"Statistics are like swimwear - what they reveal is suggestive but what they conceal is vital."6/8/2015
![]() The title quote is by Ashish Mahajan, Lancet 2007. The final sentence of the cited paragraph reads "By empowering patients to critically appraise medical data, health-care professionals will pave the road towards meaningful collaboration." I would like to argue that a level-set will benefit health professionals and medical education stakeholders as well. A primary reason for my blog, Data and Donuts, was reminiscent of the clever strategies needed to bring people into meetings with the single goal of teaching analytics to collaborative teams. Eyes would roll back and yawns stifled silent screams. The solution? Promise everyone a donut. Who doesn't love a donut? It was learned that the masses would attend unable to resist the siren song of a tasty treat. Now your eyeballs wouldn't be the only glazed orb in the room... All statistics are not created equal (or equally effective)![]() There are basically 3 different types of data. It sounds simple but if you aren't clear on what type you are dealing with you are in danger of applying the wrong analytics--leading to weak or incorrect assumptions. In healthcare we are most interested in continuous, binary (and categorical), and time to event data. Continuous data is incremental data. Think of blood pressure values, weight, height, or age--you get the picture. A good rule of thumb is to recall that one unit change (increment) means the same across the entire range of data values. Binary data is a simple yes or no. Yes I am female, no I don't have the disease in question, and no I do not smoke. You will notice that I included categorical in parentheses above as an extension of binary data category. Categorical data is actually an expansion of simple binary (yes or no) to include additional categories.
This matters because there are different statistics for different data types. In future blogs we will introduce mean differences, confidence intervals and the t-test, relative risk /risk ratios, confidence intervals and chi-square, followed by time to event incidence rate ratios, confidence intervals and Kaplan Meier or log-ranks. These tests will help to identify real differences and tests for estimating imprecision in our data sets. In other words...beware of the itsy, bitsy, teeny, weeny, yellow, polka-dot bikini!
Thoughtful discussions about content development and outcomes analytics that apply the principles and frameworks of health policy and economics to persistent and perplexing health and health care problems freelance medical content, health policy, and economics writer and insight analyst Comments are closed.
|
Sign up for our newsletter!
Browse the archive...
Thank you for making a donution!
In a world of "evidence-based" medicine I am a bigger fan of practice-based evidence.
Remember the quote by Upton Sinclair... “It is difficult to get a man to understand something, when his salary depends upon his not understanding it!” |