add a "glance" method

[acoppock]

Someone made a new regression output package built on the soon-to-be-released gt package. Estimatr isn't compatbile yet because we need a "glance" method:

library(gtsummary)
library(estimatr)

fit_1 <- lm_robust(mpg ~ hp, data = mtcars)
fit_2 <- lm_robust(mpg ~ hp, se_type = "classical", data = mtcars)

gtsummary(list(fit_1, fit_2))
Error: No glance method for objects of class lm_robust

[acoppock]

https://twitter.com/andrewheiss/status/1087408961861480448

[graemeblair]

see vincentarelbundock/modelsummary#8

[lukesonnet]

Will implement ASAP (hopefully today!) or should I wait for vincent

[graemeblair]

yes please implement! thank you!

[nfultz]

Extend via generics, same as tidy - https://github.com/r-lib/generics/blob/master/R/glance.R

[lukesonnet]

Yep, thanks!

[lukesonnet]

Added glance for lm_robust and a Rbuildignored test for gt and gtsummary. Should I add iv_robust, and just present the second stage?

[graemeblair]

I don't think it should be about first or second but what are the right set of statistics about the IV model. Over-identification test, test of weak instruments, ...? statistic and p-values?

[lukesonnet]

Ok we don't return those but I'm happy to inplement them. That said we still need to decide what goes into the main body. That's what I was asking about. Luke Sonnet PhD Candidate, Political Science UCLA

…

On Wed, Jan 23, 2019, 17:15 Graeme Blair ***@***.*** wrote: I don't think it should be about first or second but what are the right set of statistics about the IV model. Over-identification test, test of weak instruments, ...? statistic and p-values? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#282 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFDZpwFuRKLkE1qibnAFQYsguohD2Am1ks5vGQkcgaJpZM4aLz1W> .

[graemeblair]

Sorry misunderstood, what's the main body? I thought it was one row with a set of model fit statistics.

[lukesonnet]

Ah yes. That's glance. I was thinking about gtsummary and the estimates and coefficients.

[lukesonnet]

Next step is to:

• mimic glance.ivreg and include standard IV metrics
• ensure that gtsummary takes the 2sls estimates in the main body
• I'll also try to do difference in means where it has basic metrics (including design type) in the glance

I'll try to do that today and then I'll open the PR.

[vincentarelbundock]

Funny. I came here to see if a glance had been requested before only to find that my specific issue was being discussed and worked on. Thanks, I appreciate it!

Let me know if there's anything I can do to help.

[lukesonnet]

Done in #285