Figures Don't Lie but, Liars Figure

A Dissent on the Digital Divide
CONTRARY to the federal government's current thinking about the digital divide, the gap may not be closing so quickly after all, according to a new statistical analysis of data from household surveys.

Steven P. Martin, a sociologist at the University of Maryland who conducted the new analysis, argues that the government's most recent report on Americans' Internet use, which used the same data, was flawed.

In that report, released in February 2002, the National Telecommunications and Information Administration, a branch of the Commerce Department, painted a positive picture of the spread of Internet use among low-income families.

Dr. Martin, an assistant professor, contends that the data actually show the opposite of what the report highlighted.

"Computer ownership and Internet use may actually be spreading less quickly among poorer households than among richer households," he wrote in an article published this month in IT & Society, an academic journal focusing on the impact of the Internet (www.ITandsociety.org). He predicts that it may take two decades for the lowest-income groups to catch up to wealthier households.

The debate has been simmering for years. Is the disparity between rich and poor when it comes to computers and the Internet troubling enough to warrant a change in government policy or concerted help from nonprofit organizations and corporations? Or is it something that will quickly heal itself through market pressures?

The 2002 report pointed out in its executive summary that Internet use had increased faster for poor families than rich ones.

Yet earlier reports published during the Clinton administration had been less positive. The last of these, issued in 2000, said that although the have-nots had made striking gains, the disparity among groups of different income levels remained and had widened slightly in some cases.

Soon after the publication of the 2002 report, the Bush administration moved to end financing for government programs that supported community computing centers and local organizations that needed technological improvements. The cuts were opposed by technology advocates and rights groups. In the end, the programs did maintain some financing last year.

Advocates for community programs have argued that the gap between the haves and have-nots was more troubling than the administration made it out to be. "The digital divide is not abating," wrote Norris Dickard, a senior associate at the Benton Foundation, a technology advocacy group, in a report released last year. "Reduced national attention to this problem will dampen economic productivity and opportunity in low-income and rural communities."

Dr. Martin does not contend that the government was working with bad data in its 2002 report; the numbers came from Current Population Surveys, data that is collected by the Census Bureau and regularly used by social scientists. He argues rather that the authors focused on statistics that can tell two stories at once, and then chose to emphasize the positive.

As an example, he cites the government's use of annual rates of increase in Internet use. It is true, he writes, that Internet use among the poorest families - those with annual incomes of less than $15,000 - grew by an average of 25 percent from December 1998 to September 2001, while the rate among the most affluent families, with annual incomes of more than $75,000, increased 11 percent annually. But he argues that is because so few of the poorest families were using the Internet in 1998 to begin with. (Only 14 percent of the poorest families used the Internet from any location in 1998; 25 percent used it in 2001.)

The percentage of wealthiest families using the Internet in 1998 was already high, at 59 percent. By 2001, that number had risen to 79 percent. Because the number was relatively high to begin with, the growth rate inevitably slowed down.

Dr. Martin re-analyzed the data using what statisticians call odds ratios, a method that can avoid the pitfalls of looking simply at growth rates. By his measure, the odds that a family in the poorest bracket would use the Internet increased by a factor of 2.1 over those three years, while the odds for a family in the most affluent group increased by a slightly higher factor of 2.6.

Using those tools, Dr. Martin said in an interview that it could be 2013 before 90 percent of the poorest families use the Internet, be it from home or a community center or a library. Based on the numbers for computer ownership, he conjectures that it could be 2020 before 90 percent of the poorest families own PC's. (In this case, the poorest families are defined as the bottom 25 percent and the richest, the top 25 percent.) By contrast, in 2001, 88 percent of the richest families owned computers.

http://www.nytimes.com/2003/09/18/technology/circuits/18divi.html