An idea for reanalysis

Started Jul 29, 2013 | Discussions thread
Tim A2 Senior Member • Posts: 1,099
Re: An idea for reanalysis

Sorry my previous post was not worded well. I was in a hurry. My wife was calling and I know which side my bread is buttered on.

The median of a small sample is a more robust statistic than the mean. That is a given. Does that mean the median should be used instead of the mean as calculated by DPR? My guess is yes, but that is a separate issue.

Removing an equal number of high and low votes and then taking the average results in what is called a trimmed mean. To make a long story short, the value of a trimmed mean can be between that of the mean and the median. That is because the more an equal number of a sample is removed from the ends the more the mean can approach the median (if you remove enough they are equal). So the trimmed mean is very commonly used to reduce the effects of outliers (think sandbag votes). A major issue with all this, of course, is the possibilty of removing meat with the fat. If you look at it, though, as a risk vs benefit situation isn't that a moot point in this case?

If anybody here truely know statistics and how to deal with messy samples from messy populations please help us out, but don't expect DPR to pay you a consulting fee.


Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow