Advertisement

Student Leaders Question Randomization Study

Minority student leaders are attacking the methodology used in an introductory-level statistics project released last week that found upperclass houses to be racially imbalanced in a way that matches stereotypes of those houses.

In compiling their final project last spring for Statistics 100: "Introduction to Quantitative Methods," Nienke C. Grossman '99 and Mark F. Veblen '99 examined a sample of facebook pictures and assigned students to one of the three racial categories: "white, African-American or Asian/South Asian/Indian/Near Eastern."

Student leaders not only dispute the project's premise that it is possible to determine the race of an individual by looking at a facebook picture but also criticize the first-years' failure to include Hispanics and Native Americans as racial categories.

"Race and ethnicity in this country are so important to personal development because it reflects so much of people's culture and history," said Jessy J. Fernandez '99, the co-president of Minority Students' Alliance.

"When you look through a facebook and assign them to one race, you are classifying them in a way that may not be fair," said Fernandez, who is of Indian descent.

Advertisement

Sandy R. Santana '97, president of the Dominican Students' Organization, agreed.

"You have to understand that race is a social construct," he said. "The study is not right--any study done today should ask people how they want to classify themselves."

The student researchers said Dean of the College Harry R. Lewis '68 rebuffed their requests for the pertinent data. Lewis has said he does not want to release such data until the houses become fully random in the fall of 1998, saying that to do so would simply confirm stereotypes.

Lewis has refused to confirm or deny the figures provided by the students.

The students' report acknowledges the difficulty of identifying race by pictures but argues that this problem should not affect the final results unless there was "systematic bias in identifying race."

But Professor of Statistics Carl N. Morris said even the most well-intentioned researchers are sometimes affected by unconscious bias.

The professor challenged those who contest the study to try to replicate it.

"We would have to get someone who is blind to the knowledge of the houses [to get completely unbiased data]," he said. "[Critics] should get and MIT student...[to] repeat the study."

No Hispanics

Other students, such as Asian American Association (AAA) Cultural Co-chair Ellie Kim '98, criticized the lack of inclusion of Hispanics or Latin Americans in the project.

Hispanics comprise about six percent of the Class of 1999, while Native Americans make up about one percent, according to figures released by the admissions office.

Grossman acknowledged that Hispanics were not included but said the omission was intentional, not an oversight.

"We actually made a conscious decision not to include Hispanics because they are both black and white," said Grossman, who is Hispanic.

But Santana said Hispanics could not simply be classified as one race or another.

"It is just insensitive to categorize Latinos as either black or white," said Santana, who is Latino. "That shows an ignorance of who Latinos are as a community. Many consider themselves white; many would consider themselves black; many consider themselves neither."

Veblen, who is white, said it was impractical to try to pick out Hispanics by facebook pictures.

"From a visual perspective we didn't feel that we could get any kind of accurate measurement. So we concluded that it made more sense not to do it at all," said Veblen.

According to figures provided by the admissions office, nine percent of the Class of 1999 is African-American, and about 18 percent is Asian.

Grossman and Veblen's report stated that the class is 9.5 percent African-American, and 21.8 percent Asian.

Despite what they acknowledge are imperfect estimates, the students who compiled the project said their work has some informational value.

"Our goal was to look at the house populations and see how different they are," Grossman said. "You could still tell from the data we did get that there was a significant difference between the houses. We hope the masters will rejoice in this diversity and provide what is needed in each of the houses."

But AAA Co-President Sharon W. Gi '98 said invalid numbers may be worse than no numbers at all.

"I think it is a good thing to have this information out there," she said. "But if you have incorrect figures it might be more harmful.

Recommended Articles

Advertisement