Posted by Marcia Dick September 22, 2011 10:05 AM
A Somerville activist is contending the city's high-profile "Happiness Survey" over-represents the opinions of the wealthy.
In August, Somerville announced that residents reported their overall life happiness as being 7.7 on a scale of 10. The survey went out over the winter coupled with the annual city census.
Eileen Feldman of the Community Access Project didn't receive a survey herself, she wrote on the blog Somerville Voices. When she looked at the city's raw data—posted for the public in late August at somervilleresistat.blogspot.com, she found an unsettling data point. According to Feldman's calculations, one quarter of the responses came from people who said their 2010 household income was over $100,000.
Median household income in Somerville is about $70,000, SomerStat senior analyst Daniel Hadley told Feldman. And in fact, he said, there was an unintended missing link in the mail survey: It missed some residents who live in large buildings, including public housing projects and retirement homes.
The issue, Hadley said in an e-mail, is a quirk in how the city census forms work: "The main issue is that individual [city] census forms do not go to people in public housing projects."
The Elections Department is in charge of the Somerville city census. It's far less detailed than the federal census and serves essentially as a tool to establish residency for the voter rolls. To keep mailing costs low, the city decided to stuff the happiness survey in the same envelope.
However, for many buildings with eight or more units, the Elections Department simply asks the building owner to provide a list of residents, Hadley said. That's the procedure at buildings owned by the Somerville Housing Authority as well as retirement and nursing homes and the Y.
So when it comes to happiness, "our data was not as good as it would have been had we sent the form to individual households rather than the property owners," Hadley said.
To compensate, analysts performed an additional phone and e-mail survey. "The phone survey was totally random and should have reached the correct proportion of residents in public housing," Hadley said. He noted the e-mail and phone responses were "similar on many of the key questions" to the mail results.
The city received 6,167 surveys via the census form, 360 online, and 200 by phone, according to the final report. The phone calls used a random-number dialing methodology.
Analysts also used standard statistical techniques to weight the mail survey data so responses from more affluent residents didn't have undue importance.
Hadley said he was carefully considering "How to deliver the next happiness survey."