Google

A Social Program That Really Works: California's AB 2034 Program (Integrated Services for Homeless Adults with Serious Mental Illness) as an Intervention with Whelming Positive Effectiveness as Demonstrated by Rigorous Statistical Inference
 

A Social Program That Really Works: California's AB 2034 Program (Integrated Services for Homeless Adults with Serious Mental Illness) as an Intervention with Whelming Positive Effectiveness as Demonstrated by Rigorous Statistical Inference

Posted on Jul 19, 2007 6:36 am PDT  -  Contact the poster  -  Report bad item

Details

Author:   Vernon Lynn Stephens
Publish Date:   Jul 17, 2007
Pages:   6
 

Description

 

Abstract:

California's "Integrated Services for Homeless Adults with Serious Mental Illness" (AB 2034 ) is herein proven-- for the target population of the homeless-mentally-ill-- to have vast effectiveness in the reduction of days-of-homelessness, days-in-jail, days-in-a-psychiatric-hospital, and increased-days-of-work USING INFERENTIAL STATISTICS as opposed to the DESCRIPTIVE STATISTICS used in previously-published evaluations of this social service intervention.

In the attached PDF document, analysis of raw data presented in two subsequent state publications of "Effectiveness of Integrated Services for Homeless Adults with Serious Mental Illness," (2002 and 2003) are subjected to rigorous inferential assay. While it is generally true that almost always with large-scale studies inference becomes obvious from "the look" of tabular (and other) data, occasionally stubborn people DO need to see the full extent to which assertions can be inferred in such work. This condition appears to obtain with the shocking proposal by California Governor Arnold Schwarzenegger to completely eliminate this program in the 2007-2008 state budget-- despite the overweening conclusion which the data will only permit-- that on all index variables-- this program shows a substantial effect.

Using the Chi-Square statistic and derivations (necessary because the data are "off the tables" with significance), I have derived that the probability of obtaining the skew of data obtained in the 2002 report is 7.339E-48600 with size-effect making the signifcance-level run to 8.096E-54288 [the text illustrates the vast number of random trials that would be required to get these results by "happenstance."] With regard to the 2003 evaluation, the probability of this result is 3.365E-240329 with size-effect pushing the significance to 8.281E-71121. This analysis, in other words, clearly demostrates the breathtaking degree to which AB 2034 is achieving its objectives.

[An appendix is included with all the formulae used in these computations-- for the instruction of the uninstructed and the diligence of the diligent!]

 

Contact

Name: freethink@bellsouth.net
Contact the poster
 
 
Attachments