Learning Curve

There are those who argue that intelligence, as reflected in test scores, is fixed and immutable. No need for additional schooling, they say; if a student performs poorly on standardized tests, not much can help.

Kellogg School Professor Karsten Hansen begs to differ.

In an award-winning paper published recently in the Journal of Econometrics, Hansen found evidence that an additional year of schooling can indeed boost test scores by up to 2 to 4 percent.

The findings are especially timely given the federal government’s focus on standardized testing through the No Child Left Behind Act. The law has pressured educators to determine the best way to allocate education resources to meet the performance standards mandated by the act.

“Our paper taps into more philosophical aspects, because there’s a group of people out there claiming that more schooling cannot improve test scores,” Hansen says.

“It’s a little depressing to think you can’t improve your basic abilities by attending school. It’s very depressing to think it’s all over at age 6,” Hansen says. “At least the numbers in our study suggest that schooling can have an impact.”

Hansen and his co-authors, James Heckman of the University of Chicago and Kathleen Mullen of Harvard University, devised a statistical framework to analyze whether an extra year of school improved scores on standardized tests. They found a positive link between the two factors, particularly for students with lower innate abilities.

To reach their conclusions, the researchers devised a breakthrough methodology to correct for differences in students’ abilities, socioeconomic status and other variables. The process enabled them isolate the impact of schooling on test scores.

Their paper, “The Effect of Schooling and Ability on Achievement Test Scores,” received the journal’s Dennis J. Aigner Award in September. The award recognizes the best article on applied econometrics published in the journal over a two-year span. The article appeared in the journal’s July-August 2004 issue.

Hansen began his research for the paper as a post-doctoral fellow in the economics department at the University of Chicago. Now an assistant professor of marketing at the Kellogg School, he currently focuses on business rather than public policy.

But the tools Hansen helped create for the schooling study can also be brought to bear on marketing questions, he says. Researchers can use the same methodology whenever they want to study links between factors over which they have no direct control.

“Suppose you wanted to understand how a certain brand performed in different markets and how the pricing impacted sales,” Hansen says. “You could just look at the correlation between sales and prices across a number of markets, but you could be comparing a very competitive market to a less competitive market, or a large market to a small market. If you didn’t take those factors into account when you did the correlations, you could come up with the wrong inference about the exact dependence between price and sales.

“The tools for correcting for those variables are exactly the same in both economics and marketing. These are fields in which we can’t control the experiment - we can’t force consumers to do something, the way we can control variables in a lab. The best we can do is to take the data that’s already been generated and put it in a statistical framework and analyze it.”

Hansen is now turning his attention toward retail competition and the impact of big-box stores on local businesses. “Are there ways they can compete with them?” Hansen asks. “Which customers will they lose to Wal-Mart, and what can local retailers do to hold on to their remaining customers?”
Statistical analysis, which can shed light on shopping patterns, can help to answer those questions, Hansen says.

For example, he says, it is a relatively small percentage of shoppers - 20 percent - who switch from local retailers to Wal-Mart. Yet that percentage can account for a 70 percent overall loss in the local stores’ revenue.

“If local retailers can find a way to retain at least some of these customers, they may survive in the long run,” Hansen says.