of the IV. I don't think > that > > > > assumption is accurate in > > > > these particular data. > > > > > > > > I really feel like I'm missing something > here, > > > just > > > > can't figure out what. > > > > A look back at Hosmer & Lemeshow > didn't help > > > any. > > > > > > > > Thank you, > > > > > > > > Jon > > > > > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > > > <[hidden email]> wrote: > > > > > > > > > Jon, > > > > > > > > > > What is puzzling in the results? > > > > > > > > > > Have you examined the degree of > collinearity > > > among the > > > > > predictors/covariates? > > > > > > > > > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > > > <[hidden email]> wrote: > > > > > > > > > > > From: Jon Bernard > <[hidden email]> > > > > > > Subject: Logistic Regression - > Threshold > > > effect > > > > sizes? > > > > > > To: [hidden email] > > > > > > Date: Friday, September 5, 2008, > 5:42 PM > > > > > > Fellow SPSSers, > > > > > > > > > > > > I am struggling with a logistic > regression > > > issue > > > > and > > > > > > thought I'd put it out > > > > > > to the list for some clarity. > > > > > > > > > > > > The dependent variable of interest > is a > > > binary > > > > event (0 = > > > > > > Did not happen, 1 > > > > > > = Did happen). The independent > variables of > > > > interest are > > > > > > ordinal > > > > > > attitudinal survey items, with > responses on > > > a > > > > 5-point scale > > > > > > (1 = Strongly > > > > > > Disagree, 5 = Strongly Agree). I > have been > > > using > > > > logistic > > > > > > regression for > > > > > > the analysis, but either I'm > missing > > > > something in the > > > > > > results output or I'm > > > > > > using the wrong statistical > technique. I > > > have an > > > > inkling > > > > > > that the intervals > > > > > > between categories of the > independent > > > variables > > > > are not all > > > > > > equal. In other > > > > > > words, I think that the impact on > the > > > dependent > > > > variable of > > > > > > an independent > > > > > > variable score of 3 vs. one of 2 > is greater > > > than > > > > that of a > > > > > > score of 5 vs. > > > > > > one of 4. Is that clear? If so, > what would > > > you > > > > recommend > > > > > > in order to test > > > > > > that hypothesis? Is there a class > of > > > techniques > > > > designed > > > > > > particularly for > > > > > > this kind of test? > > > > > > > > > > > > Very many thanks in advance for > your > > > assistance > > > > with this > > > > > > matter. > > > > > > > > > > > > Kind regards, > > > > > > > > > > > > Jon > > > > > > > > > > > > ===================== > > > > > > 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: Logistic Regression - Threshold effect sizes?
|
|
In reply to this post by SR Millis-3
Dummy variables are only partially useful in the case of Jon, because they
do not necessarily generate ordered results (as expected). Jon: to create dummy variables in general you simply create one dummy with values (0,1) for each category minus one. The omitted category is omitted because it can be deduced from the rest. General syntax for one category: COMPUTE DUMMYCAT3=(yourvar=3). The new variable, DUMMYCAT3, will equal 1 whenever your original variable YOURVAR equals 3, and will equal 0 otherwise. Repeat this for all categories except one (for instance, you may choose the first or last category for omission). However, if you have represented your 5-category variables by four dummies each, you are not guaranteed to obtain monotonic results, i.e. that the effect increases or decreases in a monotonic way (by equal or different amounts) for each increase in the ordinal response from 1 to 5. You may well find that the effect increases for category 1 and 3, but decreases for category 2 and 4, which could be incomprehensible from a theoretical point of view and may be a random effect of your sample. You can create "incremental dummies" by assigning the value 1 to cases having one value OR LESS: COMPUTE UPTOONE=(YOURVAR=1). COMPUTE UPTOTWO=(YOURVAR LE 2). COMPUTE UPTOTHREE=(YOURVAR LE 3). COMPUTE UPTOFOUR=(YOURVAR LE 4). (it is not necessary to create UPTO FIVE, because people choosing category five can be deduced as the complement of UPTOFOUR). The increase (or decrease) in the effect between UPTOTWO and UPTOTHREE would be the specific effect of choosing 3. This ASSUMES the underlying variable represented by your ordinal IV is monotonic. If you are not sure about this, you may use simple dummies as in my first example with DUMMYCAT1 to DUMMYCAT4. Or you may try both and look at the results. In any case, try to avoid dummies where very few cases have 0 or 1 (i.e. dummies with very low or very high frequencies in each value): results could be unstable and unreliable. If, for instance, very few people chose category 5, you may group them with category 5 and define the variable in question as a four-category ordinal variable. Of course this way you sacrifice information (difference between 4 and 5) but you avoid the unpleasant consequences of a very small sample of cases in category 5. Hope this helps. Hector -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of SR Millis Sent: 05 September 2008 18:19 To: [hidden email] Subject: Fw: Re: Logistic Regression - Threshold effect sizes? Jon, If you're uncomfortable treating your ordinal variable as though it were interval, you can always "dummy code" the ordinal variable---such that the 5-category variable become 4 separate variables---and direct comparisons can be made. This can work pretty well if you don't have a lot of ordinal vairables and if you're sample size is sufficiently large. Scott Millis > --- On Fri, 9/5/08, Justin Black > <[hidden email]> wrote: > > > From: Justin Black <[hidden email]> > > Subject: Re: Logistic Regression - Threshold effect > sizes? > > To: "SR Millis" <[hidden email]> > > Date: Friday, September 5, 2008, 6:13 PM > > Scott, thank you for following up. > > > > What's puzzling to me is the seemingly omnibus > nature > > of the effect. So, I > > could say that a 1-point increase in IV1 is associated > with > > a, e.g., 40% > > increase in the likelihood of the event occurring. > But > > that assumes that a > > 1-point increase in the IV has the same effect on the > DV > > regardless of the > > baseline level of the IV. I don't think that > > assumption is accurate in > > these particular data. > > > > I really feel like I'm missing something here, > just > > can't figure out what. > > A look back at Hosmer & Lemeshow didn't help > any. > > > > Thank you, > > > > Jon > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > <[hidden email]> wrote: > > > > > Jon, > > > > > > What is puzzling in the results? > > > > > > Have you examined the degree of collinearity > among the > > > predictors/covariates? > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > <[hidden email]> wrote: > > > > > > > From: Jon Bernard <[hidden email]> > > > > Subject: Logistic Regression - Threshold > effect > > sizes? > > > > To: [hidden email] > > > > Date: Friday, September 5, 2008, 5:42 PM > > > > Fellow SPSSers, > > > > > > > > I am struggling with a logistic regression > issue > > and > > > > thought I'd put it out > > > > to the list for some clarity. > > > > > > > > The dependent variable of interest is a > binary > > event (0 = > > > > Did not happen, 1 > > > > = Did happen). The independent variables of > > interest are > > > > ordinal > > > > attitudinal survey items, with responses on > a > > 5-point scale > > > > (1 = Strongly > > > > Disagree, 5 = Strongly Agree). I have been > using > > logistic > > > > regression for > > > > the analysis, but either I'm missing > > something in the > > > > results output or I'm > > > > using the wrong statistical technique. I > have an > > inkling > > > > that the intervals > > > > between categories of the independent > variables > > are not all > > > > equal. In other > > > > words, I think that the impact on the > dependent > > variable of > > > > an independent > > > > variable score of 3 vs. one of 2 is greater > than > > that of a > > > > score of 5 vs. > > > > one of 4. Is that clear? If so, what would > you > > recommend > > > > in order to test > > > > that hypothesis? Is there a class of > techniques > > designed > > > > particularly for > > > > this kind of test? > > > > > > > > Very many thanks in advance for your > assistance > > with this > > > > matter. > > > > > > > > Kind regards, > > > > > > > > Jon > > > > > > > > ===================== > > > > 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: Logistic Regression - Threshold effect sizes?
|
|
Hi Hector and all,
Would the approach of entering both sets of dummy codes into a stepwise procedure (Forward LR) and allowing the modelling process to pick out the significant dummy codes be a valid approach (hoping that it would only pick dummy codes from the same set. Not sure what it would mean if dummy codes from different sets were to be significant....) ? I have one other question on this subject. How should missing values be treated? As a quick and dirty approach, would coding them as zeros be appropriate (simply so that cases are not excluded for the listwise procedure). Many thanks Jigs 2008/9/6 Hector Maletta <[hidden email]> > Dummy variables are only partially useful in the case of Jon, because they > do not necessarily generate ordered results (as expected). > Jon: to create dummy variables in general you simply create one dummy with > values (0,1) for each category minus one. The omitted category is omitted > because it can be deduced from the rest. General syntax for one category: > COMPUTE DUMMYCAT3=(yourvar=3). > The new variable, DUMMYCAT3, will equal 1 whenever your original variable > YOURVAR equals 3, and will equal 0 otherwise. Repeat this for all > categories > except one (for instance, you may choose the first or last category for > omission). > However, if you have represented your 5-category variables by four dummies > each, you are not guaranteed to obtain monotonic results, i.e. that the > effect increases or decreases in a monotonic way (by equal or different > amounts) for each increase in the ordinal response from 1 to 5. You may > well > find that the effect increases for category 1 and 3, but decreases for > category 2 and 4, which could be incomprehensible from a theoretical point > of view and may be a random effect of your sample. > You can create "incremental dummies" by assigning the value 1 to cases > having one value OR LESS: > COMPUTE UPTOONE=(YOURVAR=1). > COMPUTE UPTOTWO=(YOURVAR LE 2). > COMPUTE UPTOTHREE=(YOURVAR LE 3). > COMPUTE UPTOFOUR=(YOURVAR LE 4). > (it is not necessary to create UPTO FIVE, because people choosing category > five can be deduced as the complement of UPTOFOUR). > The increase (or decrease) in the effect between UPTOTWO and UPTOTHREE > would > be the specific effect of choosing 3. This ASSUMES the underlying variable > represented by your ordinal IV is monotonic. If you are not sure about > this, > you may use simple dummies as in my first example with DUMMYCAT1 to > DUMMYCAT4. Or you may try both and look at the results. > In any case, try to avoid dummies where very few cases have 0 or 1 (i.e. > dummies with very low or very high frequencies in each value): results > could > be unstable and unreliable. If, for instance, very few people chose > category > 5, you may group them with category 5 and define the variable in question > as > a four-category ordinal variable. Of course this way you sacrifice > information (difference between 4 and 5) but you avoid the unpleasant > consequences of a very small sample of cases in category 5. > Hope this helps. > > Hector > -----Original Message----- > From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of > SR > Millis > Sent: 05 September 2008 18:19 > To: [hidden email] > Subject: Fw: Re: Logistic Regression - Threshold effect sizes? > > Jon, > > If you're uncomfortable treating your ordinal variable > as though it were interval, you can always "dummy > code" the ordinal variable---such that the 5-category > variable become 4 separate variables---and direct > comparisons can be made. This can work pretty well if you > don't have a lot of ordinal vairables and if you're > sample size is sufficiently large. > > Scott Millis > > > --- On Fri, 9/5/08, Justin Black > > <[hidden email]> wrote: > > > > > From: Justin Black <[hidden email]> > > > Subject: Re: Logistic Regression - Threshold effect > > sizes? > > > To: "SR Millis" <[hidden email]> > > > Date: Friday, September 5, 2008, 6:13 PM > > > Scott, thank you for following up. > > > > > > What's puzzling to me is the seemingly omnibus > > nature > > > of the effect. So, I > > > could say that a 1-point increase in IV1 is associated > > with > > > a, e.g., 40% > > > increase in the likelihood of the event occurring. > > But > > > that assumes that a > > > 1-point increase in the IV has the same effect on the > > DV > > > regardless of the > > > baseline level of the IV. I don't think that > > > assumption is accurate in > > > these particular data. > > > > > > I really feel like I'm missing something here, > > just > > > can't figure out what. > > > A look back at Hosmer & Lemeshow didn't help > > any. > > > > > > Thank you, > > > > > > Jon > > > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > > <[hidden email]> wrote: > > > > > > > Jon, > > > > > > > > What is puzzling in the results? > > > > > > > > Have you examined the degree of collinearity > > among the > > > > predictors/covariates? > > > > > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > > <[hidden email]> wrote: > > > > > > > > > From: Jon Bernard <[hidden email]> > > > > > Subject: Logistic Regression - Threshold > > effect > > > sizes? > > > > > To: [hidden email] > > > > > Date: Friday, September 5, 2008, 5:42 PM > > > > > Fellow SPSSers, > > > > > > > > > > I am struggling with a logistic regression > > issue > > > and > > > > > thought I'd put it out > > > > > to the list for some clarity. > > > > > > > > > > The dependent variable of interest is a > > binary > > > event (0 = > > > > > Did not happen, 1 > > > > > = Did happen). The independent variables of > > > interest are > > > > > ordinal > > > > > attitudinal survey items, with responses on > > a > > > 5-point scale > > > > > (1 = Strongly > > > > > Disagree, 5 = Strongly Agree). I have been > > using > > > logistic > > > > > regression for > > > > > the analysis, but either I'm missing > > > something in the > > > > > results output or I'm > > > > > using the wrong statistical technique. I > > have an > > > inkling > > > > > that the intervals > > > > > between categories of the independent > > variables > > > are not all > > > > > equal. In other > > > > > words, I think that the impact on the > > dependent > > > variable of > > > > > an independent > > > > > variable score of 3 vs. one of 2 is greater > > than > > > that of a > > > > > score of 5 vs. > > > > > one of 4. Is that clear? If so, what would > > you > > > recommend > > > > > in order to test > > > > > that hypothesis? Is there a class of > > techniques > > > designed > > > > > particularly for > > > > > this kind of test? > > > > > > > > > > Very many thanks in advance for your > > assistance > > > with this > > > > > matter. > > > > > > > > > > Kind regards, > > > > > > > > > > Jon > > > > > > > > > > ===================== > > > > > 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 > ===================== 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: Logistic Regression - Threshold effect sizes?
