The Park Library Logo

FAQs for CAR

Q1: What do I have to do to launch CAR?

A1: You have to WANT to do it. You also have to learn how to do it. It is more than just online database searching familiar to most news librarians.

Q2: Well, where do I start learning about CAR?

A2: Read about it in trade journals. Subscribe to IRE Journal and MICAR's UPLINK newsletter. That way you'll discover the players, the projects and the pitfalls. Subscribe to some of the better electronic mailing lists such as CARR-L, NewsLib, ONLINE-NEWS to mention a few.

Q3: OK, what next?

A3: Discover who is doing investigative reporting, special projects in your organization. Read what they've produced. Track where they did their research. Was your library involved? If not, why not?

Q4. What are the most necessary parts of a CAR program?

A4: Besides well trained investigative journalists and skilled news researchers? Well, some people have identified six software features as necessary for a bare-bones CAR program. They are (a) spreadsheet, (b) database manager program, (c) Internet access, and (d) word processing (text management). These features are necessary for even the most basic CAR program and cannot be easily obtained from outside vendors (as, for example, 9-track tape conversion can be).

Q5: Which additional features should be added after the initial phase?

A5: A statistics package for starters. Computer software packages for database management (Paradox, XDB and FoxPro, among others) and statistical analysis (SPSS and SAS) are central to this type of investigative reporting. Generally speaking, a database-management program allows users to examine selectively the information that meets a specified criterion. Statistics are the way summary statements are made about the content of a database. Unfortunately, most database programs have minimal, if any, statistical-reporting capabilities. A statistical analysis program is necessary for a detailed and valid investigation of a database. On the other hand, some statistical analysis packages have some database-management capabilities. At a minimum, statistical analysis packages should perform frequencies, crosstabs, averages, t-tests, correlations and regressions (including significance and probability), depending on the size of the data, the type of software and the complexity of the analysis. The added capability of generating charts and graphs from statistical analysis is vital.

Other useful software and hardware packages include a mapping program, a programming tool for developing user interfaces for reference databases, and data archiving hardware (such as writable CD-ROMs or magneto-optical drives. These features will permit users to do more sophisticated analysis of their data, and to make use of databases as a reference tool. Other functions will continue to be farmed out except in advanced CAR programs. These include: 9-track tape reading, print document scanning, cartridge tape reading, and networking hardware and software. Additional computers should also be added at this stage.

Q6: I don't know how to use all that stuff? And I don't know who does. Where can I learn? And do I really have to?

A6: That is where the concept of "team journalism" comes into the picture. There has to be some delineation of duties. You'll need people with expertise in programming, spreadsheets and statistical analysis. We're talking about training here. You can get some of that in academic courses, in-house workshops, and CAR seminars sponsored by IRE and MICAR. It will take time and effort and money and management support.

Some news organizations are taking seriously the added responsibility of teaching their news staffs the more technical aspects of investigative reporting. Getting news staff and researchers plugged into electronic information sources requires planning and effort.

The most daunting obstacles newsrooms face in establishing a computer-assisted reporting program are not created by legal issues, nor financial issues, but rather training ones. How do you take a newsroom full of reporters*many of whom did not bother to fully learn the newspaper's word processing program, and some of whom display outright contempt for new technology*and transform them into computer-literate, high-tech muckrakers?"42

In-House Training Programs

One newspaper reporter, Pat Stith from the Raleigh, North Carolina News & Observer, created an in-house CAJ workshop for news staff interested in learning how to use the CAJ hardware and software. Stith developed an instructional manual (with worksheets) to accompany the daily two-hour sessions offered over a two-week period. Reporters get instruction and practice in using DOS, loading tapes, learning FoxPro and the computer program "tool box" to query tapes and, finally, generating reports using Report Writer. The final exercise involves applying the CAJ skills learned during the workshop. Stith displays samples of over 100 databases, discusses current newsroom projects and encourages the reporters to consider potential applications. After all, says Stith, that's the purpose of all this training.

A St. Petersburg Times news researcher, Debbie Wolfe, has developed a concept she calls "team journalism." Wolfe concentrates on "equal-partner" relationships between special-project reporters and news researchers. In addition to developing an in-house training program for computer-assisted journalism, Wolfe created job descriptions for the journalism team. Job titles include: investigative reporter/computer-assisted journalism specialist, electronic public records coordinator (primarily administrative) and electronic public records researcher. Copies of these job descriptions and the in-house training outline may be obtained from the St. Petersburg Times library.

Q7: Is there some kind of CAR methodology?

A7: Here is a useful one.

DeFleur's CAIR methodology consists of a series of steps, strategies and responsibilities central to analyzing public-records databases within a news perspective. The initial steps prior to analysis include the following:

1. Deciding on goals: After a government agency's records have been selected, the type of records must be identified and the overall purpose of the investigation clarified.

2. Initial review of records: This step begins with a specific request for government computerized records with accompanying documentation (codebooks). Charting the tape's territory and assessing overall record quality are essential. This means that the analyst must discover what variables the records contain and their precise location on the tapes, all of which must be consistent with what is described in the codebooks. Finally, the analyst must assess the consistency and completeness of the information on the tapes for all variables and all years for which the information is recorded.

3. Transferring to statistical software: For records based on limited size, desktop computers with spreadsheets can be used. With massive databases, such as those involving millions of records, more storage space (such as that on a mainframe computer) and more sophisticated statistical software are required. This step involves the conversion, usually by means of computer programming, of the raw numbers on tape to new files in a software format for statistical analysis.

4. Internal checking of codes and frequencies within each variable: After the raw data have been transferred to software, an even more detailed check on the nature and quality of the entries for each and every variable (for example, simple frequencies) in the set of records can be conducted. Knowing what entries have high and low frequencies leads to the question of why. This can be the starting place for the design of a specific analysis that will yield newsworthy information.

Q7: What kind of skills doe news librarians need to do this kind of research?

A7: Here is a kind of skills inventory test for you.

Do you have knowledge of UNIX, WAIS, web authoring tools such as perl, QPS, SQL, and web search engines?

Have you ever built or cooperated in building an internal database, relational, preferably such as Access or Paradox?

Are you skilled in creating accurate, valid digests/abstracts from report texts?

Have you collected, catalog, indexed, parsed, queried electronic government data sources?

Can you navigate government electronic data? Bulletin boards, CD-ROMs, web sites?

Are you knowledgeable about other in-house databases such as those in your circulation, advertising, photo departments?

Would you be able to cooperate in the development and maintenance of an in-house "database of databases? Plus be able to create useful interfaces for a variety of different users? This is the beginning of the "intranet" concept.

Are you prepared to archive and update large and varied types of digital material collected for many sources?

Are you skilled in training news staff as well as library/research staff in the use of different digital sources?

Have you developed policies and procedures for tracking corrections in your electronic publications?

What is your experience and skills in marketing new media products from library research?

Copyright 2003 - The Park Library - School of Journalism and Mass Communication - University of North Carolina at Chapel Hill