anyone here use JMP?

Rich Goldstein

Join Date: Mar 2014

Posts: 4466
#1

anyone here use JMP?

13 Feb 2023, 13:56

I am working with someone who uses JMP and we are doing ordered logistic regression; JMP offers something they call "expected value" but I can't quite figure out what it is; the only documentation either of us has been able to find says:

"Save Expected Value (Available only when the response is
numeric and has the ordinal modeling type.) Creates a column in the
current data table called Ord Expected. This column contains the linear
combination of the response values with the fitted response
probabilities for each row and gives the expected value."

If anyone has an idea of what this means or how this is calculated, I would greatly appreciate hearing

added in edit - the following dataex output might help:

Code:

* Example generated by -dataex-. For more info, type help dataex clear input str3 Subject byte(LL BFs) double(OrdExpected Linear Prob1 Prob2 Prob3 Prob4 Prob5) "1" 3 5 2.097943900856179 1.820143431235861 .4082699058359035 .2625658366342866 .2114464109345694 .0583861440282089 .0593317025670317 "2" 3 2 2.848690452884622 .7280573724943443 .1879771330914454 .2181211507211049 .3093786872978927 .126280187990496 .1582428408990609 "3" 2 4 2.334116924448643 1.456114744988689 .3240654378901853 .2620522009020323 .2527994389969234 .0778658432906726 .0832170789201865 "4" 5 2 2.848690452884622 .7280573724943443 .1879771330914454 .2181211507211049 .3093786872978927 .126280187990496 .1582428408990609 "5" 5 0 3.383117137636134 0 .1005375922317004 .1476735010778116 .3001566440090465 .1713987021855363 .2802335604959052 end

Note that LL is the outcome and BFs is the only predictor and "OrdExpected" is what I am trying to figure out; Linear is the linear predictor and Prob* are the probabilities of each of the 5 ordered categories

Last edited by Rich Goldstein; 13 Feb 2023, 14:15.
Tags: None
George Ford

Join Date: Aug 2014

Posts: 3152
#2

13 Feb 2023, 14:32

seems like predict or predictnl would get you that.

HTML Code:

help ologit postestimation
Comment
Rich Goldstein

Join Date: Mar 2014

Posts: 4466
#3

13 Feb 2023, 14:38

nope, unable to match using predict - so far - but maybe you can suggest something?
Comment

Joerg Luedicke (StataCorp)

StataCorp Employee

Join Date: Apr 2014
Posts: 116

13 Feb 2023, 15:29

Apparently, it is a weighted average of the outcomes where the probabilities are the weights:

Code:

* Example generated by -dataex-. For more info, type help dataex
clear
input str3 Subject byte(LL BFs) double(OrdExpected Linear Prob1 Prob2 Prob3 Prob4 Prob5)
"1" 3 5 2.097943900856179 1.820143431235861 .4082699058359035 .2625658366342866 .2114464109345694 .0583861440282089 .0593317025670317
"2" 3 2 2.848690452884622 .7280573724943443 .1879771330914454 .2181211507211049 .3093786872978927  .126280187990496 .1582428408990609
"3" 2 4 2.334116924448643 1.456114744988689 .3240654378901853 .2620522009020323 .2527994389969234 .0778658432906726 .0832170789201865
"4" 5 2 2.848690452884622 .7280573724943443 .1879771330914454 .2181211507211049 .3093786872978927  .126280187990496 .1582428408990609
"5" 5 0 3.383117137636134                 0 .1005375922317004 .1476735010778116 .3001566440090465 .1713987021855363 .2802335604959052
end

gen double exp2 = Prob1 + 2*Prob2 + 3*Prob3 + 4*Prob4 + 5*Prob5
list OrdExpected exp2

The above yields:

Code:

. list OrdExpected exp2

     +-----------------------+
     | OrdExpe~d        exp2 |
     |-----------------------|
  1. | 2.0979439   2.0979439 |
  2. | 2.8486905   2.8486905 |
  3. | 2.3341169   2.3341169 |
  4. | 2.8486905   2.8486905 |
  5. | 3.3831171   3.3831171 |
     +-----------------------+

Comment

George Ford

Join Date: Aug 2014

Posts: 3152
#5

13 Feb 2023, 15:52

are the values in the dataex from a regression on those 5 observations?
Comment
George Ford

Join Date: Aug 2014

Posts: 3152
#6

13 Feb 2023, 16:07

see what this gets you if you haven't already.