|
|
Jig,
I don't think it would be a good idea. On the one hand, it might pick some of the dummies and not others. On the other hand, there are serious objections against the stepwise procedure that have been discussed in this forum before (you may browse the archives to find them). I would advise using one or the other set of dummies according to one's conception of the corresponding IV. Hector _____ From: [hidden email] [mailto:[hidden email]] On Behalf Of J Sutar Sent: 07 September 2008 15:45 To: Hector Maletta Cc: [hidden email] Subject: Re: Logistic Regression - Threshold effect sizes? Hi Hector and all, Would the approach of entering both sets of dummy codes into a stepwise procedure (Forward LR) and allowing the modelling process to pick out the significant dummy codes be a valid approach (hoping that it would only pick dummy codes from the same set. Not sure what it would mean if dummy codes from different sets were to be significant....) ? I have one other question on this subject. How should missing values be treated? As a quick and dirty approach, would coding them as zeros be appropriate (simply so that cases are not excluded for the listwise procedure). Many thanks Jigs 2008/9/6 Hector Maletta <[hidden email]> Dummy variables are only partially useful in the case of Jon, because they do not necessarily generate ordered results (as expected). Jon: to create dummy variables in general you simply create one dummy with values (0,1) for each category minus one. The omitted category is omitted because it can be deduced from the rest. General syntax for one category: COMPUTE DUMMYCAT3=(yourvar=3). The new variable, DUMMYCAT3, will equal 1 whenever your original variable YOURVAR equals 3, and will equal 0 otherwise. Repeat this for all categories except one (for instance, you may choose the first or last category for omission). However, if you have represented your 5-category variables by four dummies each, you are not guaranteed to obtain monotonic results, i.e. that the effect increases or decreases in a monotonic way (by equal or different amounts) for each increase in the ordinal response from 1 to 5. You may well find that the effect increases for category 1 and 3, but decreases for category 2 and 4, which could be incomprehensible from a theoretical point of view and may be a random effect of your sample. You can create "incremental dummies" by assigning the value 1 to cases having one value OR LESS: COMPUTE UPTOONE=(YOURVAR=1). COMPUTE UPTOTWO=(YOURVAR LE 2). COMPUTE UPTOTHREE=(YOURVAR LE 3). COMPUTE UPTOFOUR=(YOURVAR LE 4). (it is not necessary to create UPTO FIVE, because people choosing category five can be deduced as the complement of UPTOFOUR). The increase (or decrease) in the effect between UPTOTWO and UPTOTHREE would be the specific effect of choosing 3. This ASSUMES the underlying variable represented by your ordinal IV is monotonic. If you are not sure about this, you may use simple dummies as in my first example with DUMMYCAT1 to DUMMYCAT4. Or you may try both and look at the results. In any case, try to avoid dummies where very few cases have 0 or 1 (i.e. dummies with very low or very high frequencies in each value): results could be unstable and unreliable. If, for instance, very few people chose category 5, you may group them with category 5 and define the variable in question as a four-category ordinal variable. Of course this way you sacrifice information (difference between 4 and 5) but you avoid the unpleasant consequences of a very small sample of cases in category 5. Hope this helps. Hector -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of SR Millis Sent: 05 September 2008 18:19 To: [hidden email] Subject: Fw: Re: Logistic Regression - Threshold effect sizes? Jon, If you're uncomfortable treating your ordinal variable as though it were interval, you can always "dummy code" the ordinal variable---such that the 5-category variable become 4 separate variables---and direct comparisons can be made. This can work pretty well if you don't have a lot of ordinal vairables and if you're sample size is sufficiently large. Scott Millis > --- On Fri, 9/5/08, Justin Black > <[hidden email]> wrote: > > > From: Justin Black <[hidden email]> > > Subject: Re: Logistic Regression - Threshold effect > sizes? > > To: "SR Millis" <[hidden email]> > > Date: Friday, September 5, 2008, 6:13 PM > > Scott, thank you for following up. > > > > What's puzzling to me is the seemingly omnibus > nature > > of the effect. So, I > > could say that a 1-point increase in IV1 is associated > with > > a, e.g., 40% > > increase in the likelihood of the event occurring. > But > > that assumes that a > > 1-point increase in the IV has the same effect on the > DV > > regardless of the > > baseline level of the IV. I don't think that > > assumption is accurate in > > these particular data. > > > > I really feel like I'm missing something here, > just > > can't figure out what. > > A look back at Hosmer & Lemeshow didn't help > any. > > > > Thank you, > > > > Jon > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > <[hidden email]> wrote: > > > > > Jon, > > > > > > What is puzzling in the results? > > > > > > Have you examined the degree of collinearity > among the > > > predictors/covariates? > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > <[hidden email]> wrote: > > > > > > > From: Jon Bernard <[hidden email]> > > > > Subject: Logistic Regression - Threshold > effect > > sizes? > > > > To: [hidden email] > > > > Date: Friday, September 5, 2008, 5:42 PM > > > > Fellow SPSSers, > > > > > > > > I am struggling with a logistic regression > issue > > and > > > > thought I'd put it out > > > > to the list for some clarity. > > > > > > > > The dependent variable of interest is a > binary > > event (0 = > > > > Did not happen, 1 > > > > = Did happen). The independent variables of > > interest are > > > > ordinal > > > > attitudinal survey items, with responses on > a > > 5-point scale > > > > (1 = Strongly > > > > Disagree, 5 = Strongly Agree). I have been > using > > logistic > > > > regression for > > > > the analysis, but either I'm missing > > something in the > > > > results output or I'm > > > > using the wrong statistical technique. I > have an > > inkling > > > > that the intervals > > > > between categories of the independent > variables > > are not all > > > > equal. In other > > > > words, I think that the impact on the > dependent > > variable of > > > > an independent > > > > variable score of 3 vs. one of 2 is greater > than > > that of a > > > > score of 5 vs. > > > > one of 4. Is that clear? If so, what would > you > > recommend > > > > in order to test > > > > that hypothesis? Is there a class of > techniques > > designed > > > > particularly for > > > > this kind of test? > > > > > > > > Very many thanks in advance for your > assistance > > with this > > > > matter. > > > > > > > > Kind regards, > > > > > > > > Jon > > > > > > > > ===================== > > > > 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 ===================== 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: Logistic Regression - Threshold effect sizes?
|
|
Missing values should not in general be treated as zeroes, since they just
mean "absence of information", not "negative response". The default treatment is to exclude any case where at least one variable is missing. An alternative approach is imputing a presumed value, using for instance the MISSING VALUES module of SPSS, which predicts the missing value by regression, based on the value of other variables, but this is mostly appropriate for interval variables, and much less reliable for categorical ones or for ordinal variables with just a few ordered categories. If the number of excluded cases with the default or standard procedure (listwise exclusion) is not large, you should use the standard procedure. Imputing values is to be used only as a second-best solution whenever listwise exclusion leads to the loss of a large portion of the sample, and always remembering you are no longer dealing with actual IV data but with presumed or predicted data. Hector _____ From: [hidden email] [mailto:[hidden email]] On Behalf Of J Sutar Sent: 07 September 2008 16:12 To: Hector Maletta Subject: Re: Logistic Regression - Threshold effect sizes? Thanks, appreciated. Also, what are your thoughts on my comments about missing values? 