interval can't be modifed

[oldnobody-marco]

Jacob, thanks for this great package.

Using effect_plot from the jtools-package, I am not able to modify the interval. If I change "int.type" to "prediction" it still plots the confidence, if I change "int.width" to "0.99" it still plots 0.95.

With your example:
data(mpg)
fit <- lm(cty ~ displ + year + cyl + class + fl, data = mpg[mpg$fl != "c",])
effect_plot(fit, pred = displ, interval = TRUE, int.type=c("confidence"), int.width=0.99,
plot.points = TRUE)

Marco

[oldnobody-marco]

Jacob, thanks for letting me know. re panelr: your cross-level interaction estimator (...| pvar*tvar) does not reproduce the standard FE interaction estimator - which shall measure the within-effect of tvar across between-unit categories of pvar. Problem: your formulation controlls heterogeneity in tvar only for the base-category of pvar. An easy fix would be (a) adding unit-specific means ("contextual effects") of the interaction pvar*tvar to the model, or (b) absorbing group-specific heterogeneity through adding the interaction of person-mean tvar and pvar to the model (pvar*m_tvar). and by the way: thanks for refering to our paper. if you ever find the time during an update, you could direct readers to the accepted version in the reference list: https://doi.org/10.1177/0049124120914934. (I dont mind at all if people use & cite the working paper, but in this case, it contains an error in the context of standard errors.) Marco

…

________________________ Dr. Marco Giesselmann University Zurich Institute of Sociology 8050 Switzerland +41 44 63 52373 Von: "Jacob Long" <notifications@github.com> An: "jacob-long/jtools" <jtools@noreply.github.com> Kopie: "oldnobody-marco" <marco.giesselmann@uzh.ch>, "Author" <author@noreply.github.com> Datum: 22.06.2020 17:38 Betreff: Re: [jacob-long/jtools] interval can't be modifed (#82) Closed #82 via 3ad2c4e. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.