How Do You Measure Success?

How Do You Measure Success?

Are you looking this product? Now you can get product in PDF Format,just following step by step until finish you will be guided downloading this book for free, Enjoy it.


.====> DOWNLOAD FOR FREE <====
Please follow instruction step by step until finish to get this for free. Enjoy It !!

I work with numbers and variables quite a bit, in both prevention services and healthcare.  In case you're not sure what I mean by variable, it means something you can measure which has changing values over time, place and with different people.  It is often used to measure success.  For instance, with one client I have we are successful if we can drive down the average number of times students surveyed at a local high school drink in a 30-day period.  We typically work with a school 2 or more years and measure our success each spring before spring break.

So, how do you measure SUCCESS?  (This reminds me a bit of the old commercial phrase-How do you spell relief?  ROLAIDS.  Maybe you're too young to remember this, but I'm not.)  Perhaps you measure success with a variable which is easy to measure and track.  Beware of this!  The State of Michigan is measuring its success in driving down teen alcohol use and heavy adult drinking by measuring deaths and accidents which involve alcohol.  This is part of SPF/CIG program.  It doesn't take long for me to find all kinds of problems with this.  For instance, suppose the teen isn't driving yet?  How can you measure the alcohol use of a 13 year old?  By the way, the average age a teen has his/her first drink is 13 and heavy adult drinkers typically start drinking before the age of 13. 

Another example of a poor indicator variable is measuring physician performance by the percent of patients with hypertension who keep their chronic disease under control with prescribed medication.  There are several possible flaws with this.  Most certainly there are confounding variables which make comparing two or more physicians in this area difficult.  One factor could be that one physician's patients are generally well educated and have good incomes.  Another physician might have as clients mostly Medicaid patients with generally low incomes or with poor reading skills.  Almost certainly the former physician will have a higher percentage of patients in compliance.

These are just examples of indicator variables which have been poorly chosen to measure success.  Let me give you an example of a well-chosen variable.  I would like to return to the client that I am working with in alcohol and drug prevention, Dr. Nancy Harper.  Dr. Harper directed a program to drive down frequent heavy drinking and occasional heavy drinking at Grand Valley State University from 1999 to 2006; she used a social norms theory.  In that time she was able to drive down frequent heavy drinking by nearly 50%.  Since this was the only project on campus to drive down heavy drinking, we are fairly certain that it was her campaign that made the difference.  You might object that it could be a confounding factor, such as change in the types of students that attended the college over this period.  In fact, the admission to the school became much more difficult over this period.  Yet, there are good reasons that indicate that it was the program.  One, college heavy drinking has increased significantly over this period, especially among women, as alcohol advertising and alcopops has targeted this age group.  This has also been true of some nearby colleges.  This type of comparison is known as benchmarking.  Another factor that indicates that this program was responsible for the changes is that it is based upon similar evidence-based approaches in driving down alcohol use.  Social norms theory used to drive down alcohol use has been successfully adopted at several colleges in the U.S.  Finally, I was able to prove the approach worked in a classically designed experiment with a control and experimental group. 

So, what can you glean from these examples in regards to choosing indicator variables to measure success?  First, I would work with a team of people to help choose a variable.  If the team includes someone who is a statistician or quality engineer, all the better.  Second, choose variables that have been used in similar evidence based projects.  In healthcare, many such can be found online at the American Society of Quality website or the iSixSigma web site.  Further, if you like, compare yourself to similar programs with similar patients or similar circumstances.  This can represent somewhat a control group.  Lastly, use the variable to measure a condition before any new approaches are used (baseline the variable) and then sample with it over a period several times to track changes.  If statistically significant changes continue, it is quite likely that the new approach is valid. 

One last note-you cannot prove that the new approach is directly responsible for changes without a classical statistical approach-control vs. experimental or with some other logical proof.  For instance, it was long shown that the use of tobacco was statistically significant in raising heart disease.  This was not proven, though, until the chemical process causing the diseases was understood.  The classical statistical approach couldn't be used because of ethical considerations.  It probably won't be necessary for you to try the classical approach or use a logical proof in most cases though.  Most people are fine with knowing that the program producing the changes is either directly or indirectly responsible.