Home advantage in chess.

Sorqvist, Patrik ; Halin, Niklas ; Kjellberg, Anders 等

The home team tends to win about 60% of all games across different types of sports (Jamison, 2010). Previous research has identified a number of factors that contribute to this home advantage, such as crowd size (Nevill, Newell, & Gale, 1996), crowd density (Agnew & Carron, 1994), the away team's traveling distance (Pace & Carron, 1992), biased referees (Balmer, Nevill, & Williams, 2001), and familiarity with the home field (Pollard, 2002). One methodological difficulty faced by these studies is how team quality influences with the game results. For instance, one could expect that the better team wins the game, regardless of who plays at home, but that this win is smaller when the best team plays away due to the home advantage. The home advantage may thus interact with team quality, and hence balancing the number of times two teams meet at their respective home field does not solve the problem. To control for team quality is clearly difficult when investigating the home advantage in sports like football, because team quality must be quantifiable and reliably predict the outcome of the game, although a few attempts have been made (Bray, Law, & Foyle, 2003; Madrigal & James, 1999). One sport (or competitive activity) that makes such an analysis possible is chess. In chess, each player is given a rating based on success in previous game play and the rating is a strong predictor of game outcome. The purpose of this paper was to investigate whether there is a home advantage in chess and to test if it persists when player rating is statistically controlled.

In the Swedish chess league, each team consists of eight players. When two teams compete, one player from each team plays against one (and only one) player from the other team, and the competition between the two teams is settled by summarizing the results across the eight individual chess boards. The players receive one point for a win and half a point for a draw. Hence, the two teams split a total of eight points between them, and the team that receives most points across all individual boards wins the competition. All players who participate in a chess club (whether they are elite players or not) obtain a rating score based on their previous game play. Typically, players begin with a rating of 1300 and gain rating by winning chess games (i.e., 16 rating points if against another player of the same rating, more if the win is against a player with higher rating, and less if the win is against a player with lower rating) and lose rating points by losing chess games. A typical 'master' player has a rating above 2000 and a 'world class' player has a rating above 2600. In general, the player with the highest rating in one team plays against the player with the highest rating in the other team (although this may vary within a limited degree of freedom because of team tactics). This rating system allows the researcher to keep player quality under statistical control. Other competitive activities (or sports) in which players are ranked also allow researchers to investigate whether the home advantage persists when controlling for player quality. In tennis, for instance, elite players are assigned a certain rank in relation to the other players. However, chess is particularly suitable for this analysis since all players have a rating and the players are not ranked inter-individually.

There is another reason why investigating the home advantage in chess is particularly interesting. Based on previous studies (Courneya & Carron, 1992; Jamison, 2010), one would expect that the advantage of playing at home should be very small (or even non-existent) when there is no crowd and when the role of the referee is negligible. Chess seems particularly well suited to test this prediction. First, regularly there is no audience to a chess competition (except other chess players who play separate games); and second, the referee is extremely seldom called into action, and when he is, there is little uncertainty in the judgments (e.g., the referee may help to determine whether a player has managed to make the minimum number of moves within a certain time limit, or whether a player has accidentally made an illegal move). The role of the referee is therefore negligible. These observations, amongst other factors, suggest that there should be no home advantage in chess. In this study, we used archival data to investigate (1) whether there is a home advantage in chess and (2) whether the home advantage is still present when team quality (i.e., differences in chess rating) is statistically controlled.

Method

Sample

Archival data on chess results from the Swedish chess league was used. The data was retrieved from Internet (www.schack.se) in September 2010. All results from three chess divisions ("Superettan", "Division 1" and "Division 2") played between season 2000/2001 and 2009/2010 were included in the study (not the highest division called "Elite", because in this division, all teams meet at the same time and place to play games in parallel, and hence in principle there is no actual home/away team). The total number of team competitions was 3208 (Superettan: N= 135; Division 1 : N = 1274; Division 2: N= 1799). In 299 competitions, at least one of the two teams lacked a player on at least one of the eight tables (i.e., one of the teams gained one point because the other team lacked a player). For simplicity, these competitions were excluded from the analysis (leaving 131 in Superettan, 1177 in Division 1, and 1601 in Division 2; a grand total of 2909 team competitions). Team rating was calculated by taking the arithmetic mean of the individual players' ratings in each team and competition respectively.

