For a better browsing experience, please upgrade your browser.

FiveThirtyEight

Sports

In January, the Sam Houston State University Bearkats traveled to Hammond, La., where they beat Southeastern Louisiana University, with 29 field goals, nine of them assisted. Two months later, the Bearkats hosted a rematch back in Huntsville, Texas, this time scoring 28 field goals, with an astounding 26 of them assisted.

Basketball fans know that lots of variables determine whether a shot is deemed assisted: the quality of the pass, how much time and how many dribbles separate the pass from the shot, and, most essentially, whether the shot goes in. But another factor can be just as important in assist decisions: where the pass was made. Not where as in where on the court, but where meaning at which venue.

Like errors in baseball and tackles in football, assists are subjective, and the decider’s philosophy matters. Sometimes the effect is extreme: On some courts, assist counts rise or fall dramatically, as if players suddenly learned or forgot the art of the pass. And nowhere in men’s college basketball is the effect more extreme than at Sam Houston State’s Bernard G. Johnson Coliseum.

In more than 800 team-seasons at the NCAA Division I level over the last three years, the scorers who were most generous in awarding assists were those at Sam Houston State.

“Our philosophy has always been, if the pass creates the basket, it’s an assist,” said Jason Barfield, a spokesman for Sam Houston State’s athletic department.

Sam Houston State’s twitchy assist-scoring trigger finger is well known around the Southland Conference, Barfield said. “We’ve always kind of been known as being too liberal on assists,” he said. “People laugh about it in our league.”

At the conference tournament last month, where the Bearkats lost in the final, other teams’ officials laughed as they flipped through their media guides to check which of their single-game assist records were set when playing at Sam Houston State, Barfield said. That’s a permanent record of the school’s equal-opportunity assist generosity.

Spokesmen for other teams in the conference didn’t criticize the Bearkats’ assist-counting, acknowledging that it’s a tricky stat to measure. “The assist is about the most subjective part of basketball,” said Shane M. Meling, spokesman for University of the Incarnate Word in San Antonio.

The more room for interpretation in stat collecting, the more interpretations will differ. The evidence is in assist numbers. ESPN Stats & Information supplied box-score stats for the last three seasons of Division I men’s college basketball for pairs of teams that played home games against each other in the same season. I compared the two teams’ combined assist percentage — percentage of field goals assigned an assist — in each pair of games.1

Some schools are especially fertile ground for assists. Over the last three years, scorers at five schools have consistently awarded assists on a far higher percentage of made field goals than scorers judging the same matchups at a different venue: Sam Houston State, the University of North Carolina at Asheville, Lehigh University, Stony Brook University and Michigan State University.2 Scorers at eight schools are stingy about awarding assists: the U.S. Military Academy (Army), Sacred Heart University, the University of San Francisco, Alcorn State University, Montana State University, Nicholls State University, Northern Illinois University and the University of Hawaii.3

Team4 Assist effect
Army -13.1%
Sacred Heart -13.1%
San Francisco -12.9%
Alcorn State -11.8%
Montana State -11.5%
Nicholls State -11.3%
Northern Illinois -11.3%
Hawaii -10.3%
Michigan State +9.5%
Stony Brook +9.7%
Lehigh +10.5%
UNC-Asheville +14.8%
Sam Houston State +27.5%

I also tested whether any schools were tilting the scales for their own players. None gave an unusually high boost to the home team’s assist percentage in each of the three seasons.5 Scorers at Eastern Kentucky University, though, appear to hand fewer assists to home players. (It was the only school to rank high in this category in all three seasons, so it could just be a statistical fluke.)6

The NCAA Basketball Statisticians’ Manual defines the assist, but leaves plenty of leeway for the scorer to exercise judgment: “An assist should be more than a routine pass that just happens to be followed by a field goal. It should be a conscious effort to find the open player or to help a player work free. There should not be a limit on the number of dribbles by the receiver. It is not even necessary that the assist be given on the last pass.”

“At times, statisticians have to use their judgment and knowledge on how to score a certain play,” said NCAA spokesman Ketrell Marshall. “Similar to how an official scorer in baseball has to judge a hit or error, the same applies in basketball where one statistician might give an assist on a particular play while another statistician would not.”

Some athletic departments whose men’s basketball scorers came up as outliers in the research cited the subjectivity of assists as an explanation for their stats.

