Results from Bayesian models - see [[220808_103243 prereg results - bayesian|original taskus results]] - see [[220911_073236 forest - doubly robust estimators]] - [prereg](https://docs.google.com/document/d/1xnyTYNzUVRJH1asM5Orc-QSQkwm4hZrHREYTlPM85dQ/edit) - coding - condition (0/1): control-interest vs treatment-accuracy - repetition (-0.5/0.5): novel vs repeated - veracity (-0.5/0.5): false vs true - models with covariates (7 in total) - numeric covariates are z-scored - categorical covariates have many levels and are dummies ```r # passed attention checks country N 1: IN 997 # india 2: PN 1184 # philippines # covariates covs <- c("age", "gender", "education", "income", "aot", "covid_concern", "conspiracy") ``` # country means ![[1662958646.png]] ```r # binarized 1-6 ratings and computed proportion headlines rated as true condition repetition truthB 1: interest novel 0.6738803 2: interest repeated 0.7037656 > round((0.7037656 - 0.6738803) / 0.6738803 * 100, 2) [1] 4.43 # % increase in headlines rated true for repeated headlines ``` # hypothesis 2: false headlines only ## india all effects are "significant" in the expected/correct direction ```r param b low high p_direction 1: b_Intercept 4.041 3.892 4.189 1 2: b_repetitionc 0.114 0.063 0.165 1 3: b_conditiond -0.344 -0.483 -0.205 1 4: b_repetitionc:conditiond -0.144 -0.220 -0.070 1 ``` ![[1662958496.png]] ### with covariates ```r param b low high p_direction 1: b_Intercept 3.842 3.692 3.996 1.000 2: b_repetitionc 0.121 0.065 0.177 1.000 # still significant 3: b_conditiond -0.249 -0.367 -0.130 1.000 # still significant 4: b_age -0.163 -0.224 -0.102 1.000 5: b_gender 0.035 -0.025 0.093 0.878 6: b_education 0.091 0.032 0.152 0.999 7: b_income 0.029 -0.027 0.084 0.840 8: b_aot -0.427 -0.492 -0.365 1.000 9: b_covid_concern 0.122 0.066 0.179 1.000 10: b_conspiracy 0.136 0.078 0.194 1.000 11: b_repetitionc:conditiond -0.139 -0.212 -0.067 1.000 # still significant 12: b_repetitionc:age 0.029 -0.009 0.067 0.934 13: b_repetitionc:gender 0.013 -0.024 0.051 0.760 14: b_repetitionc:education 0.000 -0.038 0.038 0.507 15: b_repetitionc:income -0.051 -0.088 -0.015 0.997 16: b_repetitionc:aot -0.028 -0.067 0.013 0.910 17: b_repetitionc:covid_concern 0.008 -0.027 0.044 0.670 18: b_repetitionc:conspiracy 0.013 -0.024 0.050 0.747 ``` ## philippines all effects are "significant" in the expected/correct direction ```r param b low high p_direction 1: b_Intercept 3.685 3.498 3.865 1 2: b_repetitionc 0.115 0.076 0.155 1 3: b_conditiond -0.478 -0.600 -0.344 1 4: b_repetitionc:conditiond -0.163 -0.219 -0.108 1 ``` ![[1662961988.png]] ### with covariates ```r param b low high p_direction 1: b_Intercept 3.759 3.581 3.937 1.000 2: b_repetitionc 0.126 0.083 0.168 1.000 # still significant 3: b_conditiond -0.468 -0.578 -0.354 1.000 # still significant 4: b_age -0.099 -0.150 -0.046 1.000 5: b_gender 0.082 0.028 0.136 1.000 6: b_education -0.071 -0.129 -0.014 0.992 7: b_income 