Code:

ologit LL BFs predict fitp1 fitp2 fitp3 fitp4 fitp5 , p
Comment
Rich Goldstein

Join Date: Mar 2014

Posts: 4466
#7

13 Feb 2023, 19:16

thank you to Joerg Luedicke who gave me a similar answer to that from tech support (and tech support also supplied a way to get CI's using -lincom-); the "odd" thing is that is what I had originally thought but I must have made a typo in my command as my values were slightly too small (e.g., 2.04 for #1))
2 likes
Comment
George Ford

Join Date: Aug 2014

Posts: 3152
#8

14 Feb 2023, 06:56

Those who spent some time thinking about your problem might like to see the answer.
Comment

Rich Goldstein

Join Date: Mar 2014
Posts: 4466

14 Feb 2023, 06:59

here is what I received from tech support:

Code:

 This is the follow-up email. I find how the “Ord Expected” variable is calculated. It is computed by averaging   1*Prob[1]+2*Prob[2]+3*Prob[3]+4*Prob[4]+5*Prob[5]  for each #BFs group.  Please run the following code to reproduce the values of the “Ord Expected” variable in Stata:  ********  use for_ologit, clear *make y =LL in order to match JMP's result generate y=q6_1b+1 summarize ologit y numq5 predict p1-p5 list in 1/10 generate Ord_expected=1*p1+2*p2+3*p3+4*p4+5*p5 bysort numq5: summarize Ord_expected margins, over(numq5) predict()  post coeflegend /* 1._predict: Pr(y==1), predict(pr outcome(1)) 2._predict: Pr(y==2), predict(pr outcome(2)) 3._predict: Pr(y==3), predict(pr outcome(3)) 4._predict: Pr(y==4), predict(pr outcome(4)) 5._predict: Pr(y==5), predict(pr outcome(5))  --------------------------------------------------------------------------------                |     Margin   Legend ---------------+---------------------------------------------------------------- _predict#numq5 |          1  0  |   .1005375  _b[1bn._predict#0bn.numq5]          1  1  |   .1385677  _b[1bn._predict#1.numq5]          1  2  |   .1879771  _b[1bn._predict#2.numq5]          1  3  |   .2498936  _b[1bn._predict#3.numq5]          1  4  |   .3240654  _b[1bn._predict#4.numq5]          1  5  |   .4082699  _b[1bn._predict#5.numq5]          1  6  |   .4982267  _b[1bn._predict#6.numq5]          1  7  |   .5882985  _b[1bn._predict#7.numq5]          1  8  |   .6728194  _b[1bn._predict#8.numq5]          1  9  |   .7474378  _b[1bn._predict#9.numq5]          1 10  |   .8098479  _b[1bn._predict#10.numq5]          2  0  |   .1476734  _b[2._predict#0bn.numq5]          2  1  |     .18353  _b[2._predict#1.numq5]          2  2  |   .2181211  _b[2._predict#2.numq5]          2  3  |   .2460839  _b[2._predict#3.numq5]          2  4  |   .2620521  _b[2._predict#4.numq5]          2  5  |   .2625658  _b[2._predict#5.numq5]          2  6  |   .2475088  _b[2._predict#6.numq5]          2  7  |   .2201601  _b[2._predict#7.numq5]          2  8  |   .1858223  _b[2._predict#8.numq5]          2  9  |   .1499087  _b[2._predict#9.numq5]          2 10  |   .1165146  _b[2._predict#10.numq5]          3  0  |   .3001565  _b[3._predict#0bn.numq5]          3  1  |   .3139164  _b[3._predict#1.numq5]          3  2  |   .3093787  _b[3._predict#2.numq5]          3  3  |   .2875196  _b[3._predict#3.numq5]          3  4  |   .2527995  _b[3._predict#4.numq5]          3  5  |   .2114464  _b[3._predict#5.numq5]          3  6  |   .1694181  _b[3._predict#6.numq5]          3  7  |   .1310173  _b[3._predict#7.numq5]          3  8  |   .0985105  _b[3._predict#8.numq5]          3  9  |   .0724854  _b[3._predict#9.numq5]          3 10  |   .0524797  _b[3._predict#10.numq5]          4  0  |    .171399  _b[4._predict#0bn.numq5]          4  1  |   .1510523  _b[4._predict#1.numq5]          4  2  |   .1262804  _b[4._predict#2.numq5]          4  3  |   .1009659  _b[4._predict#3.numq5]          4  4  |   .0778659  _b[4._predict#4.numq5]          4  5  |   .0583862  _b[4._predict#5.numq5]          4  6  |   .0428584  _b[4._predict#6.numq5]          4  7  |   .0309693  _b[4._predict#7.numq5]          4  8  |   .0221241  _b[4._predict#8.numq5]          4  9  |   .0156762  _b[4._predict#9.numq5]          4 10  |    .011043  _b[4._predict#10.numq5]          5  0  |   .2802336  _b[5._predict#0bn.numq5]          5  1  |   .2129336  _b[5._predict#1.numq5]          5  2  |   .1582428  _b[5._predict#2.numq5]          5  3  |    .115537  _b[5._predict#3.numq5]          5  4  |    .083217  _b[5._predict#4.numq5]          5  5  |   .0593317  _b[5._predict#5.numq5]          