Skip to main content

Table 4 Comparison of multiple regression, bootstrap case and model based resampling SE and CI estimates from the OA Knee data

From: The use of bootstrap methods for analysing health-related quality of life outcomes (particularly the SF-36)

Dependent   GROUP coefficient 95% CI Interval
Variable Model N SE /SE p Lower Upper Length Shape
Physical Function OLS 165 13.3 3.07 4.31 0.001 7.19 19.32 12.14 1.00
  Case    3.02 4.39   7.64 19.69 12.05 1.15
  Model    3.05 4.35   7.49 19.49 12.00 1.08
Role Physical OLS 177 -0.5 4.89 -0.11 0.915 -10.16 9.12 19.29 1.00
  Case    4.39 -0.12   -8.60 8.51 17.11 1.12
  Model    4.86 -0.11   -10.11 8.93 19.04 0.99
Bodily Pain OLS 200 14.7 3.39 4.34 0.000 8.01 21.38 13.37 1.00
  Case    3.41 4.30   7.81 21.41 13.60 0.98
  Model    3.36 4.38   8.07 21.25 13.18 0.99
General Health OLS 173 4.7 2.01 2.32 0.021 0.71 8.65 7.95 1.00
  Case    2.03 7.26   0.69 8.69 8.00 1.01
  Model    1.98 2.37   0.70 8.48 7.78 0.95
Energy OLS 185 6.5 2.46 2.64 0.009 1.65 11.36 9.72 1.00
  Case    2.50 2.60   1.75 11.63 9.88 1.08
  Model    2.46 2.65   1.49 11.04 9.54 0.90
Social Function OLS 194 9.1 3.70 2.46 0.015 1.82 16.41 14.59 1.00
  Case    3.52 2.59   2.14 16.06 13.92 1.00
  Model    3.65 2.50   1.72 16.09 14.37 0.94
Role Emotional OLS 184 9.4 6.10 1.55 0.124 -2.60 21.48 24.08 1.00
  Case    5.89 1.60   -1.90 20.85 22.75 1.01
  Model    6.03 1.57   -2.37 20.90 23.27 0.97
Mental Health OLS 191 1.1 2.15 0.51 0.613 -3.15 5.33 8.48 1.00
  Case    2.32 0.47   -3.54 5.42 8.96 0.94
  Model    2.14 0.51   -3.17 5.12 8.28 0.95
  1. Bootstrap BCa confidence intervals based 5000 bootstrap replications.