```r study topic treatment condition finished N <int> <char> <char> <char> <lgcl> <int> 1: 1 policy llm persuadeHarris TRUE 860 2: 1 policy llm persuadeTrump TRUE 931 3: 1 policy video persuadeHarris TRUE 907 4: 1 policy video persuadeTrump TRUE 960 ``` ![[1729531322.png]] ![[1729531339.png]] # models/results (prelim: i want to triple-check re-coding etc.) ## lean - treated: video (0), llm (1) - conditionC: proTrump (-0.5), proHarris (0.5) - `*_Z` variables: z-score variables ```r # spec 1 (preregistered): interaction model > m0 <- feols(lean_bidentrump_2 ~ treated * conditionC * lean_bidentrump_1Z, d0) > summ(m0, estimate = est_string) term result <char> <char> 1: (Intercept) b = 41.49 [41.02, 41.97], p < .001 2: treated b = -0.40 [-1.08, 0.28], p = .245 3: conditionC b = -2.18 [-3.13, -1.24], p < .001 4: lean_bidentrump_1Z b = 41.57 [41.09, 42.04], p < .001 5: treated × conditionC b = -0.58 [-1.94, 0.77], p = .401 # no interaction 6: treated × lean_bidentrump_1Z b = -1.01 [-1.69, -0.33], p = .003 7: conditionC × lean_bidentrump_1Z b = -0.26 [-1.21, 0.69], p = .592 8: treated × conditionC × lean_bidentrump_1Z b = -1.23 [-2.59, 0.12], p = .075 # spec 2 (not preregistered): main effect model > m1 <- feols(lean_bidentrump_2_v2 ~ treated * conditionC * lean_bidentrump_1_v2Z, d0) > summ(m1, estimate = est_string) term result <char> <char> 1: (Intercept) b = 50.24 [49.76, 50.72], p < .001 2: treated b = 0.44 [-0.25, 1.13], p = .214 # no main effect 3: conditionC b = -0.09 [-1.06, 0.88], p = .856 4: lean_bidentrump_1_v2Z b = 42.43 [41.95, 42.92], p < .001 5: treated × conditionC b = 1.25 [-0.14, 2.63], p = .077 6: treated × lean_bidentrump_1_v2Z b = -1.03 [-1.72, -0.34], p = .003 7: conditionC × lean_bidentrump_1_v2Z b = -0.27 [-1.24, 0.71], p = .592 8: treated × conditionC × lean_bidentrump_1_v2Z b = -1.26 [-2.64, 0.13], p = .075 ``` ## vote intentions ```r # all participants > m2 <- feols(vote_chance_2 ~ conditionC * vote_chance_1Z * treated, d0) > summ(m2, estimate = est_string) term result <char> <char> 1: (Intercept) b = 86.76 [86.37, 87.15], p < .001 2: conditionC b = 0.49 [-0.29, 1.27], p = .220 3: vote_chance_1Z b = 25.04 [24.66, 25.42], p < .001 4: treated b = 0.96 [0.40, 1.52], p = .001 5: conditionC × vote_chance_1Z b = -0.34 [-1.09, 0.42], p = .383 6: conditionC × treated b = -0.84 [-1.96, 0.28], p = .140 7: vote_chance_1Z × treated b = -1.57 [-2.13, -1.01], p < .001 8: conditionC × vote_chance_1Z × treated b = -0.26 [-1.38, 0.87], p = .654 # (Harris supporters in proHarris condition) or (Trump supporters in proTrump condition) > m2_votemore <- feols(vote_chance_2 ~ conditionC * vote_chance_1Z * treated, d0[vote_more == 1]) > summ(m2_votemore, estimate = est_string) term result <char> <char> 1: (Intercept) b = 87.14 [86.61, 87.66], p < .001 2: conditionC b = -0.11 [-1.16, 0.94], p = .843 3: vote_chance_1Z b = 25.29 [24.78, 25.81], p < .001 4: treated b = 0.93 [0.18, 1.69], p = .015 5: conditionC × vote_chance_1Z b = -0.88 [-1.90, 0.14], p = .093 6: conditionC × treated b = 0.07 [-1.43, 1.58], p = .923 7: vote_chance_1Z × treated b = -1.93 [-2.67, -1.19], p < .001 8: conditionC × vote_chance_1Z × treated b = 0.75 [-0.72, 2.23], p = .319 # (Harris supporters in proTrump condition) or (Trump supporters in proHarris condition) > m2_votemoreno <- feols(vote_chance_2 ~ conditionC * vote_chance_1Z * treated, d0[vote_more == 0]) > summ(m2_votemoreno, estimate = est_string) term result <char> <char> 1: (Intercept) b = 86.48 [85.88, 87.07], p < .001 2: conditionC b = 0.82 [-0.37, 2.01], p = .177 3: vote_chance_1Z b = 24.85 [24.28, 25.41], p < .001 4: treated b = 0.77 [-0.08, 1.62], p = .075 5: conditionC × vote_chance_1Z b = 0.05 [-1.08, 1.17], p = .933 6: conditionC × treated b = -1.69 [-3.39, 0.00], p = .050 7: vote_chance_1Z × treated b = -1.30 [-2.16, -0.44], p = .003 8: conditionC × vote_chance_1Z × treated b = -1.28 [-3.01, 0.44], p = .144 ```