ASA Connect

 View Only

  • 1.  Creating a equation to generate "Performance/Compliance Index measure"_Transforming qualitative data into quantitative data

    Posted 11-20-2015 08:01

    I'm creating a performance metrics for the quality of my team's work towards certain assessment document.

    Assuming we have a way to measure the value of X's, we can assign overall 'quality' of each document. 

    I set up the quality criteria to be below:

    X1: If the document is reviewed by team of 4

    X2: If the document filled out every sections

    X3: If the document is approved on time (Assuming target duration is 90 days, so we want the document to be approved within 90 days) 

    I assigned the weights to X's, and they are: X1=0.5, X2=0.25, X3=0.25. (Assuming X1 is most important, and X2, X3 are equality important)

    Ex) For certain document: 1. It is reviewed by 2 people (Out of 4) 2. 50% of the sections are filled out. 3. It took 92 days for the document to be approved. With this, we can assign, 50% to X1 since it is 2/4, and 50% to the X2. I'm having trouble assigning 92 days to X3, because they are not ratio (Or percentage). 

    My idea is to use percentage deviation, 1-(observed-expected)/(expected), so it will be 1-((92-90)/90)=.98, 98%. 

    However, if you think about possible observed duration of 200 days, this can go well above 100%. 

    Any suggestions or feedback will be appreciated

    ------------------------------
    John Lee
    Student
    ------------------------------


  • 2.  RE: Creating a equation to generate "Performance/Compliance Index measure"_Transforming qualitative data into quantitative data

    Posted 11-23-2015 09:26

    The rule you assign to X3 is arbitrary.  One arbitrary choice for such a rule is to make the outcome binary: 

       If the document is approved on time (that is, within 90 days), X3=1.  Otherwise, X3=0.

    By the way, your suggested rule,

      1 - [(observed-expected)/expected] = [(expected - observed + expected)/expected]

                                                               = [(2*expected - observed)/expected] = 2 - (observed/expected).

    If observed = 200, then X3= 2 - (200/90) = 2 - 2.222 = - 0.222... = -22.2%, which is less than 100%,

    ------------------------------
    Matthew Zack

    Original Message


  • 3.  RE: Creating a equation to generate "Performance/Compliance Index measure"_Transforming qualitative data into quantitative data

    Posted 11-23-2015 15:33

    You could reduce the weight for X3 by an increment for each day over 90. For example, if you reduced the weight by 0.01 for each day over 90, and X3 = 91, then the adjusted weight would be 0.24. You would need to decide on the increment and also if you want to let the score for X3 be negative.

    ------------------------------
    Daniel Thompson
    Training and Research Consultant - Retired
    Florida Department of Health - Retired

    Original Message


  • 4.  RE: Creating a equation to generate "Performance/Compliance Index measure"_Transforming qualitative data into quantitative data

    Posted 11-24-2015 15:04

    I think I am confused by how the definitions of X1, X2, and X3 are implemented. For example, the following definition, "X1: If the document is reviewed by team of 4", indicates that X1=1 if the condition is met, and that X1=0 if the condition is not met. So if the document is reviewed by 2 people (out of 4), then X1 should equal 0, not 50%, since the condition of "if the document is reviewed by team of 4" was clearly not met. And by the same reasoning, X2 should equal 0, not 50%, and X3 should also equal 0.

    ------------------------------
    Eric Siegel
    Biostatistician
    Univ of Arkansas for Medical Sciences of Biostatistics

    Original Message