8 Feb
2012
8 Feb
'12
4:10 p.m.
Hello all:
I have replicated this with the current (stable) and alpha releases of
Zelig. I cannot simulate quantities of interest using a mixed-effects
model estimated on data imputed with Ameila. setx() works fine, but
the sim command runs for a while and then reports it is unable to find
one of my predictor variables (variable names changed as they are
irrelevant). Code below:
model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
subset = wave4)
x.out <- setx(model.full)
simout <- sim(x.out, x = x.out)
The model estimates fine with the zelig() function, setx() works fine
in setting the explanatory variables at their means, but when I
attempt to use sim(), I get:
Error in eval(expr, envir, enclos) : object 'age' not found
A traceback produces this:
13: eval(expr, envir, enclos)
12: eval(predvars, data, env)
11: model.frame.default(object, data, xlev = xlev)
10: model.frame(object, data, xlev = xlev)
9: model.matrix.default(parsefml$fixed, data = D)
8: model.matrix(parsefml$fixed, data = D)
7: model.matrix(parsefml$fixed, data = D)
6: is.data.frame(x)
5: colnames(model.matrix(parsefml$fixed, data = D))
4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
(!is.null(x1)) as.matrix(x1))
3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
(!is.null(x1)) as.matrix(x1))
2: sim.setx.MI(model.full, x = x.out)
1: sim(model.full, x = x.out)
Any suggestions?
Thanks very much,
Trey Causey
8 Feb
8 Feb
4:18 p.m.
Hi Trey -
Is the error identical across both versions of Zelig?
On Feb 8, 2012, at 4:10 PM, Trey Causey wrote:
Hello all:
I have replicated this with the current (stable) and alpha releases of
Zelig. I cannot simulate quantities of interest using a mixed-effects
model estimated on data imputed with Ameila. setx() works fine, but
the sim command runs for a while and then reports it is unable to find
one of my predictor variables (variable names changed as they are
irrelevant). Code below:
model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
subset = wave4)
x.out <- setx(model.full)
simout <- sim(x.out, x = x.out)
The model estimates fine with the zelig() function, setx() works fine
in setting the explanatory variables at their means, but when I
attempt to use sim(), I get:
Error in eval(expr, envir, enclos) : object 'age' not found
A traceback produces this:
13: eval(expr, envir, enclos)
12: eval(predvars, data, env)
11: model.frame.default(object, data, xlev = xlev)
10: model.frame(object, data, xlev = xlev)
9: model.matrix.default(parsefml$fixed, data = D)
8: model.matrix(parsefml$fixed, data = D)
7: model.matrix(parsefml$fixed, data = D)
6: is.data.frame(x)
5: colnames(model.matrix(parsefml$fixed, data = D))
4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
(!is.null(x1)) as.matrix(x1))
3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
(!is.null(x1)) as.matrix(x1))
2: sim.setx.MI(model.full, x = x.out)
1: sim(model.full, x = x.out)
Any suggestions?
Thanks very much,
Trey Causey
-
--
Zelig Mailing List, served by HUIT
Send messages: zelig(a)lists.gking.harvard.edu
[un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
Zelig program information: http://gking.harvard.edu/zelig/
Zelig mailing list
Zelig(a)lists.gking.harvard.edu
https://lists.gking.harvard.edu/mailman/listinfo/zelig
4:20 p.m.
Yes, it's exactly the same.
On Wed, Feb 8, 2012 at 1:18 PM, Matt Owen <mowen(a)iq.harvard.edu> wrote:
Hi Trey -
Is the error identical across both versions of Zelig?
On Feb 8, 2012, at 4:10 PM, Trey Causey wrote:
Hello all:
I have replicated this with the current (stable) and alpha releases of
Zelig. I cannot simulate quantities of interest using a mixed-effects
model estimated on data imputed with Ameila. setx() works fine, but
the sim command runs for a while and then reports it is unable to find
one of my predictor variables (variable names changed as they are
irrelevant). Code below:
model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
subset = wave4)
x.out <- setx(model.full)
simout <- sim(x.out, x = x.out)
The model estimates fine with the zelig() function, setx() works fine
in setting the explanatory variables at their means, but when I
attempt to use sim(), I get:
Error in eval(expr, envir, enclos) : object 'age' not found
A traceback produces this:
13: eval(expr, envir, enclos)
12: eval(predvars, data, env)
11: model.frame.default(object, data, xlev = xlev)
10: model.frame(object, data, xlev = xlev)
9: model.matrix.default(parsefml$fixed, data = D)
8: model.matrix(parsefml$fixed, data = D)
7: model.matrix(parsefml$fixed, data = D)
6: is.data.frame(x)
5: colnames(model.matrix(parsefml$fixed, data = D))
4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
(!is.null(x1)) as.matrix(x1))
3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
(!is.null(x1)) as.matrix(x1))
2: sim.setx.MI(model.full, x = x.out)
1: sim(model.full, x = x.out)
Any suggestions?
