n_participants: 1530
candidate preference
```r
summ(feols(lean2 ~ condition * lean1Z * strategyD, data = d0, se = "HC1"))
term result sig
<char> <char> <char>
1: (Intercept) b = 34.08 [32.28, 35.88], p < .001 ***
2: conditionproPoilievre b = 10.05 [7.52, 12.57], p < .001 ***
3: lean1Z b = 32.55 [30.53, 34.57], p < .001 ***
4: strategyD b = 2.83 [0.30, 5.35], p = .028 *
5: conditionproPoilievre × lean1Z b = 1.11 [-1.51, 3.72], p = .406
6: conditionproPoilievre × strategyD b = -5.88 [-9.31, -2.46], p = .001 ***
7: lean1Z × strategyD b = -0.08 [-3.00, 2.84], p = .957
8: conditionproPoilievre × lean1Z × strategyD b = 1.72 [-1.92, 5.35], p = .354
# strategy == baseline
r
gt; summ(feols(lean2 ~ condition * lean1Z * modelF, data = d0[strategyD == 0], se = "HC1"))
term result sig
<char> <char> <char>
1: (Intercept) b = 33.15 [30.12, 36.17], p < .001 ***
2: conditionproPoilievre b = 10.92 [6.70, 15.13], p < .001 ***
3: lean1Z b = 32.46 [29.12, 35.80], p < .001 ***
4: modelFDeepSeek-V3 b = 0.91 [-3.64, 5.46], p = .694
5: modelFLlama-4 b = 2.06 [-2.28, 6.39], p = .352
6: conditionproPoilievre × lean1Z b = 2.65 [-1.62, 6.93], p = .223
7: conditionproPoilievre × modelFDeepSeek-V3 b = -0.31 [-6.67, 6.05], p = .924
8: conditionproPoilievre × modelFLlama-4 b = -2.36 [-8.37, 3.65], p = .441
9: lean1Z × modelFDeepSeek-V3 b = -0.11 [-5.08, 4.86], p = .966
10: lean1Z × modelFLlama-4 b = 0.39 [-4.57, 5.35], p = .878
11: conditionproPoilievre × lean1Z × modelFDeepSeek-V3 b = -3.57 [-10.07, 2.94], p = .282
12: conditionproPoilievre × lean1Z × modelFLlama-4 b = -0.87 [-7.09, 5.35], p = .784
```
vote choice
```r
summ(feols(vote_poilievre2 ~ condition * scale(vote_poilievre1) * strategyD, data = d0, se = "HC1"))
term result sig
<char> <char> <char>
1: (Intercept) b = 26.01 [24.50, 27.53], p < .001 ***
2: conditionproPoilievre b = 9.81 [6.74, 12.87], p < .001 ***
3: scale(vote_poilievre1) b = 41.15 [38.76, 43.55], p < .001 ***
4: strategyD b = 1.85 [-0.45, 4.14], p = .114
5: conditionproPoilievre × scale(vote_poilievre1) b = -1.64 [-4.72, 1.45], p = .299
6: conditionproPoilievre × strategyD b = -6.90 [-10.96, -2.85], p = .001 ***
7: scale(vote_poilievre1) × strategyD b = 0.12 [-3.14, 3.37], p = .944
8: conditionproPoilievre × scale(vote_poilievre1) × strategyD b = 2.10 [-2.18, 6.38], p = .336
rgt; summ(feols(vote_carney2 ~ conditionCarney * scale(vote_carney1) * strategyD, data = d0, se = "HC1"))
term result sig
<char> <char> <char>
1: (Intercept) b = 44.94 [42.94, 46.94], p < .001 ***
2: conditionCarney b = 8.43 [5.06, 11.79], p < .001 ***
3: scale(vote_carney1) b = 45.50 [43.44, 47.56], p < .001 ***
4: strategyD b = 3.09 [0.10, 6.08], p = .043 *
5: conditionCarney × scale(vote_carney1) b = -3.02 [-6.37, 0.32], p = .076 .
6: conditionCarney × strategyD b = -4.07 [-8.63, 0.48], p = .079 .
7: scale(vote_carney1) × strategyD b = -0.45 [-3.48, 2.59], p = .773
8: conditionCarney × scale(vote_carney1) × strategyD b = 3.09 [-1.41, 7.59], p = .178
```
vote likelihood
```r
# among poilievre supporters
rgt; summ(feols(vote_chance2 ~ condition * scale(vote_chance1) * strategyD, data = d0[initial_lean == "Poilievre supporter"], se = "HC1"))
term result sig
<char> <char> <char>
1: (Intercept) b = 83.94 [81.93, 85.95], p < .001 ***
2: conditionproPoilievre b = 1.55 [-0.93, 4.03], p = .220
3: scale(vote_chance1) b = 23.84 [20.60, 27.07], p < .001 ***
4: strategyD b = 0.05 [-2.31, 2.41], p = .967
5: conditionproPoilievre × scale(vote_chance1) b = -0.48 [-4.48, 3.51], p = .813
6: conditionproPoilievre × strategyD b = -1.14 [-4.55, 2.26], p = .510
7: scale(vote_chance1) × strategyD b = 1.34 [-2.32, 5.00], p = .471
8: conditionproPoilievre × scale(vote_chance1) × strategyD b = -0.10 [-5.12, 4.93], p = .969
# among carney supporters
rgt; summ(feols(vote_chance2 ~ conditionCarney * scale(vote_chance1) * strategyD, data = d0[initial_lean == "Carney supporter"], se = "HC1"))
term result sig
<char> <char> <char>
1: (Intercept) b = 86.30 [85.25, 87.35], p < .001 ***
2: conditionCarney b = 0.45 [-1.75, 2.65], p = .686
3: scale(vote_chance1) b = 24.08 [22.61, 25.55], p < .001 ***
4: strategyD b = 0.91 [-0.88, 2.70], p = .317
5: conditionCarney × scale(vote_chance1) b = -3.43 [-6.96, 0.10], p = .057 .
6: conditionCarney × strategyD b = -0.38 [-3.34, 2.58], p = .802
7: scale(vote_chance1) × strategyD b = -3.11 [-6.33, 0.11], p = .058 .
8: conditionCarney × scale(vote_chance1) × strategyD b = 3.72 [-1.48, 8.91], p = .161
```