Students of congressional voting are having a hard time figuring whether special-interest groups' campaign contributions really do "buy" votes on Capitol Hill.

A new set of studies, presented here to the annual meeting of the American Political Science Association, has deepened the controversy about the degree of influence that the growing numbers of business, union and ideological political-action committees (PACs) wield as a result of the dollars they put into House and Senate campaigns.

"No one has demonstrated very convincingly there is a link" between PAC contributions and roll-call voting on issues of importance to the PAC, said Frank Sorauf of the University of Minnesota, who is involved in a major study of PACs and was chairman today of a panel on the issue.

Sorauf said political scientists, as a group, demand a higher standard of proof than is usually found in journalists' reports or in news releases from anti-PAC organizations like Common Cause, the self-styled citizens' lobby.

Reports showing, for example, that 90 percent of the legislators who voted for a tax bill received campaign contributions from PACs supporting its passage are not adequate, Sorauf said.

Numerous political-science studies have demonstrated that party and ideology are the major determinants of how Senate and House members usually vote, along with constituency interests and committee assignments.

Unless those factors are isolated or "controlled" by sophisticated techniques of statistical analysis, it is impossible to determine the independent effect that campaign contributions may have had on the vote.

Two papers presented to Sorauf's panel demonstrate the difficulty of proving generalizations.

Diana Evans Yiannakis of Trinity College in Hartford, Conn., focused on the Chrysler Corp. loan guarantee and oil windfall profits tax bills of 1979--both heavily lobbied issues--and tried to measure the impact of 1978 campaign contributions on those votes in the House.

"The effect is small . . . , " she concluded. "In most cases, members who voted for either of the two policies did so regardless of whether or not they got contributions. By the same token, receipt of contributions from interested PACs had little effect in pulling members away from an opposing position which was determined by party or ideology."

"Perhaps the most surprising finding," Yiannakis wrote, is that statistically, at least, contributions from the United Auto Workers union, which led the lobbying campaign for the Chrysler loan, "had no effect," even though the union put $425,615 into the previous election campaign.

Quite a different conclusion was reached by Kirk F. Brown of Yale University in case studies of the 1982 House votes vetoing the Federal Trade Commission's proposed used-car rule and passing a bill to exempt doctors and other professionals from FTC regulation.

Brown said he chose those votes because critics of the PACs charged that heavy campaign contributions from the National Automobile Dealers Association and the American Medical Association determined the outcome.

He found that in both cases, even applying the most rigorous kind of statistical tests to screen out other influences, the critics had grounds for concern.

"Clearly ideology was crucial in determining a congressman's vote" on the used-car regulations veto, he said. But, he added, the size of the campaign contribution from the auto dealers had a noticeable effect on the more liberal members of Congress, who would not be expected to support the veto for ideological reasons.

"For instance, a campaign contribution of $5,000 (as opposed to no contribution) would be expected to increase the probability of a Democrat with a 75 rating from ADA Americans for Democratic Action, suggesting a liberal ideology voting for the veto by 30 percent . . . . "

The findings were similar for the vote on exempting doctors and other professionals from regulation, leading Brown to conclude: "Even when ideology and attitudes toward regulation and the FTC are statistically controlled, campaign contributions had a powerful influence on congressmen's votes . . . . "

In a paper delivered to another panel, Betty H. Zisk of Boston University reported that money seemed to be the "most powerful" determinant of referendums in the states.

In 40 of 50 referendum campaigns in four states between 1976 and 1980, the side that spent the most money won, she said.

In more than half the cases where polling was available, she found that "voter preferences were reversed in the high-spending direction in the course of the campaign, and in all but two, this was enough to change the outcome."