In a message dated 98-09-02 13:32:34 EDT, you write:
>>Even by that standard, then, they've overreached in deducing causation.
>>Neither the family communication-usage nor depression-usage correlations
>>changed significantly from pre to post, although loneliness probably
>>did.
They're claiming that since psychological state at the beginning did not
influence internet use but internet use did influence later psychological
state
then interent usage caused psychological state, not the other way around.
But, since they're comparing two correlations, the appropriate test is whether
the two correlations are statistically different. Even if the two
correlations are
sampled from the same distribution, you will frequently find one significant
and the other not when the difference is marginally significant.
>They went in and caused internet access for these people, so they presume
>that any change they then see in soc/psych is caused by internet access,
>rather than the other way around.
Well, they're not comparing internet users to non-users. They're comparing heavy users to light users, and the scientists didn't determine that; each participant determined their own usage. So conclusions on causation on that basis are just post hoc, ergo propter hoc.