add '-2*log(L)' to anova tidy renamer list

[bbolker]

We recently changed the name "deviance" to "-2*log(L)" in anova output in lme4 ("deviance" is technically incorrect, and confusing, for the value we were actually reporting). This does not affect broom directly, but it affects the cardx package, which calls broom::tidy() on anova() output from lme4. Because "-2*log(L)" isn't in the list of renamers that tidy.anova uses, this now throws a warning (and makes cardx fail its checks).

It's a little hard to implement a test for this change since lme4 is handled by broom.mixed and not broom ... you could do something hacky like this, but I'll leave it up to you to decide if you want to do that.

m <- lm(mpg ~ cyl + hp, mtcars)
aa <- anova(m)
names(aa)[5] <- "-2*log(L)"
expect_no_warning(broom::tidy(aa))

I know pushing out a new release is a hassle, but at least it's been 6 months since the last release and there are some other improvements to include in this release ... do you think you might be able to do this soon ... ?

[alexpghayes]

If we make a change here I'd advocate for a syntactic name that doesn't require backticks to interface with dplyr, etc

…

On Tue, Mar 25, 2025, 9:46 AM Ben Bolker ***@***.***> wrote: We recently changed the name "deviance" to "-2*log(L)" in anova output in lme4 ("deviance" is technically incorrect, and confusing, for the value we were actually reporting). This does not affect broom directly, but it affects the cardx package <https://github.com/insightsengineering/cardx>, which calls broom::tidy() on anova() output from lme4. Because "-2*log(L)" isn't in the list of renamers <https://github.com/tidymodels/broom/blob/main/R/stats-anova.R#L55-L88> that tidy.anova uses, this now throws a warning (and makes cardx fail its checks). It's a little hard to implement a test for this change since lme4 is handled by broom.mixed and not broom ... you could do something hacky like this, but I'll leave it up to you to decide if you want to do that. m <- lm(mpg ~ cyl + hp, mtcars)aa <- anova(m) names(aa)[5] <- "-2*log(L)" expect_no_warning(broom::tidy(aa)) I know pushing out a new release is a hassle, but at least it's been 6 months since the last release and there are some other improvements to include in this release ... do you think you might be able to do this soon ... ? ------------------------------ You can view, comment on, or merge this pull request online at: #1239 Commit Summary - 6c97bc8 <6c97bc8> add '-2*log(L)' to anova tidy renamer list File Changes (1 file <https://github.com/tidymodels/broom/pull/1239/files>) - *M* R/stats-anova.R <https://github.com/tidymodels/broom/pull/1239/files#diff-7f3b99a01cc8f29596522c10c72622e8dc56c6051800d342ca48d2a1ecddd772> (4) Patch Links: - https://github.com/tidymodels/broom/pull/1239.patch - https://github.com/tidymodels/broom/pull/1239.diff — Reply to this email directly, view it on GitHub <#1239>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADTBG23GAKTFEGC4MCEHZID2WFT3NAVCNFSM6AAAAABZX2OFO6VHI2DSMVQWIX3LMV43ASLTON2WKOZSHE2DMNZZGQYDINQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[bbolker]

FWIW what my PR does is to rename "-2*log(L)" and "X.2.log.L." (the name-mangled version) to "minus2logL", which is clunky but is indeed syntactically safe ...

(Our rationale for choosing "-2*log(L)" as the label even though it's syntactically unsafe is that it's hard to think of a short, syntactically safe alternative ...)

[simonpcouch]

Thanks for the PR! Just saw your email as well--happy to get this on CRAN in the coming days.

[github-actions]

This pull request has been automatically locked. If you believe the issue addressed here persists, please file a new PR (with a reprex: https://reprex.tidyverse.org) and link to this one.