Skip to main content

Table 2 Treatment effect estimates and confidence intervals of Facebook advertising in study 1

From: Do Online Advertisements Increase Political Candidates’ Name Recognition or Favorability? Evidence from Randomized Field Experiments

Subgroup Estimation procedure Heard of candidate Positive impression of candidate Vote for candidate Recall seeing online ad
All subjects (N = 2,948) Difference-in-means and randomization inference, no covariates −0.011
[−0.042, 0.020]
0.008
[−0.017, 0.034]
0.011
[−0.018, 0.040]
0.010
[−0.010, 0.032]
Difference-in-means and randomization inference, covariate adjustment −0.011
[−0.041, 0.018]
0.008
[−0.017, 0.034]
0.007
[−0.019, 0.034]
0.009
[−0.011, 0.031]
OLS, clustered standard errors and block fixed effects −0.010
[−0.039, 0.018]
0.010
[−0.013, 0.034]
0.016
[−0.012, 0.044]
0.006
[−0.013, 0.026]
Self-reported Facebook users only (N = 1,364) Difference-in-means, no covariates −0.010
[−0.055, 0.038]
0.000
[−0.039, 0.041]
0.012
[−0.032, 0.056]
0.012
[−0.026, 0.051]
Difference-in-means, covariate adjustment −0.012
[−0.056, 0.033]
−0.003
[−0.041, 0.037]
0.000
[−0.041, 0.039]
0.007
[−0.030, 0.046]
OLS, clustered standard errors and block fixed effects −0.020
[−0.067, 0.026]
−0.010
[−0.050, 0.030]
0.027
[−0.018, 0.072]
0.002
[−0.034, 0.038]
  1. Each cell records the estimate of the effect of being treated with online advertising on the dependent variable at the top of the column. 95 % confidence intervals are shown in brackets below each estimate. No results are statistically significant at the 0.05 level (one-tailed). The first two rows in each panel employ randomization inference to estimate confidence intervals, with the first row applying the procedure to unadjusted difference in means and the second employing covariate adjusted values. Rosenbaum (2002) 95 % confidence intervals for these results are calculated taking into account the blocked, clustered randomization scheme. The final row shows estimates employing OLS with block fixed effects, with 95 % confidence intervals calculated based on clustered standard errors