The 1990 Census remains a good reference, but more history than news. Cheer up! Here are several reliable government sources of 1990s demographic data on counties or even smaller areas. All of them are free or nominal, clean and electronic. On first glance, what they offer seems pretty obvious. The secret lies in analyzing and comparing the pieces to find patterns.

Yes, this data is estimated or incomplete. But remember: Only one person in seven got the long Census form, the source of most of what we write about the 1990 Census. In all cases below, read the documentation to understand the data's limits.

Annual county-level population estimates and components http://www.census.gov/population/www/estimates/countypop.html

The Census Bureau calculates these based on July 1 and releases the data the following March. Even better than getting the estimate is getting the underlying data used to make the estimate: births, deaths, elderly (65+), under 65, net domestic migration, net international immigration, federal citizen movement, group quarters population.

Of these, the single most valuable is net domestic migration, especially when you line up several consecutive years. That's because it's the collective verdict of people voting with their feet. People usually don't move far enough to leave a county. When they do, often it's to go to college, retire or take a better job, meaning many of them won't come back.

Annual county-level estimates for race/Hispanic origin, sex, and age

http://www.census.gov/population/www/estimates/co9094.html (Totals only)

http://www.census.gov/population/estimates/county/casrh/ (Details by five-year age group)

The Census Bureau calculates these periodically on contract and releases them about a year after the estimate date. (Current data is 1990-94; 1990-96 data is due out in July).

These differ in methods and timing from the main estimate process above, so expect differences. You can use the race/Hispanic origin/sex totals for each state and county to watch for trends.

The same data, broken down by five-year age groups, lets you see changes in a county's age structure versus the state and nation.

Annual county-level migration files, with median household income

For county-to-county and most recent state-to-state files, call Emily Gross at the IRS Statistics of Income Division, (202) 874-0944. For previous years' state-to-state files, dial the division's BBS, (202) 874-9574. Or try the often-overloaded Fedworld address: ftp://ftp.fedworld.gov/pub/irs-soi/

The IRS uses the Social Security number on each tax return to check county of residence in consecutive years. For each county, it counts how many stayed put from one year to the next, and how many moved between counties. For each county-to-county flow, it reports returns (roughly, households), personal exemptions (roughly, persons), aggregate income and, since 1992-93, median household income.

You can match a county's inbound and outbound files to calculate net flows between nearby counties, but you can also examine the files separately for distinct patterns in where people come from and go to.

Annual County Business Patterns http://www.census.gov/epcd/cbp/view/cbpview.html

The Census Bureau bases these on a survey conducted in early March. They're released 15 months after the survey year. Data shows employment, payroll and number of establishments by industry. Some data on payroll and employment is masked to prevent disclosure in small categories. But the number of establishments isn't. Neither is their distribution by size.

This data is helpful in following regional trends by industry, especially in the hard-to-track services sector: clinics, churches, day-care centers, hair salons, etc. The data includes basic groups like retailing and construction and smaller groups like "eating and drinking places."

Other useful data

Annual estimates for towns and cities http://www.census.gov/population/www/estimates/citypop.html

These are "places" or "minor civil subdivisions," in Census speak. Basically, they're any self-governing political entity smaller than a county. ("Places" covers all such things in 30 states. In 20 states in the Northeast and Midwest, a variety of local political units like townships, towns, boroughs and villages are self-governing, so the bureau lists them separately).

Estimates are based on July 1 and released the following July. The most recent estimate was July 1, 1994. (The bureau skipped the 1995 release to focus on improving its methodology) These aren't as accurate as county estimates, but are helpful for watching patterns.

Annual county totals in various Social Security programs

http://www.ssa.gov/statistics/oasdi94.html

http://www.ssa.gov/statistics/ssi_st_cty_d95/

http://www.ssa.gov/statistics/eedata93.html

(For most current data on a few counties, call Social Security Administration at (410) 965-8904).

The Social Security Administration has placed several annual county-level statistical reports on the web. They're tabular, so download and import is simple. Currency varies.

Other sources

Finally, don't forget to check for customized analyses and research done by your state demographer and state or county planning department. Also, see if your marketing department pays for access to demographic data from a company like Claritas. It's often overkill for newsrooms and easy to misuse, but it can help portray customized or ZIP code areas, where between-Census data is rare.

Paul Overberg can be reached at (703) 276 5427 or email him at 70534.3647@compuserve.com