Resolving cumulative family-related error

Interfaces brms
1

Here is my code and the error it generates(all variables have ordinal data measured on a 1-5 scale):

fit_WTA_CFC2 <-
  brm(
    WTA_MN ~ BT_MN + DT_MN + AT_MN + CFCpercpt_MN + SocInfl_MN + Complx_MN + Compt_MN,
    data = AttdnlData,
    family = cumulative(link = "probit", threshold = "flexible")
  )

Error: Family 'cumulative' requires either positive integers or ordered factors as responses.

I cannot figure out what this error means and how to overcome it.
Thanks.
Aruho

2

It looks like it’s saying that your outcome (response) variable is of the wrong type to use with family = cumulative. Based just on your code I don’t know what type of variable WTA_MN is, but according to the error message it needs to be coded as positive integers or alternatively as an ordered factor variable (apparently not just a regular factor but an ordered factor according to the error message).

3

Make sure that the 1 to 5 are literally numeric, and that there are no 0s. Also e.g., if there are NAs, that they are not in there like factors/characters such as “NA”. If you changed the 1 to 5 numbers to be factors, ensure that in the factor() function you also literally add ‘ordered = TRUE’, simply providing an order is not sufficient (which is a bit annoying!)

1 Like
4

Thank you JimBob.
Issue resolved - my data was reading as characters! I don’t know how that came to be, because other datasets imported in the same way were reading numeric.
I converted them to numeric, and that worked.
Thank you.

Aruho

2 Likes