5  6  |    .041988  _b[5._predict#6.numq5]          5  7  |   .0295549  _b[5._predict#7.numq5]          5  8  |   .0207237  _b[5._predict#8.numq5]          5  9  |    .014492  _b[5._predict#9.numq5]          5 10  |   .0101148  _b[5._predict#10.numq5] -------------------------------------------------------------------------------- */ *Compute the CIs the Ord Expected for #BFs=0 lincom  _b[1bn._predict#0bn.numq5]*1+ _b[2._predict#0bn.numq5]*2+ /// _b[3._predict#0bn.numq5]*3+_b[4._predict#0bn.numq5]*4+ /// _b[5._predict#0bn.numq5]*5 /* ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   3.383118    .127296    26.58   0.000     3.133622    3.632613 ------------------------------------------------------------------------------ */ *Compute the CIs the Ord Expected for #BFs=1 lincom  _b[1bn._predict#1.numq5]*1+ _b[2._predict#1.numq5]*2+ /// _b[3._predict#1.numq5]*3+_b[4._predict#1.numq5]*4+ /// _b[5._predict#1.numq5]*5 /* ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   3.116254   .1080101    28.85   0.000     2.904558     3.32795 ------------------------------------------------------------------------------ */ *Compute the CIs the Ord Expected for #BFs=2,...,10 forvalues i=2(1)10{ lincom _b[1bn._predict#`i'.numq5]*1+ _b[2._predict#`i'.numq5]*2+ /// _b[3._predict#`i'.numq5]*3+_b[4._predict#`i'.numq5]*4+ /// _b[5._predict#`i'.numq5]*5 } /*  ( 1)  1bn._predict#2.numq5 + 2*2._predict#2.numq5 + 3*3._predict#2.numq5 + 4*4._predict#2.numq5 + 5*5._predict#2.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   2.848691   .0894196    31.86   0.000     2.673432     3.02395 ------------------------------------------------------------------------------   ( 1)  1bn._predict#3.numq5 + 2*2._predict#3.numq5 + 3*3._predict#3.numq5 + 4*4._predict#3.numq5 + 5*5._predict#3.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   2.586169   .0749321    34.51   0.000     2.439305    2.733033 ------------------------------------------------------------------------------   ( 1)  1bn._predict#4.numq5 + 2*2._predict#4.numq5 + 3*3._predict#4.numq5 + 4*4._predict#4.numq5 + 5*5._predict#4.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   2.334117   .0673553    34.65   0.000     2.202103    2.466131 ------------------------------------------------------------------------------   ( 1)  1bn._predict#5.numq5 + 2*2._predict#5.numq5 + 3*3._predict#5.numq5 + 4*4._predict#5.numq5 + 5*5._predict#5.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   2.097944   .0668117    31.40   0.000     1.966995    2.228892 ------------------------------------------------------------------------------   ( 1)  1bn._predict#6.numq5 + 2*2._predict#6.numq5 + 3*3._predict#6.numq5 + 4*4._predict#6.numq5 + 5*5._predict#6.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   1.882872   .0699561    26.92   0.000     1.745761    2.019983 ------------------------------------------------------------------------------   ( 1)  1bn._predict#7.numq5 + 2*2._predict#7.numq5 + 3*3._predict#7.numq5 + 4*4._predict#7.numq5 + 5*5._predict#7.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   1.693322   .0727029    23.29   0.000     1.550827    1.835817 ------------------------------------------------------------------------------   ( 1)  1bn._predict#8.numq5 + 2*2._predict#8.numq5 + 3*3._predict#8.numq5 + 4*4._predict#8.numq5 + 5*5._predict#8.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |    1.53211   .0725487    21.12   0.000     1.389918    1.674303 ------------------------------------------------------------------------------   ( 1)  1bn._predict#9.numq5 + 2*2._predict#9.numq5 + 3*3._predict#9.numq5 + 4*4._predict#9.numq5 + 5*5._predict#9.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   1.399876   .0688454    20.33   0.000     1.264941     1.53481 ------------------------------------------------------------------------------   ( 1)  1bn._predict#10.numq5 + 2*2._predict#10.numq5 + 3*3._predict#10.numq5 + 4*4._predict#10.numq5 + 5*5._predict#10.numq5 = 0  ------------------------------------------------------------------------------              | Coefficient  Std. err.      z    P>|z|     [95% conf. interval] -------------+----------------------------------------------------------------          (1) |   1.295062   .0622247    20.81   0.000     1.173104     1.41702 ------------------------------------------------------------------------------ */

