Experimental Design
Our experimental design involves a coordination game we refer to as “Choose Something”. In what follows we describe our Baseline treatment.
In the Baseline condition, each subject is assigned to a computer terminal and each is shown the same six objects on her own screen. The objects are blocks (i.e., colored shapes), which the computer program identifies by the numbers (invisible to the subjects) 1 to 6. The blocks are loosely arranged in a hexagonal fashion (one object per vertex) and collectively occupy the left-hand side of the screen. After each participant has viewed the six objects on her screen, the computer program selects three of those objects – one by one – by sliding them and putting them in a column on the right-hand-side of the screen (note: the same three objects for each member of a pair); the rest of the objects subsequently disappear from the screen. Subjects are then prompted to complete a few tasks. The list below specifies the entire sequence of events.
i. Each subject is assigned to a computer terminal and everyone is shown the paper instructions. Pairs of subjects are formed at random.
ii. Subjects are presented with six objects on their screen, three of which are subsequently selected by the program and put in a column.
iii. An on-screen message prompts subjects to label those three objects [“PART A”].
iv. An on-screen message prompts subjects to estimate the probabilities of the three objects being chosen by others [“PART B”]; they are informed that good guesses will be rewarded with an additional payment.
v. An on-screen message prompts subjects to choose an object (by ticking the box located next to the label); they are reminded that their payoff will be $1 if both members of the pair chose the same object, $0 otherwise [“PART C”].
vi. Steps ii.-v. are repeated for 9 more rounds, whereby in each round a new three-object selection is implemented by the program and shown to each member of a pair. (In each round, subjects are randomly assigned to another pair, and are so informed.) No feedback is given between rounds.
vii. Payment.
We stress that the design of our Baseline treatment involves no exogenous labels. In summary, in order to easily identify the blocks (for the purposes of performing tasks iv. and v.), in PART A of each round subjects are asked to type a short text in each of three boxes beside the objects. (We note that our hypothesis-testing does not rely on such qualitative data; in fact, the labeling task simply serves the purpose of pushing subjects to think of the game in their own terms.) After that, in PART B of each round, subjects are asked to estimate the likelihood of each block being chosen by others. Finally, in PART C of each round, subjects are asked to choose a block. The entire set-up is common information among all participants.
The design of the “All-Aware” treatment is the same as the Baseline except for PART B, which presents 3 extra questions. Specifically, in PART B subjects see the following message.
«Recall that – in Part C of the experiment – you will be prompted to pick one object in order to coordinate with your partner. Now, prior to that we would like to know what you think about the other participants in this room. Please answer the following questions by moving the sliders to the desired percentages. Note that your partner will not be asked to answer these questions.
1) How likely do you think it is that the other participants have noticed the order in which the objects have been drawn by the computer program? Please move the below slider.
2) How likely do you think it is that the other participants have noticed the different colors of the objects? Please move the below slider.
3) How likely do you think it is that the other participants have noticed the different shapes of the objects? Please move the below slider.»
A few comments are due. The order in which questions 1) - 3) are presented is randomized in each round. Subjects enter their guesses by moving a slider (i.e., one slider per question) to the desired percentage, with the slider ranging from 0% to 100%. Note that the questions above are not incentivized: the purpose of these questions is to make each subject privately aware of multiple attributes. Also note that, after presenting questions 1) - 3) above, the All-Aware treatment proceeds to the incentivized belief-elicitation task described at step iv. of the Baseline; the rest of the treatment is identical to the Baseline.
Finally, we have designed three supplemental treatments to be the same as the Baseline, except for the labeling task of step iii. above (i.e., PART A): in such treatments the labeling task is omitted as the blocks are already labeled. Specifically, in PART A of the “Color-Aware”, “Shape-Aware”, and “Order-Aware” treatments, each of the available objects is displayed beside a color, shape, or order label, respectively. The purpose of such labels is to bring about common awareness of the relevant frame.