4
$\begingroup$

My experiment was in short: I had less than thirty individuals. One individual had to choose between three colours (green/red/yellow). (Very much like a multiple choice.)

Now, my question is how do I show that significantly more individuals chose red than green or yellow? Can I go with a binary test (one sample, one event) and test for one combination at the time (% that chose green vs % that chose red; % that chose green vs % that chose yellow; % that chose red vs % that chose yellow)?

$\endgroup$

1 Answer 1

1
$\begingroup$

The simplest way would be to estimate the multinomial probability of an individual choosing each color and estimate confidence intervals. An easy way to do this in R would be with the MultinomialCI package.

library(MultinomialCI)
dat <- data.frame(color = c("Green","Red","Blue"), count = c(15, 7, 8))
CI <- multinomialCI(dat$count, alpha = 0.05)
dimnames(CI) <- list(dat$color, c("Lower","Upper"))
print(CI)

If the confidence intervals do not overlap, you have a significant difference at the given $\alpha$. If you have individual-level covariates that may influence the selected group, you could use a more complicated method, like a multinomial logistic regression.

$\endgroup$
1
  • $\begingroup$ How would you report these results in a paper? $\endgroup$
    – Sara B.
    Commented Nov 21, 2015 at 19:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.