Table. Values of the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) Calculated From Models for Interrupted Time Series Analysisa.
| Model | Parameters, No. | AIC | BIC |
|---|---|---|---|
| Change in level | 3 | 1240.3 | 1249.0 |
| Change in level plus seasonality adjustment (indicator) | 14 | 1164.8 | 1205.2 |
| Change in level plus seasonality adjustment (sine and cosine pairs) | 7 | 1191.5 | 1211.7 |
| Change in level and trend | 4 | 1237.5 | 1249.0 |
| Change in level and trend plus seasonality adjustment (indicator) | 15 | 1161.8 | 1205.0 |
| Change in level and trend plus seasonality adjustment (sine and cosine pairs) | 8 | 1188.1 | 1211.1 |
| Temporal level change (gap) | 3 | 1210.9 | 1219.6 |
| Temporal level change (gap) plus seasonality adjustment (indicator) | 14 | 1135.5 | 1175.9 |
| Temporal level change (gap) plus seasonality adjustment (sine and cosine pairs) | 7 | 1161.1 | 1181.3 |
| Gap plus change in trend at 2 y after the disaster | 4 | 1212.5 | 1224.0 |
| Gap plus change in trend at 2 y after the disaster plus seasonality adjustment (indicator) | 15 | 1137.0 | 1180.3 |
| Gap plus change in trend at 2 y after the disaster plus seasonality adjustment (sine and cosine pairs) | 8 | 1162.6 | 1185.7 |
| Gap plus changes in trends at the disaster and 2 y after the disaster | 5 | 1214.5 | 1228.9 |
| Gap plus changes in trends at the disaster and 2 y after the disaster plus seasonality adjustment (indicator) | 16 | 1138.9 | 1185.0 |
| Gap plus changes in trends at the disaster and 2 y after the disaster plus seasonality adjustment (sine and cosine pairs) | 9 | 1164.5 | 1190.4 |
Poisson regression models with adjustment of scale parameter were used to estimate birth rate. Change in level modeled intercept change after March 2011. Change in level and trend modeled intercept and slope changes after March 2011. Temporal level change (ie, gap) modeled intercept change for the 2 years after March 2011. Gap plus change in trend modeled gap plus slope change 2 years after March 2011. Gap plus changes in trends modeled gap plus slope changes after March 2011 and 2 years after March 2011. Seasonality adjustment was applied by including indicator variables of the calendar month (“indicator”) or sine and cosine pairs (sin [2π × t/12], cos [2π × t/12], sin [4π × t/12], and cos [4π × t/12], where t indicates secular 12 months). None of the changes in the trend component added to the gap component were statistically significant. Taken together with this table, models including temporal level changes with seasonality adjustment are optimal and parsimonious.