Print testEstimates to more than 3 decimal places

[louise-mewton]

Hi there - I'm wanting to apply FDR correction to the p-values extracted from mitml. In order to do this, I need greater precision in the p-values printed from testEstimates. I've tried a few workarounds (with my limited programming knowledge) but none have worked for me. I wonder if this is possible? Any help here would be really appreciated.

Thanks,
Louise

[louise-mewton]

This is an example of what I have tried;

library(tryCatchLog)
sink('model_results.txt')
for(i in c(1:68)){
resp <- paste('CV',i,sep='')
for(j in c(15:18)){
V <- paste('V',j,sep='')
f <- as.formula(paste(resp,'',V,'+ sex + race + (1|family) + (1|Device)'))
print(f)
m <- with(impList,tryCatch({lmer(as.formula(paste(resp,'',V,'+ sex + race + (1|family) + (1|Device)')), control = lmerControl(optimizer ="bobyqa"))}, warning=function(w){cat("Error fitting model for:",resp,'--',V,end='\n')}))
print(testEstimates(m), digits=10)
est <- testEstimates(m)
print(confint(est), digits=10)
}
}

print(confint(est), digits=10) - by changing the number of digits here, I can get different decimal places in the confidence intervals

print(testEstimates(m), digits=10) - no matter what I enter for the digits here, the number of decimal places remains 3.

Thank you!

[simongrund1]

The print method doesn't understand digits, but you don't need that because the results are always saved with full precision in the objects produced by testEstimates(). To get all the decimal places or round them to a certain number of digits, you can look at the results in the object and round those:

library(mitml)
data(studentratings)

fml <- ReadDis + ReadAchiev ~ 1
imp <- jomoImpute(formula = fml, data = studentratings, n.burn = 500, n.iter = 100, m = 5)

implist <- mitmlComplete(imp)
fit <- with(implist, lm(ReadDis ~ ReadAchiev))
res <- testEstimates(fit)

res
# Call:
# 
# testEstimates(model = fit)
# 
# Final parameter estimates and inferences obtained from 5 imputed data sets.
# 
#              Estimate Std.Error   t.value        df   P(>|t|)       RIV       FMI 
# (Intercept)     3.678     0.160    22.955    34.610     0.000     0.515     0.375 
# ReadAchiev     -0.002     0.000    -7.106    47.290     0.000     0.410     0.319 
# 
# Unadjusted hypothesis test as appropriate in larger samples. 

round(res$estimates, 5)
#             Estimate Std.Error  t.value       df P(>|t|)     RIV     FMI
# (Intercept)  3.67821   0.16024 22.95491 34.61005       0 0.51506 0.37506
# ReadAchiev  -0.00220   0.00031 -7.10613 47.29015       0 0.41011 0.31904

Thanks for the good suggestion about the print method, though. I think it's a good idea to make them respect options("digits") in the future.

[louise-mewton]

Perfect. Thank you so much for your help with this. And thank you for an excellent package!

Louise