emmeans - estimates for levels of variable-CodePudding

With this model:

library(emmeans)
mod <- lm(conc~source   percent, data=pigs)

emmeans(mod, "percent")
 percent emmean   SE df lower.CL upper.CL
    12.9   37.6 0.93 25     35.7     39.6

I'm using the emmeans package, and it gives me the estimate for the average of percent. But what if I want to predict the percent values 8, 10 and 13? How can I specify that in the emmeans call?

CodePudding user response：

    newrg <- ref_grid(mod, at = list(source = c("fish", "soy", "skim"), percent = c(8, 10, 15)))
    predict(newrg, interval = "prediction", by = "percent")

    plot(newrg, PIs = TRUE)