Analysis

For each competition between two teams in the archival data, one team is assigned "home team" and one is assigned "away team'. This is typically because the competition takes place in the home team's clubhouse to which the away team has to travel. As the two competing teams split a pot of 8 points between them, the home team's points are completely dependent on the away team's points and vice versa. This "zero-sum game" makes a between-subject statistical comparison between home and away teams inappropriate. To overcome this problem, we randomly selected home teams from half of the 2909 competitions (a home group) and away teams from the other half (an away group). In the home group, the result of the competition was expressed from the perspective of the home team (i.e., as the difference between home team score and away team score). In the away group, the opposite difference was calculated (i.e., the difference between the score of the away team and the home team). Thus, a positive score in the home group meant that the home team had won the competition, whereas as a positive score in the away group meant that the away team had won. These difference scores thus could vary between -8 and 8. A multiple regression analysis was performed with this difference score as the dependent variable. The variable home-away team was entered as predictor in a first step, and the difference in mean rating between the teams was entered in the second step (i.e., in the home group, the difference between mean rating in the home and away team; in the away group, the difference between the away and home team).

Results

A hiearchial regression analysis was conducted to test whether there is a home advantage in chess and whether this remains when the teams' ratings are statistically controlled. Home versus away team was added as an independent variable in the first step of the analysis. There was a very small but significant difference between the home (M = 0.09, SD = 3.09) and away score (M = -0.22, SD = 3.01), [beta] = .05, [t,sub.1,2907] = 2.79, p < .01; adjusted [R.sup.2] = .002. When statistically controlling for the difference between teams in mean rating (home team, M = 13.58, SD = 142.69; away team, M = -14.47, SD = 138.10), by adding rating difference between home and away team as an independent variable in the second step of the regression model, the difference between home and away score was no longer significant, [beta] = -.02, [t.sub.1,2906] = -1.36, p =. 17, and even turned into a slight away team advantage. The rating difference was a highly significant predictor of the result of the competition, [beta] = .70, [t.sub.1,2906] = 52.84, p < .001, adjusted [DELTA][R.sup.2] = .49. There was no significant interaction between home/away team and rating difference.

Discussion

The results show that there is a home advantage in chess, but the reason for this is that the home team tends to have players with higher rating than the away team does, probably because some players have difficulty taking part in games other than at home. When differences in rating were statistically controlled, the home advantage disappeared.

The study reported here is, to our knowledge, the first systematic demonstration of a home advantage in chess and the findings have some potentially important implications. An applied implications is that competitions are not solely settled by the team's capability to play chess, but also seemingly by systematic differences in the availability of players, which leads to unfair situations whereby the success of some teams depends on having players who are free to travel. One way to solve this problem is to facilitate the possibility of players taking part in chess competitions through the Internet or similar digital techniques without the need to travel to the home team's clubhouse.

The study also has theoretical implications. The results are quite supportive of the assumption that the crowd and referee are the two major underlying causes of the home advantage in sports and other competitive activities (Courneya & Carron, 1992; Jamison, 2010). Specifically, when there is no crowd and when there is no influence from the referee, as is the case in chess, and when the two teams have players of equal quality as ensured by statistical procedures in the present study, there is no home advantage. The absence of a home advantage when team quality is statistically controlled also, arguably, rules out some other possible causes to the home advantage such as self-fulfilling prophecies (i.e., the tendency to play better when at home just because one knows that home teams in various sports have an advantage). Future studies could experimentally test whether a crowd would cause a home advantage in chess.

References

Agnew, G. A., & Carton, A. V. (1994). Crowd effects and the home advantage. International Journal of Sport Psychology, 25, 53-62.

Balmer, N. J., Nevill, A. M., & Williams, A. M. (2001). Home advantage in the Winter Olympics (1908-1998). Journal of Sports Sciences, 19, 129-139.

Bray, S. R., Law, J., & Foyle, J. (2003). Team quality and game location effects in English professional soccer. Journal of Sport Behavior, 26, 319-334.

Courneya, K. S., & Carron, A. V. (1992). The home advantage in sport competition: A literature review. Journal of Sport and Exercise Psychology, 14, 13-27.

Jamison, J. P., (2010). The home field advantage in athletics: A meta-analysis. Journal of Applied Social Psychology, 40, 1819-1848.

Madrigal, R., & James, J. (1999). Team quality and the home advantage. Journal of Sport Behavior, 22, 381-398.

Nevill, A. M., Newell, S. M., & Gale, S. (1996). Factors associated with the home advantage in English and Scottish soccer matches. Journal of Sports Sciences, 14, 181-186.

Pace, A., & Carron, A. V. (1992). Travel and the National Hockey League. Canadian Journal of Sports Sciences, 17, 60-64.

Pollard, R. (2002). Evidence of a reduced home advantage when a team moves to a new stadium. Journal of Sports Sciences, 20, 969-973.

Patrik Sorqvist

University of Gavle, Sweden

Swedish Institute for Disability Research, Linkoping University, Sweden

Niklas Halin and Anders Kjellberg

University of Gavle, Sweden

Address correspondence to: Patrik Sorqvist, Department of Building, Energy and Environmental Engineering, University of Gavle, SE-801 76 Gavle, Sweden. Email: patrik.sorqvist@hig.se