help with Running regressions on wellbeing?

Dancingjebus3

Need help with running some regressions on wellbeing, I plan to try show which of the following are the best indicators of happiness/wellbeing: Human development index, Gdp per capita, consumption and possibly a fourth(open to suggestions)

The data is coming from: http://hdrstats.undp.org/en/indicators/
http://pwt.econ.upenn.edu/php_site/pwt63/pwt63_form.php

Data for happiness is coming from http://www.worldvaluessurvey.org/

My problem now is finding out how to actually successfully run a regression for what I want to do, (another issue i found is dealing with the data from WVS as it has negative values in it as happiness level is measured between +5 and -5) If you have a suggestion to which countries i should use to measure these please let say.

Any help on how I can successfully do this really would be greatly appreciated.. If any more information that I could provide would help please say so and ill add it as soon as I can..
Thanks in advance DancingJ
Feel free to PM if you like..

I'd check out the repec page of Andrew Oswald, he has a lot of work out on this topic.
http://ideas.repec.org/e/pos21.html

These people also work in the area:
http://ideas.repec.org/f/pma305.html
http://ideas.repec.org/e/pdi128.html

Time Magazine

Dancingjebus3 wrote: »

My problem now is finding out how to actually successfully run a regression for what I want to do

What statistical package are you using?

another issue i found is dealing with the data from WVS as it has negative values in it as happiness level is measured between +5 and -5)

For undergraduate econometrics, this doesn't matter an awful lot. However you might consider changing your regression from OLS to Ordered Probit. If this sounds familiar: good, look into it. If it doesn't: it's probably too advanced.

If you have a suggestion to which countries i should use to measure these please let say.

You should probably restrict to advanced Western economies.

Any help on how I can successfully do this really would be greatly appreciated.. If any more information that I could provide would help please say so and ill add it as soon as I can..
Thanks in advance DancingJ
Feel free to PM if you like..

1. You are going to have a serious problem of multicollinearity. Consumption, GDP and HDI are so highly correlated (you're talking along the lines of [latex]\rho[/latex] = 0.95 here) that it will be hard to exactly tease out what matters most.
2. You will need to include controls for age, marital status, health, etc.
3. You might consider changing the "what matters most" bit to health or wealth. It's a bit more interesting (imho) and is more econometrically valid.

Flamed Diving

Very useful data here:

http://www.qog.pol.gu.se/

See the codebook:

http://www.qog.pol.gu.se/data/QoG_Codebook_vJune09.pdf


It's a bunch of various datasets that were used in published works. Covering mainly socio-political issues, but they are quantitative and most were used in economics papers, I would give it a look. There are a lot of potential ind variables in there. However, I would recommend choosing just one dataset, instead of picking variables from many. If you are an undergrad, just use the Economist Intelligence Unit dataset. It should be fairly straightforward to use.