Last edited by Rich Goldstein; 14 Feb 2023, 07:02. Reason: added in edit - the above is clearly not formed correctly but I don't know how to fix - I hope there is enough there to pick it out; note that the set-up from tech support is slightly different from t

Comment

Rich Goldstein

Join Date: Mar 2014
Posts: 4466

#10

14 Feb 2023, 07:08

here is another try:

Code:

This is the follow-up email. I find how the “Ord Expected” variable is calculated. It is computed by averaging
             1*Prob[1]+2*Prob[2]+3*Prob[3]+4*Prob[4]+5*Prob[5]
 for each #BFs group.
 Please run the following code to reproduce the values of the “Ord Expected” variable in Stata:
 ********
 use for_ologit, clear
*make y =LL in order to match JMP's result
generate y=q6_1b+1
summarize
ologit y numq5
predict p1-p5
list in 1/10
generate Ord_expected=1*p1+2*p2+3*p3+4*p4+5*p5
bysort numq5: summarize Ord_expected
margins, over(numq5) predict()  post coeflegend
/*
1._predict: Pr(y==1), predict(pr outcome(1))
2._predict: Pr(y==2), predict(pr outcome(2))
3._predict: Pr(y==3), predict(pr outcome(3))
4._predict: Pr(y==4), predict(pr outcome(4))
5._predict: Pr(y==5), predict(pr outcome(5))
 
