Boyd's World-> ISR's -- 1996 Season -- Score-Gathering Methodology and Resolution About the author, Boyd Nation

ISR's -- 1996 Season -- Score-Gathering Methodology and Resolution

Predendum: I'm going to leave the essay below intact, because the rules that I describe in it are the ones that I will continue to use going forward to older seasons to resolve disputed scores. However, the plea that I mentioned to SID's for further data was quite productive. I got a response from about 25% of the SID's contacted, for which I am grateful, but one of those in particular was extremely productive, as one sharp-eyed individual noted something that I had missed. Over the course of the last year, some unsung hero at the NCAA office has apparently spent some time with a filing cabinet and a quite prodigious digitalization project. Looking up the records for pre-web seasons now gives you scans of the paper records submitted at the time, going back to the mid-60's. My favorite that I've found so far is the 1965 Grambling State report, which is literally a scan of a Western Union telegram. This, over time, will be wonderfully valuable, as I should now be able over time to re-create seasons as far back as that goes. With that in hand, I've now updated the 1996 scores with only a few gaps. As far as I can tell, I'm only missing 6 games for Cornell, 3 for Howard, and 1 for Texas Southern, with it being quite likely that all of those are outside Division I, so they wouldn't affect the ratings.

The foundation of this data came from records collected by an old Usenet friend named Rick Rollins, collected at the time and shared with me somewhere around 1999 or 2000 (as you can tell, this project took a while). Rick's records were, of course, incomplete and occasionally wrong, but they served as an invaluable starting point and are still the only source I have for a lot of games involving teams with incomplete or poorly published records. Many thanks go to Rick for that start.

It's odd to think of things that took place when I was on my third adult job and had two kids as being in need of historical research, but that very much describes the process that's needed to pull these together. In that light, it has some resemblance to another hobby of mine, genealogy. Records for some things are fantastically thorough, sometimes in areas where you might not expect it - for example, there are quite detailed records of every Cincinnati score back to 1886, some areas where records exist but require some substantial digging, some where records are published but in unfriendly ways (SID's, if you're using issuu.com to publish your guides, that's great, but check that "Downloadable" box), and some where the institution doesn't even really exist any more. 1996 was the last year of baseball for Colgate, Wyoming and New Hampshire; in the Colgate case, I can't find records of the ten games needed to be able to do even a cursory rating of them, although given what I have found, I suspect they played a short, mostly non-D1 schedule that year. Records that exist can be misleading or flat out wrong. This isn't complaining - it's actually part of the appeal sometimes - but it's an explanation for why the record is and presumably always will be a little less than complete or perfect.

Because of the resulting incompleteness, there are some situations where I had to develop rules for resolution. In all of these cases, where I talk about sources, I'm referring to something published by the team, typically a media guide or record book; I have these sources for around 170 teams at this point, which will help when I tackle 1995.

Here are the rules that I used to resolve each type of identified error:

Score disagreements with no sources: I have no idea how, but I have three games involving Butler, two against Illinois-Chicago and one against North Carolina A&T, where I have no sources (presumably referring to scores I got from the Rollins archive) but have differing scores from the data sets for the two teams. For these games, I used the closer of the two scores so as to minimize the margin of victory advantage given in the ISR's.

Score disagreements with one source: In these cases, I'm using the score from the official source, playing the odds that it's more likely to be correct, with the exception of cases where there are disagreements as to who actually won the game, in which case I obviously use the score that leads to the correct won-loss records. Here's a table of those games:

Date Score 1   Score 2 Source School
326 Winthrop 11 North Carolina A&T 4   Winthrop 13 North Carolina A&T 4 Winthrop
219 Appalachian State 10 North Carolina A&T 4   Appalachian State 9 North Carolina A&T 4 Appalachian State
407 Ball State 6 Eastern Michigan 4   Ball State 7 Eastern Michigan 4 Eastern Michigan
323 Bowling Green State 6 Vermont 4   Bowling Green State 7 Vermont 5 Bowling Green State
421 Princeton 10 Columbia 5   Princeton 10 Columbia 6 Columbia
424 Columbia 6 Siena 5   Siena 5 Columbia 3 Columbia
507 Drexel 6 St. Joseph's 2   Drexel 8 St. Joseph's 2 St. Joseph's
516 Virginia Commonwealth 10 East Carolina 0   Virginia Commonwealth 10 East Carolina 3 East Carolina
407 Ball State 6 Eastern Michigan 4   Ball State 7 Eastern Michigan 4 Eastern Michigan
308 Florida Atlantic 15 Monmouth 2   Monmouth 20 Florida Atlantic 15 Florida Atlantic
423 Maryland-Baltimore County 9 Georgetown 5   Maryland-Baltimore County 9 Georgetown 8 Maryland-Baltimore County
225 St. Joseph's 5 Georgetown 1   St. Joseph's 6 Georgetown 1 St. Joseph's
313 St. Louis 11 Grambling State 5   St. Louis 21 Grambling State 5 St. Louis
311 Illinois 26 Pace 11   Illinois 27 Pace 11 Illinois
406 St. Joseph's 3 Rhode Island 2   St. Joseph's 5 Rhode Island 4 St. Joseph's
327 Rider 5 St. Joseph's 2   Rider 6 St. Joseph's 2 St. Joseph's
428 Detroit Mercy 6 Wisconsin-Milwaukee 0   Detroit Mercy 9 Wisconsin-Milwaukee 8 Wisconsin-Milwaukee
511 Wisconsin-Milwaukee 8 Butler 5   Wisconsin-Milwaukee 9 Butler 5 Wisconsin-Milwaukee
411 Toledo 4 Bowling Green State 1   Bowling Green State 4 Toledo 1 Bowling Green State
424 Eastern Michigan 12 Toledo 4   Eastern Michigan 12 Toledo 0 Eastern Michigan

