A very stubborn gender gap
We’ve seen it over and over again, study after study – when it comes to math, boys perform better than girls. But what’s not clear is why. In recent years, measures have been taken in the U.S. to attempt to close this gender gap, such as the No Child Left Behind law that required schools to report test scores separated by gender, thereby making such gaps more obvious and a target of intervention.
And by some measures the gap has been closing. A 2008 study published in Science found that on state standardized tests, girls and boys performed at the same level. “So the media take-away from that was, there’s no more gender gap in math, everything is better,” says Joseph Cimpian, an Associate Professor of Economics and Education Policy at New York University. He and his colleagues wanted to investigate this further and see if recent interventions and reports of the gap closure were in fact reflected in early age math performance.
Examining the gender gap
To do this, the researchers compared two data sets collected 12 years apart. The Early Childhood Longitudinal Study-Kindergarten Class of 1998-1999 (ECLS-K:1999) measured math ability of boys and girls beginning in kindergarten and retested them each year through Grade 3. The newest cohort, ECLS-K:2011 collected similar data from kindergarteners and available reports include retests through Grade 2.
The researchers compared the two cohorts to see if any improvements could be seen across the 12-year separation, but unfortunately and somewhat surprisingly, that was not the case. “The thing that was really striking about this was the similarity of the patterns across these two cohorts that were separated by 12 years,” says Cimpian. In both cohorts, a gender gap in math performance was seen as early as kindergarten and that gap got worse with each year. By Grade 2, other than students scoring at the bottom 15 percent of the class, boys were outperforming girls at every other level, with girls making up only 20 percent at the top one percent of the grade.
The ECLS-K cohorts provide another interesting measure. Along with math performance, they also assessed the levels to which the teachers believed their students to be performing. And at nearly every performance level, teachers were found to be underrating girls’ performances. When put head to head, girls that were achieving the same as boys on the ECLS-K assessments and had similar classroom behavior were consistently rated below their male counterparts.
“It’s alarming to think that almost as soon as boys and girls enter formal schooling, there’s a bias against girls.”
And while most of the teachers in both cohorts were primarily women, Grade 3 of the ECLS-K:1999 dataset allowed for a male-female teacher comparison. “What we found was that the male teachers were not underrating girls,” says Cimpian, “it was the female teachers that were rating the girls as less able.” And while explaining that finding won’t be an easy one, previous studies – like one that found math-anxious female teachers had an effect on girls’ but not boys’ performances – may help us begin to unpack the relationship between teachers and the gender gap.
Different findings between tests
But how do these findings mesh with those of the 2008 study and more recent ones that find no or shrinking gender gaps? The answer likely lies in the tests themselves. Different findings between state standardized tests and those like the ECLS-K aren’t infrequent, probably because the two types of tests are quite different. First, teachers often teach to the state tests, designing their curricula with the goal of having their students perform their best on the tests. And state assessments are often less challenging than ECLS-K-type tests.
Secondly, the ECLS-K tests also adjust to the student, with those performing well receiving progressively harder questions and those at the lower performance levels receiving easier questions to match their ability. “There are really no ceiling effects with this kind of test and also no floor effects either,” says Cimpian, meaning there really is no limit to how difficult or simple the questions can be, they constantly adjust to measure a student’s ability as accurately as possible.
And as the effects of the new national math curriculum, the Common Core math standards, start being analyzed, it may have an effect on observations of the gender gap. “I think it will be interesting to see what happens with the Common Core assessments because they’re closer to the kinds of problems that you would see on the ECLS-K,” says Cimpian, “they tend to be a little more challenging and of a different format than you would see in a state standardized test.”
Now whether that will produce a gap in state tests or not remains unclear. “Perhaps we won’t see a gap because teachers will be adjusting their teaching practices to the Common Core math assessments,” says Cimpian.
Now, Cimpian and his colleagues are looking into ways to change the teacher biases against girls. During a recent discussion with teachers on this very topic, one teacher told the researchers that she didn’t understand why they were even studying the gender gap since her state tests showed that her boy and girl students were scoring at the same levels. “She said, ‘So the girls can perform just as well as the boys if they try hard enough,’” says Cimpian. But as soon as she said it, she immediately recognized that she had just shown her own bias as did the other teachers in the room. A moment that really highlighted how implicit these biases can be.
“It’s alarming to think that almost as soon as boys and girls enter formal schooling, there’s a bias against girls,” says Cimpian, “and that bias is influencing their achievement and most likely their confidence and interest in this field.” He points out that we look at higher education and try to see what’s pushing women out of science, technology, engineering, and mathematics fields and careers. “And we should look there, we definitely should,” says Cimpian, “but this is also saying we need to look really early on, as soon as kindergarten, maybe earlier.”