Friday, June 1, 2012

How should we assign weights to various aspects in the OECD's 'better life' index?

The second edition of the OECD’s ‘Better Life’ index is now available. As with the first edition, the index is presented in an interactive form.  Users are able to compare the quality of life in different OECD countries by assigning their own weights to a range of factors that are relevant to assessment of well-being.

When the first edition of the index came out I played around with it to see how different weighting systems and inclusion of additional information on factors, such as perceptions of freedom and corruption, might affect the comparisons. In order to make much difference I had to make some fairly extreme assumptions. My conclusion was that all well-being indicators tend to tell similar stories at a national level.

In this post I want to give some further consideration to the question of what weighting systems might be most appropriate. I want to allow for the possibility that because some of the factors included in the index measure similar aspects of well-being, there may be potential for users to inadvertently give excessive weight to the factors that seem important to them. For example, some people who assign a high weight to life satisfaction might also give high weight to factors that could be expected to have a strong influence on life satisfaction (e.g. jobs and work-life balance). That might mean that they end up giving excessive weight to factors that are strongly correlated with life satisfaction, at the expense of other factors (e.g. education and the environment).

The approach I have adopted in deciding on the weightings was to begin by giving highest weight (5/5) to safety (low crime), life satisfaction and the environment and then to assign weights to other factors depending on whether I consider their influence to be adequately reflected in the factors already covered. I have assigned a weight of 4/5 to income because I think high income adds to quality of life in ways that are not fully taken into account when people are asked whether they are satisfied with life (the value of the future security that a high income can provide). I have assigned a 4/5 weight to education because the contribution of education to quality of life may not be adequately reflected in either life satisfaction or income. Similar reasoning applies to community (covering support networks and volunteering). A weight of 3/5 is assigned to housing, jobs and health because these factors are already taken into account to a considerable extent in measures of life satisfaction and income. I have assigned a relatively low weighting (2/5) to civic engagement because important aspects of civic engagement are reflected in other factors (e.g. safety) and voter turnout seems to me to be a poor indicator of civic engagement in countries with compulsory voting. Finally, I have assigned a rating of 1/5 to work-life balance (rather than 0/5) because it might possibly cover some aspects of the quality of life that are not adequately reflected in other factors.

The ranking of countries according to the weights I have assigned can be found by clicking here. (The OECD offers a facility to embed it, but I am not clever enough to use it.)

The ranking according to my weights differs somewhat from the ranking if equal weights are assigned to all factors. The top five countries under my weights are: Australia, Switzerland, United States, Canada and Norway. With equal weighting, the top five countries are: Australia, Norway, United States, Sweden and Denmark.

However, this difference in rankings doesn’t mean a great deal because the difference in ratings of the top ranked countries is small. That doesn’t surprise me - all well-being indicators tend to tell similar stories at a national level! 


Anonymous said...

Winton, I'm picturing the authors of same feverishly recording all the interactive modelling now being done - by IP address as to country of origin, then by factors considered important. That would be a fascinating follow up, I think.

Also, the safety item links to murder and assault rates? From a personal perspective I'd probably rate that more on the basis of home invasion (if we stick to crime) or (more widely) on food and product legislation/regulation.


Winton Bates said...

Hi kvd
I wouldn't be surprised if someone in the OECD is recording the interactive modelling. I wonder what they would make of the low weight I gave to work-life balance.

It would be interesting to see the results of a properly designed survey of how weighting systems differ for people in different countries. It would not be easy to design the survey to net out the overlap between the factors. Perhaps respondents could be asked to rate the importance of each aspect relative to life satisfaction.

I agree with you about home invasion. Food safety regulation is also important. So isn't food regulation enforced more effectively?

Unknown said...

Hi Winton
In Scotland Oxfam have created a measure of prosperity using weights derived from public consultation - a very tricky process, but important in that it put people at the centre of the process (compared to academics or others making arbitrary decisions as in other such initiatives). We made a particular effort to engage with seldom heard voices.
The results (including description of the methods of the consultation) are available here:

Winton Bates said...

Hi Katherine
I like the idea of asking people what is important to their quality of life. As you are aware there are still many problem. One that concerns me is that different people might have different priorities.

I have observed in a few studies comparing well-being of people in different regions within countries that average levels of life satisfaction seem to be relatively low in some regions with relatively high average incomes e.g. major cities. There may be several reasons for this, but I wonder whether an important factor might be a tendency for people to live in cities because they give higher priority to income at the expense of other factors affecting the quality of their lives. This could be quite rational. For example, they might put up with long commutes, cramped housing etc. in order to achieve greater financial security for their families and hence a better quality of life over the longer term.