0

Apologies if I'm making any silly errors, I'm pretty new to R. I have been searching for the answer to my question, but haven't got very far!

I need to report standardized coefficients for the predictors in my model. I think I have figured out how to do this for my model of response time, by scaling all variables (type and speed are categorical variables with two levels coded with 1 and -1).

e.g. 

RT.lmer = lmer(scale(RT) ~ scale(type) + scale(speed) + (1|subject) + (1|word), RTdata)

and then asking for the coefficients:

fixef(RT.lmer)

However, I can't figure out how to do this for my accuracy model. This is the regular model:

ACC.lmer = lmer(Acc ~ type + speed + (1|subject) + (1|word), fulldata, family=binomial)

It won't allow me to use scale on the accuracy scores. If I just scale the predictors, this does not give me the correct coefficients because I am getting values greater than 1.

Thanks in advance for your help!

Laura

alexforrence
  • 2,724
  • 2
  • 27
  • 30
LauraSpeed
  • 1
  • 1
    I'm going to suggest that this is a statistical, not a programming, question. It's not immediately obvious (to me at least) how to extend the concept of standardized coefficients to GLMs. (See e.g. [this mailing list post](https://stat.ethz.ch/pipermail/r-sig-mixed-models/2010q4/004974.html).)Voting to migrate to [CrossValidated](http://stats.stackexchange.com). – Ben Bolker Jan 21 '15 at 12:41
  • If you already have your `type` and `speed` as coded values, there's no reason to apply `scale()` to them. – alexwhitworth Mar 06 '15 at 03:52

0 Answers0