 |
 |
The old saying about figures don't lie but liars can figure is exemplified by your message beautifully. My professor in Statistics gave a good example of false conclusions - he demonstrated from available data that there was a 100% correlation between the increase of salaries of teachers in New York State and the increase in sales of liquor over a period of a few years. Conclusion ? One possible - if you want to increase sales of liquor then give the teachers a raise. Bob Anibal -----Original Message----- From: Baldwin Robertson <[log in to unmask]> To: Multiple recipients of list PARKINSN <[log in to unmask]> Date: Friday, October 24, 1997 2:55 AM Subject: Re: Response to PD Survey >Mary, > >I responded to your survey as soon as I read it because it took me only a >few minutes. I did it, however, even though surveys of this kind are >usually not reliable. It is not that the questions you asked were >deficient in any way, and I believe you would be objective in evaluating >the answers. > >It is because retrospective studies are usually not a reliable way to >establish correlations. I say usually because if everyone on the list >answered and gave the same answer, you would indeed have uncovered >something of value. But, results seldom turn out that way. Usually the >correlations one might observe in retrospective studies have no validity >and are useful only as a suggestion of one thing to look for in a more >reliable subsequent study. > >This point was made an article by William Feller in the throwaway >periodical 'Scientific Research' many years ago. I remember it well >because it impressed me so much. Feller is the author of the book for a >course on probability that I took. He was the first to present the subject >in the now accepted way. The book went through a number of printings. He >is well respected. > >His article in 'Scientific Research' was short and to the point. He >probably chose that magazine because he wanted as many people as possible >to get the message, which no doubt is widely known by probability theorists >and mathematical statisticians, but seems frequently to be ignored by >epidemiologists. > >Feller said that if you consider 100 uncorrelated random variables >generated by a computer and calculated the correlations between all pairs >of them, you would find at least one pair with a convincingly large >numerical correlation. I do not recall the exact number, but I think it is >something like 25 per cent or more, maybe even 50 per cent. Statisticians >would say that value of the numerical correlation is 'statistically >significant.' There are also a surprisingly large number of pairs for >which the calculated correlations are large enough that someone who didn't >know the source of the data might say that there is convincing evidence of >a correlation. Again I do not recall the exact numbers, but I think they >are in the 5 percent to 10 percent range. All this for random numbers with >absolutely no real (i.e. causal) correlation. > >I am not an expert in statistics, and so I do not have definitive >information to give you about this. I certainly am not skilled in >designing a careful study. But I think Feller's point leads one to be >cautious about the interpretation of retrospective studies. > >Baldwin, 63/4 >8x25/100 carbi/levo-dopa, 5 mg Eldepryl per day >[log in to unmask] > > >---------- >From: Mary Sheehan >Sent: Thursday, October 23, 1997 11:06 PM >To: Multiple recipients of list PARKINSN >Subject: Response to PD Survey > >My PD started 11 years ago with a frozen shoulder. Bradykinesia and >micrographia followed, but the doctors attributed my symptoms to the >effects >of the frozen shoulder. It wasn't until 21 months ago that I was diagnosed >with PD. After 11 years with PD, I am still in stage 1. I take only 5mg of >Eldepryl daily and have recently started to take Mirapex. > >I am telling you this so you will understand my motivation for doing the >survey. I have been amazed at the different rates at which PD progresses, > the different symptoms in each person and the different ways that each >person is medicated. I came up with a premise that the progression and >symptoms of PD must relate to either personal lifestyle or to how PD >presents >in an individual .... or possibly to a combination of both. I had hoped to >use the resources of this list to find a pattern that made the progression >of >PD more predictable. I have done some successful research in the past, >although not in the medical field, and felt there were clues out there >waiting to be uncovered. > >Unfortunately, with only 18 responses, it was impossible to draw any >conclusions. I had hoped for 100 responses, or more, out of the 1500 >people >on the list. What really upset me was the responses i received as to why >people didn't respond. >" I'm too busy to reply". The whole survey couldn't have taken more than 5 >minutes to fill out. Most questions required only one-word answers. >"I'm tired of all the surveys on this list." In the year and a half I've >been on this list I recall only two persons doing a survey. I answer >surveys >in the small hope that maybe something will come out of it that will help. >"The PD organizations can"t get together". So let them do their research >separately. I have great respect for the researchers at the Parkinson's >Institute and for my neuro who also does research and whom I thought would >be >interested in what we had to say. > >I'm sorry to be so wordy. I put in several hours of work on the survey and >was annoyed to find my efforts denigrated. I'm probably only having a bad >day, but I'm disappointed because I honestly thought we might put our heads >together and discover something. I will shut up now and in the future will >avoid initiating surveys . Thanks for letting me vent. > >Mary