In Part 3 of the continuing analysis of turnovers, I’ll look at the impact of turnover margin (TOM) on the win/loss margin (WLM).

Summary:Basis: Only the 66 teams that automatically qualify (AQ) for the BCS bowls are included for the last ten years – 2000 to 2009. (See discussion below for why this change to AQ instead of all FBS was made.)

Note: In Part 1, a statistical analysis concluded: Luck is primarily responsible for TOM of approximately 80% of FBS football teams. For the other 20%, team performance (good or bad) is primarily responsible for TOM. Teams within +/- 4.0 TOM per year are primarily dealing with luck. For teams with more than +4.0 TOM per year, good performance is the primary factor for the TOM. For teams with more than -4.0 TOM per year, poor performance is the primary factor for the TOM.

Details Here:  http://mgoblog.com/diaries/turnover-analysis-part-2-do-turnovers-turnar…

In Part 2, a statistical analysis concluded: Phil Steele is wrong. Turnovers do NOT equal turnaround. The teams Steele isolates (those with double-digit turnovers) are the teams whose TOM is primarily due to performance and not luck. Therefore his basic premise is incorrect. In addition, the percentage of teams that “turnaround” the next season is approximately the same when TOM is completely ignored.

Details Here:   http://mgoblog.com/diaries/turnover-analysis-part-1-it-all-just-luck-1

The Switch To Automatic Qualifiers Only:It is obvious (to even the most casual observer) that teams in the AQ conferences are significantly better than the teams in non-AQ conferences. Over the past decade, 71% of AQ teams have winning records but only 30% of non-AQ teams have winning records. The reason for the low winning record of non-AQ teams is they are significantly overmatched in their OOC games against AQ teams.

TOM becomes less of a factor when the two teams playing have vastly different abilities. I looked at the same TOM data including all FBS teams and the correlation was not nearly as high. This is exactly as expected. Non-AQ teams often have +TOM in their OOC and still lose. Including non-AQ teams in this analysis would distort the results.

I did not attempt to eliminate the data for non-AQ games (i.e. subtract all the non-AQ data for TOM and WLM). The data for TOM and WLM includes ALL games but only the 66 AQ teams were included in the analysis.

The Gory Details:The first table provides yearly and total over the decade for: 1) the number of teams with the indicated TOM; 2) of those, the number of teams with a positive WLM; and 3) the percentage of teams with a positive WLM in that category.

	+5TOM	+WLM	%		+TOM	+WLM	%		-TOM	+WLM	%		-5 TOM	+WLM	%
2009	18	17	94%		42	37	88%		24	7	29%		14	3	21%
2008	24	19	79%		37	28	76%		29	15	52%		18	7	39%
2007	20	19	95%		38	34	89%		28	11	39%		14	4	29%
2006	22	18	82%		36	31	86%		30	14	47%		16	5	31%
2005	23	21	91%		32	26	81%		34	13	38%		17	4	24%
2004	21	18	86%		38	30	79%		28	11	39%		16	6	38%
2003	19	17	89%		40	31	78%		26	11	42%		15	3	20%
2002	31	30	97%		42	39	93%		24	11	46%		18	5	28%
2001	19	18	95%		38	33	87%		27	11	41%		14	3	21%
2000	27	25	93%		40	34	85%		24	7	29%		15	4	27%
Total	224.0	202.0	90%		383.0	323.0	84%		274.0	111.0	41%		157.0	44.0	28%

The second table provides yearly and total over the decade for: 1) the number of teams with the indicated TOM; 2) of those, the number of teams with a positive WLM of 5.0 or greater; and 3) the percentage of teams with a positive WLM of 5.0 or greater in that category.

					0 to +4				0 to -4				-5
	+5TOM	+4WLM	%		TOM	+4WLM	%		TOM	+4WLM	%		TOM	+4WLM	%
2009	18	12	67%		24	7	29%		10	1	10%		14	2	14%
2008	24	13	54%		13	7	54%		11	4	36%		18	2	11%
2007	20	16	80%		18	8	44%		14	3	21%		14	1	7%
2006	22	13	59%		14	9	64%		14	4	29%		16	1	6%
2005	23	15	65%		9	1	11%		17	4	24%		17	3	18%
2004	21	12	57%		17	6	35%		12	3	25%		16	2	13%
2003	19	10	53%		21	9	43%		11	4	36%		15	0	0%
2002	31	22	71%		11	1	9%		6	2	33%		18	1	6%
2001	19	10	53%		19	7	37%		13	3	23%		14	0	0%
2000	27	16	59%		13	6	46%		9	1	11%		15	0	0%
Total	224.0	139.0	62%		159.0	61.0	38%		117.0	29.0	25%		157.0	12.0	8%

One More Thing– The question of cause and effect is bound to come up because I have only shown a correlation. "Correlation does not imply causation" is a phrase used in science and statistics to emphasize that correlation between two variables does not automatically imply that one causes the other (though correlation is necessary for causation and can indicate possible causes or areas for further investigation).

One question to ask for causation, “Is the variable relevant?” The answer is obvious “TOM is relevant to the outcome of football games.” Some people have experienced a correlation between the clothes they wear and their team winning a football game. But, of course, there can be no cause and effect – the clothes a person wears is NOT relevant to the outcome of a football game (sorry to burst anyone’s bubble).

It is also obvious that there is no “direct cause’ and effect because teams win games with even large negative TOM (Michigan was -4 TOM versus Florida in the 2007 Citrus Bowl).

However, a cause may be classified as a "contributory cause," if the presumed cause precedes the effect, and altering the cause alters the effect. It does not require that all those subjects which possess the contributory cause experience the effect. It does not require that all those subjects which are free of the contributory cause be free of the effect. In other words, a contributory cause may be neither necessary nor sufficient but it must be contributory.

Thus, TOM meets the definition of a contributory cause.