Where is all this data we've been looking at leading? Photos by Alan Versaw and Jeff McCoy.
This is the part where the guy who spent the last week lobbing numbers your direction is supposed to stand up and tell you what it all means. The trouble with that is that I'm pretty confused myself about what it all means. But, I do think there are still a few takeaways we can gather from the data.
Let's look first at the correlations between Liberty Bell places and State places. The table below summarizes those and serves as the jumping off point for a thought or two:
The raw correlations are just that, what the correlation was between Liberty Bell time ranking and State time ranking for each of the top 25 from Liberty Bell who finished the state meet race. As you can see, those numbers are all over the place, ranging from 0.04 to 0.65. Perfect correlation would score a 1, but we don't expect perfect correlation between Liberty Bell and State time rankings for a number of reasons--health of athletes, variations inherent in six weeks of space between the two races, accumulated fatigue, differences in how different races develop, and so on.
The adjusted correlations reflect a recalculated correlation with obvious outliers removed. The trouble here, though, is that "obvious" is a subjective standard. What seemed obvious to me may not have been obvious to another, or vice versa. And, obvious or not, are we messing with something we shouldn't be messing with by removing outliers in the first place? I never removed more than three outliers for any year and gender, and mostly removed only one or two. Removing outliers brought the correlations closer together, but I'm not sure it tells us much of anything.
Paradoxically, correlations for the NPEC/BCRP course are weaker for boys and stronger for girls than other State courses, at least after removing outliers.
It was worth noting that almost all scatterplots of Liberty Bell vs. State placings displayed a reasonably linear pattern. There was, of course, scatter to be observed in each case, but the overall pattern of the paired places was almost uniformly linear.
About the only summative thing I can say is that correlation varies a lot between years and genders. This analysis might have turned out being a whole lot more informative if it were done with all Liberty Bell finishers, but I didn't have the tools at hand to make that a manageable analysis.
We'll first note that the average times listed for Liberty Bell are varsity-only average times. Adding sub-varsity runners to the pile only adds to the confusion. Temperatures and barometric pressures for Liberty Bell are reported conditions for Littleton at 5 PM on the day of each race. Temperatures and barometric pressures for State are reported conditions for Fort Collins, Aurora, or Colorado Springs (as appropriate) at 11 AM on the day of each race.
Liberty Bell, of course, has been a constant course over the years 2008-14. You can readily see the impacts of heat and density altitude on Liberty Bell times. You might also notice that the best average times for Liberty Bell came from 2008. Even though conditions were superior in 2014, this year's average times did not approach the 2008 or 2009 averages for either boys or girls.
So, let's begin this part of the discussion with that. Why would 2008 and 2009 average times be superior to 2014 average times? We could start with the raw count of teams. There were 71 Liberty Bell teams in 2008 and 90 in 2014. That could imply a dilution of the field, though it would not necessarily have to mean that. It was probably a somewhat more elite set of out-of-state teams in 2008 than 2014 (think Los Alamos near the peak of their cross country power, for example). It could also be that 2014 was cool enough that the temperature, while it did produce a very favorable density altitude, was a little below optimal for 5K race performance. And, that may have as much to do with warm-up variables as it has to do with race variables. It may mean that the adrenaline factor for Liberty Bell has diminished slightly over the last five to six years. It may mean athletes were superior, overall, in 2008 and 2009. It almost certainly means some combination of some of the above, plus other factors not yet hypothesized.
I cannot begin to explain why girls times in 2009 were so much faster than 2008. That will have to stand as a mystery for now. Doubtless it has something to do with overall field strength and how races went out, but it's hard to be more specific than suggestion of broad possibilities.
You will also note something I've mentioned seeing before. 2010 was the heat and density altitude (overlapping, but distinguishable, concepts) year of misery for Liberty Bell. While both boys and girls were slower overall than any other year in the analysis, the girls were slower by a much larger percentage of total race time than the boys. A similar phenomenon can be observed, and has already been mentioned, for the 2014 State race.
At this point, the evidence is anecdotal, but suggestive nevertheless. It's worth studying more deeply if conditions of heat and/or increasing density altitude have a magnified impact on female runners versus male runners. If it turns out that they do, the question of why comes next.
Turning our attention now toward state, we already knew both the Arapahoe County Fairgrounds and NPEC/BCRP courses are tougher courses than the Fossil Ridge course. The data bear out that truth and quantify the difference a little. It would appear further, and this has been the subject of some debate, that the NPEC/BCRP course is clearly a more difficult course than the Arapahoe County Fairgrounds course. Even with superior density altitude conditions, NPEC/BCRP has run slower than Arapahoe County Fairgrounds. The net uphill characteristic of the NPEC/BCRP course is doubtless also a factor here.
We could further hypothesize that if we ever do see a low barometric pressure day combined with any sort of temperature over 60F, we probably haven't yet seen the high end of average times at NPEC/BCRP. Thus far, the State meets at this facility have led a charmed life where barometric pressure is concerned. Getting three in a row with an 11AM barometric pressure of 30.11 or higher is a little on the side of exceptional.
***********
Honestly, I'd hoped to have something more to compelling to tell you at the end of all this work of looking at Liberty Bell and State data. On the one hand, I apologize for that. On the other hand, that is simply a lesson of dealing with data. Sometimes data is inconclusive.
At this point, I'd welcome--even invite--observations from those of you who've taken an interest in these questions along with me over the last week. If I've missed something obvious, or missed asking an important question, please let me know about that. And please share your thoughts on the forum topic for this article so that the community can join in the conversation.