Investors looking for information about Qiao Xing Universal Telephone on Yahoo Finance, the popular Internet site, would learn that stock analysts expect earnings to rise and rate the shares a "buy."

What they may not learn, however, is that the Chinese telecommunications-equipment maker, with a market value of just $312 million, paid the analysts' research firms tens of thousands of dollars in cash or stock to write those reports. As paid-for research firms -- also known as sponsored-research or issuer-paid research firms -- they receive fees from various sources, including companies seeking analyst coverage of their stock.

One of the two research firms that rate Qiao Xing has been paid $33,000 for four reports since May 2005, while the other owns about $20,000 of stock in the company.

Thomson Financial, which supplies earnings estimates and investment ratings to Yahoo Finance and dozens of databases available to institutional investors, is among a number of financial-data providers that include information from paid-for research firms.

The practice is controversial. Since 2003, when big Wall Street brokerage firms paid $1.4 billion in a landmark settlement agreement aimed at eliminating potential conflicts of interest between analysts' research and investment-banking deals, there has been pressure on firms to portray their research as "independent."

Paid-for research firms typically follow tiny, little-known companies the big brokerage firms ignore. Without them, there likely would be no research reports at all on these companies. Critics of paid-for research say it is less reliable, because an analyst could be influenced to be more bullish or bearish in his or her report, depending on who pays for it. Others note that paid-for research isn't that different from companies paying credit-rating firms that assess their debt.

Financial-data providers, such as Thomson, Bloomberg and Reuters, say they are committed to using only independent analysts' reports for the earnings estimates and stock ratings they distribute, though there are exceptions.

Thomson says seven paid-for research firms contribute earnings estimates and ratings on a total of about 145 companies; currently, two of the research firms, J.W. Dutton Associates and Forun Technologies, regularly contribute their opinions on a total of 117 companies.

"They were grandfathered in," says Mike Thompson, director of research at Thomson. "There didn't seem to be any reason to exclude them," before the 2003 settlement. The data provider instituted a policy some three years ago not to use any paid-for research when it compiles estimates or ratings going forward, though it would continue to distribute full research reports from the firms separately.

Paid-for research constitutes only a tiny fraction of all the earnings estimates and ratings that the big financial-data companies put out. Thomson, for instance, says earnings estimates from Dutton and Forun make up just 0.003% of the 37,215 estimates in its North American universe, and just 0.001% of the 98,616 estimates in its global universe.

At Bloomberg, fewer than half a dozen such sponsored-research firms contribute data for earnings estimates and ratings, and they have to pass certain quality-control standards, according to a person familiar with Bloomberg's business.

Jerry Czarzasty, head of global estimates at Reuters Estimates, says the company doesn't use any paid-for research in its estimates or ratings, but does distribute full reports to subscribers.

The full reports include disclosure information about potential conflicts of interest. Paying clients of various Thomson services, for instance, can access the reports and see the disclosures denoting they are paid-for research firms. Thomson also mentions the firms' paid-for status in footnotes in its data files.

But investors who peruse Thomson's data on Yahoo Finance and other Web sites don't see those disclosures.

Yahoo Finance is exploring "possible solutions to best inform our audience when research is funded by the companies covered," says Peggy White, general manager of Yahoo Finance. "Our goal...is to provide users with the information they need to make informed investing decisions, while at the same time remaining fully transparent about the firms providing the news, data and research on our site."

According to statistics from Dutton Associates, one of the paid-for firms, it currently has buy-type ratings -- ranging from "speculative-buy" to "strong buy" -- on 77.5% of the stocks it covers. About 18% are neutrals and 2.9% are "avoid."

In comparison, of all ratings compiled by Thomson Financial for U.S. stocks and American depositary shares, 46.5% are "buy" ratings, 46.2% "hold" and 7.3% "sell."

Those who can access the full reports and read the fine-print disclosures will see that Dutton, an El Dorado, Calif.-based firm, received $33,000 from Qiao Xing to write four research reports, starting on May 18, 2005. According to the disclaimer in a research report, Forun Tech Research, a Plainsboro, N.J., firm, has owned 1,500 Qiao Xing shares since December 2004, as well as warrants to buy more shares.

Other companies whose data come from paid-for research include Paulson Capital, a tiny investment bank with a market value of $35 million whose shares are listed on Nasdaq.

Trent Davis, president of Paulson Capital, said the boutique investment bank commissioned research from Dutton because shareholders "told us we needed to be more proactive about our stock." None of the big Wall Street firms followed it.

Some say the paid-for research isn't that different from other reports.

"Really, on a very basic level, what research isn't paid for?" says Todd Essary, chief executive of Investrend Research, member of a consortium of about 15 paid-for research firms that use a set of standards to publish reports.