Re: st: ICE with Interval Error #459 multiple imputation

[A Loumiotis <antonis.loumiotis@gmail.com>]

To clarify, you do not need to create bound variables if you use
ologit to impute sclifeca2.  Bound variables are needed only if you
use intreg to impute sclifeca2.

Antonis

On Tue, Aug 21, 2012 at 9:25 PM, Nick Cox <njcoxstata@gmail.com> wrote:
> The initial assignments to 999 are puzzling here.
>
> This code can be condensed to
>
> foreach v in sclifeca2 sclifeca3 sclifeca4 {
>      gen `v'll1 = cond(`v' == ., 1, `v')
>      gen `v'ul7 = cond(`v' == ., 7, `v')
> }
>
> Nick
>
> On Tue, Aug 21, 2012 at 6:56 PM, Laura Kudrna <laura.kudrna@gmail.com> wrote:
>> Dear Antonis,
>>
>> Thank you very much for solving my problem!  Using the ologit prefix
>> combined with your example for how to set my upper and lower limits
>> solved everything.
>>
>> For anyone else who struggles, I've included my syntax below for
>> defining the upper and lower limits of my variables sclifeca2,
>> sclifeca3, and sclifeca4 to 7 and 1, respectively.
>>
>> Best wishes,
>> Laura
>>
>> gen sclifeca2ll1 = 999
>> replace sclifeca2ll1 = sclifeca2 if sclifeca2 !=.
>> replace sclifeca2ll1 = 1 if sclifeca2 == .
>>
>> gen sclifeca2ul7 = 999
>> replace sclifeca2ul7 = sclifeca2 if sclifeca2 !=.
>> replace sclifeca2ul7 = 7 if sclifeca2 == .
>>
>> gen sclifeca3ll1 = 999
>> replace sclifeca3ll1 = sclifeca3 if sclifeca3 !=.
>> replace sclifeca3ll1 = 1 if sclifeca3 == .
>>
>> gen sclifeca3ul7 = 999
>> replace sclifeca3ul7 = sclifeca3 if sclifeca3 !=.
>> replace sclifeca3ul7 = 7 if sclifeca3 == .
>>
>> gen sclifeca4ul7 = 999
>> replace sclifeca4ul7 = sclifeca4 if sclifeca4 !=.
>> replace sclifeca4ul7 = 7 if sclifeca4 == .
>>
>> gen sclifeca4ll1 = 999
>> replace sclifeca4ll1 = sclifeca4 if sclifeca4 !=.
>> replace sclifeca4ll1 = 1 if sclifeca4 == .
>>
>>
>> On Mon, Aug 13, 2012 at 7:49 AM, A Loumiotis
>> <antonis.loumiotis@gmail.com> wrote:
>>> Perhaps you could also try to use ologit instead of intreg to impute
>>> sclifeca2.  Make sure you use the o. prefix, ie,
>>> ice o.sclifeca2 ..., saving(imputed_dataset) m(5)
>>>
>>> On Mon, Aug 13, 2012 at 9:20 AM, A Loumiotis
>>> <antonis.loumiotis@gmail.com> wrote:
>>>> Hi Laura,
>>>>
>>>> I'm not sure if you are setting up your lower and upper variables
>>>> correctly because you said that when the observation is not missing
>>>> both upper and lower bounds take the value of 7.  This is not correct.
>>>>  They must be equal to the non-missing value.  Here is an example of
>>>> how they should look:
>>>>
>>>> input sclifeca2 ll1 ul7
>>>> 4 4 4
>>>> . 1 7
>>>> 2.5 2.5 2.5
>>>> 3 3 3
>>>> . 1 7
>>>> end
>>>>
>>>> Best,
>>>> Antonis
>>>>
>>>>
>>>>
>>>>
>>>> On Fri, Aug 10, 2012 at 9:32 PM, Laura Kudrna <laura.kudrna@gmail.com> wrote:
>>>>> Hi,
>>>>>
>>>>> Thank you for your response.  Apologies about ICE, I'm new to the
>>>>> listserve: ICE comes from //www.homepages.ucl.ac.uk/~ucakjpr/stata/
>>>>>
>>>>> To answer your question about the bounds, yes: when I created the
>>>>> upper limit variable ul7 that has no missing values and I include it
>>>>> in the ice code, the upper and lower bounds remain 7.  But for
>>>>> example, when I impute my variable sclifeca2  "Life Satisfaction"
>>>>> which has a scale of 1 to 7 I get imputed values ranging from -0.5 to
>>>>> 9.  I am using multiple imputation to impute all the predictor and
>>>>> explanatory variables at once.
>>>>>
