Unless I’ve missed something, the authors conclusions are incorrect with regard to both the relationship of the EPO era to the previous era, and of the EPO era to the post EPO (Perneger) era.
First, look at the last three lines in Table 1, where the average speeds in these three eras are provided: 44.16 kph, 49.34 kph, 49.34 kph, respectively. These raw data indicate that the average speed in the EPO era was considerably faster than in the previous era, and did not change in the post-EPO era.
Now go to the fourth paragraph in the Results section, where they discuss these values and how they correct for them, using a year of competition b-value and a distance b-value. The first reflects that the raw data indicate a steady increase of 0.16 kph per year over the entire period of the study; the second is based on a decrease of 0.03 kph per additional km of ITT length. Comparing the EPO era with the pre-EPO era, these factors are 4.60 (representing about a 30 year difference between the mean of the pre-EPO era and the mean of the EPO era), and -0.096 (because the average distance in the EPO era was about 3.2 km longer than in the pre-EPO era, see Table 1). The first factor in effect reduces the difference between the pre-EPO and EPO era, by correcting for the steady increase in speeds over time; the second factor increases it, by correcting for the longer distances in the EPO era, which would reduce average speeds.
After making these corrections, they find that the difference in average speeds between the two eras is greatly reduced, to about 1.5% (45.59 vs. 44.92). This value, however, since it already includes correction for the steady increase, indicates that speeds during the EPO era increased by 1.5% above and beyond the steady increase. This is about what one would expect if there were roughly a 5% increase in power, and would certainly be consistent with doping.
Now consider the EPO vs. post-EPO era comparison. The two raw values are identical; no difference at all in average speeds. This can also be seen in Fig. 2, where the curve flattens out after 1990. They claim in the text, though, that when corrections are made for year of competition and length of ITT, the post-EPO speeds go up, to about 1.8% more than for the EPO era.
Several problems here. First, as with the EPO vs. pre-EPO comparison, this difference is above and beyond the steady increase, so again, it would suggest an effect of doping. But more important, their conclusion doesn’t make sense. As they state in the text, the year of competition correction factor is -0.83, which corresponds to about 5 years. It ought to be about double that, corresponding to ten years between the mean years of these eras. But the negative value means that the correction goes the other way, that is, the speeds in the post EPO era are reduced relative to the EPO era, when one takes into account the steady increase. Moreover, they claim only a very small correction for distance, though the average distance for the post EPO era was even less than for the pre-EPO era, so the correction should have been even larger.
All of that is bad enough. But then at the end of the discussion in this section, they state, “Notice that the slight performance differences we obtained between the various comparison groups did not yield significant results. Still, they are indicative of the linear progress in speed we discussed previously.” As I just pointed out, these differences exist after the speeds are corrected for the linear progress in speed. They are above and beyond this.
So unless I’m missing something—and I welcome anyone to correct me if I am--their data actually do show an increase in speeds from the pre-EPO era to the EPO era above and beyond the steady increase in speeds seen throughout the pre-EPO era. The increase is not large, and given all the scatter, not significant, but as I discussed earlier, given the power vs. speed relationship in time trialing, one would not expect it to be very large. And second, there is no increase in speed going from the pre-EPO to post-EPO or Perneger era. So these data seem to be quite consistent with doping in the 1990s that has not decreased since, though again, because these changes are not significant, one certainly could not rule out that doping has decreased.