Thanks very much,
Trey Causey
-
--
Zelig Mailing List, served by HUIT
Send messages: zelig(a)lists.gking.harvard.edu
[un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
Zelig program information: http://gking.harvard.edu/zelig/
Zelig mailing list
Zelig(a)lists.gking.harvard.edu
https://lists.gking.harvard.edu/mailman/listinfo/zelig
4:33 p.m.
The issue may be with how Zelig interfaces with Amelia's mi class. Please attempt the
following (modified) zelig command:
model.full <- zelig(
attitude ~ age + married + male + income + tag(1 | state),
model = "ls.mixed",
data = imp$imputations$imp1,
REML = FALSE,
subset = wave4
)
Notice that the difference is in the data parameter:
data = imp$imputations$imp1
If this produces an error, we can narrow down the problem further. Otherwise, the issue is
likely a bug in Zelig's implementation for multilevel models.
- Matt
On Feb 8, 2012, at 4:20 PM, Trey Causey wrote:
Yes, it's exactly the same.
On Wed, Feb 8, 2012 at 1:18 PM, Matt Owen <mowen(a)iq.harvard.edu> wrote:
> Hi Trey -
> Is the error identical across both versions of Zelig?
>
> On Feb 8, 2012, at 4:10 PM, Trey Causey wrote:
>
>> Hello all:
>>
>> I have replicated this with the current (stable) and alpha releases of
>> Zelig. I cannot simulate quantities of interest using a mixed-effects
>> model estimated on data imputed with Ameila. setx() works fine, but
>> the sim command runs for a while and then reports it is unable to find
>> one of my predictor variables (variable names changed as they are
>> irrelevant). Code below:
>>
>> model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
>> state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
>> subset = wave4)
>>
>> x.out <- setx(model.full)
>>
>> simout <- sim(x.out, x = x.out)
>>
>> The model estimates fine with the zelig() function, setx() works fine
>> in setting the explanatory variables at their means, but when I
>> attempt to use sim(), I get:
>>
>> Error in eval(expr, envir, enclos) : object 'age' not found
>>
>> A traceback produces this:
>>
>> 13: eval(expr, envir, enclos)
>> 12: eval(predvars, data, env)
>> 11: model.frame.default(object, data, xlev = xlev)
>> 10: model.frame(object, data, xlev = xlev)
>> 9: model.matrix.default(parsefml$fixed, data = D)
>> 8: model.matrix(parsefml$fixed, data = D)
>> 7: model.matrix(parsefml$fixed, data = D)
>> 6: is.data.frame(x)
>> 5: colnames(model.matrix(parsefml$fixed, data = D))
>> 4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 2: sim.setx.MI(model.full, x = x.out)
>> 1: sim(model.full, x = x.out)
>>
>> Any suggestions?
>>
>> Thanks very much,
>> Trey Causey
>> -
>> --
>> Zelig Mailing List, served by HUIT
>> Send messages: zelig(a)lists.gking.harvard.edu
>> [un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
>> Zelig program information: http://gking.harvard.edu/zelig/
>> Zelig mailing list
>> Zelig(a)lists.gking.harvard.edu
>> https://lists.gking.harvard.edu/mailman/listinfo/zelig
>
4:40 p.m.
Yes, that works. Thanks very much!
On Wed, Feb 8, 2012 at 1:33 PM, Matt Owen <mowen(a)iq.harvard.edu> wrote:
The issue may be with how Zelig interfaces with
Amelia's mi class. Please attempt the following (modified) zelig command:
model.full <- zelig(
attitude ~ age + married + male + income + tag(1 | state),
model = "ls.mixed",
data = imp$imputations$imp1,
REML = FALSE,
subset = wave4
)
Notice that the difference is in the data parameter:
data = imp$imputations$imp1
If this produces an error, we can narrow down the problem further. Otherwise, the issue
is likely a bug in Zelig's implementation for multilevel models.
- Matt
On Feb 8, 2012, at 4:20 PM, Trey Causey wrote:
Yes, it's exactly the same.
On Wed, Feb 8, 2012 at 1:18 PM, Matt Owen <mowen(a)iq.harvard.edu> wrote:
> Hi Trey -
> Is the error identical across both versions of Zelig?
>
> On Feb 8, 2012, at 4:10 PM, Trey Causey wrote:
>
>> Hello all:
>>
>> I have replicated this with the current (stable) and alpha releases of
>> Zelig. I cannot simulate quantities of interest using a mixed-effects
>> model estimated on data imputed with Ameila. setx() works fine, but
>> the sim command runs for a while and then reports it is unable to find
>> one of my predictor variables (variable names changed as they are
>> irrelevant). Code below:
>>
>> model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
>> state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
>> subset = wave4)
>>
>> x.out <- setx(model.full)
>>
>> simout <- sim(x.out, x = x.out)
>>
>> The model estimates fine with the zelig() function, setx() works fine
>> in setting the explanatory variables at their means, but when I
>> attempt to use sim(), I get:
>>
>> Error in eval(expr, envir, enclos) : object 'age' not found
>>
>> A traceback produces this:
>>
>> 13: eval(expr, envir, enclos)
>> 12: eval(predvars, data, env)
>> 11: model.frame.default(object, data, xlev = xlev)
>> 10: model.frame(object, data, xlev = xlev)
>> 9: model.matrix.default(parsefml$fixed, data = D)
>> 8: model.matrix(parsefml$fixed, data = D)
>> 7: model.matrix(parsefml$fixed, data = D)
>> 6: is.data.frame(x)
>> 5: colnames(model.matrix(parsefml$fixed, data = D))
>> 4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 2: sim.setx.MI(model.full, x = x.out)
>> 1: sim(model.full, x = x.out)
>>
>> Any suggestions?
>>
>> Thanks very much,
>> Trey Causey
>> -
>> --
>> Zelig Mailing List, served by HUIT
>> Send messages: zelig(a)lists.gking.harvard.edu
>> [un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
>> Zelig program information: http://gking.harvard.edu/zelig/
>> Zelig mailing list
>> Zelig(a)lists.gking.harvard.edu
>> https://lists.gking.harvard.edu/mailman/listinfo/zelig
>
4:59 p.m.
Note that you lose the aggregated standard errors from doing
summary(zelig-output-from-a-MI-list-object).
On Wed, Feb 8, 2012 at 1:40 PM, Trey Causey <tcausey(a)uw.edu> wrote:
Yes, that works. Thanks very much!
On Wed, Feb 8, 2012 at 1:33 PM, Matt Owen <mowen(a)iq.harvard.edu> wrote:
-
--
Zelig Mailing List, served by HUIT
Send messages: zelig(a)lists.gking.harvard.edu
[un]subscribe Options:
http://lists.gking.harvard.edu/mailman/listinfo/zelig
Zelig program information: http://gking.harvard.edu/zelig/
Zelig mailing list
Zelig(a)lists.gking.harvard.edu
https://lists.gking.harvard.edu/mailman/listinfo/zelig
The issue may be with how Zelig interfaces with
Amelia's mi class.
Please attempt the following (modified) zelig command:
model.full <- zelig(
attitude ~ age + married + male + income + tag(1 |
state),
model = "ls.mixed",
data = imp$imputations$imp1,
REML = FALSE,
subset = wave4
)
Notice that the difference is in the data parameter:
data = imp$imputations$imp1
If this produces an error, we can narrow down the problem further.
Otherwise, the
issue is likely a bug in Zelig's implementation for
multilevel models.
- Matt
On Feb 8, 2012, at 4:20 PM, Trey Causey wrote:
> Yes, it's exactly the same.
>
> On Wed, Feb 8, 2012 at 1:18 PM, Matt Owen <mowen(a)iq.harvard.edu> wrote:
>> Hi Trey -
>> Is the error identical across both versions of Zelig?
