Table 1 Regression output from interrupted time series model investigating the effect of time, period, and Hillary attention on the relative frequency of misogynistic language among tweets about Hillary Clinton.
Effect | Estimate | SE | t | p |
|---|---|---|---|---|
Time | 1.392 | 0.421 | 3.306 | 0.001** |
Period (campaign vs. before announcing) | − 1478.165 | 250.875 | − 5.892 | < 0.001*** |
Period (campaign vs. after election) | 1223.325 | 298.974 | 4.092 | < 0.001*** |
Time x period interaction (campaign vs. before announcing) | − 4.379 | 0.627 | − 6.985 | < 0.001*** |
Time x period interaction (campaign vs. after election) | − 2.015 | 0.700 | − 2.878 | 0.004** |
Hillary attention | 0.009 | 0.004 | 2.038 | 0.042* |
AR1 component (Φ) | 0.231 |
