Table 1.
Parameters and options under consideration for computing the tested algorithms.
Fig 1.
T-graphs for Extreme Correct Latencies Treatment (Parameter 1).
Options are coded as follows: 1 (No treatment – No), 2 (fixed value trimming – FT), 3 (fixed value winsorizing—FW), 4 (10% statistical trimming—ST), 5 (10% statistical winsorizing—SW), 6 (10% statistical inverse trimming—IvT). An arrow points from one option to another if the first option outperforms significantly the second. For example, 10% statistical trimming (Node “4. ST”) is outperformed by all other treatments in terms of Validity. Effect sizes are reported in S1 and S2 Tables.
Fig 2.
T-graphs for Error Treatment (Parameter 2).
Options are coded as follows: 1 (Ignore), 2 (Exclude), 3 (Recode with correct M + 2SD – Rec2SD), 4 (Separate – Separ), 5 (Recode 600 with correct M + 600 – Rec600). Effect sizes are reported in S3 and S4 Tables.
Fig 3.
T-graphs for IAT Score Formula (Parameter 3).
Options are coded as follows: 1 (D), 2 (G), 3 (Worse Performance Rule—WPR), 4 (Mini Differences—MD), 5 (10% statistical Trimming on Mini Differences—MDT), 6 (10% statistical Winsorizing on Mini Differences—MDW), and 6 (10% statistical Inverse Trimming on Mini differences—MDIvT). Effect sizes are reported in S5 and S6 Tables.