Can test scores change?

Introduction: Researchers tested whether reminding people of a belief that their group does poorly on a test affects how well they perform.

Participants: 120 women at a large public university who had completed a calculus course and rated math as important to them. Their mean age was 19.6. They reported their race/ethnicity as 48% White, 22% Asian, 17% Hispanic or Latino and 13% Black.

Method: Each participant took the same difficult 20-question math test alone. Participants were randomly assigned to read one of two descriptions before starting. Condition A said: 'This test has shown differences between men's and women's scores in past research.' Condition B said: 'This test has been shown to be fair; men and women have scored equally in past research.' Afterward, participants were told the true purpose of the study.

Results and Discussion: Women in Condition A answered a mean of 41% of the questions correctly, compared with 58% for women in Condition B, a statistically significant difference. Women in Condition A also reported more worry during the test.

Introduction: Researchers tested whether a short online lesson about the brain could change students' grades.

Participants: About 6,300 ninth-grade students at 40 public high schools across the United States took part with parental consent. The schools served a wide range of communities, and 43% of the students came from low-income families.

Method: In the fall, students were randomly assigned to complete two 25-minute online lessons. The experimental lessons taught that the brain forms stronger connections when people work on challenging problems, and that ability can grow with effort and good strategies. The control lessons taught facts about brain functions without saying anything about whether ability can change. Researchers collected each student's end-of-year grade point average (GPA) on a 0–4 scale.

Results and Discussion: Students who completed the experimental lessons earned slightly higher GPAs than students in the control group (a difference of 0.05 points; statistically significant, with a small effect size of 0.06). Among students who started high school with lower grades, the difference was larger (2.48 versus 2.37). Students in the experimental group were also more likely to sign up for harder math courses the next year.

Introduction: Researchers examined average scores on the same reasoning test across several generations in one country.

Participants: Test records for about 500,000 18-year-old men who took a reasoning test as part of required military registration in one European country between 1952 and 2002. Women were not tested.

Method: Every year, registrants took the same nonverbal reasoning test. Researchers rescaled the scores so that the 1952 average was 100 and compared average scores across years. They also gathered national records on schooling and average adult height, a common sign of childhood nutrition.

Results and Discussion: Average scores rose steadily (see the table). Because a population's genes change very little in 50 years, the researchers concluded that changes in the environment, such as better nutrition, more years of schooling and higher family incomes, were the most likely causes.