Skip to main content

Table 3 Comparison of outlying participants and errors identified by the multivariate and univariate outlier detection approaches in the first iteration of the data evaluation process, regardless of specific method and whether covariate adjustment was applied

From: The utility of multivariate outlier detection techniques for data quality evaluation in large studies: an application within the ONDRI project

 

Neuropsychology

n = 161; p = 53

Gait

n = 148; p = 29

Multi. & Uni.

Multi.

Uni.

Multi. & Uni.

Multi.

Uni.

Outlying Participants

44

44

133

25

43

42

Number of Errors

3a

8

3b

3

5

3b

  1. aAll outlying participants identified by multivariate methods were also identified by univariate methods. However, not all univariate methods identified the participant as an outlier on the variable with the error
  2. bOutlying participants identified by univariate methods only were not verified.