2008/9/7 Hector Maletta <[hidden email]> Jig, I don't think it would be a good idea. On the one hand, it might pick some of the dummies and not others. On the other hand, there are serious objections against the stepwise procedure that have been discussed in this forum before (you may browse the archives to find them). I would advise using one or the other set of dummies according to one's conception of the corresponding IV. Hector _____ From: <mailto:[hidden email]> [hidden email] [mailto: <mailto:[hidden email]> [hidden email]] On Behalf Of J Sutar Sent: 07 September 2008 15:45 To: Hector Maletta Cc: <mailto:[hidden email]> [hidden email] Subject: Re: Logistic Regression - Threshold effect sizes? Hi Hector and all, Would the approach of entering both sets of dummy codes into a stepwise procedure (Forward LR) and allowing the modelling process to pick out the significant dummy codes be a valid approach (hoping that it would only pick dummy codes from the same set. Not sure what it would mean if dummy codes from different sets were to be significant....) ? I have one other question on this subject. How should missing values be treated? As a quick and dirty approach, would coding them as zeros be appropriate (simply so that cases are not excluded for the listwise procedure). Many thanks Jigs 2008/9/6 Hector Maletta <[hidden email]> Dummy variables are only partially useful in the case of Jon, because they do not necessarily generate ordered results (as expected). Jon: to create dummy variables in general you simply create one dummy with values (0,1) for each category minus one. The omitted category is omitted because it can be deduced from the rest. General syntax for one category: COMPUTE DUMMYCAT3=(yourvar=3). The new variable, DUMMYCAT3, will equal 1 whenever your original variable YOURVAR equals 3, and will equal 0 otherwise. Repeat this for all categories except one (for instance, you may choose the first or last category for omission). However, if you have represented your 5-category variables by four dummies each, you are not guaranteed to obtain monotonic results, i.e. that the effect increases or decreases in a monotonic way (by equal or different amounts) for each increase in the ordinal response from 1 to 5. You may well find that the effect increases for category 1 and 3, but decreases for category 2 and 4, which could be incomprehensible from a theoretical point of view and may be a random effect of your sample. You can create "incremental dummies" by assigning the value 1 to cases having one value OR LESS: COMPUTE UPTOONE=(YOURVAR=1). COMPUTE UPTOTWO=(YOURVAR LE 2). COMPUTE UPTOTHREE=(YOURVAR LE 3). COMPUTE UPTOFOUR=(YOURVAR LE 4). (it is not necessary to create UPTO FIVE, because people choosing category five can be deduced as the complement of UPTOFOUR). The increase (or decrease) in the effect between UPTOTWO and UPTOTHREE would be the specific effect of choosing 3. This ASSUMES the underlying variable represented by your ordinal IV is monotonic. If you are not sure about this, you may use simple dummies as in my first example with DUMMYCAT1 to DUMMYCAT4. Or you may try both and look at the results. In any case, try to avoid dummies where very few cases have 0 or 1 (i.e. dummies with very low or very high frequencies in each value): results could be unstable and unreliable. If, for instance, very few people chose category 5, you may group them with category 5 and define the variable in question as a four-category ordinal variable. Of course this way you sacrifice information (difference between 4 and 5) but you avoid the unpleasant consequences of a very small sample of cases in category 5. Hope this helps. Hector -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of SR Millis Sent: 05 September 2008 18:19 To: [hidden email] Subject: Fw: Re: Logistic Regression - Threshold effect sizes? Jon, If you're uncomfortable treating your ordinal variable as though it were interval, you can always "dummy code" the ordinal variable---such that the 5-category variable become 4 separate variables---and direct comparisons can be made. This can work pretty well if you don't have a lot of ordinal vairables and if you're sample size is sufficiently large. Scott Millis > --- On Fri, 9/5/08, Justin Black > <[hidden email]> wrote: > > > From: Justin Black <[hidden email]> > > Subject: Re: Logistic Regression - Threshold effect > sizes? > > To: "SR Millis" <[hidden email]> > > Date: Friday, September 5, 2008, 6:13 PM > > Scott, thank you for following up. > > > > What's puzzling to me is the seemingly omnibus > nature > > of the effect. So, I > > could say that a 1-point increase in IV1 is associated > with > > a, e.g., 40% > > increase in the likelihood of the event occurring. > But > > that assumes that a > > 1-point increase in the IV has the same effect on the > DV > > regardless of the > > baseline level of the IV. I don't think that > > assumption is accurate in > > these particular data. > > > > I really feel like I'm missing something here, > just > > can't figure out what. > > A look back at Hosmer & Lemeshow didn't help > any. > > > > Thank you, > > > > Jon > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > <[hidden email]> wrote: > > > > > Jon, > > > > > > What is puzzling in the results? > > > > > > Have you examined the degree of collinearity > among the > > > predictors/covariates? > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > <[hidden email]> wrote: > > > > > > > From: Jon Bernard <[hidden email]> > > > > Subject: Logistic Regression - Threshold > effect > > sizes? > > > > To: [hidden email] > > > > Date: Friday, September 5, 2008, 5:42 PM > > > > Fellow SPSSers, > > > > > > > > I am struggling with a logistic regression > issue > > and > > > > thought I'd put it out > > > > to the list for some clarity. > > > > > > > > The dependent variable of interest is a > binary > > event (0 = > > > > Did not happen, 1 > > > > = Did happen). The independent variables of > > interest are > > > > ordinal > > > > attitudinal survey items, with responses on > a > > 5-point scale > > > > (1 = Strongly > > > > Disagree, 5 = Strongly Agree). I have been > using > > logistic > > > > regression for > > > > the analysis, but either I'm missing > > something in the > > > > results output or I'm > > > > using the wrong statistical technique. I > have an > > inkling > > > > that the intervals > > > > between categories of the independent > variables > > are not all > > > > equal. In other > > > > words, I think that the impact on the > dependent > > variable of > > > > an independent > > > > variable score of 3 vs. one of 2 is greater > than > > that of a > > > > score of 5 vs. > > > > one of 4. Is that clear? If so, what would > you > > recommend > > > > in order to test > > > > that hypothesis? Is there a class of > techniques > > designed > > > > particularly for > > > > this kind of test? > > > > > > > > Very many thanks in advance for your > assistance > > with this > > > > matter. > > > > > > > > Kind regards, > > > > > > > > Jon > > > > > > > > ===================== > > > > 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 ===================== 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 Hector Maletta
Dear all,
As Jon Peck suggested quite long ago about using the autocript to automatically run Syntax/ macros when SPSS starts. I've tried to get the script to link to my macro but fail to do it. Would really appreciate if any one could help me out. Is there any way to convert Syntax/ Macros to script so that I can put the converted version to the autoscript. Another option is how to write a link from autoscript to a certain set of Syntax/ Macros. The need for this is to simplify the job for those who use SPSS in my office, they do not need to know about, run the macros when SPSS starts. All the tools should be built-in. Thank you very much, Hop ===================== 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 |
SV: Convert Syntax to Script
|
|
Hi Hop!
I think you can find what you are looking for here: http://www.spsstools.net/Scripts/SyntaxDoc/ConvertSyntaxToScript.txt All the best Wilhelm Landerholm | +46-735-460000 Queue Sweden AB - A Management Science Company | www.qsweden.com -----Ursprungligt meddelande----- Från: SPSSX(r) Discussion [mailto:[hidden email]] För Thien Hop Skickat: den 8 september 2008 05:14 Till: [hidden email] Ämne: Convert Syntax to Script Dear all, As Jon Peck suggested quite long ago about using the autocript to automatically run Syntax/ macros when SPSS starts. I've tried to get the script to link to my macro but fail to do it. Would really appreciate if any one could help me out. Is there any way to convert Syntax/ Macros to script so that I can put the converted version to the autoscript. Another option is how to write a link from autoscript to a certain set of Syntax/ Macros. The need for this is to simplify the job for those who use SPSS in my office, they do not need to know about, run the macros when SPSS starts. All the tools should be built-in. Thank you very much, Hop ===================== 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 |
|
|
Dear Wilhelm Landerholm,
Thank you very much for your recommendation. I successfully converted the macros, I works well when I manually run the script but I really cannot get it to automatically run (I incorporated in Autoscript already & Enable Autoscripting is selected in Options). Again, is there any way around to get script or syntax to atomatically run on start-up of SPSS? Thank you ever so much! Hop ----- Original Message ----- From: "Wilhelm Landerholm (Svensson) | Queue Sweden AB" <[hidden email]> To: "'Thien Hop'" <[hidden email]>; <[hidden email]> Sent: Monday, September 08, 2008 4:48 PM Subject: SV: Convert Syntax to Script Hi Hop! I think you can find what you are looking for here: http://www.spsstools.net/Scripts/SyntaxDoc/ConvertSyntaxToScript.txt All the best Wilhelm Landerholm | +46-735-460000 Queue Sweden AB - A Management Science Company | www.qsweden.com -----Ursprungligt meddelande----- Från: SPSSX(r) Discussion [mailto:[hidden email]] För Thien Hop Skickat: den 8 september 2008 05:14 Till: [hidden email] Ämne: Convert Syntax to Script Dear all, As Jon Peck suggested quite long ago about using the autocript to automatically run Syntax/ macros when SPSS starts. I've tried to get the script to link to my macro but fail to do it. Would really appreciate if any one could help me out. Is there any way to convert Syntax/ Macros to script so that I can put the converted version to the autoscript. Another option is how to write a link from autoscript to a certain set of Syntax/ Macros. The need for this is to simplify the job for those who use SPSS in my office, they do not need to know about, run the macros when SPSS starts. All the tools should be built-in. Thank you very much, Hop ===================== 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 |
|
|
Have you looked at production jobs?