--------------------------------------------------------------------------------
               |     Margin   Legend
---------------+----------------------------------------------------------------
_predict#numq5 |
         1  0  |   .1005375  _b[1bn._predict#0bn.numq5]
         1  1  |   .1385677  _b[1bn._predict#1.numq5]
         1  2  |   .1879771  _b[1bn._predict#2.numq5]
         1  3  |   .2498936  _b[1bn._predict#3.numq5]
         1  4  |   .3240654  _b[1bn._predict#4.numq5]
         1  5  |   .4082699  _b[1bn._predict#5.numq5]
         1  6  |   .4982267  _b[1bn._predict#6.numq5]
         1  7  |   .5882985  _b[1bn._predict#7.numq5]
         1  8  |   .6728194  _b[1bn._predict#8.numq5]
         1  9  |   .7474378  _b[1bn._predict#9.numq5]
         1 10  |   .8098479  _b[1bn._predict#10.numq5]
         2  0  |   .1476734  _b[2._predict#0bn.numq5]
         2  1  |     .18353  _b[2._predict#1.numq5]
         2  2  |   .2181211  _b[2._predict#2.numq5]
         2  3  |   .2460839  _b[2._predict#3.numq5]
         2  4  |   .2620521  _b[2._predict#4.numq5]
         2  5  |   .2625658  _b[2._predict#5.numq5]
         2  6  |   .2475088  _b[2._predict#6.numq5]
         2  7  |   .2201601  _b[2._predict#7.numq5]
         2  8  |   .1858223  _b[2._predict#8.numq5]
         2  9  |   .1499087  _b[2._predict#9.numq5]
         2 10  |   .1165146  _b[2._predict#10.numq5]
         3  0  |   .3001565  _b[3._predict#0bn.numq5]
         3  1  |   .3139164  _b[3._predict#1.numq5]
         3  2  |   .3093787  _b[3._predict#2.numq5]
         3  3  |   .2875196  _b[3._predict#3.numq5]
         3  4  |   .2527995  _b[3._predict#4.numq5]
         3  5  |   .2114464  _b[3._predict#5.numq5]
         3  6  |   .1694181  _b[3._predict#6.numq5]
         3  7  |   .1310173  _b[3._predict#7.numq5]
         3  8  |   .0985105  _b[3._predict#8.numq5]
         3  9  |   .0724854  _b[3._predict#9.numq5]
         3 10  |   .0524797  _b[3._predict#10.numq5]
         4  0  |    .171399  _b[4._predict#0bn.numq5]
         4  1  |   .1510523  _b[4._predict#1.numq5]
         4  2  |   .1262804  _b[4._predict#2.numq5]
         4  3  |   .1009659  _b[4._predict#3.numq5]
         4  4  |   .0778659  _b[4._predict#4.numq5]
         4  5  |   .0583862  _b[4._predict#5.numq5]
         4  6  |   .0428584  _b[4._predict#6.numq5]
         4  7  |   .0309693  _b[4._predict#7.numq5]
         4  8  |   .0221241  _b[4._predict#8.numq5]
         4  9  |   .0156762  _b[4._predict#9.numq5]
         4 10  |    .011043  _b[4._predict#10.numq5]
         5  0  |   .2802336  _b[5._predict#0bn.numq5]
         5  1  |   .2129336  _b[5._predict#1.numq5]
         5  2  |   .1582428  _b[5._predict#2.numq5]
         5  3  |    .115537  _b[5._predict#3.numq5]
         5  4  |    .083217  _b[5._predict#4.numq5]
         5  5  |   .0593317  _b[5._predict#5.numq5]
         5  6  |    .041988  _b[5._predict#6.numq5]
         5  7  |   .0295549  _b[5._predict#7.numq5]
         5  8  |   .0207237  _b[5._predict#8.numq5]
         5  9  |    .014492  _b[5._predict#9.numq5]
         5 10  |   .0101148  _b[5._predict#10.numq5]
--------------------------------------------------------------------------------
*/
*Compute the CIs the Ord Expected for #BFs=0
lincom  _b[1bn._predict#0bn.numq5]*1+ _b[2._predict#0bn.numq5]*2+ ///
        _b[3._predict#0bn.numq5]*3+_b[4._predict#0bn.numq5]*4+ ///
               _b[5._predict#0bn.numq5]*5
/*
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   3.383118    .127296    26.58   0.000     3.133622    3.632613
------------------------------------------------------------------------------
*/
*Compute the CIs the Ord Expected for #BFs=1
lincom  _b[1bn._predict#1.numq5]*1+ _b[2._predict#1.numq5]*2+ ///
        _b[3._predict#1.numq5]*3+_b[4._predict#1.numq5]*4+ ///
               _b[5._predict#1.numq5]*5
/*
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   3.116254   .1080101    28.85   0.000     2.904558     3.32795
------------------------------------------------------------------------------
*/
*Compute the CIs the Ord Expected for #BFs=2,...,10
forvalues i=2(1)10{
lincom _b[1bn._predict#`i'.numq5]*1+ _b[2._predict#`i'.numq5]*2+ ///
        _b[3._predict#`i'.numq5]*3+_b[4._predict#`i'.numq5]*4+ ///
               _b[5._predict#`i'.numq5]*5
}
/*
 ( 1)  1bn._predict#2.numq5 + 2*2._predict#2.numq5 + 3*3._predict#2.numq5 + 4*4._predict#2.numq5 + 5*5._predict#2.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   2.848691   .0894196    31.86   0.000     2.673432     3.02395
------------------------------------------------------------------------------
 
