Reciprocating Bill
Posts: 4265
Joined: Oct. 2006
|
DaveScot reports on the mental health of Republicans vs. Democrats:
| Quote | Crazy to be Republican? Hardly…
DaveScot
I’ve been reading a lot of claims about how it’s batsht crazy to support Governor Palin for VP. Yet she managed to enchant & energize the Republican party. I thought I’d check to see which party really has more crazy people in it.
November 2007 Gallup poll shows Republicans by a wide margin across all age, gender, income, and education levels report significantly better mental health than Democrats and Independents... |
As one commenter points out, "reported mental health" isn't the same thing as mental health.
More importantly, DaveScot's inability to read statistics is on display. In the original article, the construction of a regression model establishing the independent contributions of each of the variables surveyed is described, as represented in a table of regression coefficients. To obtain the shared variance between a given independent variable (economic status, party identification, etc.) and the dependent variable (reported mental health) one squares the appropriate regression coefficient.
The r value reported for party affiliation is .139. r^2 is therefore .019321. We'll round up to .02 - which indicates that party affiliation accounts for a whopping 2% of the variance of reported mental health in this model. ETA: Gallup reports this as an "independent and highly significant impact of being a Republican on mental health." Bullshit on several counts - it isn't "highly significant" and the dependent variable was not "mental health."
The summary statement "The table shows that income, education, gender, church attendance, and being a Republican are significantly related to self-reported mental health -- each such relationship occurring even when the impact of the other variables is taken into account" repeats the common error of conflating statistical significance with practical significance.
In summary, reported mental health (as disclosed to a stranger through a telephone survey) isn't the same as mental health. And party affiliation accounts for 2% of the variance of this piss-poor dependent variable.
(I also had to laugh at the report that several regression coefficients were "significant at the .000 level.")
Yep, yer a genius, Dave.
--------------
Myth: Something that never was true, and always will be.
"The truth will set you free. But not until it is finished with you."
- David Foster Wallace
"Here’s a clue. Snarky banalities are not a substitute for saying something intelligent. Write that down."
- Barry Arrington
|
