Quick: You’re walking by a store window and you see a sign that says, “20% off the original price plus an additional 25% off the already reduced sale price.” So, how much is the discount” Consumers often mistakenly think the total discount is 45% off the original price when, in fact, the true discount is 40%.

A thought-provoking new study from the October issue of the Journal of Consumer Research explores why consumers frequently think a double discount is a better deal than a single discount of the same total magnitude.

“Retailers frequently use the strategy of double discounts for their regular promotions or to induce customers to open a credit card account with them. Such errors in peoples’ judgments of the net effect of multiple price discounts . . . have implications for a variety of marketing settings including advertising, promotion, pricing, and public policy,” write Haipeng (Allan) Chen (University of Miami) and Akshay R. Rao (University of Minnesota).

Prior studies have shown that even math teachers frequently have trouble calculating percentages. In the first experiment, the researchers found that 59 percent of the respondents – students at a large university – erroneously added the two percentages to calculate the overall discount. Only 26 percent of students got the answer right.

As not everyone was prone to the miscalculation effect, the researchers then sought to identify the situations that help counter calculation error. They first incentivized one group of participants, offering $2 for correct answers. The rate of computational error was 26 percent, compared to 44 percent for students who were not offered money for correct answers.

Participants were also less prone to computational error when the problems were easier or the results seemingly illogical. Students presented with a base price of $100 from which to calculate the double discount were less likely to make an error than students presented with a base price of $80. Similarly, the researchers asked participants to calculate either a double percentage decrease of 70% and 45% or a price increase of the same percentages.

Those who mistakenly added the figures in the decrease condition were confronted with a decrease of 115%, and were more likely to make errors than those confronted with a 115% increase (explained as a change in gasoline prices).

“Since this computational error can potentially influence peoples’ judgment in a variety of settings, the economic impact of such errors on consumer welfare may be substantial,” Chen and Rao write.

The researchers also point to policy implications of computational error outside of consumer settings, such as when a 70% increase followed by a 60% decrease in statewide test scores is considered positive, even though the net effect on test scores is a 32% decrease.