You have 50 approximately normally distributed samples, both measuring a value between 0 and 100, and two methods of measuring them.
One of the methods (let's call it the Election Result Method, ERM) is supposedly perfect as it gathers all of the data. The other is a sampling of the data (let's call it the exit poll method, EPM) which has a margin-of-error about plus or minus 1 percent*.
In thirty of these cases (the anomolous cases), the ERM yielded a result outside the confidence interval of the EPM. Twenty-six of the cases gave a value above the confidence interval, while 4 of the cases gave a value below the confidence interval.
Missing info includes:
1) What confidence interval was used in determining the margin of error?
2) How many and which of the 30 anomolous cases resulted in a change between 49 and 51%
3) Probably most important--is there a public list of all the data involved somewhere to torture a 500 level stats class with?
2007-06-03
09:05:24
·
3 answers
·
asked by
Jon
3
in
Politics & Government
➔ Elections
The company that collected the data was called Edison Media Research.
Edison-Mitofsky performed an independent analysis of the data:
http://uscountvotes.org/ucvAnalysis/US/USCountVotes_Re_Mitofsky-Edison.pdf
But is the data publicly available?
2007-06-03
12:56:28 ·
update #1