|
Field
Planned Number of Clusters
|
Before
Component 1: 45 treated and 45 control constituencies
Component 2: 15 video debates (overlaps with T above), 15 pure controls (overlaps with C above), 21 radio airtime, 21 radio information control
|
After
Component 1: 45 treated and 45 control constituencies
Component 2: 15 video debates (overlaps with T above), 15 pure controls (overlaps with C above), 20 radio platform
|
|
Field
Planned Number of Observations
|
Before
Pooling both components: 132 constituencies in total, with at least 264 candidates (exact number of candidates per constituency TBD from NEC official registration), 1,800 voters
|
After
Polling both components: 132 constituencies in total, with at least 264 candidates (exact number of candidates per constituency TBD from NEC official registration), 1,800 voters
|
|
Field
Intervention (Hidden)
|
Before
The first component, general barriers to scaling, will be evaluated on a sample of 45 treatment and 45 control constituencies. In treatment areas, SFCG will hold public screenings in two locations per constituency, and intensify dissemination through a combination of radio re-broadcasts, mass text messaging and DVD distribution to local cinema halls (which are small and informal venues whose main business is charging fees to watch sporting events and films). To evaluate whether the scaled version achieves similar results to the 2012 pilot, we will survey voters on Election Day in both treatment and control constituencies, sampled from voter registration centers (VRCs) that are selected to host screenings (more similar to the pilot model) and to comparable untargeted VDCs (to capture broader dissemination and spillover effects).
For dissemination to spread more broadly throughout the constituency, though, voters must be willing to incur some costs to access the debates and cinema halls must find it worth their while to play them. Regarding voters, we will distribute free admission passes in communities at varying ordeal costs from the debate screening sites and track take up rates. For the cinema halls, monitors will unobtrusively track how many times the halls choose to play a free DVD of the debate (under no incentives), how much they charge, and how many tickets they sell.
The second component focuses on candidate incentives to participate in initiatives that inform voters and operates on an "on demand" sample of 72 constituencies. SFCG will disseminate an information nudge to candidates in all 72 constituencies and track expressions of interest. The information nudge conveys three things: the value of debates in informing voters about the policy positions and professional qualifications of candidates from different parties; the fact that SFCG is currently hosting debates in other constituencies and looking for additional sites; and instructions for candidates who are interested in having their constituency considered for a debate site. Interested candidates have one week to complete two steps: first they must phone SFCG to express their personal interest in participating. Then they must coordinate with rival party candidates to agree to a date and place where all commit to showing up for a debate, and each must independently contact SFCG to confirm these details. This provides a low cost measure of individual candidate demand and a higher cost measure of coordination across parties, which we will correlate with individual popularity and partisan advantage to evaluate a simple participation model.
A second lottery and information nudge introduces exogenous variation in the presence of a guaranteed broadcast platform and sees if this shifts candidate demand. A computer lottery will allocate 15 constituencies to a SFCG video screening (this is included in the 45 total treatments described in component 1), 15 to pure control (also included above), 21 to an hour of free radio airtime, and 21 to radio information control. SFCG will contact all candidates a second time to inform them of the lottery assignments. SFCG then commences its standard implementation protocol in the first 15. Candidates in the radio sample are informed that SFCG has paid for airtime for a one hour policy debate on a specific local radio station which will broadcast so long as at least one candidate shows up. Candidates in the 21 radio information control constituencies will be informed that their constituency was not selected for an extra debate and provided with the contact details of a local radio station should they wish to organize their own debate. Note that we have exogenous variation in whether candidates previously participated in a debate (from earlier experiments), and can test whether this has positive or negative effects on their willingness to participate again.
|
After
The first component, general barriers to scaling, will be evaluated on a sample of 45 treatment and 45 control constituencies. In treatment areas, SFCG will hold public screenings in two locations per constituency, and intensify dissemination through a combination of radio re-broadcasts, mass text messaging and DVD distribution to local cinema halls (which are small and informal venues whose main business is charging fees to watch sporting events and films). To evaluate whether the scaled version achieves similar results to the 2012 pilot, we will survey voters on Election Day in both treatment and control constituencies, sampled from voter registration centers (VRCs) that are selected to host screenings (more similar to the pilot model) and to comparable untargeted VDCs (to capture broader dissemination and spillover effects).
For dissemination to spread more broadly throughout the constituency, though, voters must be willing to incur some costs to access the debates and cinema halls must find it worth their while to play them. Regarding voters, we will distribute free admission passes in communities at varying ordeal costs from the debate screening sites and track take up rates. For the cinema halls, monitors will unobtrusively track how many times the halls choose to play a free DVD of the debate (under no incentives), how much they charge, and how many tickets they sell.
The second component focuses on candidate incentives to participate in initiatives that inform voters and operates on an "on demand" sample of 72 constituencies. A call center affiliated with SFCG was established to contact candidates in all 72 constituencies and track expressions of interest. Call center workers conveyed three things: the value of debates in informing voters about the policy positions and professional qualifications of candidates from different parties; the fact that SFCG is currently hosting debates in other constituencies and looking for additional sites; and instructions for candidates who are interested in having their constituency considered for a debate site. Interested candidates have one week to complete two steps: first they must phone SFCG to express their personal interest in participating. Then they must coordinate with as many rival party candidates as possible to agree to a date and place where all commit to showing up for a debate, and each must independently contact SFCG to confirm these details. This provides a low cost measure of individual candidate demand and a higher cost measure of coordination across parties, which we will correlate with individual popularity and partisan advantage to evaluate a simple participation model.
A second set of lotteries and information nudges introduce exogenous variation in the presence of a guaranteed broadcast platform and sees if this shifts candidate demand. Roughly half of the 72 on demand constituencies were covered by a local radio station. Of these, 20 were assigned to a guaranteed radio dissemination platform. A computer lottery allocated 15 of the 72 constituencies to a SFCG video screening (this is included in the 45 total treatments described in component 1) and 15 to pure control (also included above). A subset of 30 constituencies had an equal chance of being randomly assigned to the radio platform, video screening and control condition. This subset forms the basis of the embedded experiment to evaluate the distinct platforms against each other. Call center workers contacted all candidates a second time to inform them of the lottery assignments. SFCG then commences its standard implementation protocol in the video 15. Candidates in the radio sample were informed that SFCG has paid for airtime for a one hour policy debate on a specific local radio station which will broadcast so long as at least one candidate shows up. Candidates in remaining constituencies were informed that their constituency was not selected for an extra debate. Note that we have exogenous variation in whether candidates previously participated in a debate (from earlier experiments), and can test whether this has positive or negative effects on their willingness to participate again.
|