Risk-averse freshmen have always attempted to outguess their classmates by strategically listing those River Houses of average or low popularity as their first choice Houses in an attempt to avoid rooming at the Quad. This year is no different.

Results of a telephone poll of 106 randomly chosen freshmen this week showed that freshmen applying for Houses yesterday faced the tough dilemma of deciding whether they'd rather maximize their chances of living at the River, by picking an "undesirable" River House, or gamble an undesirable housing assignment for the possibility of living in a glamour House.

"The general trends of unpopularity that tend to occur year after year have been turned upside down, "Bruce Collier, assistant to the dean of the Faculty, said yesterday.

Collier conducted the housing lotteries of the previous two years.

This year's form permitted freshmen only three choices. After maximizing first choices, the housing office computer goes quickly through second and third choices--and then randomly assigns any roommate groups who could not be placed in one of the three to a remaining unfilled House.

If a freshman's goal was to avoid the Quad, he'd do best to list those River Houses he thought would fill up latest in the assignment process, Collier said.

The plummet to seventh and eighth places, respectively, of traditionally favored Lowell and Adams Houses, confirms these predictions.

Only 2 per cent of the freshmen surveyed selected North or South House as their first choice House, possibly the result of adverse publicity the Quad has received.

Poll data suggest that the Houses fall into two rough categories. The first five Houses--Leverett, Kirkland, Dunster, Quincy and Winthrop, in descending order--will be the first-choice House of nearly three-quarters of the freshmen, unless they altered their housing forms following publication of the survey results.

Kenneth W. Wachter, assistant professor of Statistics, said yesterday the differences between Houses like Leverett and Kirkland Houses, on the other hand, and Lowell and Adams on the other, are "substantially greater than could be ascribed to sampling error, given the sample size."