Back to History Current Version

When Castes Collide: A Field Experiment in India

Last registered on July 24, 2018


Trial Information

General Information

When Castes Collide: A Field Experiment in India
Initial registration date
December 20, 2016

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
December 20, 2016, 8:26 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated
July 24, 2018, 1:30 PM EDT

Last updated is the most recent time when changes to the trial's registration were published.



Primary Investigator


Other Primary Investigator(s)

Additional Trial Information

Start date
End date
Secondary IDs
52% of upper caste brahmins in India continue to practice untouchability. These group relations persist despite castes living in close proximity (albeit in segregated villages). Integration may be a natural policy response, though (a) little is known about how effectively contact can reduce prejudice; and (b) even less is known about how the type of contact might mediate effects, despite Allport's (1954) famous hypothesis being that contact should work only if certain conditions are met. This field experiment then tests an integrative program in rural India, within which participants are randomised to different conditions, intended to affect the "type" of contact. To do this, I use different incentive structures: half the participants are paid monetary incentives that encourage competition, whilst the rest get incentives that encourage collaboration. The full design allows me to distinguish between the effects of program participation, exposure to other castes, incentives, and the interaction between incentives and exposure (a test of whether the type of contact mediates its effects).
External Link(s)

Registration Citation

Lowe, Matt. 2018. "When Castes Collide: A Field Experiment in India." AEA RCT Registry. July 24.
Former Citation
Lowe, Matt. 2018. "When Castes Collide: A Field Experiment in India." AEA RCT Registry. July 24.
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details


Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
- Individual and Collective (Team) Voting on Field Trip Participants (for individual-level caste preference and its translation to group decisions)
- Caste Implicit Association Tests (for subconscious caste biases)
- Trading Exercise (for caste bias in willingness to interact/cooperate for economic surplus)
- Future Team Choice (for caste bias in willingness to interact/network formation)
- Trust Games (for generalised caste bias).
- Social Networks (for network formation)
- Goldberg Paradigm Vignettes (for generalised caste prejudice/attributions)
- Match-by-Match Observations (for within-match conflict/cooperation/team cohesion)
- Individual and Team Performance (for efficiency effects) and Allocation (for within-match prejudice)
Primary Outcomes (explanation)
The outcomes are largely revealed and stated preference measures of various aspects of caste prejudice/bias. In this case, though data on subcaste (jati) will be collected, the primary focus is on three broad caste groups: General, OBC and SC/ST. It is these three groups that will be used for stratification in the randomisation, and it is these three groups which will be used for the primary analysis of effects on bias/prejudice towards own vs. "other" castes.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Several waves of month-long cricket tournaments will be organised, with the tournaments serving as a means of integrating men (aged 14 to 30) from the full spectrum of caste groups.

For each tournament, sign-ups will be randomised to:

- Control vs. Treatment
Control men will not play in the tournament.

- Homogeneous vs. Mixed Teams
Those treated will be randomised to either homogeneous-caste or mixed teams.

- Individual vs. Team Pay
Participating teams will be randomised to be paid match-by-match monetary incentives that are either Individual (encouraging competition) or Team-based (encouraging collaboration).
Experimental Design Details
For each tournament, eligible men can sign up as players, umpires (referees), or both. Umpires will be paid Rs. 40 for each match they adjudicate.

After a 5-day sign-up period, we return to administer ability testing (of bowling, batting and fielding ability) for all those that signed up. These ability measures permit testing of whether contact effects depend on the ability of those a person is exposed to.


Individual-level randomisation is carried out after ability testing is complete.

Those in the Control group are split caste-wise (General, OBC and SC/ST), and then each group is randomly split into two (Individual vs. Team Pay). These six groups are then randomly ordered. Each list then serves as a priority ordered set of "backups" for any cases in which those in the Treatment group cannot attend matches they have been scheduled to play in. (For example, if an OBC player with Team Pay cannot play in his match, we attempt to replace him with the Control group member who is atop of the OBC-Team Pay backup list.) This protocol serves as a means of preserving the contact and incentive treatments as far as possible, even in the case of imperfect attendance of matches. Provided backups are not used too frequently (and provided we do not go too far down the list), a large enough Control group is also preserved in order to test the effect of program participation.

After each match, there is also an "income lottery": 1 person will randomly win Rs. 100, and 1 person will randomly win Rs. 50. This enables me to isolate income effects.


