Long-term media effects and macro-level contexts: How crisis versus routine periods moderate the effects of suicide coverage on suicide rate change