“While the NCAA Statisticians’ Manual provides several examples of what is and isn’t an assist, and those rules are followed, to quote Justice Potter Stewart, I know it when I see it,” said Daniel Snowden, athletic department associate director for media and public relations at the University of Mississippi, which is one of the leaders in my measure of awarding assists in the last two seasons. He added, “It does not surprise me we are at the top of the conference.”

Snowden said he and the Rebels’ official scorer are “very proactive” about awarding stats other than points, such as assists and blocks. “I don’t believe any team deliberately underreports assists, or blocks for that matter, but I have noticed that some teams are less liberal in awarding assists. When I have mentioned this in the past, the response is usually the scorer forgot to add it. Rarely, if ever, have I had someone not agree that a play wasn’t an assist or a block. Basketball is an extremely fast-paced sport to stat on a computer, and sometimes things are simply missed.”

“We try and be objective for both teams and maybe I’m a little more liberal than some places,” said Mike Gore, who does most scoring for home basketball games at the University of North Carolina at Asheville, which was one of the top venues for assist rates. “I’ve been doing this 28 years and every place is subjective,” he added.

Schools on either end of the assist spectrum think other scorers are too generous or too stingy with assists. “It’s an extra keystroke, so sometimes people are being lazy when they don’t give the assist,” Barfield said. “When you go on the road, you’ll see obvious assists that aren’t marked for one reason or another.”

He’s baffled by road box scores in which the Bearkats are credited with five assists on more than 20 field goals.7 “That’s almost an impossibility,” Barfield said. “Our guards are very involved in ball movement. Very few of our shots are where the guard just dribbles it up and takes a shot. Probably once a year we’ll play a game where we get the box score back and see something like that.”

Dave Guffey, a spokesman for the University of Montana, which awards a relatively low assist rate at home, said, “I think there are at least two teams in our conference that over-count assists.”

Some spokespeople cited factors other than their scoring tendencies — such as statistical flukes, or playing styles differing at home and away. One of the venues we found to be suppressing assist rates is Sacred Heart University’s William H. Pitt Center.

“My guess is it’s more of an anomaly than anything else,” said Chris O’Connor, associate athletic director for external affairs at Sacred Heart. He said several different people have scored games over the last three years, reducing the likelihood of a rampant anti-assist philosophy. He also pointed out that the analysis, restricted to games between teams that played in each others’ arenas in the same season, excluded some home dates where visitors got lots of assists: Long Island University’s Jason Brickman had 12 at Sacred Heart this year, but the Blackbirds played LIU only once so the game didn’t count toward this assist analysis.8

Told that his players’ assist percentage relative to opponents’ was much higher at home than on the road, Kevin Lorincz, director of athletic communications at Rutgers University, said, “I would be really surprised if our home stats weren’t markedly better, given that we shared the ball and played much better at home.”9 Lorincz added, in an email, that opponents wouldn’t be shy if they thought their players were getting shortchanged in visits to Piscataway, N.J.: “You develop relationships with your fellow [sports information directors] and being the team that ‘loads up its box scores’ would be an uncomfortable and short-lived exercise.”

In special cases, though, teams might do just that. “In my experience that typically happens when there’s a direct benefit in a particular stat — for instance when a team or player is near the top of conference, national or all-time school leaders,” Lorincz said. “We had a center a few years back that was a tremendous shot-blocker. We didn’t give him any blocks he didn’t deserve, but we certainly didn’t miss any either.”

Lorincz added, “I’m not denying that home stat crews can be somewhat optimistic at times.”

In their home game last month against Southeastern Louisiana, Sam Houston State players got the benefit of optimistic scoring. Many of their 26 assists were clear-cut, catch-and-shoot situations.10 Four passes credited as assists, though, ended with a Bearkat catching the ball near the three-point line. In each case, the player then took several seconds and dribbled through the defense before scoring from near or in the paint — once on a reverse layup. On a fifth occasion, Kaheem Ransom caught a pass from James Thomas behind the three-point line, waited two seconds for a pick to be set, then took a few dribbles before shooting from elsewhere in three-point territory, 4.5 seconds after receiving the pass. Thomas got credit for an assist.

“Whether the guy takes one step after or four steps on a break, if the pass was good enough to set up a basket, it’s an assist,” Barfield said.

Barfield gave examples of types of passes that his scorers see as assists but others might not. For instance, he believes an outlet pass setting up a streaking guard for a score should count no matter where the guard catches it, or how many dribbles he takes afterwards. “That [basket] doesn’t happen if you don’t pass the ball down the floor,” Barfield said. “I would give an assist there, where on the road you might not see an assist there.”