There is also randomisation at the match-level:

- I first create a random fixture schedule subject to the constraint that each team plays 8 matches, never playing the same team twice.
This gives randomness in the quality of opponents. The pre-determined quality of a given opponent can then be used as an IV for whether a team wins a given match (permitting a test of whether "winning", or achieving a common goal, increases team cohesion and reduces caste prejudice).

- Each match is randomised to have either a General or OBC or SC/ST umpire.
This randomisation is not of primary interest, but given that some allocation of umpires to matches is needed, random allocation comes at no extra cost.


The matches played are limited overs cricket – with 7 overs (1 over = 6 balls) batting, 7 overs bowling (meaning that matches cannot go on indefinitely. In practice, each match lasts roughly 1 hour).

- Individual Pay: when teams are in this treatment group, each player gets Rs. 2.5 per run scored when batting, Rs. 35 per out when bowling. Put another way, each player’s pay depends only on his own performance.

- Team Pay: each player gets Rs. 1/2 per run scored by any team player, and Rs. 7 per out by any team player. Unlike in Individual Pay all players now get the same pay regardless of the match’s outcome.

Payouts are made immediately after each match finishes. After the tournament is over, the 1st, 2nd and 3rd best teams and 1st, 2nd and 3rd best players get additional trophies and monetary prizes.

In the hope of encouraging attendance, players will also each receive a Rs. 10 attendance bonus per match, and all players that attend at least 6 matches are entered into a lottery to win a cricket bat.
Randomization Method
Randomisation done in office by a computer
Randomization Unit
Individual, Team, Match
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
120-160 Teams
Sample size: planned number of observations
900-1200 Men
Sample size (or number of clusters) by treatment arms
Per tournament (for 6 to 8 tournaments) (rough, depends on number of sign-ups):
50 men in Control / 100 men (=20 teams) in Treatment
10 Teams Individual Pay / 10 Teams Team Pay
1/3rd of teams (to closest integer) homogeneous-caste (cross-randomised with incentives)
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number


Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information


Is the intervention completed?
Intervention Completion Date
July 03, 2017, 12:00 +00:00
Data Collection Complete
Data Collection Completion Date
July 28, 2017, 12:00 +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
160 cricket teams + 461 "pure control" individuals
Was attrition correlated with treatment status?
Final Sample Size: Total Number of Observations
1261 men
Final Sample Size (or Number of Clusters) by Treatment Arms
461 "pure control" individuals 800 assigned to cricket leagues, of which: 280 had homogeneous-caste team 52 had 1 other-caste on their team 174 had 2 other-castes on their team 180 had 3 other-castes on their team 114 had 4 other-castes on their team Or alternatively, of 800 assigned to cricket leagues: 8 had 35-40% of opponents from other-caste 62 had 40-50% 112 had 50-60% 321 had 60-70% 179 had 70-80% 122 had 80-90% 1 had 90-100%
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Integration is a common policy used to reduce discrimination, but different types of integration may have different effects. This paper estimates the effects of two types of integration: collaborative and adversarial. I recruited 1,261 young Indian men from different castes and randomly assigned them either to participate in month-long cricket leagues or to serve as a control group. Players faced variation in collaborative contact, through random assignment to homogeneous-caste or mixed-caste teams,
and adversarial contact, through random assignment of opponents. Collaborative contact reduces discrimination, leading to more cross-caste friendships and 33% less own-caste favoritism when voting to allocate cricket rewards. These effects have efficiency consequences, increasing both the quality of teammates chosen for a future match, and cross-caste trade and payouts in a real-stakes trading exercise. In contrast, adversarial contact generally has no, or even harmful, effects. Together these findings show that the economic effects of integration depend on the type of contact.
Lowe, Matt (2018) “Types of Contact: A Field Experiment on Collaborative and Adversarial Caste Integration,” Working Paper.
I estimate the effects of collaborative and adversarial intergroup contact. I randomly assigned Indian men from different castes to participate in cricket leagues or to serve as a control group. League players faced variation in collaborative contact, through random assignment to homogeneous-caste or mixed-caste teams, and adversarial contact, through random assignment of opponents. Collaborative contact increases cross-caste friendships and efficiency in trade, and reduces own-caste favoritism. In contrast, adversarial contact generally reduces cross-caste interaction and efficiency. League participation reduces intergroup differences, suggesting that the positive aspects of intergroup contact more than offset the negative aspects in this setting.
Lowe, Matt. 2021. "Types of Contact: A Field Experiment on Collaborative and Adversarial Caste Integration." American Economic Review, 111 (6): 1807-44.

Reports & Other Materials