Table 9.
Observed familywise error rate (fwer, %, that matches the pcer for ) and 95% confidence intervals for scenarios xv and xvi, used to investigate effect of different sample sizes, for the three different methods for dealing with nuisance variables.
| N | Without pca |
With pca |
||||
|---|---|---|---|---|---|---|
| Simple residuals | Huh–Jhun | Theil | Simple residuals | Huh–Jhun | Theil | |
| 100 | 96.60 (95.29–97.56) | 4.80 (3.64–6.31) | 5.00 (3.81–6.53) | 59.20 (56.12–62.21) | 5.00 (3.81–6.53) | 5.10 (3.90–6.64) |
| 200 | 42.20 (39.17–45.29) | 5.50 (4.25–7.09) | 5.40 (4.16–6.98) | 22.30 (19.83–24.98) | 5.00 (3.81–6.53) | 4.50 (3.38–5.97) |
| 300 | 25.10 (22.51–27.88) | 5.00 (3.81–6.53) | 5.40 (4.16–6.98) | 12.20 (10.31–14.37) | 5.40 (4.16–6.98) | 5.10 (3.90–6.64) |
| 400 | 18.00 (15.74–20.50) | 4.30 (3.21–5.74) | 5.10 (3.90–6.64) | 11.80 (9.95–13.95) | 4.50 (3.38–5.97) | 5.70 (4.43–7.31) |
| 500 | 11.50 (9.67–13.63) | 6.10 (4.78–7.76) | 4.10 (3.04–5.51) | 9.50 (7.83–11.48) | 6.80 (5.40–8.53) | 5.10 (3.90–6.64) |
| 600 | 10.80 (9.02–12.88) | 5.20 (3.99–6.76) | 5.00 (3.81–6.53) | 9.30 (7.65–11.26) | 4.70 (3.55–6.19) | 4.70 (3.55–6.19) |
| 700 | 11.20 (9.39–13.31) | 4.20 (3.12–5.63) | 4.40 (3.29–5.86) | 7.20 (5.76–8.97) | 5.70 (4.43–7.31) | 5.50 (4.25–7.09) |
| 800 | 10.10 (8.38–12.12) | 5.50 (4.25–7.09) | 4.20 (3.12–5.63) | 7.00 (5.58–8.75) | 4.80 (3.64–6.31) | 5.90 (4.60–7.54) |
| 900 | 8.40 (6.84–10.28) | 5.00 (3.81–6.53) | 4.00 (2.95–5.40) | 7.70 (6.20–9.52) | 5.20 (3.99–6.76) | 5.20 (3.99–6.76) |
| 1000 | 7.70 (6.20–9.52) | 4.30 (3.21–5.74) | 5.90 (4.60–7.54) | 5.00 (3.81–6.53) | 6.20 (4.87–7.87) | 4.70 (3.55–6.19) |