Eyeballs and brains are very good at finding
A) Patterns that don't exist
B) Correlations that only exist in the eye of the beholder
- Statistical tests for randomness have merit, eyeballing something and saying 'this looks just like...' has as much merit as something that doesn't have a lot of merit at all. (Apologies to those who didn't encounter UK comedy when that sort of comment meant something).
'Here is a chart that resembles a stock chart, but it was produced by a random process' has little real merit - to produce the chart, for a start, the coin should have been made of a non conductive, non magnetic substance, it should have been 'tossed' by a randomly generated force in a vacuum, and about 30 other serious attempts to ensure randomness should have been made. Secondly the results should be analysed statistically to determine the 'non-random' nature of the result... anyone taking the Open University's Biology course 'Intro to cutting the b*****s off a Marmoset' (or whatever it's called) will be taught the basics of determining whether something is statistically significant or not. In a random poll of T2W members I have determined that the number of members capable of deciding if something is statistically significant is statistically INsignificant. (I'd have to look it up again, and at least I knew it once over!)
Suffice it to say, that I am aware I will be burned as a witch in due course, but you can (with the aid of a 'statisitics for dummies' style reader designed for 16 yr old exam candidates) improve understanding no end by using valid - time tested over multiple disciplines - statistical measurements to decide if something is right or not.
Isn't that better than eyeballing it and saying 'a bit over 6 ft guv' ?
Dave
A) Patterns that don't exist
B) Correlations that only exist in the eye of the beholder
- Statistical tests for randomness have merit, eyeballing something and saying 'this looks just like...' has as much merit as something that doesn't have a lot of merit at all. (Apologies to those who didn't encounter UK comedy when that sort of comment meant something).
'Here is a chart that resembles a stock chart, but it was produced by a random process' has little real merit - to produce the chart, for a start, the coin should have been made of a non conductive, non magnetic substance, it should have been 'tossed' by a randomly generated force in a vacuum, and about 30 other serious attempts to ensure randomness should have been made. Secondly the results should be analysed statistically to determine the 'non-random' nature of the result... anyone taking the Open University's Biology course 'Intro to cutting the b*****s off a Marmoset' (or whatever it's called) will be taught the basics of determining whether something is statistically significant or not. In a random poll of T2W members I have determined that the number of members capable of deciding if something is statistically significant is statistically INsignificant. (I'd have to look it up again, and at least I knew it once over!)
Suffice it to say, that I am aware I will be burned as a witch in due course, but you can (with the aid of a 'statisitics for dummies' style reader designed for 16 yr old exam candidates) improve understanding no end by using valid - time tested over multiple disciplines - statistical measurements to decide if something is right or not.
Isn't that better than eyeballing it and saying 'a bit over 6 ft guv' ?
Dave
Similar threads
