The CT Campaign Cash database was developed by The Connecticut Mirror to make public campaign fund-raising and expense records more accessible.
The data in this database are pulled from the State Elections Enforcement Commission's website. Because it's an imperfect process to pull from different parts of the state's system and stitch them together, some records, either individual receipt records or filing documents, are lost along the way. As a result, this database is not complete.
Here's some more detail on why we chose this approach.
We pull in transaction records, committee registrations and document filings from different parts of the state's web search tool. When deciding how to pull in transactions (expenses and receipts), we had two options: Use records we pull from the SEEC web tool, or use a pair of spreadsheets containing all of these records that SEEC update a week or more after certain filing deadlines.
Option A, the transaction data we pull from the web, can't be perfectly matched up to the committees they belong to because the web interface doesn't include numeric committee IDs with each transaction, and we have to rely on committee names. Committee names can be re-used, meaning two committees have the same name, either in different election cycles, or when a candidate has an exploratory committee and decides to start a candidate committee. Committee names are also prone to typos, or slight variations like "Smith for CT" and "Smith for Conn." As a result, some of these transactions that can't be linked to committees automatically are dropped on the cutting room floor. Since there are hundreds of thousands of transactions, manually fixing these errors isn't practical.
Option B was to use reports the state publishes about a week or more after filing deadlines. The great benefit of those reports is they include numeric committee IDs that allow complete matching of transactions to committees. But we'd have to wait a week or more after filing deadlines to update the transactions in our database.
We went with Option A, choosing more current data over more complete data because waiting on a state report weeks before the election isn't acceptable for our purposes.