"Non standard" output from Rfit (npsm package) in R version 3.6.3

General
1

Hallo, I am using the Rfit function from the npsm package in R. I have moved from version 3.5.2 to version 3.6.3 . With version 3.6.3 , for some variables, I get a "non standard" output from univariate regression.

For instance, in the example below STPIT (an anxiety score) is the DV and ETA (age) is the IV.

model1 <-rfit (STPIT~ETA, data=dataset)
summary(model1)
Call:
rfit.default(formula = STPIT ~ ETA, data = dataset)

Coefficients:
Estimate Std. Error t.value p.value
[1,] 18.000000 5.375171 3.3487 0.0009165 ***
[2,] 0.000000 0.067963 0.0000 1.0000000

Signif. codes: 0 ‘’ 0.001 ‘’ 0.01 ‘’ 0.05 ‘.’ 0.1 ‘ ’ 1

Multiple R-squared (Robust): 0
Reduction in Dispersion Test: 0 p-value: 1

If I run the same analysis in R version 3.5.2 I get standardly-labeled output with the variable and the intercept.

Call:
rfit.default(formula = STPIT ~ ETA, data = dataset)

Coefficients:
Estimate Std. Error t.value p.value
(Intercept) 1.8000e+01 5.3752e+00 3.3487 0.0009165 ***
ETA -9.7124e-15 6.7963e-02 0.0000 1.0000000

Signif. codes: 0 ‘’ 0.001 ‘’ 0.01 ‘’ 0.05 ‘.’ 0.1 ‘ ’ 1
Overall Wald Test: 0 p-value: 1

Can someone help?

Paola

2

Hi @paola,
In R-4.0.0 I get the "standard" output you refer to using this example code:

library(npsm)
help(Rfit)

data(baseball)
str(baseball)
model1 <- rfit(weight~height, data=baseball)
summary(model1)

Maybe your packages need to be updated for your newer version of R?

HTH

3

Dear @DavoWW, thank you so much for your feedback!
I checked that the package was updated and it was. However, in doing some further checking I realized that the weird output comes from the 64 bit version of R 3.6.3 (I checked this too: my computer is 64 bit). For some unknown reason, if I use the 32 bit version, the output is "standard"...
Paola

4

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.