The green curve is a loess curve, which simply smooths the relationship to indicate the basic trend.
The fact that HDI increases with Unbelief percentage does not mean that greater unbelief necessarily causes greater HDI; the causality may run the other way, or both variables may be caused by some other variable, or there may be complicated feedback between the two variables, and probably several other causal factors (which is what I would imagine is the truth).
The identified countries, clockwise from top left, are United Arab Emirates, USA, Ireland, New Zealand, Australia, Norway, Sweden, Estonia, Azerbaijan, Mongolia, Vietnam, Zimbabwe, Dem.Rep.Congo.
2 comments:
I see the possibility of two different processes at work here. If we look at the way the curve goes, it's fairly steep till unbelief = .25 or so. From .25 on, the data would appear to be fit very well by a straight line. The whole thing could be fit by two straight lines, one for the data up to unbelief = .25, and another for unbelief > .25. There's at least one point of inflection, possibly two.
Though, spotting points of inflection in a curve with this much scatter in the data is rather tricky, to say the least. I worked with curve-fitting software a few years ago on a project in which we found data which could be fit best with two curves joined at a point of inflection. However, the data had very little scatter around the curves we fit, and there was a plausible physical reason for the point of inflection (the data were from a physical process). One doesn't typically get this tightness of fit with social and psychologcial data.
Info is out of this world, I would love to read more.
womans health magazine
Post a Comment