--- In [email protected], "L B Shriver" <[EMAIL PROTECTED]> wrote: > > Whoa, too embarrassing! I predicted, but NOT in print. But in a way, I did, too. > > Here is the quote from Survival In Paradise (Thursday, August 26, 1993): > > "My own prediction for the course was characteristically conservative. [Yes, it would have > been more impressive to have lodged it with an independant monitoring oranization, or at > least to have put it into print, but since the paper didn't come out until after the course, > this was the best I could do.] I anticipated that the results of the course would be mixed: a > reduction in crime statistics of about 17%, enough so that the Movement would at least > save face, or would feel internally vindicated for its efforts, but not enough to clearly > convince others that we had accomplished the equivalent of the Eddington expediition." > > Just so the chronology is understood, etc: in this publication, soon after the course, I was > admitting that I had apparently low-balled the stats. This was because the first reports of > the Movement were claiming 25%, or thereabouts. Then they scaled back to > 17%, or thereabouts. Then they started massaging. > > My impression, based on a few conversations with my friend Mark _____ (last name > forgotten) was that they were digging around for anything they could toss into the > equations to beef up the percentage of reduction. It was somewhat reminiscent of the > previous discussion of the problem about restoring the claim that the rest during TM is > twice as deep as the rest during sleep. > > To recap: my guesstimation was indeed intutional, and based on the circumstance that the > pundits were not actually going to appear for the course and the numbers were going to > be a bit low. Initial, preliminary claims of 25% were downgraded (although THIS did not > happen until after I published, making my "after the fact" prediction somewhat prescient). > Then they massaged, and got the stats back up. > > By the way, I don't remember any sudden, sharp drop in crime stats in DC. Actually, there > was a mass gang banger murder at a swimming pool during the first week or so, and > violent homocides were probably up. > > Anyone else remember about that? > > L B S > > > > > > 5 > > --- In [email protected], akasha_108 <[EMAIL PROTECTED]> wrote: > > > > --- In [email protected], "L B Shriver" <[EMAIL PROTECTED]> > > > They spent a LONG TIME massaging the data after the study. They had > > predicted 25% > > > reduction in crime. Their first published result was 16 or 17%, > > which was EXACTLY WHAT I > > > HAD PREDICTED, IN PRINT, PRIOR TO THE COURSE. > > > > On what basis? Analysis? Guesstimate? > > > > > > > > Then they when back to the massage table, and eventually they > > claimed 25%. I knew one > > > of the guys who was doing the massaging, a graduate student named > > Mark_____ (last > > > name forgotten, for the moment.) > > > > > > ok, now its starting to make more sense. I would guess that the > > original study used ARIMA,testing over say a 4 month period, 1 month > > pre and most, 2 months course, and showed the 17% drop. > > > > Though this would be curious methodologically in that ARIMA models > > typically require 5-6 seasons of data to reliable. > > > > > > THEN, later, when they decided that they wanted to test for weather > > effects, I am guessing that a 3-5 year regression model was used to > > control for weather in predicting crime rates. No ME variable. They > > then used this estimated weather normalized crime rate to replace the > > pre and post months data in the ARIMA model. > > > > If thats what they did, it is funky - see adjacent post. They could > > have adjusted for weather in the ARIMA model, unless they were using a > > very simple bi-variate form of the model that only allowed testing > > the intervnetion variable (ME) against crime. If so, thats weak > > methodology. A multivariate form of the ARIMA (or regression) should > > have been used to control for weather and all the other crime factors. > > > > And its been stated over and over that they did test for multiple > > crime factors. Could it be that they simply tested all the independent > > crime variables ONLY in the longer term weather/crime model, and NOT > > in the actual impact analysis in the ARIMA model? If so, it smells so > > funky, very band-aidish. And prone to lots of hidden analyst > > "discretion" (cough) (cherry-picking). > > > > I am beginning to wonder if the absence of copies of the study are > > intentional - to hide such funky analysis (if above is what they did > > -- though thats the only expalanation of their methodolgy that fits > > the bits and pieces people have contributed.) > > > > They should put the data out on a web site and let analysts go at it. > > Using methodologies of their choice. The reason they have not made the > > data accessible speaks volumes. One only hides data, when it won't > > hold up to independent analysis. > > > > Or anyone could reconstruct the dataset. Crime data, socialogical > > variables, weather data, etc are all accessible -- though it would > > take a bit of effort to dig it out for near 20 years ago. And the ME > > participation numbers should be accessable -- or recallable by > > participants. > > > Murder stats were up by 2x the average one weekend and down by 1/2 the average the next due to that gang thing. All in all, murder was slightly elevated for the entire period. The other three "violent crime" indicators were reduced(in absolute numbers, IIRC). The weather pattern thing enhanced the findings to significance for 3 indicators but really didn't touch the murder rate, again IIRC.
------------------------ Yahoo! Groups Sponsor --------------------~--> Get fast access to your favorite Yahoo! Groups. Make Yahoo! your home page http://us.click.yahoo.com/dpRU5A/wUILAA/yQLSAA/JjtolB/TM --------------------------------------------------------------------~-> To subscribe, send a message to: [EMAIL PROTECTED] Or go to: http://groups.yahoo.com/group/FairfieldLife/ and click 'Join This Group!' Yahoo! Groups Links <*> To visit your group on the web, go to: http://groups.yahoo.com/group/FairfieldLife/ <*> To unsubscribe from this group, send an email to: [EMAIL PROTECTED] <*> Your use of Yahoo! Groups is subject to: http://docs.yahoo.com/info/terms/