>>
>> On Feb 8, 2012, at 4:10 PM, Trey Causey wrote:
>>
>>> Hello all:
>>>
>>> I have replicated this with the current (stable) and alpha releases of
>>> Zelig. I cannot simulate quantities of interest using a mixed-effects
>>> model estimated on data imputed with Ameila. setx() works fine, but
>>> the sim command runs for a while and then reports it is unable to find
>>> one of my predictor variables (variable names changed as they are
>>> irrelevant). Code below:
>>>
>>> model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
>>> state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
>>> subset = wave4)
>>>
>>> x.out <- setx(model.full)
>>>
>>> simout <- sim(x.out, x = x.out)
>>>
>>> The model estimates fine with the zelig() function, setx() works fine
>>> in setting the explanatory variables at their means, but when I
>>> attempt to use sim(), I get:
>>>
>>> Error in eval(expr, envir, enclos) : object 'age' not found
>>>
>>> A traceback produces this:
>>>
>>> 13: eval(expr, envir, enclos)
>>> 12: eval(predvars, data, env)
>>> 11: model.frame.default(object, data, xlev = xlev)
>>> 10: model.frame(object, data, xlev = xlev)
>>> 9: model.matrix.default(parsefml$fixed, data = D)
>>> 8: model.matrix(parsefml$fixed, data = D)
>>> 7: model.matrix(parsefml$fixed, data = D)
>>> 6: is.data.frame(x)
>>> 5: colnames(model.matrix(parsefml$fixed, data = D))
>>> 4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>>> (!is.null(x1)) as.matrix(x1))
>>> 3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>>> (!is.null(x1)) as.matrix(x1))
>>> 2: sim.setx.MI(model.full, x = x.out)
>>> 1: sim(model.full, x = x.out)
>>>
>>> Any suggestions?
>>>
>>> Thanks very much,
>>> Trey Causey
>>> -
>>> --
>>> Zelig Mailing List, served by HUIT
>>> Send messages: zelig(a)lists.gking.harvard.edu
>>> [un]subscribe Options:
http://lists.gking.harvard.edu/mailman/listinfo/zelig
>>
Zelig program information: http://gking.harvard.edu/zelig/
>> Zelig mailing list
>> Zelig(a)lists.gking.harvard.edu
>> https://lists.gking.harvard.edu/mailman/listinfo/zelig
>
5:39 p.m.
Yes, it seems there is an issue in constructing the design matrix for the
"ls.mixed" model. I'm creating an issue on github here:
https://github.com/zeligdev/ZeligMixed/issues/1
________________________________________
From: monkeykupo(a)gmail.com [monkeykupo(a)gmail.com] On Behalf Of Olivia Lau
[olivia.lau(a)post.harvard.edu]
Sent: Wednesday, February 08, 2012 4:59 PM
To: Trey Causey
Cc: Owen, Matthew; zelig(a)lists.gking.harvard.edu
Subject: Re: [zelig] sim(), Amelia, and quantities of interest with ls.mixed
Note that you lose the aggregated standard errors from doing
summary(zelig-output-from-a-MI-list-object).
On Wed, Feb 8, 2012 at 1:40 PM, Trey Causey
<tcausey@uw.edu<mailto:tcausey@uw.edu>> wrote:
Yes, that works. Thanks very much!
On Wed, Feb 8, 2012 at 1:33 PM, Matt Owen
<mowen@iq.harvard.edu<mailto:mowen@iq.harvard.edu>> wrote:
The issue may be with how Zelig interfaces with
Amelia's mi class. Please attempt the following (modified) zelig command:
model.full <- zelig(
attitude ~ age + married + male + income + tag(1 | state),
model = "ls.mixed",
data = imp$imputations$imp1,
REML = FALSE,
subset = wave4
)
Notice that the difference is in the data parameter:
data = imp$imputations$imp1
If this produces an error, we can narrow down the problem further. Otherwise, the issue
is likely a bug in Zelig's implementation for multilevel models.
- Matt
On Feb 8, 2012, at 4:20 PM, Trey Causey wrote:
-
--
Zelig Mailing List, served by HUIT
Send messages: zelig@lists.gking.harvard.edu<mailto:zelig@lists.gking.harvard.edu>
[un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
Zelig program information: http://gking.harvard.edu/zelig/
Zelig mailing list
Zelig@lists.gking.harvard.edu<mailto:Zelig@lists.gking.harvard.edu>
https://lists.gking.harvard.edu/mailman/listinfo/zelig
Yes, it's exactly the same.
On Wed, Feb 8, 2012 at 1:18 PM, Matt Owen
<mowen@iq.harvard.edu<mailto:mowen@iq.harvard.edu>> wrote:
> Hi Trey -
> Is the error identical across both versions of Zelig?
