CODEWITHSUNDEEP
remmeans
heatherr
heatherr

Reputation: 143

Trying to look at interaction plot in emmeans - Error caused by missing argument in emmip?

I'm following the vignette for examining interactions using emmeans here https://cran.r-project.org/web/packages/emmeans/vignettes/interactions.html

But I'm using my own dataset (all.det) - which does have significant interactions. all.det contains 1621 observations of 12 variables, 3 of which are factors. I did try to create a re-producible example, but haven't figured out how to create one with all interactions significant. So I'm hoping this description will be enough for someone to point out where I am going wrong.

I am using

library(data.table)
library(car)
library(emmeans)

I run an anova using this code

DistanceKm is a numeric and Method, IDGroup and Sightability are factors.

model = lm(DistanceKm ~ Method * IDGroup * Sightability,
           data=all.det[(IDGroup == "Whale" | IDGroup == "Dolphin")
                        & DistanceKm <=5])
Anova(model, type="II")

All the interactions are significant, so I follow this up with an interaction plot as advised in the vignette

emmip(model, Method ~ Sightability | IDGroup)

But I get these errors

Error in if (!all(chk == tbl)) stop("Data appear to be randomized -- ",  : 
  missing value where TRUE/FALSE needed
Error in ref_grid(object, ...) : 
  Perhaps a 'data' or 'params' argument is needed

I did try to create a subset of my data

sub=all.det[(IDGroup == "Whale" | IDGroup == "Dolphin") & DistanceKm <=5]

And then re-ran the model, followed with an Anova

model=lm(DistanceKm ~ Method * IDGroup * Sightability, data=sub)
Anova(model, type="II")

I get the same Anova results (as expected), but now I'm also able to produce the interaction plot without errors.

emmip(model, Method ~ Sightability | IDGroup)

I just get a single warning instead about a missing value

Am I missing something when calling emmip?

Upvotes: 1

Views: 1816

Answers (1)

Andrew Almonte
Andrew Almonte

Reputation: 53

I had a similar warning, then found I had mislabled one of my factors in the dataframe. Check your dataframe and make sure there aren't any NA values where there shouldn't be.

Upvotes: 2

Related Questions