-----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Thien Hop Sent: Monday, September 08, 2008 6:34 AM To: [hidden email] Subject: Re: Convert Syntax to Script Dear Wilhelm Landerholm, Thank you very much for your recommendation. I successfully converted the macros, I works well when I manually run the script but I really cannot get it to automatically run (I incorporated in Autoscript already & Enable Autoscripting is selected in Options). Again, is there any way around to get script or syntax to atomatically run on start-up of SPSS? Thank you ever so much! Hop ----- Original Message ----- From: "Wilhelm Landerholm (Svensson) | Queue Sweden AB" <[hidden email]> To: "'Thien Hop'" <[hidden email]>; <[hidden email]> Sent: Monday, September 08, 2008 4:48 PM Subject: SV: Convert Syntax to Script Hi Hop! I think you can find what you are looking for here: http://www.spsstools.net/Scripts/SyntaxDoc/ConvertSyntaxToScript.txt All the best Wilhelm Landerholm | +46-735-460000 Queue Sweden AB - A Management Science Company | www.qsweden.com -----Ursprungligt meddelande----- Från: SPSSX(r) Discussion [mailto:[hidden email]] För Thien Hop Skickat: den 8 september 2008 05:14 Till: [hidden email] Ämne: Convert Syntax to Script Dear all, As Jon Peck suggested quite long ago about using the autocript to automatically run Syntax/ macros when SPSS starts. I've tried to get the script to link to my macro but fail to do it. Would really appreciate if any one could help me out. Is there any way to convert Syntax/ Macros to script so that I can put the converted version to the autoscript. Another option is how to write a link from autoscript to a certain set of Syntax/ Macros. The need for this is to simplify the job for those who use SPSS in my office, they do not need to know about, run the macros when SPSS starts. All the tools should be built-in. Thank you very much, Hop ===================== 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 |
|
|
In reply to this post by Thien Hop
The way to run a script on startup depends on your SPSS version. Prior to 16, you need to add a function to your autoscript.sbs file with a special name: Application_Create
and make sure that it is enabled in that file (look at the list at the top after you save the file). In 16, there is no application create event. In 17, there is a new, more flexible mechanism separate from the autoscript mechanism. If you place a script named StartClient_.py or StartClient_.wwd in the scripts subdirectory of your SPSS Statistics installation, it (or both) will be executed at the start of a session. If you switch to the Server and a script named StartServer_.py or StartServer_.wwd is in that directory, it will be executed. Having said this, though, the original purpose may be better served by converting macros to Python either as programs or as extension commands. Any libraries they use are loaded automatically when needed, so no special action is required in order to get them defined. HTH, Jon Peck -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Thien Hop Sent: Monday, September 08, 2008 6:34 AM To: [hidden email] Subject: Re: [SPSSX-L] Convert Syntax to Script Dear Wilhelm Landerholm, Thank you very much for your recommendation. I successfully converted the macros, I works well when I manually run the script but I really cannot get it to automatically run (I incorporated in Autoscript already & Enable Autoscripting is selected in Options). Again, is there any way around to get script or syntax to atomatically run on start-up of SPSS? Thank you ever so much! Hop ----- Original Message ----- From: "Wilhelm Landerholm (Svensson) | Queue Sweden AB" <[hidden email]> To: "'Thien Hop'" <[hidden email]>; <[hidden email]> Sent: Monday, September 08, 2008 4:48 PM Subject: SV: Convert Syntax to Script Hi Hop! I think you can find what you are looking for here: http://www.spsstools.net/Scripts/SyntaxDoc/ConvertSyntaxToScript.txt All the best Wilhelm Landerholm | +46-735-460000 Queue Sweden AB - A Management Science Company | www.qsweden.com -----Ursprungligt meddelande----- Från: SPSSX(r) Discussion [mailto:[hidden email]] För Thien Hop Skickat: den 8 september 2008 05:14 Till: [hidden email] Ämne: Convert Syntax to Script Dear all, As Jon Peck suggested quite long ago about using the autocript to automatically run Syntax/ macros when SPSS starts. I've tried to get the script to link to my macro but fail to do it. Would really appreciate if any one could help me out. Is there any way to convert Syntax/ Macros to script so that I can put the converted version to the autoscript. Another option is how to write a link from autoscript to a certain set of Syntax/ Macros. The need for this is to simplify the job for those who use SPSS in my office, they do not need to know about, run the macros when SPSS starts. All the tools should be built-in. Thank you very much, Hop ===================== 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 |
|
|
Hi Peck, Antro and Wilhelm Landerholm,
Thank you very much for your advices. I now can successfully load my macros to run automatically. This will be a great move! Python is something new to me but I will try to learn more about it as Peck mentioned it is very useful. Warmest regards, Hop On Mon, Sep 8, 2008 at 9:17 PM, Peck, Jon <[hidden email]> wrote: > The way to run a script on startup depends on your SPSS version. Prior to > 16, you need to add a function to your autoscript.sbs file with a special > name: Application_Create > and make sure that it is enabled in that file (look at the list at the top > after you save the file). > > In 16, there is no application create event. > > In 17, there is a new, more flexible mechanism separate from the autoscript > mechanism. If you place a script named StartClient_.py or StartClient_.wwd > in the scripts subdirectory of your SPSS Statistics installation, it (or > both) will be executed at the start of a session. If you switch to the > Server and a script named StartServer_.py or StartServer_.wwd is in that > directory, it will be executed. > > Having said this, though, the original purpose may be better served by > converting macros to Python either as programs or as extension commands. > Any libraries they use are loaded automatically when needed, so no special > action is required in order to get them defined. > > HTH, > Jon Peck > > > -----Original Message----- > From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of > Thien Hop > Sent: Monday, September 08, 2008 6:34 AM > To: [hidden email] > Subject: Re: [SPSSX-L] Convert Syntax to Script > > Dear Wilhelm Landerholm, > > Thank you very much for your recommendation. I successfully converted the > macros, I works well when I manually run the script but I really cannot get > it to automatically run (I incorporated in Autoscript already & Enable > Autoscripting is selected in Options). > > Again, is there any way around to get script or syntax to atomatically run > on start-up of SPSS? > > Thank you ever so much! > > Hop > ----- Original Message ----- > From: "Wilhelm Landerholm (Svensson) | Queue Sweden AB" <[hidden email]> > To: "'Thien Hop'" <[hidden email]>; <[hidden email]> > Sent: Monday, September 08, 2008 4:48 PM > Subject: SV: Convert Syntax to Script > > > Hi Hop! > > I think you can find what you are looking for here: > > http://www.spsstools.net/Scripts/SyntaxDoc/ConvertSyntaxToScript.txt > > All the best > > Wilhelm Landerholm | +46-735-460000 > Queue Sweden AB - A Management Science Company | www.qsweden.com > > > -----Ursprungligt meddelande----- > Från: SPSSX(r) Discussion [mailto:[hidden email]] För Thien Hop > Skickat: den 8 september 2008 05:14 > Till: [hidden email] > Ämne: Convert Syntax to Script > > Dear all, > > As Jon Peck suggested quite long ago about using the autocript to > automatically run Syntax/ macros when SPSS starts. I've tried to get the > script to link to my macro but fail to do it. > > Would really appreciate if any one could help me out. Is there any way to > convert Syntax/ Macros to script so that I can put the converted version to > the autoscript. Another option is how to write a link from autoscript to a > certain set of Syntax/ Macros. > > The need for this is to simplify the job for those who use SPSS in my > office, they do not need to know about, run the macros when SPSS starts. > All > the tools should be built-in. > > Thank you very much, > > Hop > > ===================== > 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: Fw: Re: Logistic Regression - Threshold effect sizes?
|
|
In reply to this post by SR Millis-3
Thanks again Scott.
