Table 1.
Results of previous studies on the associations between Google Trends search volumes and suicide statistics.
Table 2.
Final search terms and lengths (in months) and missing values (in weeks) of the respective Google Trends time series.
Table 3.
Reliability of individual and averaged time series for selected Google Trends search terms.
Table 4.
Fitted models and outliers in the individual and averaged Google Trends search volume time series per country.
Fig 1.
Heat map of cross-correlation coefficients for the US data (pro-suicide terms A).
Numbers in column headers refer to lags in months. Idealized patterns reflect what could be expected if search volumes predicted suicide rates. In the observed patterns, frames highlight statistically significant (p < .05) coefficients.
Fig 2.
Heat map of cross-correlation coefficients for the US data (pro-suicide terms B).
Numbers in column headers refer to lags in months. Idealized patterns reflect what could be expected if search volumes predicted suicide rates. In the observed patterns, frames highlight statistically significant (p < .05) coefficients.
Fig 3.
Heat map of cross-correlation coefficients for the US data (suicide prevention terms).
Numbers in column headers refer to lags in months. Idealized patterns reflect what could be expected if search volumes predicted suicide rates. In the observed patterns, frames highlight statistically significant (p < .05) coefficients.
Fig 4.
Heat map of cross-correlation coefficients for the German data.
Suizid, Selbstmord, Freitod = ‘suicide’ in English; Depressionen = ‘depression’ (plural in German); Selbstmord Forum = ‘suicide chat’. Numbers in column headers refer to lags in months. Idealized patterns reflect what could be expected if search volumes predicted suicide rates. In the observed patterns, frames highlight statistically significant (p < .05) coefficients.
Fig 5.
Heat map of cross-correlation coefficients for the Austrian data.
Suizid, Selbstmord = ‘suicide’ in English; Depressionen = ‘depression’ (plural in German). Numbers in column headers refer to lags in months. Idealized patterns reflect what could be expected if search volumes predicted suicide rates. In the observed patterns, frames highlight statistically significant (p < .05) coefficients.
Fig 6.
Heat map of cross-correlation coefficients in the Swiss data.
Selbstmord = ‘suicide’ in English; Depressionen = ‘depression’ (plural in German). Numbers in column headers refer to lags in months. Idealized patterns reflect what could be expected if search volumes predicted suicide rates. In the observed patterns, frames highlight statistically significant (p < .05) coefficients.
Table 5.
Significant cross-correlation coefficients of individual and averaged time-series data.