5. Uniformity Tests • rankingQ

In this vignette, we show how to detect the presence of random responses using the uniformity test, which shows that given respondent-level item order randomization, recorded responses will follow a uniform distribution in the absence of random responses.

library(rankingQ)
#> Registered S3 method overwritten by 'GGally':
#>   method from   
#>   +.gg   ggplot2

Recorded Responses

In an earlier vignette, we have shown how to perform the bias correction on ranking questions using the anchor question. But how do we know if there is a need for bias correction in the first place?

Distribution of rankings of recorded responses can show whether there are indeed random responses. First, what are recorded responses?

Uniformity Test

In this example, given there are 4! = 24 possible ways to rank the items, the proportion of recorded responses for each possible ranking should converge to 1/24 = 0.0417, in the absence of random responses.

Validity of the Anchor Question

Checking for uniformity also validates the usage of the anchor question.