Friday, February 6, 2015

Insanity of the Marzano Evaluation Redux...

My "Insanity" post re: the Marzano evaluation system seems to be the only "popular" post I have so I thought it might be time to revisit it--this time in fewer words.  I encourage everyone to sign up at his website to access the database themselves.  Then go through and take a look.  Before we begin, I urge you to keep in mind the GIGO paradigm:



Sorting by p, you quickly see a lot of data that seems to have no value reported; I'm fairly certain that with a sample size of 8, a p of <0.0000 is not going to happen. 


 And, eyeballing the scrollbar, it would seem that ~66-75% of Marzano's database is statistically INsignfiicant.  ("...in Statistics 'significant' means probably true (not due to chance)") Combined with the portion above (all those zeros...), I personally find this to be, well, damning, for lack of a better word.



Now, before putting in these last two pictures, I would like to once again quote Marzano's description of the p values:  "Basically, if the value in this column is less than .05, the effect size reported for the study can be considered statistically significant at a significance level of 5% (that is, the probability of observing such a result by chance is less than 1 in 20). In other words, a reasonable inference can be made that ... the reported effect size represents a real change in student learning."

Behold the "better than a coin flip" point:


And, this is definitely my favorite:  Seven results that are 100% likely to be random fluctuations. 



It's the epitome of hypocrisy that in this day and age of "data driven decision making" we're relying on this type/quality of research to make decisions.  It's like our politicians and educrats have never even heard of the phrase "peer review."  Again: 




To Marzano's credit, he's stated his research is being misused.  But I don't see him making a big fuss in telling states like Florida to stop basing their evaluations on his work, i.e., stop buying his books and materials.



BONUS:  This is one of my favorite pictures, when you sort it by N (C), that is, the size of the control.  You used a study that had ONE kid as a "control"?  Really?  I mean, REALLY?  You don't need to be a statistician to realize that public policy should not even be influenced, yet alone based on this analysis.




No comments:

Post a Comment