Table 2. Fitted Models and Excess Number of Suicides, April 2017 and April to June 2017.
| Time Series | Best-Fitting Time Series Model Before April 2017a | Stationary R2 | Ljung-Box Q Statistic (df) | Estimated Excess No. (SE) of Suicidesb | |||
|---|---|---|---|---|---|---|---|
| April 2017 Only | % Increase (SE), No. | April-June 2017 | % Increase (SE), No. | ||||
| 10-19 y Age group | |||||||
| All | ARIMA(0,1,1)(0,1,1) | 0.583 | 7.85 (16) | 38.16 (14.13)c | 14.63 (5.42)c | 94.41 (28.14)d | 13.30 (3.97)d |
| Male sex | ARIMA(0,1,1)(0,1,1) | 0.595 | 8.01 (16) | 27.90 (13.04)e | 14.23 (6.65)e | 66.03 (25.35)c | 12.44 (4.77)c |
| Female sex | ARIMA(1,1,2)(0,1,1) | 0.661 | 17.75 (15) | 15.98 (6.35)e | 27.08 (10.76)e | 36.96 (12.51)c | 21.74 (7.36)c |
| Shooting with firearm suicide method, all | ARIMA(0,1,1)(0,1,1) | 0.625 | 18.18 (16) | 2.82 (9.04) | 2.39 (7.65) | 6.48 (17.22) | 2.07 (5.49) |
| Hanging suicide method, all | ARIMA(0,1,2)(0,1,1) | 0.545 | 18.68 (15) | 34.72 (9.17)d | 33.62 (8.88)d | 79.83 (17.49)d | 28.86 (5.89)d |
| Male individuals | |||||||
| 20-29 y age group | ARIMA(0,1,1)(0,1,1) | 0.630 | 25.88 (16) | −21.10 (23.60) | −3.96 (4.43) | 49.41 (45.00) | 3.10 (2.82) |
| ≥30+ y age group | ARIMA(0,1,1)(0,1,1) | 0.655 | 17.58 (16) | 9.52 (60.67) | 0.41 (2.59) | 211.35 (118.05) | 2.97 (1.66) |
| Female individuals | |||||||
| 20-29 y age group | ARIMA(0,1,1)(0,1,1) | 0.681 | 15.34 (16) | 0.59 (9.80) | 0.50 (8.35) | 25.29 (17.85) | 7.17 (5.06) |
| ≥30+ y age group | ARIMA(0,1,1)(0,1,1) | 0.615 | 19.07 (16) | −6.66 (27.59) | −0.96 (3.97) | 38.76 (53.28) | 1.84 (2.54) |
Abbreviation: ARIMA, autoregressive integrated moving average.
a
The pre–April 2017 data were checked for innovative and additive outliers, which were integrated into the models when necessary. ARIMA(p,d,q)(P,D,Q) time series model, where P = number of time lags, which affect current values autoregressively; d = number of times past values were subtracted from current values to reduce nonstationarity in the time series; and q = number of current and past random noise terms, which affect current values. P, D, and Q are the respective parameters of a seasonal ARIMA model with a periodicity of 12.
b
Numbers indicate associations of 13 Reasons Why with suicide rates in the respective period.
c
P < .01.
d
P < .001.
e
P < .05.