![[20241107195325.png]] ```r Statistics (16.05% NAs) N Ndist Mean SD Min Max Skew Kurt 1936 1933 41.14 6.85 29.07 71.91 0.69 4.63 Quantiles 1% 5% 10% 25% 50% 75% 90% 95% 99% 30.32 31.08 31.35 37.3 42.29 43.88 48.02 52.18 63.16 ``` What if we median split? Focus those who completed follow-up under the median no. of days ```r > med <- 42.29 # median days > summ(lm(lean_bidentrump_3 ~ condition + scale(lean_bidentrump_1), data = d0[days <= med])) term result <char> <char> 1: (Intercept) b = 46.68 [45.53, 47.83], p < .001 2: conditionproTrump b = 1.49 [-0.17, 3.15], p = .079 # almost significant 3: scale(lean_bidentrump_1) b = 40.88 [40.05, 41.71], p < .001 # for simplicity, coded treatment as increase vote (1), decrease vote (0) > summ(lm(vote_chance_3 ~ conditionVoteMore + scale(vote_chance_1), data = d0[days <= med])) term result <char> <char> 1: (Intercept) b = 88.67 [87.43, 89.91], p < .001 2: conditionVoteMore b = 0.09 [-1.66, 1.84], p = .919 # not at all 3: scale(vote_chance_1) b = 19.23 [18.35, 20.10], p < .001 > summ(lm(vote_harris_3 ~ condition + scale(vote_harris_1), data = d0[days <= med])) term result <char> <char> 1: (Intercept) b = 0.49 [0.47, 0.50], p < .001 2: conditionproTrump b = -0.03 [-0.06, 0.00], p = .020 # significant 3: scale(vote_harris_1) b = 0.45 [0.44, 0.47], p < .001 > summ(lm(vote_trump_3 ~ condition + scale(vote_trump_1), data = d0[days <= med])) term result <char> <char> 1: (Intercept) b = 0.43 [0.41, 0.46], p < .001 2: conditionproTrump b = 0.02 [-0.02, 0.05], p = .348 # not significant 3: scale(vote_trump_1) b = 0.43 [0.41, 0.45], p < .001 ``` # other stuff ```r > summ(lm(lean_bidentrump_3 ~ (lean_bidentrump_2 + lean_bidentrump_1) * days, data = d0)) term result <char> <char> 1: (Intercept) b = 2.70 [-2.45, 7.84], p = .304 2: lean_bidentrump_2 b = -0.03 [-0.38, 0.31], p = .862 3: lean_bidentrump_1 b = 0.98 [0.64, 1.33], p < .001 4: days b = -0.01 [-0.14, 0.11], p = .823 5: lean_bidentrump_2 × days b = 0.01 [0.00, 0.02], p = .034 6: lean_bidentrump_1 × days b = -0.01 [-0.02, 0.00], p = .037 > summ(lm(vote_chance_3 ~ (vote_chance_2 + vote_chance_1) * days, data = d0)) term result <char> <char> 1: (Intercept) b = -8.17 [-22.45, 6.12], p = .262 2: vote_chance_2 b = 0.45 [0.04, 0.87], p = .032 3: vote_chance_1 b = 0.62 [0.24, 1.00], p = .001 4: days b = 0.59 [0.25, 0.92], p = .001 5: vote_chance_2 × days b = 0.00 [-0.01, 0.01], p = .840 6: vote_chance_1 × days b = -0.01 [-0.01, 0.00], p = .257 > summ(lm(vote_harris_3 ~ (vote_harris_2 + vote_harris_1) * days, data = d0)) term result <char> <char> 1: (Intercept) b = 0.02 [-0.06, 0.09], p = .703 2: vote_harris_2 b = 0.21 [-0.14, 0.55], p = .244 3: vote_harris_1 b = 0.74 [0.39, 1.08], p < .001 4: days b = 0.00 [0.00, 0.00], p = .444 5: vote_harris_2 × days b = 0.01 [0.00, 0.02], p = .088 6: vote_harris_1 × days b = -0.01 [-0.02, 0.00], p = .067 > summ(lm(vote_trump_3 ~ (vote_trump_2 + vote_trump_1) * days, data = d0)) term result <char> <char> 1: (Intercept) b = 0.09 [0.01, 0.17], p = .025 2: vote_trump_2 b = 0.26 [-0.14, 0.67], p = .202 3: vote_trump_1 b = 0.61 [0.21, 1.02], p = .003 4: days b = 0.00 [0.00, 0.00], p = .294 5: vote_trump_2 × days b = 0.01 [0.00, 0.02], p = .074 6: vote_trump_1 × days b = -0.01 [-0.02, 0.00], p = .095 ```