x

We are happy to announce that all trial registrations will now be issued DOIs (digital object identifiers). For more information, see here.
Context Effects in Strategic Environments: An Experimental Study
Last registered on May 13, 2019

Pre-Trial

Trial Information
General Information
Title
Context Effects in Strategic Environments: An Experimental Study
RCT ID
AEARCTR-0004129
Initial registration date
April 30, 2019
Last updated
May 13, 2019 11:35 PM EDT
Location(s)

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Affiliation
University of Haifa
Other Primary Investigator(s)
PI Affiliation
University of Haifa
PI Affiliation
Tel Aviv University
Additional Trial Information
Status
In development
Start date
2019-05-07
End date
2020-01-31
Secondary IDs
ISF 664/17
Abstract
The goal of this study is to examine whether some the most well-known decision biases show up in strategic environments. The decision biases we explore are the attraction effect, the compromise effect and the duplication effect. These biases, that are part of a larger group of behavioral biases known as context effects, have been well-documented in individual decision making but received little attention in interactive environments. In a lab experiment, we design two-player games in which different strategies are added to one of the players' set of strategies. The player whose strategy set is manipulated is placed in a position where his strategic choice may reflect one of the above context effects. Using a between-subject analysis, we will examine these effects in two types of 2x2 simultaneous-move games: coordination games and games in which one player has a dominating strategy. We will measure the direct effect on the player whose strategy set is manipulated and also measure the indirect effect as reflected through the strategic choices of the opponent. In other words, we test not only whether players with the extended strategy set react to the context effects but also whether the players, who face those with the extended set, anticipate the behavioral bias and respond to it.
External Link(s)
Registration Citation
Citation
Arad, Ayala, Benjamin Bachi and Amnon Maltz. 2019. "Context Effects in Strategic Environments: An Experimental Study." AEA RCT Registry. May 13. https://doi.org/10.1257/rct.4129-1.0.
Former Citation
Arad, Ayala, Benjamin Bachi and Amnon Maltz. 2019. "Context Effects in Strategic Environments: An Experimental Study." AEA RCT Registry. May 13. https://www.socialscienceregistry.org/trials/4129/history/46375.
Experimental Details
Interventions
Intervention(s)
We investigate two types of two-player matrix-form games. The first is coordination games (with two equilibria) and the second is games where one player has a dominating strategy (one equilibrium). Overall, we have 8 games which belong to two families (4 coordination games and 4 with a dominating strategy). Within a family of games the differences between them are only in the monetary payoffs and the "location" of the equilibria. For each of the 8 games, the control is the 2x2 "base game" where each player has two pure strategies and he must choose one of them. The treatments (interventions), which we also call "extensions", are based on the same 2x2 game with one additional strategy to the row player (so that it becomes a 3x2 game). This strategy may be dominated by another strategy in the game, extreme compared to the other strategies or identical to one of the strategies, reflecting the three context effects we are interested in examining (attraction, compromise and duplication effect, respectively). The analysis will be based on comparing the distribution of choices of the different strategies in the control (base game) and the treatments (extensions) for each one of the 8 games. More specifically, we will compare the proportion of those choosing the strategy which is predicted to be chosen more often by the context effects (which we also call the "target") in the extensions to the proportion that choose it in the base game. We will hold such a comparison for each extension separately, both for specific games and for all 4 games of the same type. Our interest lies in whether or not this comparison shows differences in choice frequencies that are along the lines of the documented context effects in studies of individual choice.
Intervention Start Date
2019-05-07
Intervention End Date
2020-01-31
Primary Outcomes
Primary Outcomes (end points)
Primary outcome variables are choices of strategies of the row player in the control game compared to treatments. More specifically, the proportion of choices of the "target" strategy in the base games compared to their corresponding extension games. These outcome variables will be examined and compared on game-specific and context-specific levels as well as on more aggregate levels. These outcomes will allow testing whether the three context effects we examine show up for players who are introduced with the manipulation that is expected to generate these effects (according to standard individual behavior findings).
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary outcome variables are choices of strategies of the column player in the control game compared to treatments. More specifically, the proportion of choices of the best response to the "target" strategy in the base games compared to its proportion in the extension games. These outcome variables will be examined and compared on game-specific and context-specific levels as well as on more aggregate levels. These outcomes will allow testing whether the three context effects we examine are anticipated by players who are playing against those who have their choice set manipulated.
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
Participants play 36 simultaneous games in the lab; in each game they play against a randomly selected participant from the same session (who remains anonymous) and their roles as column or row player change throughout the experiment. These 36 games include: 8 base games (the controls, which are 2x2 games), 3 different extensions (3x2 games) for each base game, and 4 "distraction games". There are two versions of these 36 games with the roles of column and row players reversed across versions to allow participants to play against each other (we will have an even number of participants in each session - half will play each version - so that everyone in the session is matched with an opponent). To control for the order in which the 36 games are presented, each version has two opposing orders. For every one of our 8 games, participants in a given version either play the control 2x2 matrix or the three 3x2 extended games (i.e., between-subject design); in no case will a participant play a base game and one of the same game's extensions in the same role. Moreover, the base game is always separated from its extensions by a number of other games and two extensions of the same type are also separated by at least two other games (other extensions or base games). Four games that are completely unrelated to the goal of our study and are not even described in matrix form (the "distraction games") are scattered in between the 32 games of interest to assist in keeping participants fresh and in order to increase the separation between games that share similar features.
Experimental Design Details
Not available
Randomization Method
Randomization done by a computer.
Randomization Unit
Individual level.
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
160-250 university students
Sample size: planned number of observations
160-250 university students
Sample size (or number of clusters) by treatment arms
80-125 in the control for each game and 80-125 in each of the corresponding treatments.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
Tel Aviv University
IRB Approval Date
2019-04-30
IRB Approval Number
N/A