Binary logistic regression
Locked
 
|
Is there any way to get binary logistic regression output in normal units (like probit) rather than logits?
|
Administrator
|
Kurt, are you talking about the coefficients, the predicted probabilities, or something else? Please clarify.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
[I do not see the original post.] The UCLA site gives an example of running a Probit analysis using PLUM with a probit link: http://www.ats.ucla.edu/stat/spss/dae/probit.htm I don't think it is likely that you will find a logistic regression that offers the probit transformation instead of logit. The ancillary statistics that suit logit do not suit probit. -- Rich Ulrich > Date: Tue, 12 Jul 2016 16:04:32 -0700 > From: [hidden email] > Subject: Re: Binary logistic regression > To: [hidden email] > > Kurt, are you talking about the coefficients, the predicted probabilities, or > something else? Please clarify. > > > > Kurt Taube wrote > > Is there any way to get binary logistic regression output in normal units > > (like probit) rather than logits? > |
|
|
In reply to this post by Bruce Weaver
The B weights. The weights are logits, from which logistic probabilities may be obtained. I am asking if weights interpretable in terms of a normal, rather than logistic, distribution are obtainable.
|
|
Administrator
|
Some Googling turned up the following pages that you may find helpful. They suggest that B_probit = B_logit / X, where X is somewhere between 1.6 and 1.8.
http://andrewgelman.com/2006/06/06/take_logit_coef/ http://data.princeton.edu/wws509/stata/c3s7.html HTH.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
Re: Binary logistic regression
|
|
In reply to this post by Kurt Taube
yes.
convert the logits back to probability p = logit/(1+logit) then convert p to z = probit by looking up z in standard normal tables p lgt z lgt/z 0.60 0.41 0.25 1.60 0.75 1.10 0.67 1.63 0.90 2.20 1.28 1.71 0.95 2.94 1.64 1.79 0.99 4.60 2.33 1.98 the ratio gets higher as p increase because normal drops like a stone as exp(-z^2) whereas logit drops like a leaf as exp(-z) best Diana > On 12 Jul 2016, at 23:02, Kurt Taube <[hidden email]> wrote: > > Is there any way to get binary logistic regression output in normal units > (like probit) rather than logits? > > > > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Binary-logistic-regression-tp5732698.html > Sent from the SPSSX Discussion mailing list archive at Nabble.com. > > ===================== > To manage your subscription to SPSSX-L, send a message to > [hidden email] (not to SPSSX-L), with no body text except the > command. To leave the list, send the command > SIGNOFF SPSSX-L > For a list of commands to manage subscriptions, send the command > INFO REFCARD ________________________________________ Professor Diana Kornbrot Work University of Hertfordshire College Lane, Hatfield, Hertfordshire AL10 9AB, UK +44 (0) 170 728 4626 [hidden email] http://dianakornbrot.wordpress.com/ http://go.herts.ac.uk/Diana_Kornbrot skype: kornbrotme Home 19 Elmhurst Avenue London N2 0LT, UK +44 (0) 208 444 2081 ===================== 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 |
Administrator
|
I believe you mean something like this. MATRIX. SAVE MAKE(201,4,0)/ OUTFILE * / VARIABLES p logit z lgt_z. END MATRIX. COMPUTE logit=($CASENUM-101)/20. COMPUTE p= EXP(logit)/(1 + EXP(logit)). COMPUTE z=IDF.NORMAL(p,0,1). COMPUTE lgt_z=logit/z.
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
Re: Binary logistic regression
|
|
yes. i do it in excel - much easier can send file if needed
best diana > On 13 Jul 2016, at 17:31, David Marso <[hidden email]> wrote: > > I believe you mean something like this. > MATRIX. > SAVE MAKE(201,4,0)/ OUTFILE * / VARIABLES p logit z lgt_z. > END MATRIX. > > COMPUTE logit=($CASENUM-101)/20. > COMPUTE p= EXP(logit)/(1 + EXP(logit)). > COMPUTE z=IDF.NORMAL(p,0,1). > COMPUTE lgt_z=logit/z. > > > Kornbrot, Diana wrote >> yes. >> convert the logits back to probability p = logit/(1+logit) >> then convert p to z = probit by looking up z in standard normal tables >> >> p lgt z lgt/z >> 0.60 0.41 0.25 1.60 >> 0.75 1.10 0.67 1.63 >> 0.90 2.20 1.28 1.71 >> 0.95 2.94 1.64 1.79 >> 0.99 4.60 2.33 1.98 >> >> the ratio gets higher as p increase because normal drops like a stone as >> exp(-z^2) whereas logit drops like a leaf as exp(-z) >> best >> Diana >> >> >>> On 12 Jul 2016, at 23:02, Kurt Taube < > >> ktaube@ > >> > wrote: >>> >>> Is there any way to get binary logistic regression output in normal units >>> (like probit) rather than logits? >>> >>> >>> >>> -- >>> View this message in context: >>> http://spssx-discussion.1045642.n5.nabble.com/Binary-logistic-regression-tp5732698.html >>> Sent from the SPSSX Discussion mailing list archive at Nabble.com. >>> >>> ===================== >>> To manage your subscription to SPSSX-L, send a message to >>> > >> LISTSERV@.UGA > >> (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 >> >> ________________________________________ >> Professor Diana Kornbrot >> Work >> University of Hertfordshire >> College Lane, Hatfield, Hertfordshire AL10 9AB, UK >> +44 (0) 170 728 4626 > >> d.e.kornbrot@.ac > >> http://dianakornbrot.wordpress.com/ >> http://go.herts.ac.uk/Diana_Kornbrot >> skype: kornbrotme >> Home >> 19 Elmhurst Avenue >> London N2 0LT, UK >> +44 (0) 208 444 2081 >> >> ===================== >> To manage your subscription to SPSSX-L, send a message to > >> LISTSERV@.UGA > >> (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 > > > > > > ----- > Please reply to the list and not to my personal email. > Those desiring my consulting or training services please feel free to email me. > --- > "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." > Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" > -- > View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Binary-logistic-regression-tp5732698p5732715.html > Sent from the SPSSX Discussion mailing list archive at Nabble.com. > > ===================== > To manage your subscription to SPSSX-L, send a message to > [hidden email] (not to SPSSX-L), with no body text except the > command. To leave the list, send the command > SIGNOFF SPSSX-L > For a list of commands to manage subscriptions, send the command > INFO REFCARD ________________________________________ Professor Diana Kornbrot Work University of Hertfordshire College Lane, Hatfield, Hertfordshire AL10 9AB, UK +44 (0) 170 728 4626 [hidden email] http://dianakornbrot.wordpress.com/ http://go.herts.ac.uk/Diana_Kornbrot skype: kornbrotme Home 19 Elmhurst Avenue London N2 0LT, UK +44 (0) 208 444 2081 ===================== 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 |
|
Administrator
|
I meant the missing EXP functions and the IDF.NORMAL.
Using your sample points doesn't kick back p using your formula
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
|
|
In reply to this post by Kornbrot, Diana
Thank you all for your assistance. I knew I could adjust logistic output (e.g., by dividing the logit by approximately 1.7) to estimate normal parameters. I wanted to know if I could obtain normal parameters directly, and though I can come very close, the answer might be no. (I haven't tried PLUM yet.)
KT |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |