Difference between revisions of "Political Contributions By Venture Capitalists"
Line 27: | Line 27: | ||
Requests to the server become very slow when the page number is high. At a restriction rate of 1000 records/minute and 200 records per page, the page requests are slower than the restriction (i.e. no throttling back is required) beyond about page 20. This maybe due to a lack of database caching (particularly, the php scripts may be requesting data from the first page each time). This is particularly problematic when requesting the candidates list, which for most research purposes will be needed as a starting point for subsequent requests. | Requests to the server become very slow when the page number is high. At a restriction rate of 1000 records/minute and 200 records per page, the page requests are slower than the restriction (i.e. no throttling back is required) beyond about page 20. This maybe due to a lack of database caching (particularly, the php scripts may be requesting data from the first page each time). This is particularly problematic when requesting the candidates list, which for most research purposes will be needed as a starting point for subsequent requests. | ||
− | One work-around this problem is to request candidates by state. The <tt>state_postal_code</tt> variable, requested as <tt>state</tt> unless in a sort, (presumably) follows the [http://www.usps.com/ncsc/lookups/usps_abbreviations.html USPS state abbreviation convention]. The codes provided in [http://www.edegan.com/repository/USPS-StateCodes.txt USPS-StateCodes.txt]. | + | One work-around this problem is to request candidates by state. The <tt>state_postal_code</tt> variable, requested as <tt>state</tt> unless in a sort, (presumably) follows the [http://www.usps.com/ncsc/lookups/usps_abbreviations.html USPS state abbreviation convention]. The codes provided in [http://www.edegan.com/repository/USPS-StateCodes.txt USPS-StateCodes.txt]. Sept '09 estimates of number of candidates in the dataset are about 120,000. Given 200 per page, that is about 600 pages. Given 50 states, we expect about 12 pages per state with variation. |
Revision as of 19:34, 29 September 2009
This project proposes examining political contributions by Venture Capitalists to address two basic questions:
- Are political contributions by venture capitalists systematic and meaningful?
- Are contributions related to a competition effect between private and government-sponsored venture capitalists? This might be attributable to either:
- A Crowding-Out Effect
- That the presence of GVCs gives entrepreneurs an outside option in negotiation with PVCs, reducing their bargaining surplus
Data on Political Contributions
This project uses data from FollowTheMoney.Org. The Basic documentation on the API states that requests should be of the form:
http://api.followthemoney.org/api_method.php?key=123456789ABCDEF
where the key is the user assigned key. Error codes are returned as XML in the format:
<error code="101" text="access denied - restricted api" origin="National Institute on Money in State Politics" process_timestamp="08-21-2006 17:20:43" update_timestamp="08-17-2006 03:15:00">
A full list of methods that can be called through the API, with links to their documentation, is provided. Of particular interest to this research are (required request fields are in brackets):
- candidates.list.php - (key) Provides a list of all candidates, including their IDs
- candidates.contributions.php - (key,candidate_id) Provides details on all contributions, including contributor name
- contributors.name_search.php - (key,contributor_name) Provides details on contributors
- ballot_measures.list.php - (key) Provides details on ballot measures
- ballot_measures.committees.php - (key) Provides details on ballet measure committees, including committee ID
- ballot_measures.committees.contributions.php - (key,committee_id) Provides details on contributions to specific ballot measures.
Requests to the server become very slow when the page number is high. At a restriction rate of 1000 records/minute and 200 records per page, the page requests are slower than the restriction (i.e. no throttling back is required) beyond about page 20. This maybe due to a lack of database caching (particularly, the php scripts may be requesting data from the first page each time). This is particularly problematic when requesting the candidates list, which for most research purposes will be needed as a starting point for subsequent requests.
One work-around this problem is to request candidates by state. The state_postal_code variable, requested as state unless in a sort, (presumably) follows the USPS state abbreviation convention. The codes provided in USPS-StateCodes.txt. Sept '09 estimates of number of candidates in the dataset are about 120,000. Given 200 per page, that is about 600 pages. Given 50 states, we expect about 12 pages per state with variation.