0.002 -0.057 0.061 0.524 8: b_aot -0.458 -0.514 -0.406 1.000 9: b_covid_concern 0.003 -0.054 0.057 0.537 10: b_conspiracy 0.165 0.110 0.217 1.000 11: b_repetitionc:conditiond -0.160 -0.217 -0.103 1.000 # still significant 12: b_repetitionc:age -0.001 -0.031 0.028 0.530 13: b_repetitionc:gender 0.020 -0.010 0.050 0.908 14: b_repetitionc:education 0.002 -0.030 0.034 0.554 15: b_repetitionc:income 0.017 -0.015 0.049 0.847 16: b_repetitionc:aot -0.035 -0.065 -0.005 0.990 17: b_repetitionc:covid_concern -0.003 -0.034 0.028 0.578 18: b_repetitionc:conspiracy -0.007 -0.036 0.023 0.679 ``` # hypothesis 1: all headlines - only CONTROL group, because H1 interaction effect is significant ## india ```r param b low high p_direction 1: b_Intercept 4.175 4.070 4.276 1.00 2: b_veracityc 0.278 0.130 0.424 1.00 3: b_repetitionc 0.113 0.069 0.157 1.00 4: b_veracityc:repetitionc -0.005 -0.084 0.073 0.55 ``` ![[1662963699.png]] ### with covariates ```r param b low high p_direction 1: b_Intercept 4.008 3.904 4.111 1.000 2: b_veracityc 0.326 0.176 0.474 1.000 # significant 3: b_repetitionc 0.112 0.068 0.156 1.000 # significant 4: b_age -0.134 -0.208 -0.062 1.000 5: b_gender 0.021 -0.048 0.087 0.731 6: b_education 0.064 -0.010 0.138 0.954 7: b_income 0.053 -0.016 0.120 0.936 8: b_aot -0.372 -0.450 -0.298 1.000 9: b_covid_concern 0.102 0.036 0.168 0.998 10: b_conspiracy 0.137 0.067 0.207 1.000 11: b_veracityc:repetitionc -0.006 -0.086 0.076 0.555 # not significant 12: b_veracityc:age 0.057 0.008 0.104 0.989 13: b_veracityc:gender -0.015 -0.060 0.029 0.758 14: b_veracityc:education 0.031 -0.017 0.079 0.904 15: b_veracityc:income 0.033 -0.011 0.077 0.929 16: b_veracityc:aot 0.173 0.124 0.222 1.000 17: b_veracityc:covid_concern 0.010 -0.033 0.053 0.675 18: b_veracityc:conspiracy 0.037 -0.010 0.084 0.937 ``` ## philippines ```r param b low high p_direction 1: b_Intercept 3.894 3.775 4.008 1.000 2: b_veracityc 0.420 0.221 0.618 1.000 3: b_repetitionc 0.100 0.062 0.136 1.000 4: b_veracityc:repetitionc -0.030 -0.099 0.039 0.805 ``` ![[1662966433.png]] ### with covariates ```r param b low high p_direction 1: b_Intercept 3.949 3.836 4.064 1.000 2: b_veracityc 0.401 0.200 0.594 1.000 # significant 3: b_repetitionc 0.100 0.062 0.138 1.000 # significant 4: b_age -0.090 -0.152 -0.029 0.999 5: b_gender 0.066 0.004 0.128 0.984 6: b_education -0.057 -0.121 0.011 0.950 7: b_income 0.026 -0.037 0.094 0.783 8: b_aot -0.334 -0.397 -0.268 1.000 9: b_covid_concern 0.079 0.022 0.141 0.996 10: b_conspiracy 0.163 0.102 0.224 1.000 11: b_veracityc:repetitionc -0.030 -0.099 0.043 0.805 # not significant 12: b_veracityc:age 0.020 -0.030 0.070 0.776 13: b_veracityc:gender -0.096 -0.146 -0.046 1.000 14: b_veracityc:education 0.059 0.005 0.113 0.984 15: b_veracityc:income 0.040 -0.015 0.094 0.927 16: b_veracityc:aot 0.228 0.174 0.279 1.000 17: b_veracityc:covid_concern 0.055 0.006 0.105 0.987 18: b_veracityc:conspiracy -0.084 -0.133 -0.033 1.000 ```