Health of Mind and BodyPhysical illness is measured in Britain and the United
States by the number of physical illnesses or conditions
(like diabetes, angina, stroke, asthma, arthritis, etc.).^7 In Aus-
tralia it is measured by the physical components of the SF36
questionnaire, entered with a lag.
To answer our original question we can turn all the con-
tinuous variables into discrete variables. So in Table 6.1 we
examine the effect of eliminating
- Poverty (defined as the lowest 20% of incomes)
- Unemployment
- Physical illness (defined as the lowest 20% of physi-
cal health), and - Mental illness (defined as in the text above, or in
the BHPS as the lowest 20% on the GHQ- 12).
None of these variables is the same as low life- satisfaction,
but all of them contribute to it. Let us see by how much.
The approach in Table 6.1 is simple.^8 In the first column
we ask by how much a person’s probability of misery is in-
creased if they have each characteristic like poverty or de-
pression, other things equal. The numbers are regression co-
efficients estimated by OLS with all variables shown being
dummy variables (1, 0).^9 In the second column, we record
what proportion of the total population have the characteris-
tic in question. In the third column we multiply the effect of
the characteristic by its prevalence, which provides the answer
to our original question: How much misery would be eliminated
if we eliminated the characteristic in question (ceteris paribus)?
The results are remarkable. In the United States, a person
with diagnosed mental illness is 0.10 points of probability
more likely than otherwise to be in misery. Of the total pop-
ulation, 22% have diagnosed mental illness. So if there were