He also thinks centers don’t get enough credit for kicking the ball out for a three after collecting an offensive rebound. “For whatever reason, that’s a play where the center does not get an assist, or the forward who gets a rebound will not get an assist there,” Barfield said. “I can’t explain why.”

Barfield points out that Sam Houston’s scorer, Paul Ridings, applies his philosophy consistently, lifting opponents’ assist rates, too. “It’s not just our numbers” that are relatively high, Barfield said.

And he’s right. In a typical Division I game not played on a neutral floor, the home team’s assist rate is 5 percentage points higher than the road team’s — perhaps a reflection of both a slight lean toward home players by the scorer and home teams simply playing better. But teams that played both home and away games against Sam Houston State in the same season over the last three years averaged assist rates that were 20 percentage points higher in Huntsville than when they hosted the Bearkats.11

Season Bearkats’ assist boost Visitors’ boost Total boost
2010-11 +31.3% +25.6% +29.1%
2011-12 +32.0% +16.7% +24.7%
2012-13 +37.2% +13.8% +25.8%
2013-14 +36.5% +28.1% +32.0%

Subjectivity isn’t the only downside of the assist stat. It gives equal credit to passes of varying value: Some do much of the work for shooters, while others leave the shooter to make, say, a long jump shot.12 The passer also  gets credit only if his teammate hits the shot — if he misses, or is fouled and hits free throws, there’s no assist.

“I believe if the shooter goes to the line and makes both free throws, then the passer should be awarded the assist because his pass led directly to points on the board,” said Lance Fleming, a spokesman for Abilene Christian University’s athletic department. “Help get that written in the rulebook.”

A generous assist-scorer will only get a college player so far, since most professional scouts go by video, rather than relying solely on box scores. A working paper by University of Maryland Baltimore County economists found that in relatively weak conferences such as Southland, college assist stats have no statistically significant effect on NBA draft position.

A big assist number could get a player past the first screen, though. And leagues outside the U.S. have to rely more on stats than on in-person scouting. Mike Laninga, director of athletic communications at the University of Illinois at Chicago, said he could imagine scouts saying, “Obviously if you had eight assists per game rather than five assists per game, whoa, that’s something we need to look at.”

Some players and coaches notice which places are the best and worst for their stats. At Montana, where assists are hard to come by, “Our players have grumbled in the past, but not in recent years,” Guffey said.

Visiting players know when they’re going to Sam Houston, “they’re going to get some assists they wouldn’t normally get,” Barfield said. Opposing coaches have commented that “point guards like to come and play against us.”

Laninga said this wouldn’t fly at UIC, where assist rates are relatively high.13 “Our coaching staff told them, you don’t even get a stat sheet after the game. If a starter ever asked for stats, there would be hell to pay,” he said.

The only people who complain about college stats, in his experience, are players’ parents.

Footnotes

  1. The simplest way to compare teams is to examine their home assist rates, where assist rate is percentage of made field goals that are credited with assists (assists divided by field goals made). But that doesn’t account for different styles of play: Some teams set up more of their shots with passes, while other teams have players who generate their own shots.

    One step toward controlling for the team’s style of play is to subtract its road assist rate from its home assist rate. But teams in college hoops don’t all play balanced home and away schedules. So the difference in their home and away assist rate may reflect differences in opponents’ defensive schemes.

    So I isolated my comparison to the most similar pairs of games I could find: those between the same two teams. And since most of these were home and away games, I didn’t include pairs with neutral games, in case those had different properties than away games. These pairs of games are nearly always, at the college level, between teams from the same conference.

    I isolated the analysis to teams that had at least five pairs of home and away games in each of the last three seasons: 279 teams in all. Then I calculated the average of each team’s assist percentage and of its combined assist percentage, and compared those figures for home and away games.

    One weakness of this method is that a team that plays mostly against teams with home scorers who are, say, especially generous with assists might look stingy because of the tough comparison. However, it’s unlikely this affected the results because the outliers were spread out over many different conferences. Another possibility is that some teams play a different style at home than on the road, even against the same opponents. ^

  2. This is based on averaging overall assist percentage — for both teams in our home-road pairs — in each game and subtracting the away percentage from the home percentage. All five schools ranked in the top 15 percent of schools in our sample by this measure in each of the last three seasons. ^
  3. All eight schools ranked in the bottom 15 percent of schools by this measure in each of the last three seasons.