I tried this method and now have to make sense of the output. I entered one ordinal covariate and one binary DV, selecting the option for contrasts as you explained below. The results give the following Exp(B)'s (all of them significant at p < .001): var1(1) = 2.704 var1(2) = 2.198 var1(3) = 1.649 var1(4) = 1.189 Could I interpret this as suggesting that the impact of a one-point positive change in response to the item is greater when the baseline value is 1 than when it is 2, 3, or 4? Thank you, Jon On Fri, Sep 5, 2008 at 6:36 PM, SR Millis <[hidden email]> wrote: > Jon, > > If you're using SPSS binary logistic regression, enter your ordinal > variable as a covariate. Then, click on the Categorical button. This will > take you to a new window where you indicate which covariates are > categorical. The Contrast should be "Indicator." Then run the analysis. > > If you're using Stata software, you use the xi option. > > Scott Millis > > > > > --- On Fri, 9/5/08, Jon Bernard <[hidden email]> wrote: > > > From: Jon Bernard <[hidden email]> > > Subject: Re: Fw: Re: Logistic Regression - Threshold effect sizes? > > To: "SR Millis" <[hidden email]> > > Cc: [hidden email] > > Date: Friday, September 5, 2008, 6:24 PM > > Thank you Scott. Would you recommend any resources for a > > quick study on > > this technique? > > > > Jon > > > > On Fri, Sep 5, 2008 at 6:18 PM, SR Millis > > <[hidden email]> wrote: > > > > > Jon, > > > > > > If you're uncomfortable treating your ordinal > > variable > > > as though it were interval, you can always "dummy > > > code" the ordinal variable---such that the > > 5-category > > > variable become 4 separate variables---and direct > > > comparisons can be made. This can work pretty well if > > you > > > don't have a lot of ordinal vairables and if > > you're > > > sample size is sufficiently large. > > > > > > Scott Millis > > > > > > > --- On Fri, 9/5/08, Justin Black > > > > <[hidden email]> wrote: > > > > > > > > > From: Justin Black > > <[hidden email]> > > > > > Subject: Re: Logistic Regression - Threshold > > effect > > > > sizes? > > > > > To: "SR Millis" > > <[hidden email]> > > > > > Date: Friday, September 5, 2008, 6:13 PM > > > > > Scott, thank you for following up. > > > > > > > > > > What's puzzling to me is the seemingly > > omnibus > > > > nature > > > > > of the effect. So, I > > > > > could say that a 1-point increase in IV1 is > > associated > > > > with > > > > > a, e.g., 40% > > > > > increase in the likelihood of the event > > occurring. > > > > But > > > > > that assumes that a > > > > > 1-point increase in the IV has the same > > effect on the > > > > DV > > > > > regardless of the > > > > > baseline level of the IV. I don't think > > that > > > > > assumption is accurate in > > > > > these particular data. > > > > > > > > > > I really feel like I'm missing something > > here, > > > > just > > > > > can't figure out what. > > > > > A look back at Hosmer & Lemeshow > > didn't help > > > > any. > > > > > > > > > > Thank you, > > > > > > > > > > Jon > > > > > > > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > > > > <[hidden email]> wrote: > > > > > > > > > > > Jon, > > > > > > > > > > > > What is puzzling in the results? > > > > > > > > > > > > Have you examined the degree of > > collinearity > > > > among the > > > > > > predictors/covariates? > > > > > > > > > > > > > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > > > > <[hidden email]> wrote: > > > > > > > > > > > > > From: Jon Bernard > > <[hidden email]> > > > > > > > Subject: Logistic Regression - > > Threshold > > > > effect > > > > > sizes? > > > > > > > To: [hidden email] > > > > > > > Date: Friday, September 5, 2008, > > 5:42 PM > > > > > > > Fellow SPSSers, > > > > > > > > > > > > > > I am struggling with a logistic > > regression > > > > issue > > > > > and > > > > > > > thought I'd put it out > > > > > > > to the list for some clarity. > > > > > > > > > > > > > > The dependent variable of interest > > is a > > > > binary > > > > > event (0 = > > > > > > > Did not happen, 1 > > > > > > > = Did happen). The independent > > variables of > > > > > interest are > > > > > > > ordinal > > > > > > > attitudinal survey items, with > > responses on > > > > a > > > > > 5-point scale > > > > > > > (1 = Strongly > > > > > > > Disagree, 5 = Strongly Agree). I > > have been > > > > using > > > > > logistic > > > > > > > regression for > > > > > > > the analysis, but either I'm > > missing > > > > > something in the > > > > > > > results output or I'm > > > > > > > using the wrong statistical > > technique. I > > > > have an > > > > > inkling > > > > > > > that the intervals > > > > > > > between categories of the > > independent > > > > variables > > > > > are not all > > > > > > > equal. In other > > > > > > > words, I think that the impact on > > the > > > > dependent > > > > > variable of > > > > > > > an independent > > > > > > > variable score of 3 vs. one of 2 > > is greater > > > > than > > > > > that of a > > > > > > > score of 5 vs. > > > > > > > one of 4. Is that clear? If so, > > what would > > > > you > > > > > recommend > > > > > > > in order to test > > > > > > > that hypothesis? Is there a class > > of > > > > techniques > > > > > designed > > > > > > > particularly for > > > > > > > this kind of test? > > > > > > > > > > > > > > Very many thanks in advance for > > your > > > > assistance > > > > > with this > > > > > > > matter. > > > > > > > > > > > > > > Kind regards, > > > > > > > > > > > > > > Jon > > > > > > > > > > > > > > ===================== > > > > > > > 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: Logistic Regression - Threshold effect sizes?
|
|
In reply to this post by Hector Maletta
Thank you Hector.
Both of the approaches you recommended produced useful results. I am still working out the interpretation - see my email to Scott from a few minutes ago - but I'm getting there. As you might expect, the first type of dummy coding you demonstrated ("COMPUTE DUMMYCAT3=(yourvar=3).") produced results identical to those from the categorical contrasts method Scott recommended. Thanks again. Jon On Fri, Sep 5, 2008 at 7:31 PM, Hector Maletta <[hidden email]>wrote: > Dummy variables are only partially useful in the case of Jon, because they > do not necessarily generate ordered results (as expected). > Jon: to create dummy variables in general you simply create one dummy with > values (0,1) for each category minus one. The omitted category is omitted > because it can be deduced from the rest. General syntax for one category: > COMPUTE DUMMYCAT3=(yourvar=3). > The new variable, DUMMYCAT3, will equal 1 whenever your original variable > YOURVAR equals 3, and will equal 0 otherwise. Repeat this for all > categories > except one (for instance, you may choose the first or last category for > omission). > However, if you have represented your 5-category variables by four dummies > each, you are not guaranteed to obtain monotonic results, i.e. that the > effect increases or decreases in a monotonic way (by equal or different > amounts) for each increase in the ordinal response from 1 to 5. You may > well > find that the effect increases for category 1 and 3, but decreases for > category 2 and 4, which could be incomprehensible from a theoretical point > of view and may be a random effect of your sample. > You can create "incremental dummies" by assigning the value 1 to cases > having one value OR LESS: > COMPUTE UPTOONE=(YOURVAR=1). > COMPUTE UPTOTWO=(YOURVAR LE 2). > COMPUTE UPTOTHREE=(YOURVAR LE 3). > COMPUTE UPTOFOUR=(YOURVAR LE 4). > (it is not necessary to create UPTO FIVE, because people choosing category > five can be deduced as the complement of UPTOFOUR). > The increase (or decrease) in the effect between UPTOTWO and UPTOTHREE > would > be the specific effect of choosing 3. This ASSUMES the underlying variable > represented by your ordinal IV is monotonic. If you are not sure about > this, > you may use simple dummies as in my first example with DUMMYCAT1 to > DUMMYCAT4. Or you may try both and look at the results. > In any case, try to avoid dummies where very few cases have 0 or 1 (i.e. > dummies with very low or very high frequencies in each value): results > could > be unstable and unreliable. If, for instance, very few people chose > category > 5, you may group them with category 5 and define the variable in question > as > a four-category ordinal variable. Of course this way you sacrifice > information (difference between 4 and 5) but you avoid the unpleasant > consequences of a very small sample of cases in category 5. > Hope this helps. > > Hector > -----Original Message----- > From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of > SR > Millis > Sent: 05 September 2008 18:19 > To: [hidden email] > Subject: Fw: Re: Logistic Regression - Threshold effect sizes? > > Jon, > > If you're uncomfortable treating your ordinal variable > as though it were interval, you can always "dummy > code" the ordinal variable---such that the 5-category > variable become 4 separate variables---and direct > comparisons can be made. This can work pretty well if you > don't have a lot of ordinal vairables and if you're > sample size is sufficiently large. > > Scott Millis > > > --- On Fri, 9/5/08, Justin Black > > <[hidden email]> wrote: > > > > > From: Justin Black <[hidden email]> > > > Subject: Re: Logistic Regression - Threshold effect > > sizes? > > > To: "SR Millis" <[hidden email]> > > > Date: Friday, September 5, 2008, 6:13 PM > > > Scott, thank you for following up. > > > > > > What's puzzling to me is the seemingly omnibus > > nature > > > of the