On Tue, 23 Sep 2008 22:48:50 -0700, Michael Sylvester,PhD observed: >I must confess that I never learned how to use that table that >was at the last section of those stat texts.
They also appeared in research methods texts and in other texts like Mike D'Amato's (1970) "Experimental Psychology". >It just seemed like a bunch of meaningless numbers to me. I'm not sure I know what you mean by "meaningless numbers". The number 20 means one thing if it refers to how many dollars you've won on a lottery ticket and it means something else if it refers to the number of minutes you have to prepare for your next class. The meaning of a number depends upon how you use it and the context of its usage ($20 is very different from the 20 on a footbal jersey). >How were those supposed to be used anyway? If you have a copy of D'Amato handy, see pages 25-25 in the section titled "Using a Table of Random Numbers". Short explanation: imagine you have three experimental conditions "A", "B" and "C". Assign numbers to these letters (A=1, B=2, C=3). You need to randomly assign subjects to these three conditions; how do you do so? In essence, you have to come up with a series of 1, 2, and 3s that represent the order of assignment of subjects with the constraint that each number can only appear 20 times. You go to your random number table, starting in the first position and check to see if the number is either 1, 2, or 3 (if it is some other number, skip to the next column, row, or block; presumably your random table is set up so that each of the 10 digits has an equal probably of appearing in all positions). If the number is 1, 2, or 3, then this is the condition that the first subject is assigned to. Repeat this process for the remaining 59 subjects. >I preferred the method of putting names in a salami paper bag, I have never heard of "salami paper". Is it paper made out of salami? >or in a Mexican sombrero or a Cuban campesino hat and >drawing out names with the first name going to the experimental >group and the econd pick from the hat going to the control group. This assumes that you have the names of all of the subjects. In the planning stage of most experiements this is not the case (or if one is working with animals; it's a good idea that you not name your animal subjects for the same reason you shouldn't name the fish/animal you're going to have for dinner) >Did someone really spend time generating a table of random >numbers? Yes. See: http://en.wikipedia.org/wiki/Random_number_table The Rand Coroporation made a project of this and its "A Million Random Digits with 100,000 Normal Deviates" became the source of random numbers for a variety of textbooks (e.g., Dixon and Massey's "Introduction to Statistical Analysis"). See: http://www.rand.org/pubs/monograph_reports/MR1418/ >Must be the ultimate research paradigm. Indeed. Physical processes like tossing fair dice or using a fair roulette wheel or the "bubble and balls" mechanism used for state lotteries and so on will provide true random numbers especially if a large sample is taken but, though one might think that computers can simulate such a process, it turns that this is extraordinarily difficult to do and only "pseudorandom" numbers can be produced (e.g., see: http://www.maths.abdn.ac.uk/~igc/tch/mx4002/notes/node78.html http://www.maths.abdn.ac.uk/~igc/tch/mx4002/notes/node77.html ). Although researchers may use random numbers for random assignment and such, there are many other uses, especially in cryptography; see: http://en.wikibooks.org/wiki/Cryptography/Random_Quality http://www.ietf.org/rfc/rfc1750.txt http://www.std.com/~cme/P1363/ranno.html -Mike Palij New York University [EMAIL PROTECTED] --- To make changes to your subscription contact: Bill Southerly ([EMAIL PROTECTED])
