Missing happiness data

Here is the third installment of my self-important "series" on happiness and the environment (here are the first and second). This post considers the missing data for Vanuatu and other countries that are included in the happy list.

The WSJ’s Numbers Guy digs (er, dug, it appeared 10 days ago) into the Vanuatu’s happiness score and finds it lacking (Putting a number on happiness):

Vanuatu received its designation last week from the New Economics Foundation, a U.K. think tank, in a report laying out its numerical "Happy Planet Index." The index used a formula to quantify the happiness of citizens, using the assumption that a "happy" person is someone who lives a long time, identifies themselves as generally satisfied, and uses an appropriate amount of natural resources. That last bit was included because the think tank was trying to measure the countries that best balanced their citizens' happiness and ecological concerns -- a subtlety lost in some of the press coverage.

The group relied on outside sources for the individual statistics, before applying its own calculations. For instance, the satisfaction scores came from a range of surveys conducted around the globe, most notably the World Values Survey, an international research network which has surveyed people in over 80 countries. The environmental impact numbers originate primarily with an Oakland, Calif., non-profit organization called Global Footprint Network.

In the case of the World Values Survey, there were no data available for Vanuatu. That survey reached fewer than half of all nations, and Vanuatu wasn't among them. Even if the archipelago had been included, it would have been tough to get a representative sample: There is fewer than one phone line for every 10 people, according to the CIA World Factbook.

To fill in that gap, New Economics extrapolated Vanuatu's happiness score from happiness surveys in Africa and Asia, and made some adjustments based on the unique demographics of Vanuatuans.

So following that questionable adjustment, New Economics arrived at this winning formula for Vanuatu: Its people live an average of 68.6 years, are satisfied at a level of 7.4 -- on a scale of 1 to 10 -- and have an environmental footprint of just 1.1. By contrast, the numbers for the sad, natural-resource-hogging U.S. are 77.4, 7.4 and 9.5.

The rankings would have differed greatly had the group sought merely to determine which nations are happiest, rather than coming up with a happiness "efficiency" score that took into account resource consumption. Nic Marks, who devised the report for New Economics, emailed me a spreadsheet that left the resource factor out, and looked only at life expectancy and satisfaction. Suddenly, life looked a lot better in the U.S.: Americans ranked 19th in those terms, two spots ahead of the U.K. and 30 spots ahead of Vanuatu. Switzerland was first, with Denmark second.

So, Vanuatu’s score was “imputed.” Data imputation can be a good idea. Sometimes data can be missing and it can be replaced without screwing up the analysis. For example, if we are trying to understand travel behavior and income is missing, and if it can be predicted in a half-way decent way, then the researcher doesn’t have to toss the observation.

But it almost always a bad idea when the missing variable is the object of study. For example, if income is missing and we’re trying to predict income there is a pretty good chance that the predicted income value will be way off the mark. The Census Bureau imputes income for certain analyses and then it is used by researchers as a dependent variable (sometimes naively). It is usually good practice to toss observations in the data that have missing values for dependent variables.

Vanuatu and other countries that don't have happiness scores should be tossed.

Someday: Part four is on happiness and the environment.