I read the
article in question. It is poorly designed because the variable of interest, deep stop time, does not vary systematically while holding shallow stop time constant. The net result is that it is very difficult to determine the impact of 1 min vs 2 min vs 3 min vs 5 min deep stop and how they differ from a control with 0 min deep stop. It would have been nice to have had deep stops of 0,1,2,3,4,5 min all tested with 0 shallow stop time; then repeated at 1 min shallow stop time, then 2 min, etc. Then, a solid statistical analysis could have been done and would have yielded meaningful results.
So, I took their Bubble Score Index data and plotted it relative to a composite variable which is the simple sum of deep stop time and shallow stop time. The result is a statistically significant correlation coefficient (r = -0.59544, df=13, p<.02, 2-tailed). This means that a large proportion of the variance in the data is accounted for by simply taking more stop time, whether deep or shallow. In other words,
the more time you spend at a safety stop, the lower (better) the Bubble Score Index.
More detailed data analysis is needed to determine whether one particular stop profile is better than another. The data, as presented in the article, do not allow such analysis. There is no indication of between- or within-diver variability or within group variability.