According to what I was told, the category "0 reduced class size" represents students who transferred into district, an entirely different population than the other students who were tracked and one which is known to have lower achievement levels. They cannot legitimately be included in a study to observe the effects of reduced class sizes. Rather than assign schools or students to groups the district reduced class sizes uniformly across the district. There are no legitimate comparison groups. This is descriptive data comparing radically different populations used to make statements about causality. For those of you who don't know statistics this is something that should never be done and especially not by such a large organization with its own in-house research group.
Michael Atherton Prospect Park "Anderson, Sue" wrote: > -----Original Message----- > From: David Brauer [mailto:[EMAIL PROTECTED]] > Sent: Friday, September 06, 2002 11:22 AM > To: Anderson, Sue > Subject: Re: MPS Webpage Information > > Sue - can you send this to the issues list? [EMAIL PROTECTED] > > Thanks for looking it over! > > Best, > David > > on 9/6/02 10:32 AM, Anderson, Sue at [EMAIL PROTECTED] wrote: > > > David, > > Although no "statistical expert" (and it's been many years since my > college > > classes), the questions I would have after reviewing the website data are > > these: > > > > + No coefficients of variation are shown in the relatively simplified > > graphs. As these are significantly affected by the number of observations > > (students) in each category, it would be expected that for some groups > these > > might be quite large (small number of students) vs. the larger groups. > This > > could result in "overlapping data" which would indicate no real > statistical > > differences across groups or "years of reduced class size". The large > sample > > size (3022 students) as a whole would probably show the trends, since this > > is a statistically valid sample size, but individual sub-categories (years > > or "race") might not show a significant trend. > > > > + There is no category of "Caucasian" or "Other", so technically a claim > > cannot be made that "All races" benefit, as not "All races" are > represented. > > > > > > + A point made previously in other contexts is that these studies are > > conducted using various methodologies: some seem to track the "same > > students" through the timeframe (in this case 9 years apparently, > > 1990-1999); others seem to track "classes" or "schools" which may be more > > dynamic. There is no indication as to the method used in this study, and > no > > indication of how "random" the sampling was. I would "guess" that there > > would be wide variations in MBST test results between samplings at > different > > schools in the District (ref. recent Star Trib articles and other info > > published by the District). So without more information it's difficult to > > judge if there really is sufficient "randomness" to draw conclusions > across > > the whole MPS spectrum. > > > > + No indication is given as to the age categories this data was taken > from, > > nor is there any analysis of "self-selection factors" which may be > occuring > > over the 9-year timeframe. The starting student number is the same as the > > ending student number apparently. The study indicates it is a > "longitudinal > > study", but there is no additional data. Over the 9 years, if the "same > > group" was tracked, students dropping out at differential rates across > > "races" would tend to impact the data. If it's statistically "random" > across > > schools and district, demographic variations within the MPS students would > > impact the data. > > > > If this research has been published in a more complete form (preferably in > a > > refereed journal), many of these questions may be answered. As it stands, > > the website makes "claims" that wouldn't seem to be substantiated by what > we > > see. > > > > > > Sue Anderson, live in Phillips , work in Fridley _______________________________________ Minneapolis Issues Forum - A Civil City Civic Discussion - Mn E-Democracy Post messages to: [EMAIL PROTECTED] Subscribe, Unsubscribe, Digest option, and more: http://e-democracy.org/mpls
