Posted by tecwrg on 3/4/2015 7:39:00 AM (view original):
When you're dealing with large volumes of discrete data points, you sometimes have to group and summarize the data in order to look for trends, trends that you would not be able to easily see (or see at all) when looking at the discrete data points.
When BL looks at his 600 data points, he sees nothing. Yet there is a trend that's seen when the data is grouped and summarized, not just by the eyeball test (looking at the numbers themselves), but by statistical correlation.
Or are you denying that over the recent past that strikeouts have been steadily going up while scoring has been steadily going down?
No, grouping it together doesn't help you see a trend. Each team's Ks and run scoring are independent of other teams Ks and run scoring. By grouping them together, you miss that.
For example, if you looked at the team by team numbers for a two team league you might see this:
2014 Team A - 700 runs, 900 strikeouts
2013 Team A - 800 runs, 900 strikeouts
2012 Team A - 850 runs, 900 strikeouts
2014 Team B - 750 runs, 1200 strikeouts
2013 Team B - 750 runs, 1000 strikeouts
2012 Team B - 750 runs, 900 strikeouts
Grouping those teams together would give the appearance of a strong correlation when there clearly isn't one. If strikeout totals impact run scoring, we'd HAVE to see it in a team level. Otherwise, it isn't happening.