Table 3.
Simulation results using regular logistic regression to analyze the data generated from the proposed model (top) and from a model with no effects of Markov chains on the outcome (bottom).
| Covariate in the joint model | Parameter and true value | Estimate | Bias | SDa | SEb | Coverage probability |
|---|---|---|---|---|---|---|
| Logistic regression model fitted to the simulated data from scenario (i) (β ≠ 0, α ≠ 0, α2 ≠ 0, δ1 ≠ 0, δ2 ≠ 0) | ||||||
|
| ||||||
| Intercept | β0 = -1.45 | -1.21 | 0.24 | 0.37 | 0.37 | 0.88 |
| x1 (continuous) | β1 = 0.57 | 0.53 | -0.04 | 0.09 | 0.10 | 0.93 |
| x2 (binary, initial state) | β2 = -0.26 | -0.54 | -0.28 | 0.40 | 0.41 | 0.91 |
|
| ||||||
| Logistic regression model fitted to the simulated data from scenario (ii) (β ≠ 0, α = 0, α2 = 0, δ1 ≠ 0, δ2 ≠ 0) | ||||||
|
| ||||||
| Intercept | β0= -1.45 | -1.51 | -0.06 | 0.37 | 0.38 | 0.96 |
| x1 (continuous) | β1= 0.57 | 0.59 | 0.02 | 0.10 | 0.10 | 0.96 |
| x2 (binary, initial state) | β2 = -0.26 | -0.27 | -0.01 | 0.41 | 0.42 | 0.96 |
a
Standard deviation of the point estimates.
b
Standard error, obtained from the squared root of the average of the estimated variance for each run.