>>>>> I replicated the STATA code precisely from the ICE help for interval.
>>>>>
>>>>> ICE help for interval states:  ...intlist has the syntax varname:llvar
>>>>> ulvar [, varname:llvar ulvar ...], where each varname is an
>>>>> interval-censored variable, each llvar contains the lower bound (a)
>>>>> for varname and each ulvar contains the upper bound (b)
>>>>>
>>>>> My syntax is: ice ll1 ul7 sclifeca2 .....(other variables).....,
>>>>> saving(imputed_dataset) m(5) interval(sclifeca2:ll1 ul7).
>>>>>
>>>>> I still can't understand why I am getting the error message and I
>>>>> would be most grateful if you have any other suggestions on how to fix
>>>>> it.
>>>>>
>>>>> Best wishes,
>>>>> Laura
>>>>>
>>>>>
>>>>> On Fri, Aug 10, 2012 at 7:32 AM, A Loumiotis
>>>>> <antonis.loumiotis@gmail.com> wrote:
>>>>>> If the variable that you are imputing with interval regression is not
>>>>>> missing then the lower and upper bounds should equal that non missing
>>>>>> value.  Is this true in your case?  See -help ice- especially the
>>>>>> interval option.
>>>>>>
>>>>>> -ice- is a user written program available from SJ-9-3.  Please state
>>>>>> where user written programs come from as explained in Statalist FAQ.
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Fri, Aug 10, 2012 at 12:29 AM, Laura Kudrna <laura.kudrna@gmail.com> wrote:
>>>>>>> Hi,
>>>>>>>
>>>>>>> I am having the same error message / problem as the post below from
>>>>>>> the STATA message boards - Error #459.  It occurs when using the
>>>>>>> interval option with ICE.
>>>>>>>
>>>>>>> I generated two variables, ll1 and ul7, containing values of 1 and 7
>>>>>>> respectively to set the upper and lower bound for my variable
>>>>>>> sclifeca2.
>>>>>>>
>>>>>>> My STATA code is:
>>>>>>>
>>>>>>> ice ll1 ul7 sclifeca2 sclifeca3 sclifeca4 sclddr2 sclddr3 sclddr4
>>>>>>> iincome2 iincome3 iincome4 education2 occupation2 unemployed2
>>>>>>> unemployed3 sex2 age2 age3 age4 partner2 partner3 partner4 vehicles2
>>>>>>> vehicles3 vehicles4 holiday2 holiday3 holiday4 secprop2 secprop3
>>>>>>> secprop4 weight2 weight3 weight4, saving(imputed_dataset) m(5)
>>>>>>> interval(sclifeca2:ll1 ul7)
>>>>>>>
>>>>>>> My error message is:
>>>>>>> Error #459 encountered while running -uvis-
>>>>>>> I detected a problem with running uvis with command intreg on response sclifeca2
>>>>>>> and covariates sclifeca3 sclifeca4 sclddr2 sclddr3 sclddr4 iincome2
>>>>>>> iincome3 iincome4 education2 occupation2 unemployed2 unemployed3 sex2
>>>>>>> ag
>>>>>>>> e2 age3 age4 partner2 partner3 partner4 vehicles2 vehicles3 vehicles4 holiday2 holiday3 holiday4 secprop2 secprop3 secprop4 weight2 weight
>>>>>>>> 3 weight4.
>>>>>>>
>>>>>>> The offending command resembled:
>>>>>>> uvis intreg sclifeca2 sclifeca3 sclifeca4 sclddr2 sclddr3 sclddr4
>>>>>>> iincome2 iincome3 iincome4 education2 occupation2 unemployed2
>>>>>>> unemployed3
>>>>>>>> sex2 age2 age3 age4 partner2 partner3 partner4 vehicles2 vehicles3 vehicles4 holiday2 holiday3 holiday4 secprop2 secprop3 secprop4 weight2
>>>>>>>>  weight3 weight4 , gen([imputed])
>>>>>>>
>>>>>>> you may wish to try the -persist- option to persist beyond this error.
>>>>>>> dumping current data to ./_ice_dump.dta
>>>>>>> something that should be true of your data is not
>>>>>>>
>>>>>>> When I use the "persist" option, the data is imputed, but doesn't set
>>>>>>> the upper and lower bounds of my variable.
>>>>>>>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/