>
> On Feb 8, 2012, at 4:10 PM, Trey Causey wrote:
>
>> Hello all:
>>
>> I have replicated this with the current (stable) and alpha releases of
>> Zelig. I cannot simulate quantities of interest using a mixed-effects
>> model estimated on data imputed with Ameila. setx() works fine, but
>> the sim command runs for a while and then reports it is unable to find
>> one of my predictor variables (variable names changed as they are
>> irrelevant). Code below:
>>
>> model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
>> state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
>> subset = wave4)
>>
>> x.out <- setx(model.full)
>>
>> simout <- sim(x.out, x = x.out)
>>
>> The model estimates fine with the zelig() function, setx() works fine
>> in setting the explanatory variables at their means, but when I
>> attempt to use sim(), I get:
>>
>> Error in eval(expr, envir, enclos) : object 'age' not found
>>
>> A traceback produces this:
>>
>> 13: eval(expr, envir, enclos)
>> 12: eval(predvars, data, env)
>> 11: model.frame.default(object, data, xlev = xlev)
>> 10: model.frame(object, data, xlev = xlev)
>> 9: model.matrix.default(parsefml$fixed, data = D)
>> 8: model.matrix(parsefml$fixed, data = D)
>> 7: model.matrix(parsefml$fixed, data = D)
>> 6: is.data.frame(x)
>> 5: colnames(model.matrix(parsefml$fixed, data = D))
>> 4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 2: sim.setx.MI(model.full, x = x.out)
>> 1: sim(model.full, x = x.out)
>>
>> Any suggestions?
>>
>> Thanks very much,
>> Trey Causey
>> -
>> --
>> Zelig Mailing List, served by HUIT
>> Send messages:
zelig@lists.gking.harvard.edu<mailto:zelig@lists.gking.harvard.edu>
>> [un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
>> Zelig program information: http://gking.harvard.edu/zelig/
>> Zelig mailing list
>> Zelig@lists.gking.harvard.edu<mailto:Zelig@lists.gking.harvard.edu>
>> https://lists.gking.harvard.edu/mailman/listinfo/zelig
>
5:45 p.m.
Yes -- in the past I had been using the MIextract/MIcombine functions
from the mitools package, as Zelig has never been able to combine
imputed datasets for mixed effects models. Thanks for the reminder on
the QoIs, though.
On Wed, Feb 8, 2012 at 2:39 PM, Owen, Matthew <mowen(a)iq.harvard.edu> wrote:
Yes, it seems there is an issue in constructing the
design matrix for the "ls.mixed" model. I'm creating an issue on github
here:
https://github.com/zeligdev/ZeligMixed/issues/1
________________________________________
From: monkeykupo(a)gmail.com [monkeykupo(a)gmail.com] On Behalf Of Olivia Lau
[olivia.lau(a)post.harvard.edu]
Sent: Wednesday, February 08, 2012 4:59 PM
To: Trey Causey
Cc: Owen, Matthew; zelig(a)lists.gking.harvard.edu
Subject: Re: [zelig] sim(), Amelia, and quantities of interest with ls.mixed
Note that you lose the aggregated standard errors from doing
summary(zelig-output-from-a-MI-list-object).
On Wed, Feb 8, 2012 at 1:40 PM, Trey Causey
<tcausey@uw.edu<mailto:tcausey@uw.edu>> wrote:
Yes, that works. Thanks very much!
On Wed, Feb 8, 2012 at 1:33 PM, Matt Owen
<mowen@iq.harvard.edu<mailto:mowen@iq.harvard.edu>> wrote:
The issue may be with how Zelig interfaces with
Amelia's mi class. Please attempt the following (modified) zelig command:
model.full <- zelig(
attitude ~ age + married + male + income + tag(1 | state),
model = "ls.mixed",
data = imp$imputations$imp1,
REML = FALSE,
subset = wave4
)
Notice that the difference is in the data parameter:
data = imp$imputations$imp1
If this produces an error, we can narrow down the problem further. Otherwise, the issue
is likely a bug in Zelig's implementation for multilevel models.
- Matt
On Feb 8, 2012, at 4:20 PM, Trey Causey wrote:
-
--
Zelig Mailing List, served by HUIT
Send messages: zelig@lists.gking.harvard.edu<mailto:zelig@lists.gking.harvard.edu>
[un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
Zelig program information: http://gking.harvard.edu/zelig/
Zelig mailing list
Zelig@lists.gking.harvard.edu<mailto:Zelig@lists.gking.harvard.edu>
https://lists.gking.harvard.edu/mailman/listinfo/zelig
Yes, it's exactly the same.
