Comments (5)
alexanderrobitzsch
commented on June 17, 2024
I guess that it directly pertains `multiwayvcov::cluster.vcov()` because
`miceadds::glm.cluster()` is just a wrapper around `stats::glm()`and
`multiwayvcov::cluster.vcov()`. Nevertheless, I can catch the case of
coefficients with `NA` and produce a warning message. See the source code
https://github.com/alexanderrobitzsch/miceadds/blob/master/R/glm.cluster.R
from miceadds.
tra6sdc
commented on June 17, 2024
Hi, do I need to post this bug to the multiwayvcov R package? I also note that I get the error even when there are no NA parameter estimates.
> library(miceadds)
> df<-structure(list(Zone = structure(c(1L, 2L, 1L, 1L, 1L, 2L, 2L, +
3L, 3L), .Label = c("A", "B", "C"), class = "factor"), Y = c(2L, +
3L, 5L, 4L, 3L, 5L, 7L, 8L, 10L), Cat1 = structure(c(1L, 1L, +
1L, 1L, 1L, 2L, 2L, 2L, 2L), .Label = c("Fact1", "Fact2"), class = "factor"), +
Cat2 = structure(c(1L, 1L, 2L, 2L, 2L, 2L, 3L, 3L, 3L), .Label = c("Level1", +
"Level2", "Level3"), class = "factor")), class = "data.frame", row.names = c(NA,-9L))
> Fitglm<-glm(Y~Cat1+Cat2,df,family="gaussian")
summary(Fitglm)
Call: glm(formula = Y ~ Cat1 + Cat2, family = "gaussian", data = df)
Deviance Residuals: 1 2 3 4 5 6 7 8 9
-0.5000 0.5000 1.0000 0.0000 -1.0000 0.0000 -1.3333 -0.3333 1.6667
Coefficients:
Estimate Std. Error t value Pr(>\|t\|)
(Intercept) 2.5000 0.8466 2.953 0.0318 *
Cat1Fact2 1.0000 1.3824 0.723 0.5019
Cat2Level2 1.5000 1.0929 1.372 0.2283
Cat2Level3 4.8333 1.7623 2.743 0.0407 *
--- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Dispersion parameter for gaussian family taken to be 1.433333)
Null deviance: 55.5556 on 8 degrees of freedom
Residual deviance: 7.1667 on 5 degrees of freedom AIC: 33.491
Number of Fisher Scoring iteraations 2
> Fitglm.cluster<-glm.cluster(df,Y~Cat1+Cat2,cluster="Zone",family="gaussian")
> summary(Fitglm.cluster)
Error in `[<-`(`*tmp*`, , "z value", value = csmod[, "Estimate"]/csmod[, :
subscript out of bounds
> length(Fitglm.cluster$glm_res$coefficients) [1] 4
> Fitglm.cluster$glm_res$coefficients
(Intercept) Cat1Fact2 Cat2Level2 Cat2Level3
2.500000 1.000000 1.500000 4.833333
> length(Fitglm.cluster$vcov) [1] 16
> Fitglm.cluster$vcov
(Intercept) Cat1Fact2 Cat2Level2 Cat2Level3
(Intercept) 3.000000e-01 1.973730e-16 -3.000000e-01 -5.666667e-01
Cat1Fact2 7.012097e-31 7.012097e-31 -1.402419e-30 -4.715021e-16
Cat2Level2 -3.000000e-01 -1.973730e-16 3.000000e-01 5.666667e-01
Cat2Level3 -5.666667e-01 -7.894919e-16 5.666667e-01 1.781481e+00
from miceadds.
alexanderrobitzsch
commented on June 17, 2024
I replicated your bug. You were indeed right. The summary.glm.cluster produced a bug. I fixed the bug in miceadds 3.0-3.
from miceadds.
tra6sdc
commented on June 17, 2024
Thanks Alexander. I emailed the author of multiwayvcov who sent this reply. I expect you are aware of the diagnosis of the issue since you report it as being fixed? Thanks again.
“Stephen,
It looks like whatever's going wrong is happening in summary.glm.cluster(); multiwayvcov is producing a variance-covariance matrix as desired. I narrowed down the error to the assignment on line 35 in glm.cluster.R in miceadds; the division operation is fine if you pull it out and walk through the function line-by-line:
csmod[, "Estimate"]/csmod[, "Std. Error"]
(Intercept) Cat1Fact2 Cat2Level2 Cat2Level3
4.564355e+00 1.194197e+15 2.738613e+00 3.621230e+00
So it looks like the "subscript out of bounds" that's incorrect is "z value", which isn't a valid subscript in the matrix csmod (the generated summary matrix uses t instead of z). Everything works just fine if I change the family of the glm link function to something that uses z scores instead, such as poisson (I mean, the model is trash, but the code works). It looks like this version of summary.glm.cluster has never been tested against gaussian glm models. I don't know anything about miceadds, so I don't know if this is a big deal or not, but it isn't a problem with multiwayvcov.”
Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10
…________________________________
From: Alexander Robitzsch <[email protected]>
Sent: Wednesday, October 24, 2018 1:47:45 PM
To: alexanderrobitzsch/miceadds
Cc: Stephen Clark; Author
Subject: Re: [alexanderrobitzsch/miceadds] subscript out of bounds (#14)
I replicated your bug. You were indeed right. The summary.glm.cluster produced a bug. I fixed the bug in miceadds 3.0-3.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub<#14 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AgqC4czCq3aTtC6AVRISyiZMCs3E02Vnks5uoGFxgaJpZM4XxbUn>.
from miceadds.
alexanderrobitzsch
commented on June 17, 2024
Yes, this was the reason. I did not expect a change of labeling in summary.glm.
from miceadds.
Related Issues (20)
- micombine.cor with Fisher transform? HOT 2
- mi.anova intercept? HOT 3
- groupwise imputation with ml.lmer? HOT 3
- summary() subscript out of bounds HOT 2
- mice.impute.bygroup - outs of bounds HOT 9
- glm.cluster yielding wrong p-values HOT 3
- not able to impute using ml.lmer in v 3.3? HOT 3
- glm.cluster reliant on/generates NULL wgt__ value HOT 3
- Error when using 2lonly.function with polyreg HOT 2
- datlist2mids fails after subsetting if factors don't still contain all levels HOT 1
- pooling error with nested imputations HOT 1
- Memory limit error with weighted imputation HOT 1
- Error in weighted predictive mean matching method HOT 1
- different variable imputed values using same predictor variable - 2l.pmm HOT 3
- `mice.impute.ml.lmer` on large three-level dataset: `"binary"` logistic model returns error, 'hangs' when adding random slopes or interactions HOT 1
- Is it possible to convert mids objects to an Amelia object? HOT 2
- post hoc test for mi.anova
- Question on using mice.impute.imputeR.lmFun HOT 2
- Possible error in example for ml.lmer documentation HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from miceadds.