. 2015 Mar;135(3):460–468. doi: 10.1542/peds.2014-2707

TABLE 1.

Model Fit Statistics for Age-Period-Cohort Model of Adolescent Sleep in the United States

Model Parameter	Change in Deviance (Degrees of Freedom)
Model Parameter	≥7 Hours of Sleep/Night	Adequate Sleep
Age	–	–
Age-drift	411.5 (1)^a	219.53 (1)^a
Age-cohort	123.7 (4)^a	14.29 (4)^a
Age-period-cohort	37.2 (4)^a	17.22 (4)^a
Age-period	−54.1 (–4)^a	−5.29 (–4)
Age-drift	−106.7 (–4)^a	−26.23 (–4)^a

^{^a}

Fit statistics are based on an iterative model building strategy, beginning with an age effect and then testing whether the model is significantly improved in fit when drift is added as a parameter, then cohort, then age, period, and cohort. Parameters are then removed from the full age-period-cohort parameterization to determine whether the model fit significantly decreases. The values in the table represent the change in deviance; significant positive parameters indicate that the model has significantly improved, and significant negative parameters indicate that the model has significantly worsened.