 ( 1)  1bn._predict#3.numq5 + 2*2._predict#3.numq5 + 3*3._predict#3.numq5 + 4*4._predict#3.numq5 + 5*5._predict#3.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   2.586169   .0749321    34.51   0.000     2.439305    2.733033
------------------------------------------------------------------------------
 
 ( 1)  1bn._predict#4.numq5 + 2*2._predict#4.numq5 + 3*3._predict#4.numq5 + 4*4._predict#4.numq5 + 5*5._predict#4.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   2.334117   .0673553    34.65   0.000     2.202103    2.466131
------------------------------------------------------------------------------
 
 ( 1)  1bn._predict#5.numq5 + 2*2._predict#5.numq5 + 3*3._predict#5.numq5 + 4*4._predict#5.numq5 + 5*5._predict#5.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   2.097944   .0668117    31.40   0.000     1.966995    2.228892
------------------------------------------------------------------------------
 
 ( 1)  1bn._predict#6.numq5 + 2*2._predict#6.numq5 + 3*3._predict#6.numq5 + 4*4._predict#6.numq5 + 5*5._predict#6.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   1.882872   .0699561    26.92   0.000     1.745761    2.019983
------------------------------------------------------------------------------
 
 ( 1)  1bn._predict#7.numq5 + 2*2._predict#7.numq5 + 3*3._predict#7.numq5 + 4*4._predict#7.numq5 + 5*5._predict#7.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   1.693322   .0727029    23.29   0.000     1.550827    1.835817
------------------------------------------------------------------------------
 
 ( 1)  1bn._predict#8.numq5 + 2*2._predict#8.numq5 + 3*3._predict#8.numq5 + 4*4._predict#8.numq5 + 5*5._predict#8.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |    1.53211   .0725487    21.12   0.000     1.389918    1.674303
------------------------------------------------------------------------------
 
 ( 1)  1bn._predict#9.numq5 + 2*2._predict#9.numq5 + 3*3._predict#9.numq5 + 4*4._predict#9.numq5 + 5*5._predict#9.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   1.399876   .0688454    20.33   0.000     1.264941     1.53481
------------------------------------------------------------------------------
 
 ( 1)  1bn._predict#10.numq5 + 2*2._predict#10.numq5 + 3*3._predict#10.numq5 + 4*4._predict#10.numq5 + 5*5._predict#10.numq5 = 0
 
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         (1) |   1.295062   .0622247    20.81   0.000     1.173104     1.41702
------------------------------------------------------------------------------
*/          
 
*********
 
Both -generate Ord_expected=1*p1+2*p2+3*p3+4*p4+5*p5- and -margins, over(numq5) predict()  post coeflegend- can compute the values of the “Ord Expected” variable. I apply -lincom- to the result of -margins- to compute the confidence intervals of the “Ord Expected” variable. You can compare my calculations with the JMP’s result.

Comment

George Ford

Join Date: Aug 2014

Posts: 3152
#11

14 Feb 2023, 07:25

Cool. Turned out to be pretty straightforward.
Comment

Announcement