effect. So, I > > > could say that a 1-point increase in IV1 is associated > > with > > > a, e.g., 40% > > > increase in the likelihood of the event occurring. > > But > > > that assumes that a > > > 1-point increase in the IV has the same effect on the > > DV > > > regardless of the > > > baseline level of the IV. I don't think that > > > assumption is accurate in > > > these particular data. > > > > > > I really feel like I'm missing something here, > > just > > > can't figure out what. > > > A look back at Hosmer & Lemeshow didn't help > > any. > > > > > > Thank you, > > > > > > Jon > > > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > > <[hidden email]> wrote: > > > > > > > Jon, > > > > > > > > What is puzzling in the results? > > > > > > > > Have you examined the degree of collinearity > > among the > > > > predictors/covariates? > > > > > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > > <[hidden email]> wrote: > > > > > > > > > From: Jon Bernard <[hidden email]> > > > > > Subject: Logistic Regression - Threshold > > effect > > > sizes? > > > > > To: [hidden email] > > > > > Date: Friday, September 5, 2008, 5:42 PM > > > > > Fellow SPSSers, > > > > > > > > > > I am struggling with a logistic regression > > issue > > > and > > > > > thought I'd put it out > > > > > to the list for some clarity. > > > > > > > > > > The dependent variable of interest is a > > binary > > > event (0 = > > > > > Did not happen, 1 > > > > > = Did happen). The independent variables of > > > interest are > > > > > ordinal > > > > > attitudinal survey items, with responses on > > a > > > 5-point scale > > > > > (1 = Strongly > > > > > Disagree, 5 = Strongly Agree). I have been > > using > > > logistic > > > > > regression for > > > > > the analysis, but either I'm missing > > > something in the > > > > > results output or I'm > > > > > using the wrong statistical technique. I > > have an > > > inkling > > > > > that the intervals > > > > > between categories of the independent > > variables > > > are not all > > > > > equal. In other > > > > > words, I think that the impact on the > > dependent > > > variable of > > > > > an independent > > > > > variable score of 3 vs. one of 2 is greater > > than > > > that of a > > > > > score of 5 vs. > > > > > one of 4. Is that clear? If so, what would > > you > > > recommend > > > > > in order to test > > > > > that hypothesis? Is there a class of > > techniques > > > designed > > > > > particularly for > > > > > this kind of test? > > > > > > > > > > Very many thanks in advance for your > > assistance > > > with this > > > > > matter. > > > > > > > > > > Kind regards, > > > > > > > > > > Jon > > > > > > > > > > ===================== > > > > > 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 > ===================== 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: Logistic Regression - Threshold effect sizes?
|
|
Of course, Jon, my formulation of the dummies is algebraically equivalent to
Scott's. Hector _____ From: Jon Bernard [mailto:[hidden email]] Sent: 09 September 2008 17:19 To: Hector Maletta; [hidden email] Subject: Re: Logistic Regression - Threshold effect sizes? Thank you Hector. Both of the approaches you recommended produced useful results. I am still working out the interpretation - see my email to Scott from a few minutes ago - but I'm getting there. As you might expect, the first type of dummy coding you demonstrated ("COMPUTE DUMMYCAT3=(yourvar=3).") produced results identical to those from the categorical contrasts method Scott recommended. Thanks again. Jon On Fri, Sep 5, 2008 at 7:31 PM, Hector Maletta <[hidden email]> wrote: Dummy variables are only partially useful in the case of Jon, because they do not necessarily generate ordered results (as expected). Jon: to create dummy variables in general you simply create one dummy with values (0,1) for each category minus one. The omitted category is omitted because it can be deduced from the rest. General syntax for one category: COMPUTE DUMMYCAT3=(yourvar=3). The new variable, DUMMYCAT3, will equal 1 whenever your original variable YOURVAR equals 3, and will equal 0 otherwise. Repeat this for all categories except one (for instance, you may choose the first or last category for omission). However, if you have represented your 5-category variables by four dummies each, you are not guaranteed to obtain monotonic results, i.e. that the effect increases or decreases in a monotonic way (by equal or different amounts) for each increase in the ordinal response from 1 to 5. You may well find that the effect increases for category 1 and 3, but decreases for category 2 and 4, which could be incomprehensible from a theoretical point of view and may be a random effect of your sample. You can create "incremental dummies" by assigning the value 1 to cases having one value OR LESS: COMPUTE UPTOONE=(YOURVAR=1). COMPUTE UPTOTWO=(YOURVAR LE 2). COMPUTE UPTOTHREE=(YOURVAR LE 3). COMPUTE UPTOFOUR=(YOURVAR LE 4). (it is not necessary to create UPTO FIVE, because people choosing category five can be deduced as the complement of UPTOFOUR). The increase (or decrease) in the effect between UPTOTWO and UPTOTHREE would be the specific effect of choosing 3. This ASSUMES the underlying variable represented by your ordinal IV is monotonic. If you are not sure about this, you may use simple dummies as in my first example with DUMMYCAT1 to DUMMYCAT4. Or you may try both and look at the results. In any case, try to avoid dummies where very few cases have 0 or 1 (i.e. dummies with very low or very high frequencies in each value): results could be unstable and unreliable. If, for instance, very few people chose category 5, you may group them with category 5 and define the variable in question as a four-category ordinal variable. Of course this way you sacrifice information (difference between 4 and 5) but you avoid the unpleasant consequences of a very small sample of cases in category 5. Hope this helps. Hector -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of SR Millis Sent: 05 September 2008 18:19 To: [hidden email] Subject: Fw: Re: Logistic Regression - Threshold effect sizes? Jon, If you're uncomfortable treating your ordinal variable as though it were interval, you can always "dummy code" the ordinal variable---such that the 5-category variable become 4 separate variables---and direct comparisons can be made. This can work pretty well if you don't have a lot of ordinal vairables and if you're sample size is sufficiently large. Scott Millis > --- On Fri, 9/5/08, Justin Black > <[hidden email]> wrote: > > > From: Justin Black <[hidden email]> > > Subject: Re: Logistic Regression - Threshold effect > sizes? > > To: "SR Millis" <[hidden email]> > > Date: Friday, September 5, 2008, 6:13 PM > > Scott, thank you for following up. > > > > What's puzzling to me is the seemingly omnibus > nature > > of the effect. So, I > > could say that a 1-point increase in IV1 is associated > with > > a, e.g., 40% > > increase in the likelihood of the event occurring. > But > > that assumes that a > > 1-point increase in the IV has the same effect on the > DV > > regardless of the > > baseline level of the IV. I don't think that > > assumption is accurate in > > these particular data. > > > > I really feel like I'm missing something here, > just > > can't figure out what. > > A look back at Hosmer & Lemeshow didn't help > any. > > > > Thank you, > > > > Jon > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR Millis > > <[hidden email]> wrote: > > > > > Jon, > > > > > > What is puzzling in the results? > > > > > > Have you examined the degree of collinearity > among the > > > predictors/covariates? > > > > > > > > > > > > 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 Fri, 9/5/08, Jon Bernard > > <[hidden email]> wrote: > > > > > > > From: Jon Bernard <[hidden email]> > > > > Subject: Logistic Regression - Threshold > effect > > sizes? > > > > To: [hidden email] > > > > Date: Friday, September 5, 2008, 5:42 PM > > > > Fellow SPSSers, > > > > > > > > I am struggling with a logistic regression > issue > > and > > > > thought I'd put it out > > > > to the list for some clarity. > > > > > > > > The dependent variable of interest is a > binary > > event (0 = > > > > Did not happen, 1 > > > > = Did happen). The independent variables of > > interest are > > > > ordinal > > > > attitudinal survey items, with responses on > a > > 5-point scale > > > > (1 = Strongly > > > > Disagree, 5 = Strongly Agree). I have been > using > > logistic > > > > regression for > > > > the analysis, but either I'm missing > > something in the > > > > results output or I'm > > > > using the wrong statistical technique. I > have an > > inkling > > > > that the intervals > > > > between categories of the independent > variables > > are not all > > > > equal. In other > > > > words, I think that the impact on the > dependent > > variable of > > > > an independent > > > > variable score of 3 vs. one of 2 is greater > than > > that of a > > > > score of 5 vs. > > > > one of 4. Is that clear? If so, what would > you > > recommend > > > > in order to test > > > > that hypothesis? Is there a class of > techniques > > designed > > > > particularly for > > > > this kind of test? > > > > > > > > Very many thanks in advance for your > assistance > > with this > > > > matter. > > > > > > > > Kind regards, > > > > > > > > Jon > > > > > > > > ===================== > > > > 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 ===================== 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: Fw: Re: Logistic Regression - Threshold effect sizes?
