🚀 go-pugleaf

kym@kymhorsell.com wrote:
[...]

Here's the kind of thing you can find with google searching.

The EU gather stats on dietary salt consumption (as do many countries
checking dietary iodine in salt -- an Aussie invention I understand).

You can run an OLS against life expectancy and see if there's a
statistically significant link. Even on this randomly-selected
data there is such a link.

Here's the output:

 eusaltcons: table I Zbigniew Szybiski et al.
 eusaltcons: salt cons g.dat
 life.dat:  highest life exp
 life.dat:  1990-95
 life.dat:  the economist pocket world in figures p64
 eusaltcons vs life.dat
 NOTES:
 52 missing x values discarded
avx	9.250000
sdx	1.705137 (= 18.433918% of avx)
rangex	7.900000..13.500000 (= 60.540540% of avx)
avy	75.500000
sdy	1.414214 (= 1.873131% of avy)
rangey	72.000000..77.000000 (= 6.622517% of avy)
 2 (25.0%) points > 1 \sigma
y = -0.778166*x + 82.698
sdx	4.82286
rss	1.9153
se	0.564993
r	-0.938246
limits for beta at 90.0% CI
tc = 1.94317 at 6 d.f.
beta in -0.778166 +- 0.22764 = [-1.00581, -0.550525]
T-tests on beta:
H0 beta == 0.000000 against H1 beta != 0.000000
calculated t = -6.64253 at 6 d.f.
|t| > tc (1.94317 2-sided); reject H0
H0 beta == 0.000000 against H1 beta < 0.000000
t < tc (-1.43975 left tail); reject H0
Probabilities:
P(beta!=0.000000) = 0.999438
P(beta<0.000000) = 0.999719
limits for alpha at 90.0% CI
tc = 1.94317 at 6 d.f.
alpha in 82.698 +- 2.14115 = [80.5569, 84.8392]
r2 = 0.880307
tag                          x         y        yp
france                     7.9        77   76.5505
germany                    8.2        76   76.3171
denmark                    8.2        76   76.3171
belgium                    8.3        76   76.2393
portugal                     9        75   75.6945
italy                      9.2        76   75.5389
uk                         9.7        76   75.1498
poland                    13.5        72   72.1928

This shows that there is a variation in life expectancy across
the countries that has a 88% corressponding variation  in
dietary salt consumption.

It finds that for each g of salt consumped per day there is
a corresponding reduction of .77 years of life expectancy.
The probability that this -ve correlation is due to chance
alone is almost 0. (I.e. Pr(beta<0) is 99.9%!)

Since this is aus.science we can't leave even this simple test
there.

The UK is an outlier in the data. It may be skewing the results.
So let's remove it:

avx	9.185714
sdx	1.813782 (= 19.745685% of avx)
rangex	7.900000..13.500000 (= 60.964233% of avx)
avy	75.428574
sdy	1.498189 (= 1.986235% of avy)
rangey	72.000000..77.000000 (= 6.628788% of avy)
 3 (42.9%) points > 1 \sigma
y = -0.797121*x + 82.7507
sdx	4.79881
rss	1.08096
se	0.464963
r	-0.964963
limits for beta at 90.0% CI
tc = 2.01505 at 5 d.f.
beta in -0.797121 +- 0.195241 = [-0.992362, -0.601879]
T-tests on beta:
H0 beta == 0.000000 against H1 beta != 0.000000
calculated t = -8.22695 at 5 d.f.
|t| > tc (2.01505 2-sided); reject H0
H0 beta == 0.000000 against H1 beta < 0.000000
t < tc (-1.47589 left tail); reject H0
Probabilities:
P(beta!=0.000000) = 0.999568
P(beta<0.000000) = 0.999784
limits for alpha at 90.0% CI
tc = 2.01505 at 5 d.f.
alpha in 82.7507 +- 1.82806 = [80.9226, 84.5788]
r2 = 0.931152
tag                          x         y        yp
france                     7.9        77   76.4534
germany                    8.2        76   76.2143
denmark                    8.2        76   76.2143
belgium                    8.3        76   76.1346
portugal                     9        75   75.5766
italy                      9.2        76   75.4172
poland                    13.5        72   71.9896


The fit now is even better. 93% of the variation in
salt consumption has a corresponding variation in life
expectancy. I.e. for each 1 g of salt consumed per day
there is .79 less years of life expectancy.

As I said, this is only a very cursory "study". :)

But there are many many similar studies around. You can do any similar
thing yourself to check whether you can find evidence there is
or is not a problem with excessive levels of salt consumption.

Using data on heart attack, stoke, cancer or various specialisations
of these (e.g. stomach cancer in particular) you can narrow down
what may be the specific problems that cause life expectancy to
have the above prima facie link. 

-- 
R Kym Horsell <kym@kymhorsell.com>

If your ideas are any good you'll have to ram them down people's throats. 
  -- Howard Aiken