Abstract
The first experimental data includes four groups; thus, four survey links have been designed for each group, which was sent via email. The approved target population is twelve thousand Georgia State University affiliates among faculty, staff, and students. We randomly picked three thousand for each group. Six thousand were randomly selected from the faculty and staff, and six thousand from the students. In the survey, we asked for generic information on individuals' characteristics such as gender, income, US native or non-native, and occupation. We provided information on the impact of carbon emission on the environment and human lives and how we can alter that by switching energy-based fuels. In the second and more comprehensive dataset, we reduced the groups to three, based on our power analysis on the first attempt. Instead of an across nations comparison (US vs. EU and China), we performed a state-level comparison. We also added a question about their political affiliation to control for prior beliefs. We executed the experiment-at the same time-in two different states, one Democrat and one Republican, but in the same region and with similar geographical boundaries. We selected Arizona (as a red state having a republican governor) and New Mexico (as a blue state having a Democrat governor). Arizona uses 9 percent renewable energy, and New Mexico utilizes only 8 percent. To control for subjects' judgment about the norm, we compared the energy use from each of the selected states, with two different pioneer states using renewable, one blue and one red, within the same range percentage usage. We chose South Dakota with 38 percentage and Maine with 36 (at the time of execution). We wanted to rule out the subjects' prior beliefs and preferences on each state they were about to compare with that design. One can use these data to replicate the analysis of the reference paper entitled "Impact of the Social Norms on Energy Reform Petitions: Cross-State Comparisons." Or since the collected data is random, there could be some side analysis on the control variables in the dataset (wage, education, political affiliation, age, gender, birthplace, marital status, and participating in an environmentally friendly Act).