Contribution
It separates rivalry from competition and shows that rivalry belongs to the relationship. Earlier work treated competition as a property of the situation: whether goals are opposed, how scarce the resource is. Here the same teams, in the same conference with the same rules and stakes, feel very different intensities toward different opponents. That difference is predicted by their shared history.
Key insight
Competition leaves a residue. Contests long since decided keep shaping how much the next one matters, so competition is “path-dependent” (p. 948). Even matches matter more than similar status: once the historic head-to-head balance is in the model, similarity in all-time winning percentage drops out (Model 16).
Assumptions
- Teams as unitary actors, justified by the homogeneity of NCAA teams.
- Sportswriters as informants. Each school’s student sportswriters rated how much “your team” sees each conference opponent as a rival, on a 0–10 scale. No definition of rivalry was given, to capture the lay concept.
- Defensive statistics as effort. Effort is assumed to show in defence more than in shooting accuracy.
- Generality across levels (individuals, groups, firms) is argued, not tested.
Key results
-
Sample (pp. 950–951). 421 sportswriters at the 73 universities in six major conferences. After screening, 362 remained at 71 schools; mean ICC .93. Players and coaches at 11 schools (134 and 23) agreed with the sportswriters, r = .89. 778 perceiver–target pairs, mean rivalry 5.02.
-
Where the variance lies (social relations model, pp. 952–953).
- relationship: 50.4%;
- target: 26.2%;
- perceiver: 4.6%;
- error: 18.7%.
Relationship variance exceeded perceiver plus target in every conference. Example: Oregon State rated Oregon 10.0, with its own mean 5.50 and Oregon’s mean as target 5.25.
-
Reciprocity. Across 389 dyads, r = .64; .85 with perceiver and target effects removed. No predictor of within-dyad asymmetry was found.
-
Antecedents (Table 3, team fixed effects, n = 318 dyads).
- distance: t = −8.80;
- same state: t = 7.26;
- status difference, all-time: t = −3.47;
- academic-quality difference: t = −3.20;
- games played, controlling distance: t = 3.03;
- historic competitiveness: t = 5.00.
-
History over recency. With both in the model, historic status similarity and historic competitiveness stayed significant and their recent counterparts did not (Models 13–15).
-
Target effects. High-status teams drew rivalry (r = .73 with all-time winning percentage; R² = .77 for the full model). One-sided rivalry runs from lower-status teams to higher-status ones.
-
Consequences (563 games, 2005–06).
- attendance: t = 3.75;
- points per 100 possessions fell, t = −2.14 (1.19% per SD of rivalry);
- blocked shots rose, t = 1.99 (4.47% per SD);
- steals and free-throw accuracy: n.s.
The null on free throws argues against “choking”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Rivalry varies mostly at the level of the particular pair of competitors, not with either competitor’s attributes | strong in this setting: round-robin design, social relations model, consistent in all six conferences | Part I |
| C2 | Rivalry grows with similarity, repeated competition and, above all, a history of even contests | moderate: cross-sectional; scheduling could follow rivalry | Part II |
| C3 | Competition is path-dependent: past contests keep raising the stakes of present ones | moderate: historic predictors beat recent ones; objective-stakes explanations are not excluded for every predictor | Part II, Models 13–16 |
| C4 | Rivalry increases effort | weak: small effects on defensive statistics, offence and defence confounded at game level | Part III |
| C5 | Rivalry is largely reciprocated; one-sided rivalry runs toward higher status | moderate for teams in major conferences | Parts I–II |
Concepts
- rivalry: “a subjective competitive relationship that an actor has with another actor that entails increased psychological involvement and perceived stakes of competition for the focal actor, independent of the objective characteristics of the situation” (p. 945).
- relationship effect: in the social relations model, the part of A’s rating of B not explained by A’s general tendency or B’s general pull.
- competitiveness index: closeness of the historic head-to-head record to 50–50.
Connections
- Berendt and Uhrich (THEORY-1113) take their definition of rivalry from this paper.
- Tyler and Cobbs (LIT-1682), Unfetched, document one-sided fan rivalries. Here, among teams, one-sidedness is rare and follows status.
- Intergroup bias. It cites Tajfel et al. 1971 (LIT-1697) and Jetten et al. on similarity and threat. Rivalry rising with similarity fits the distinctiveness-threat line that self-categorization theory (THEORY-1114) folds in; the paper does not use the theory.
- Observers. Attendance rising with rivalry is read as rivalry “spreading to those indirectly involved” (p. 961), and how observers feel rivalry is listed as a future direction. The fans are not studied.
Bearing on the record
- CASE-098 (a rivalry that outlives its carriers). It supports the case by giving a mechanism. Rivalry is measurably a property of the pair, and it is driven by a history of contests most current players never played in. All-time records beat recent form. That is the inheritance the case describes, measured in the institution rather than the fans. It does not show who carries it: sportswriters’ and players’ current ratings are the only measure. A dated note is added to the case.
- QUESTION-298 (is perceived reciprocity real?). A partial answer at the level of teams. Measured from both sides, rivalry was largely returned (r = .85 net of team effects), and one-sided rivalry ran from lower to higher status. If fans’ perceptions follow their teams’, the perceived reciprocity Berendt and Uhrich found is mostly accurate, with an error expected from fans of lower-status clubs. The question asked about fans, so it stays open; a note is added to it.
- The owner’s thesis. Little. The paper assumes that group members who identify will be moved by their group’s rivalry (p. 950), which is the thesis’s picture, but tests nothing about members.
- No THEORY filed.
- No ML instruction.
Limitations
- Objective stakes are not fully excluded. Proximity may matter because nearby teams compete for recruits and fans; the authors say so. The historic-competitiveness result is the one they say stakes cannot explain.
- Cross-sectional. Conferences may schedule rivals more often, so games played may follow rivalry.
- Informants, not members. The ratings are sportswriters’, validated at 11 schools.
- Small consequences, perhaps from a ceiling on athletes’ motivation.
- Sport only. The authors note intense loyalty in college athletics may make rivalry unusually strong.
Open questions
- Do fans’ rivalries have the same structure: pair-level, reciprocated, status-driven when one-sided? Tyler and Cobbs’s data could answer it.
- Do respectful and hostile rivalries (Bird and Johnson against British Airways and Virgin) have different antecedents? The authors raise it.