|
|
In reply to this post by Jon Bernard-2
Jon,
Could you provide all of the SPSS output? SPSS is not my "main" statistical software so I am less familiar with its conventions. Thanks, Scott Millis --- On Tue, 9/9/08, Jon Bernard <[hidden email]> wrote: > From: Jon Bernard <[hidden email]> > Subject: Re: Fw: Re: Logistic Regression - Threshold effect sizes? > To: [hidden email] > Date: Tuesday, September 9, 2008, 3:57 PM > Thanks again Scott. > > I tried this method and now have to make sense of the > output. I entered one > ordinal covariate and one binary DV, selecting the option > for contrasts as > you explained below. > > The results give the following Exp(B)'s (all of them > significant at p < > .001): > var1(1) = 2.704 > var1(2) = 2.198 > var1(3) = 1.649 > var1(4) = 1.189 > > Could I interpret this as suggesting that the impact of a > one-point positive > change in response to the item is greater when the baseline > value is 1 than > when it is 2, 3, or 4? > > Thank you, > > Jon > > > > On Fri, Sep 5, 2008 at 6:36 PM, SR Millis > <[hidden email]> wrote: > > > Jon, > > > > If you're using SPSS binary logistic regression, > enter your ordinal > > variable as a covariate. Then, click on the > Categorical button. This will > > take you to a new window where you indicate which > covariates are > > categorical. The Contrast should be > "Indicator." Then run the analysis. > > > > If you're using Stata software, you use the xi > option. > > > > Scott Millis > > > > > > > > > > --- On Fri, 9/5/08, Jon Bernard > <[hidden email]> wrote: > > > > > From: Jon Bernard <[hidden email]> > > > Subject: Re: Fw: Re: Logistic Regression - > Threshold effect sizes? > > > To: "SR Millis" > <[hidden email]> > > > Cc: [hidden email] > > > Date: Friday, September 5, 2008, 6:24 PM > > > Thank you Scott. Would you recommend any > resources for a > > > quick study on > > > this technique? > > > > > > Jon > > > > > > On Fri, Sep 5, 2008 at 6:18 PM, SR Millis > > > <[hidden email]> wrote: > > > > > > > Jon, > > > > > > > > If you're uncomfortable treating your > ordinal > > > variable > > > > as though it were interval, you can always > "dummy > > > > code" the ordinal variable---such that > the > > > 5-category > > > > variable become 4 separate variables---and > direct > > > > comparisons can be made. This can work > pretty well if > > > you > > > > don't have a lot of ordinal vairables > and if > > > you're > > > > sample size is sufficiently large. > > > > > > > > Scott Millis > > > > > > > > > --- On Fri, 9/5/08, Justin Black > > > > > <[hidden email]> wrote: > > > > > > > > > > > From: Justin Black > > > <[hidden email]> > > > > > > Subject: Re: Logistic Regression - > Threshold > > > effect > > > > > sizes? > > > > > > To: "SR Millis" > > > <[hidden email]> > > > > > > Date: Friday, September 5, 2008, > 6:13 PM > > > > > > Scott, thank you for following up. > > > > > > > > > > > > What's puzzling to me is the > seemingly > > > omnibus > > > > > nature > > > > > > of the effect. So, I > > > > > > could say that a 1-point increase > in IV1 is > > > associated > > > > > with > > > > > > a, e.g., 40% > > > > > > increase in the likelihood of the > event > > > occurring. > > > > > But > > > > > > that assumes that a > > > > > > 1-point increase in the IV has the > same > > > effect on the > > > > > DV > > > > > > regardless of the > > > > > > baseline level of the IV. I > don't think > > > that > > > > > > assumption is accurate in > > > > > > these particular data. > > > > > > > > > > > > I really feel like I'm missing > something > > > here, > > > > > just > > > > > > can't figure out what. > > > > > > A look back at Hosmer & > Lemeshow > > > didn't help > > > > > any. > > > > > > > > > > > > Thank you, > > > > > > > > > > > > Jon > > > > > > > > > > > > On Fri, Sep 5, 2008 at 6:00 PM, SR > Millis > > > > > > <[hidden email]> wrote: > > > > > > > > > > > > > Jon, > > > > > > > > > > > > > > What is puzzling in the > results? > > > > > > > > > > > > > > Have you examined the degree > of > > > collinearity > > > > > among the > > > > > > > predictors/covariates? > > > > > > > > > > > > > > > > > > > > > > > > > > > > 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 Fri, 9/5/08, Jon > Bernard > > > > > > <[hidden email]> wrote: > > > > > > > > > > > > > > > From: Jon Bernard > > > <[hidden email]> > > > > > > > > Subject: Logistic > Regression - > > > Threshold > > > > > effect > > > > > > sizes? > > > > > > > > To: > [hidden email] > > > > > > > > Date: Friday, September > 5, 2008, > > > 5:42 PM > > > > > > > > Fellow SPSSers, > > > > > > > > > > > > > > > > I am struggling with a > logistic > > > regression > > > > > issue > > > > > > and > > > > > > > > thought I'd put it > out > > > > > > > > to the list for some > clarity. > > > > > > > > > > > > > > > > The dependent variable > of interest > > > is a > > > > > binary > > > > > > event (0 = > > > > > > > > Did not happen, 1 > > > > > > > > = Did happen). The > independent > > > variables of > > > > > > interest are > > > > > > > > ordinal > > > > > > > > attitudinal survey > items, with > > > responses on > > > > > a > > > > > > 5-point scale > > > > > > > > (1 = Strongly > > > > > > > > Disagree, 5 = Strongly > Agree). I > > > have been > > > > > using > > > > > > logistic > > > > > > > > regression for > > > > > > > > the analysis, but either > I'm > > > missing > > > > > > something in the > > > > > > > > results output or > I'm > > > > > > > > using the wrong > statistical > > > technique. I > > > > > have an > > > > > > inkling > > > > > > > > that the intervals > > > > > > > > between categories of > the > > > independent > > > > > variables > > > > > > are not all > > > > > > > > equal. In other > > > > > > > > words, I think that the > impact on > > > the > > > > > dependent > > > > > > variable of > > > > > > > > an independent > > > > > > > > variable score of 3 vs. > one of 2 > > > is greater > > > > > than > > > > > > that of a > > > > > > > > score of 5 vs. > > > > > > > > one of 4. Is that > clear? If so, > > > what would > > > > > you > > > > > > recommend > > > > > > > > in order to test > > > > > > > > that hypothesis? Is > there a class > > > of > > > > > techniques > > > > > > designed > > > > > > > > particularly for > > > > > > > > this kind of test? > > > > > > > > > > > > > > > > Very many thanks in > advance for > > > your > > > > > assistance > > > > > > with this > > > > > > > > matter. > > > > > > > > > > > > > > > > Kind regards, > > > > > > > > > > > > > > > > Jon > > > > > > > > > > > > > > > > ===================== > > > > > > > > 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 ===================== 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 |