Table 3.
Assumed Prior | Model Type |
||
---|---|---|---|
Unconstrained | Exact Constraints | Bayesian Constraints | |
Main Prior | |||
(1) Mean squared bias | 0.0042 | 0.0070 | 0.0000 |
(2) Sampling variance | 0.0587 | 0.0219 | 0.0220 |
(3) Constraint variance | 0.0025 | 0.0025 | 0.0025 |
MSD | 0.0654 | 0.0315 | 0.0245 |
High-Bias Prior | |||
(1) Mean squared bias | 0.0017 | 0.0361 | 0.0000 |
(2) Sampling variance | 0.0587 | 0.0219 | 0.0221 |
(3) Constraint variance | 0.0114 | 0.0114 | 0.0114 |
MSD | 0.0719 | 0.0694 | 0.0335 |
High-Bias, High-Variance Prior | |||
(1) Mean squared bias | 0.0029 | 0.0408 | 0.0000 |
(2) Sampling variance | 0.0587 | 0.0219 | 0.0220 |
(3) Constraint variance | 0.0479 | 0.0479 | 0.0479 |
MSD | 0.1094 | 0.1106 | 0.0699 |
Mean squared deviation (MSD) = sum of rows (1), (2), and (3). Row (1) = point estimate of β0 under Bayesian prior. Row (2) = asymptotic variance for the each regression estimate of β0, given by [SE(β0)]2. Row (3) = variance in β0 due to uncertainty about the true population constraint value.