Commentary on Magnitude-Based Inference Under Attack Martin Buchheit Sportscience 18, 48, 2014 (sportsci.org/2014/mb.htm) |
This item
highlights clearly the advantages of the MBI approach and presents welcome
reassuring rebuttals of the Welsh and Knight criticisms. Scientists and practitioners
should not rely on the all-too simple yes-or-no approach of statistical
significance when making decisions about an effect. What really matters to us
is the magnitude of the effect and the likelihood that it represents
something useful. While many roads lead to Rome, the MBI approach clearly
meets our needs. It has therefore changed the sports discipline for me and many others, both in academic
and applied settings. The freely-available
spreadsheets at the Sportscience site are extremely helpful and have allowed
many scientists and practitioners to run stats on their data efficiently,
which is remarkable. I don’t know of any stats package that comes close to
these spreadsheets in terms of availability, ease of
use and, of course, relevance of analysis. Will Hopkins and Alan Batterham
should be congratulated for their constant efforts to develop and democratize
the MBI approach. While the Welsh and Knight paper could be seen as an
obstacle for the further development of MBI, it has actually allowed Will and
Alan to better defend and highlight the quality and interest of their work. Published November 2014 |