On Wed, Feb 8, 2012 at 1:18 PM, Matt Owen
<mowen@iq.harvard.edu<mailto:mowen@iq.harvard.edu>> wrote:
> Hi Trey -
> Is the error identical across both versions of Zelig?
>
> On Feb 8, 2012, at 4:10 PM, Trey Causey wrote:
>
>> Hello all:
>>
>> I have replicated this with the current (stable) and alpha releases of
>> Zelig. I cannot simulate quantities of interest using a mixed-effects
>> model estimated on data imputed with Ameila. setx() works fine, but
>> the sim command runs for a while and then reports it is unable to find
>> one of my predictor variables (variable names changed as they are
>> irrelevant). Code below:
>>
>> model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
>> state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
>> subset = wave4)
>>
>> x.out <- setx(model.full)
>>
>> simout <- sim(x.out, x = x.out)
>>
>> The model estimates fine with the zelig() function, setx() works fine
>> in setting the explanatory variables at their means, but when I
>> attempt to use sim(), I get:
>>
>> Error in eval(expr, envir, enclos) : object 'age' not found
>>
>> A traceback produces this:
>>
>> 13: eval(expr, envir, enclos)
>> 12: eval(predvars, data, env)
>> 11: model.frame.default(object, data, xlev = xlev)
>> 10: model.frame(object, data, xlev = xlev)
>> 9: model.matrix.default(parsefml$fixed, data = D)
>> 8: model.matrix(parsefml$fixed, data = D)
>> 7: model.matrix(parsefml$fixed, data = D)
>> 6: is.data.frame(x)
>> 5: colnames(model.matrix(parsefml$fixed, data = D))
>> 4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 2: sim.setx.MI(model.full, x = x.out)
>> 1: sim(model.full, x = x.out)
>>
>> Any suggestions?
>>
>> Thanks very much,
>> Trey Causey
>> -
>> --
>> Zelig Mailing List, served by HUIT
>> Send messages:
zelig@lists.gking.harvard.edu<mailto:zelig@lists.gking.harvard.edu>
>> [un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
>> Zelig program information: http://gking.harvard.edu/zelig/
>> Zelig mailing list
>> Zelig@lists.gking.harvard.edu<mailto:Zelig@lists.gking.harvard.edu>
>> https://lists.gking.harvard.edu/mailman/listinfo/zelig
>
5:48 p.m.
Also, it may be that the values contained in:
imp$imputations
are invalid. That is, that because the issue is that the "age" variable cannot
be found, there is something wrong with an object stored in the "imputations"
list of the Amelia object.
This would probably have to be checked by manual inspection, or a one-liner such as:
Map(is.data.frame, imp$imputations)
Without seeing the actual object returned by Amelia, it would be hard to tell.
________________________________________
From: zelig-bounces(a)lists.gking.harvard.edu [zelig-bounces(a)lists.gking.harvard.edu] On
Behalf Of Owen, Matthew [mowen(a)iq.harvard.edu]
Sent: Wednesday, February 08, 2012 5:39 PM
To: Olivia Lau; Trey Causey
Cc: zelig(a)lists.gking.harvard.edu
Subject: Re: [zelig] sim(), Amelia, and quantities of interest with ls.mixed
Yes, it seems there is an issue in constructing the design matrix for the
"ls.mixed" model. I'm creating an issue on github here:
https://github.com/zeligdev/ZeligMixed/issues/1
________________________________________
From: monkeykupo(a)gmail.com [monkeykupo(a)gmail.com] On Behalf Of Olivia Lau
[olivia.lau(a)post.harvard.edu]
Sent: Wednesday, February 08, 2012 4:59 PM
To: Trey Causey
Cc: Owen, Matthew; zelig(a)lists.gking.harvard.edu
Subject: Re: [zelig] sim(), Amelia, and quantities of interest with ls.mixed
Note that you lose the aggregated standard errors from doing
summary(zelig-output-from-a-MI-list-object).
On Wed, Feb 8, 2012 at 1:40 PM, Trey Causey
<tcausey@uw.edu<mailto:tcausey@uw.edu>> wrote:
Yes, that works. Thanks very much!
