Function to conduct a statistical test with the null hypothesis that there is no difference between the correlation coefficients between list experiment and endorsement experiment data.
comp.listEndorse(
y.endorse,
y.list,
treat,
n.draws = 10000,
alpha = 0.05,
endorse.mean = FALSE,
method = "pearson"
)
comp.listEndorse
returns a list with four elements: the
correlation statistic (rho or tau) for the treatment group as
cor.treat
, the correlation statistic for the control group as
cor.control
, the p.value for the statistical test comparing the two
correlation statistics as p.value
, and the bootstrapped confidence
interval of the difference as ci
.
A numerical matrix containing the response data for the endorsement experiment.
A numerical vector containing the response data for a list experiment.
A numerical vector containing the binary treatment status for the experiments. The treatment assignment must be the same for both experiments to compare across experiments.
Number of Monte Carlo draws.
Confidence level for the statistical test.
A logical value indicating whether the mean endorsement experiment response is taken across questions.
The method for calculating the correlation, either Pearson's rho or Kendall's tau.
Graeme Blair, UCLA, graeme.blair@ucla.edu and Kosuke Imai, Princeton University, kimai@princeton.edu
This function allows the user to calculate the correlation between list and endorsement experiment data within the control group and the treatment group, and to conduct a statistical test with the null hypothesis of no difference between the two correlation coefficients.
Blair, Graeme, Jason Lyall and Kosuke Imai. (2014) ``Comparing and Combining List and Experiments: Evidence from Afghanistan." American Journal of Political Science. available at http://imai.princeton.edu/research/comp.html