Skip to main content

Table 2 Goodness of fit for the 1- and 2-state Hidden Markov Models for seven patients

From: Major depressive disorder as a nonlinear dynamic system: bimodality in the frequency distribution of depressive symptoms over time

ID Model Θ1 Θ2 Θ2 - Θ1 logl AIC BIC nfree BC
A 1 4.87    −158.89 319.77 322.42 1 .36
  2 4.87 4.87 0.00 −158.89 327.77 340.99 5  
B 1 4.94    −164.52 331.05 333.69 1 .43
  2 4.94 4.94 0.00 −164.52 339.03 352.25 5  
C 1 4.21    −210.47 422.93 425.58 1 .53
  2 3.26 6.81 3.55 −158.53 327.06 340.28 5  
D 1 5.87    −300.57 603.14 605.78 1 .63
  2 3.02 7.73 4.71 −175.20 360.39 373.62 5  
E 1 5.53    −306.42 614.83 617.48 1 .75
  2 0.17 6.65 6.48 −150.89 311.77 325.00 5  
F 1 0.58    −147.10 296.20 298.84 1 .90
  2 0.00 2.20 2.20 −72.26 154.51 167.73 5  
G 1 1.20    −311.46 624.91 627.55 1 .94
  2 0.03 7.18 7.15 −51.94 113.89 127.11 5  
  1. Note. Θ mean number of symptoms per mode, Θ2 Θ1 = distance between the modes, logl log likelihood, AIC Akaike’s Information Criterion, BIC Bayesian Information Criterion, nfree number of freely estimated parameters, BC Bimodality Coefficient