- see [[220331_174150 absolute count badness|badness absolute count]], [[220331_173513 summed badness|summed badness]] # parameters - bad/goodness measures: fc, mbfc (factual, bias, mean(factual, bias), min(factual_bias)), afm (reliable, bias, mean(reliable, bias), min(reliable, bias)) (see [[different measures of domain quality]]) - goodness thresholds: $[5, 10, 15...90, 95]$ - each domain with value **below** threshold is counted as $1$ (domains above threshold not counted) - outcome: **fraction** bad domain (range: $[0, 1]$) - $\frac{\text{no. of bad domains retweeted}}{\text{total no. of retweets including those without domains}}$ - winsorize user summed values (both at t1 and t0): $[0.85, 0.90, 0.95, 0.99, 1.00]$ (1.00 = no winsorizing) - winsorizing applied **before** *filling in missing values or NaN (0/0: no retweets at all) with 0* # results - models - fraction bad tweets, **OLS** model: `feols(t1 ~ conditionC * t0LC | block, dt1, vcov = "HC1")` - `conditionC` $[-0.5, 0.5]$ - `t0` (mean-centered) time0/pre-campaign covariate not shown because all estimates are highly significant - `block` as fixed effect - red: statistically significant effect (p < .05) ![[_frac_winsorize-1.00.png]] ![[_frac_winsorize-0.99.png]] ![[_frac_winsorize-0.95.png]] ![[_frac_winsorize-0.90.png]] ![[_frac_winsorize-0.85.png]]