At the NICAR database library, we get a lot of calls that sound like this:
"I finally got some data from our local police force, but I don't know what's on it. Can I just send it to you and you tell me what it is?"
Unfortunately, the answer is no. Without some basic information about data on the tape, disk or CD-ROM, it's nearly impossible to decipher what the agency delivered.
Whether you're analyzing the data yourself or sending it to NICAR, the best approach for avoiding this dilemma is to first do a little homework, then write a detailed Freedom of Information request to make sure you get the proper paperwork.
The right questions
Before you even order the data, call the tech support or data person for the agency - whomever is in charge of data collection. Try not to talk to the public relations person because they usually have no idea what a database is, never mind which ones the agency oversees.
Once you get to the right person, ask what the exact name of the dataset is, request a record layout in advance (which will tell you exactly what fields are in the dataset) and ask how they keep the records.
Many agencies are now using common computer programs such as Access or Foxpro, but others will want to send you the information on a 9-track tape, 3480 cartridge or Jaz cartridge.
Nine-track tapes will be the most likely media you might run into.
Here's a checklist of questions you should ask when ordering data on nine-track tape (thanks to Brant Houston's book, "Computer-Assisted Reporting: A Practical Guide.")
What is the tape's density? 6,250 bpi, 1,600 bpi or other? What is the record length? Is it fixed or variable? Are the records standard or defined? What is the blocking factor or block size? Is the tape labeled or unlabeled? Is the database in ASCII or EBCDIC? What kind of machine is producing the tape? IBM, Hewlett-Packard, Digital Equipment Corporation or other? How many records are contained on the tape? Has the information been compressed or packed?
If you don't have the capability of converting data from these formats, NICAR can usually help. But please give us a call to make sure before you send anything to us.
In addition to the above information, NICAR also likes to have an agency contact, just in case we have questions.
The FOIA letter
Now you are ready to write your FOIA request. Here is a list of 10 things you should include in your eFOIA to better your chances:
1. Mention the exact name of the dataset. 2. Request all the fields collected in the database. 3. Request all the code sheets used with the database. 4. Request the record layout. 5. Ask for a record count so you can make sure you've gotten the entire database. 6. Ask for a printout of the first 100 records. 7. If the dataset includes more than one table, ask for a relational chart of how all the tables join. 8. If the record layout does not include a description of the fields, ask for a data dictionary, which gives more details. 9. Specify how you would like your data. You should know what is available from your discussion with the tech person. Say "We would like this data in dbf format on a CD-Rom" or "We would like this data in text format on a zip disk." Sometimes you will have to supply the medium. 10. If it is state data, cite the local Sunshine Law. If it is a federal database, the FOIA citation is U.S. Freedom of Information Act, 5 U.S.C. 522.
Be prepared to argue over field elimination and price. The more you know about the data to begin with, the stronger position you'll have when haggling.
Dawn Fallik may be reached by e-mail at dawn@nicar.org