On Wed, Feb 8, 2012 at 1:33 PM, Matt Owen
<mowen@iq.harvard.edu<mailto:mowen@iq.harvard.edu>> wrote:
The issue may be with how Zelig interfaces with
Amelia's mi class. Please attempt the following (modified) zelig command:
model.full <- zelig(
attitude ~ age + married + male + income + tag(1 | state),
model = "ls.mixed",
data = imp$imputations$imp1,
REML = FALSE,
subset = wave4
)
Notice that the difference is in the data parameter:
data = imp$imputations$imp1
If this produces an error, we can narrow down the problem further. Otherwise, the issue
is likely a bug in Zelig's implementation for multilevel models.
- Matt
On Feb 8, 2012, at 4:20 PM, Trey Causey wrote:
-
--
Zelig Mailing List, served by HUIT
Send messages: zelig@lists.gking.harvard.edu<mailto:zelig@lists.gking.harvard.edu>
[un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
Zelig program information: http://gking.harvard.edu/zelig/
Zelig mailing list
Zelig@lists.gking.harvard.edu<mailto:Zelig@lists.gking.harvard.edu>
https://lists.gking.harvard.edu/mailman/listinfo/zelig
-
--
Zelig Mailing List, served by HUIT
Send messages: zelig(a)lists.gking.harvard.edu
[un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
Zelig program information: http://gking.harvard.edu/zelig/
Zelig mailing list
Zelig(a)lists.gking.harvard.edu
https://lists.gking.harvard.edu/mailman/listinfo/zelig
Yes, it's exactly the same.
On Wed, Feb 8, 2012 at 1:18 PM, Matt Owen
<mowen@iq.harvard.edu<mailto:mowen@iq.harvard.edu>> wrote:
> Hi Trey -
> Is the error identical across both versions of Zelig?
>
> On Feb 8, 2012, at 4:10 PM, Trey Causey wrote:
>
>> Hello all:
>>
>> I have replicated this with the current (stable) and alpha releases of
>> Zelig. I cannot simulate quantities of interest using a mixed-effects
>> model estimated on data imputed with Ameila. setx() works fine, but
>> the sim command runs for a while and then reports it is unable to find
>> one of my predictor variables (variable names changed as they are
>> irrelevant). Code below:
>>
>> model.full <- zelig(attitude ~ age + married + male + income + tag(1 |
>> state), model = "ls.mixed", data = imp$imputations, REML = FALSE,
>> subset = wave4)
>>
>> x.out <- setx(model.full)
>>
>> simout <- sim(x.out, x = x.out)
>>
>> The model estimates fine with the zelig() function, setx() works fine
>> in setting the explanatory variables at their means, but when I
>> attempt to use sim(), I get:
>>
>> Error in eval(expr, envir, enclos) : object 'age' not found
>>
>> A traceback produces this:
>>
>> 13: eval(expr, envir, enclos)
>> 12: eval(predvars, data, env)
>> 11: model.frame.default(object, data, xlev = xlev)
>> 10: model.frame(object, data, xlev = xlev)
>> 9: model.matrix.default(parsefml$fixed, data = D)
>> 8: model.matrix(parsefml$fixed, data = D)
>> 7: model.matrix(parsefml$fixed, data = D)
>> 6: is.data.frame(x)
>> 5: colnames(model.matrix(parsefml$fixed, data = D))
>> 4: qi.mer(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 3: qi(object[[1]], simpar = simpar, x = as.matrix(x), x1 = if
>> (!is.null(x1)) as.matrix(x1))
>> 2: sim.setx.MI(model.full, x = x.out)
>> 1: sim(model.full, x = x.out)
>>
>> Any suggestions?
>>
>> Thanks very much,
>> Trey Causey
>> -
>> --
>> Zelig Mailing List, served by HUIT
>> Send messages:
zelig@lists.gking.harvard.edu<mailto:zelig@lists.gking.harvard.edu>
>> [un]subscribe Options: http://lists.gking.harvard.edu/mailman/listinfo/zelig
>> Zelig program information: http://gking.harvard.edu/zelig/
>> Zelig mailing list
>> Zelig@lists.gking.harvard.edu<mailto:Zelig@lists.gking.harvard.edu>
>> https://lists.gking.harvard.edu/mailman/listinfo/zelig
>
4483
days inactive
4483
days old
8 comments
4 participants
participants (4)
-
Matt Owen -
Olivia Lau -
Owen, Matthew
-
Trey Causey