On 8/13/03 4:06 PM, "Peter Boucher" <[EMAIL PROTECTED]> wrote:

>
> I went all the way out to 1988 to see what happens on both sides of the
> year 1985, which Tim was interested in (that year's rate being lower
> than in 1976).
>
> 1968 -- 25.76
> 1969 -- 29.68
> 1970 -- 31.84
> 1971 -- 34.20
> 1972 -- 34.66
> 1973 -- 35.38
> 1974 -- 33.32
> 1975 -- 32.32
> 1976 -- 30.74
> 1977 -- 28.56
> 1978 -- 28.30
> 1979 -- 29.96
> 1980 -- 30.54
> 1981 -- 30.82
> 1982 -- 30.96
> 1983 -- 29.36
> 1984 -- 28.54
> 1985 -- 29.64


>
> BTW, IMHO, this data doesn't suggest that there was anything special
> about 1976, either -- it's just smack in the center of a 5 year downward
> trend.
>
> Peter

I deleted 1986-88 because those years were clearly some new phenomenon
occurring (the crack epidemic onset, probably). The average of the years
before 1976 (and it makes no difference whether 76 itself is included or
not) is 32. The average for the years after is 30.

If the fluctuations are anything of a trend (that is, something other than
random--or at least non-patterned), I would read it as one "wave" from 1968
to 1978, with its peak at 1973, and another "wave" from 1978 to 1984 or
1985, with its peak at 1981-1982. The second wave had a lower peak than the
first.

If one had no knowledge of any factors causing the fluctuations, and just
looked at a graph of the data, I doubt that one would select 1976 (or 1977)
as years of any particular causative significance. The "flex" years--i.e.,
the years at which the trend changes direction--are 1973, 1978, and 1984.

If no analysis of other potentially causative factors is done, I think it's
entirely defensible to say that these raw data don't support the contention
that an event in 1976 or 1977 caused a decrease--certainly not any *lasting*
decrease.




--
Bob Woolley
St. Paul, MN
[EMAIL PROTECTED]


A government which robs Peter to pay Paul can always depend on
the support of Paul.

   --George Bernard Shaw

Reply via email to