CODEWITHSUNDEEP
rpsych
Oscar Kjell
Oscar Kjell

Reputation: 1651

R psych, cohen.d error Error in `.rowNamesDF<-`(x, value = value) : invalid 'row.names' length

I get the below error when trying to compute effect size using cohen.d function in psych. Please see example data below. 

library(psych)
x <- c(1, 2, 3, 4, 5, 6, 7, 8)
y <- c(1, 1, 1, 1, 2, 2, 2, 2)
xy <- data.frame(x,y)
names(xy) <- c("x", "group")
CohensD <- cohen.d(xy, "group", alpha=.05)

Error in .rowNamesDF<-(x, value = value) : invalid 'row.names' length

Upvotes: 1

Views: 1669

Answers (3)

Panda
Panda

Reputation: 9

You should use the effsize package, instead of psych.

install.packages("effsize")

library(effsize)

Then use the cohen.d of the effsize.

x <- c(1, 2, 3, 4, 5, 6, 7, 8)

y <- c(1, 1, 1, 1, 2, 2, 2, 2)

cohen.d(x,y)

Upvotes: -1

William Revelle
William Revelle

Reputation: 1230

That is indeed a bug that I introduced when adding Mahalabonis distances to cohen.d. The fix is available in psych_1.9.12.21 which is available on my server. You can install it from there:

install.packages("psych",repos="http://personality-project.org/r",type="source")

I will release this patched version to CRAN once it comes back up next week. In the interim, if you don't want the patched version, just use a data set with more than two columns (as pointed out by A. Suliman).

Upvotes: 1

A. Suliman
A. Suliman

Reputation: 13125

Probably there is a bug as it works fine in 1.8.12, also it generates the same error with 

cohen.d(sat.act[1:8, c('education','gender')], "gender")

but not with 

cohen.d(sat.act[1:8, c('education','gender','age')], "gender")

So, should work fine with a data frame has more than two columns 

x <- c(1, 2, 3, 4, 5, 6, 7, 8)
y <- c(1, 1, 1, 1, 2, 2, 2, 2)
xy <- data.frame(x=x, group=y, z=rnorm(8))
cohen.d(xy, "group")

Call: cohen.d(x = xy, group = "group")
Cohen d statistic of difference between two means
  lower effect upper
x  0.53   3.58  6.60
z -0.71   0.88  2.37

Multivariate (Mahalanobis) distance between groups
[1] 3.6
r equivalent of difference between two means
   x    z 
0.87 0.40

Upvotes: 2

Related Questions