Somerville’s Happiness Survey
National and local governments are increasingly developing happiness measures to enhances their performance measurement and decision-making around resource allocation. Somerville, Massachusetts was an early innovator at the municipal level. This case study reviews the history of Somerville’s approach, the details of their methodology, and important lessons revealed by the survey’s implementation.
Somerville was the first municipality in the United States to include questions specifically targeting happiness and wellbeing measurement in its annual citizen survey. Somerville included the happiness-focused questions for the first time in the 2011 version of its bi-annual city “census”.
The administration of the program is undertaken by the City’s SomerStat division. Somerville’s Happiness Survey is administered by a division named SomerStat. The SomerStat division was inspired by Baltimore’s CitiStat performance management approach and was initiated in 2003 as an initiative of Mayor Joe Curtatone.
Several national efforts influenced the designers of Somerville’s program, including the United Kingdom’s attempt to track wellbeing, as well as Bhutan’s focus on gross national happiness. The survey’s designers consulted academic research and reports. They also sought input from Harvard psychologist Daniel Gilbert, who authored the 2006 best seller “Stumbling on Happiness.” So far, Somerville has conducted the happiness questions in its 2011, 2013, and 2015 citizen census survey’s.
For the 2011 version of the survey, respondents rated their momentary happiness (“How happy do you feel right now?”) on a scale of one to ten, and did the same for their overall happiness (“How satisfied are you with your life in general?”). Responses were guaranteed anonymity.
The City sent the survey to all of the city’s approximately 27,000 households. The response rate was 22% with 6,167 completed surveys in the first year. The City also undertook a more detailed telephone survey, using a randomized sample of Somerville’s 76,000 residents.
These happiness scores were analyzed to uncover the effects of policies and investments in areas such as transportation and parking, education, policing, and residential tree density.
For the 2013 edition of the happiness survey, the SomerStat team made some changes to strengthen the reliability of the results. Instead of a broad blast that undercut the statistical validity of the results, the SomerStat team randomly selected approximately 400 residents for intensive engagement. The random sample of residents was taken from census records of addresses kept by the Somerville Elections Department.
The 2013 edition’s engagement of the selected surveys included:
- An initial mailed two-page paper survey in English, Spanish, and Portuguese.
- A second survey mailing with a two-dollar bill enclosed to spark curiosity and reciprocity. Then they mailed it again, this time with a two-dollar bill enclosed.
- A final reminder postcard encouraging survey completion.
The response rate increased from 22 to 48%, with a total of 194 completed surveys. While the count of responses was lower, the procedure followed increased the statistical validity of the results, though the SomerStat division did have to adjust for the under-representation of the city’s African-American community.
The 2015 edition of the surveys were sent to a random sample of 500 residences in Somerville and were received back with a 35% response rate (185 completed surveys). The random sample of 500 Somerville residences was sent the happiness survey by mail in March and April of 2015. Residents subsequently received a reminder postcard, as well as a “last chance” reminder with additional copies of the survey.
Pointers & Learnings
- Somerville’s implementation was very cost-effective because they tacked on a few questions to existing citizen survey. According to public reports, for the later versions of the survey, the total materials and postage cost was around $4,000.
- It is important to set realistic community expectations about the improved performance surveying happiness will yield, as well as reasonable expectations of the analytical insights. Somerville officials have repeatedly publicly stressed that it is easier to begin measuring happiness than it is to determine which combination of policy levers will best enhance it.
- Think ahead at what policy effects you want to uncover and make sure to properly assign those data attributes to survey responses. For example, Somerville’s analysis found that residents living in the pilot area for a “zero-sort” recycling program were significantly happier with the city’s recycling efforts than those who had to sort paper and plastics into different bins. As a result, Somerville took the zero-sort program citywide. In order to uncover this effect, the SomerStat team had to think ahead an make sure that survey responses could be encoded as being inside or outside the pilot’s geography, while still maintaining the promise of anonymity.
- Engage peer communities on coordinating survey questions and analysis. Having a benchmark of similar communities will help yield additional insight about policy effects. The staff administrating Somerville’s census have indicated they are hoping for additional adoption of the happiness and wellbeing-focused questions to facilitate peer comparisons.
You can review Somerville Happiness Survey Responses from 2011 to 2015 on their open data portal.