2 ms·
I suggest people actually look at the plots in the paper. As an example, on P55, they show the effects on subjective well being for many metrics (life satisfact
by BeetleB 8y ago
I suggest people actually look at the plots in the paper. As an example, on P55, they show the effects on subjective well being for many metrics (life satisfaction, loneliness, etc). Note that a number of these metrics have a large enough standard deviation that it spans both the positive and negative range.
Most news outlets focus only on the mean (most of which are positive). But when your standard deviation is large enough to change the sign, don't put much stock in the outcome. If there's one thing that was hammered into my head when I took statistics, it's that a confidence interval of (-1, 5) does not mean the true value is more likely close to 2 (or positive, or whatever). There's no valid reason to focus on the positive values more, when your confidence interval crosses the zero point.