Why sometimes do we have inaccurate results from our quality assessments?

Why do we often focus our work on something wrong, but do not measure the return on our work investment and thus fail to prioritize and continually re-prioritize as we deliver our work.

Important concepts that are key to understand accuracy and conduct work properly includes:
  • Population sampling – two common types include attribute data and variable data.  Attribute data is counted and is a pass/fail or yes/no result (e.g. number of errors, number of dump trucks).  Variable data can be broken down, it is based on a type of continuous scale, ranges exist and compliance can be rated (e.g. time, weight, etc). 
Note 1: ensure sample size is appropriately sized to align with the trade-offs that you must make (e.g. larger sampling gives more accuracy but takes more time, money, resources, etc).
Note 2: if you don't have the experience then ensure you get strong support from subject matter experts (SMEs):
    • Tolerances and control limits – tolerance is the variation that is acceptable in relation to the project quality standards.  When a variation exceeds the control limits this is considered unacceptable and some type of corrective action must be taken. 

    Another important concept is the ability to know where to prioritize your efforts, manage your expectations and the probability of success in terms of what type of work you are doing when trying to fix something that is not going according to plan (also known as something that has variance to what you expected to happen).  A simple example how to group two types of work variances include:
    • random cause – these are always present and individually are insignificant (people learn to live with them).  Addressing these types of variances takes a great effort for a small value that will be received  (e.g. train was as early as 2 minutes and as late as 5 minutes over the past week);
    • special cause – these are generally unusual or atypical events that are often caused by a random unpredictable reason (not a flaw in the overall process) (e.g. train was late one day 15 minutes due to a mechanical breakdown).
    Note: so based on the variance causes above, you will generally get more "bang for your buck" when addressing special cause variances.  The approaches to the two types of variances will also be very different so ensure to understand that.