Score disagreements with two sources: This happens more often than you might think. I'm using the closer score, with the exception of cases where there are disagreements as to who actually won the game, in which case I obviously use the score that leads to the correct won-loss records. Here's a table of those games:

Date Score 1   Score 2
323 Akron 11 Bethune-Cookman 3   Akron 12 Bethune-Cookman 3
126 Florida International 1 Bethune-Cookman 0   Florida International 5 Bethune-Cookman 1
127 Florida International 5 Bethune-Cookman 2   Florida International 8 Bethune-Cookman 1
501 Bradley 4 Western Illinois 2   Bradley 5 Western Illinois 2
406 North Carolina-Asheville 8 Maryland-Baltimore County 6   North Carolina-Asheville 9 Maryland-Baltimore County 6
301 Arizona State 7 Stanford 2   Arizona State 8 Stanford 1
323 Baylor 8 Texas Tech 0   Baylor 9 Texas Tech 0
301 Hawaii 6 Cal State Sacramento 5   Hawaii 8 Cal State Sacramento 5
409 Xavier 7 Cincinnati 3   Xavier 7 Cincinnati 4
227 St. Joseph's 8 Delaware State 2   St. Joseph's 8 Delaware State 3
219 Grand Canyon 6 Texas Tech 3   Texas Tech 6 Grand Canyon 3
320 Texas-Pan American 10 Miami, Ohio 2   Texas-Pan American 9 Miami, Ohio 2
224 Purdue 5 Texas Christian 0   Purdue 9 Texas Christian 5

There are a number of games where I only have a score from one team's data, either from their source document or from my (sometimes inherited) historical records. In those cases, I'm using the score, since the most likely explanation by far is just a records gap for the missing team. I don't see any benefit to listing those games; there are around 265 of them.

Finally, despite my best efforts, there are a number of teams where I can't find records of as many games as are listed in the historical win-loss records that I have from the team, typically gleaned from official conference records. It's possible, and even likely in a lot of cases, that the missing games were against non-Division I opponents, but I can't guarantee that there aren't gaps in the scores that underlie the ratings. I'm writing to SID's in search of additional scores and will update as new material comes in. Here's a list of the teams with missing games, in addition to teams where I've never been able to find a W-L record for 1996:

Team W-L Record   Record from Scores
Alcorn State 0 0   16 23
C. W. Post 15 24   13 24
Centenary 23 34   22 34
Central Connecticut State 17 25   17 23
Colgate 0 0   0 7
Coppin State 19 21   13 18
Cornell 17 27   11 23
Delaware 44 12   42 12
Florida A&M 14 34   12 32
Grambling State 0 0   6 12
Howard 14 35   10 29
Illinois-Chicago 23 26   21 26
Iona 26 21   20 15
Jackson State 0 0   13 17
LeMoyne 17 19   14 14
Long Island 22 19   21 19
Manhattan 9 30   5 21
Maryland-Baltimore County 9 28   9 27
Maryland-Eastern Shore 4 45   4 41
Mississippi Valley State 0 0   2 16
Monmouth 25 22   25 21
Niagara 14 20   8 17
Nicholls State 17 36   17 35
North Carolina A&T 4 38   3 34
North Carolina State 42 19   41 19
Prairie View A&M 0 0   0 23
Princeton 26 21   25 20
Rice 42 23   40 23
Southern 34 7   18 5
Southern Utah 0 0   7 29
St. Mary's 14 37   13 38
St. Peter's 18 24   12 21
Tennessee Tech 28 29   27 29
Tennessee-Martin 14 30   13 30
Texas Southern 0 0   6 17
Texas-Arlington 23 35   20 35
Troy State 41 13   40 13
Valparaiso 18 29   18 28
Wagner 14 24   13 23

In addition to all of that, off-field situations can change the official won-loss record of a team after the fact, and it's hard to know exactly how to handle that. Ideally, a rating system should reflect what actually happened on the field, but the reason that eligibility issues are punished is because they distort what happens on the field. There were two situations in 1996 that involved forfeits. Loyola Marymount subsequently forfeited all but four of their wins for the year. I can't find any record stating why, although it fits the pattern for use of an ineligible player. In those cases, the score records show the original game score. Without more detail, I used the original game scores, so that Loyola Marymount and their opponents are rated as they were on the field. Feel free to mentally adjust or dismiss their rating if you wish.

The other situation is different and a bit more mysterious. Texas-Arlington lists a forfeit weekend against Sam Houston State for the last weekend of the regular season, while SHSU has no games listed for that weekend. Both have the seasons fully listed in old media guides but with no explanation. I'd love to hear the story if anyone knows it, but I've left the games out of the record, since a set of 9-0, 7-0, and 9-0 doesn't really tell us anything useful about the teams involved.

Boyd's World-> ISR's -- 1996 Season -- Score-Gathering Methodology and Resolution About the author, Boyd Nation