    If there were no consistency from season to season, we’d expect roughly one school to rank in the top 15 percent, and another in the bottom 15 percent each season. There were 279 schools for which we had stats in each of the three seasons, for at least five pairs of games in each season. So about 42 schools ranked in the top 15 percent in 2011-12. By chance alone we’d expect 15 percent of these, or about six, to rank in the top 15 percent again in 2012-13. And again by chance we’d expect 15 percent of those, or roughly one, to rank in the top 15 percent in 2013-14. That there are instead five and eight schools, respectively, in the top and bottom 15 percent in each season suggests a meaningful finding. ^

  4. Average over last three seasons, of average effect on total assist rate of playing at home against teams also faced on the road. Teams shown ranked in bottom or top 15 percent in each of the three seasons. ^
  5. This is based on calculating a team’s net assist percentage — its assist percentage minus its opponent’s — for both the home and road game in each matchup, then subtracting the road net assist percentage from the home percentage. I averaged that over all the pairs of games in each sample for the teams studied, then searched for teams that ranked in the top 15 percent in each season. This stat, by the way, had almost no correlation (R=0.015) with our measure of a team’s assist generosity — suggesting whether teams were more likely to give assists overall wasn’t related to whether they were more generous to their own players than to opponents. ^
  6. As mentioned in an earlier footnote, we’d expect about one school each to rank in the top and bottom 15 percent of each of our measures. ^
  7. They had two road games with six or fewer assists and 19 or more field goals in 2012-13. ^
  8. Brickman also had 12 or more assists in two home games against Sacred Heart over the last two seasons — plus 13 this season in a visit to Sam Houston State. ^
  9. Rutgers wasn’t included in the list of outliers because it had fewer than five pairs of home-away games in 2011-12 and 2013-14. ^
  10. I watched a video cut of the Bearkats’ assists using Synergy Sports Technology. ^
  11. I took a closer look at Sam Houston State’s stats because of the Bearkats’ status as outliers among assists outliers. Using data from Sports Reference over the last four seasons, I pooled the Bearkats’ Division I games into two groups: those that were part of home-away pairs against the same opponent in the same season, and those that weren’t.

    For the first group, I ran three regressions using the dummy variable of home or away games: one with the games’ total assist rate, one with the Bearkats’ assist rate and one with their opponents’. All yielded highly significant results (p<10^-4). The first indicated that teams’ combined assist rates rise by 26 percentage points at Sam Houston State. The second indicated that the Bearkats’ assist rate rises by 35 percentage points at home. And the third showed opponents’ assist rate rises by 20 percentage points when visiting Sam Houston State.

    The second group was all Bearkats games that didn’t fit neatly into home-away pairs — many of these were against nonconference opponents or Southland teams that played the Bearkats just once before the conference tournament. I didn’t include these sorts of games in most analyses because other factors such as team matchups could come into play. Nonetheless, as a check, I ran similar regressions. Since I didn’t have paired games to compare, I checked three variables:

    1. total assist rate for these games minus the average of the total assist rate in all other games played by the Bearkats and their opponent in that game;
    2. the Bearkats’ assist rate in that game, minus the average of their assist rate in the rest of their games, and the assist rate yielded by their opponents in the rest of their games;
    3. their opponents’ assist rate in that game, minus the average of that team’s assist rate in the rest of their games, and the assist rate yielded by the Bearkats in the rest of their games.

    For each I ran a regression, with a dummy variable for home, away and neutral-site Sam Houston State games. The analyses found, with a high degree of significance (p<10^-7), that Sam Houston State’s assist rate was 33 percentage points higher at home than on the road and 27 points higher at home than on neutral courts; that the Bearkats’ opponents’ assist rates climbed by 24 and 27 percentage points when playing at Sam Houston State relative to their home court or a neutral court, respectively; and that the total assist rate was 28 and 27 points higher at Bernard G. Johnson Coliseum than on the road or on a neutral court.

    The story from the two sets of games was consistent, suggesting that the effect isn’t isolated to Sam Houston State’s conference opponents. ^

  12. A charting project by 82games.com found that in the NBA, passes that would likely have counted as assists boosted shooting percentage on close shots by more than three times the boost for three-point shots. ^
  13. UIC ranked just below the top 15 percent in our sample in 2011-12 and 2012-13. ^

Filed under , , ,

comments Add Comment

Powered by WordPress.com VIP