search

pitakakariki / simr

Power Analysis of Generalised Linear Mixed Models by Simulation
69 stars 19 forks source link

powerSim error: no applicable method for 'simulate' applied to an object of class "c('double', 'numeric')" #151

Closed xiaoransun closed 5 years ago

xiaoransun commented 5 years ago

Hi,

I've been trying to replicate the procedure in the Arend & Schäfer (2019) Psych Methods paper, especially the R codes at Table 3.

However, by exactly following the R codes, the powerSim still cannot work, returning the warning message This appears to be an "observed power" calculation. And further, using lastResult()$errors, I got the error message: no applicable method for 'simulate' applied to an object of class "c('double', 'numeric')"

Strange enough, is that when I tried using simulate and doSim on the simulated lmer model, there was no error message.

Could you tell why that was? Any thoughts would be appreciated!

pitakakariki commented 5 years ago

That paper's begind a paywall and I don't have access to it. Could you please email a copy to the email address on the cran page? https://cran.r-project.org/web/packages/simr/

xiaoransun commented 5 years ago

Thanks for quick reply, Peter! Sorry I didn't realize you wasn't accessible to the paper. Just sent! Even not with the simulation R code, but with my own lmer model, I actually got the same error message as well. I tried to reinstall lme4 on my R but the error message still occurs.

pitakakariki commented 5 years ago

My guess is that instead of:

sim.ef <- powerSim(model, fixed("x", "kr"), alpha=alpha.S, nsim=10)

You've typed:

sim.ef <- powerSim(model, fixed("x", "kr"), alpha.S, nsim=10)
xiaoransun commented 5 years ago

I double checked my code and it was sim.ef <- powerSim(model, fixed("x", "kr"), alpha=alpha.S, nsim=10) (I didn't miss the alpha= part). Do you think it could be due to my R version (3.5.0)?

The error message seems to indicate something wrong with the simulate function. Therefore I tried the source code of this package where this function was called: simulate( object, newparams=simParams, newdata=simData, family=family(object), weights=w)[[1]] And it ran well, actually.

Did I miss any other places where the package (especially the powerSim) function calls simulate?

pitakakariki commented 5 years ago

I ran the code from the paper and it works fine. R version shouldn't matter, 3.5.0 is very recent.

Maybe copy your script into the comments and I'll see if I can spot anything.

xiaoransun commented 5 years ago

I just realized what it is now-- a really silly spot. Problem is I exactly copied the code from the paper: sim.ef <- powerSim(model, fixed("x", "kr"), alpha=alpha.S, nsim=1,000) And there is this evil comma 1,000. 🤷‍♀️ Hope nobody else would get caught up in it.

Thanks so much! I really appreciate your time and help.

pitakakariki commented 5 years ago

Yep, journals are awful at formatting code!