Intervention (Hidden)
Methodological principles
Participants will self-select into our human-intelligence tasks (HIT) advertised on the internet and conducted via Amazon Mechanical Turk, as a means of earning money. Ultimately, an individual will make two decisions during the experiment which then, together with the decisions of other participants, determine the financial reward the participants receive. The financial reward is essential for two reasons. First, the participants (the so called “Mturkers”) are completing HITs to earn a living or at least to supplement their income. Second, studies in experimental economics have to be fully incentivized to obtain valid results. Fully incentivized means that the participants need to have the same incentives in the experiment as individuals have in the real world or in the models to be tested.
Main aim and participant choices
The aim for the participant is to earn some money by helping a group of participants (the electorate) to determine via majority vote, which of two equally likely events has occurred. The event is the outcome of a coin flip. Before deciding to vote for heads, tails, or abstain. A participant can decide to buy information, which allows the participant to update and learn, which of the two outcomes is more likely (with a probability of 2/3 versus 1/3). After the information acquisition has occurred (or not if the participant decided not buy), she has the choice to either vote for heads, tails or to abstain. Once this occurred, the HIT ends. The researches then tally the votes in an electorate and check if the electorate was able to determine the correct event. The participant will then be paid via the Amazon-Mechanical-Turk website.
The payment a participant receives consists of US$ 2 for reading the instructions and answering the control questions, minus the cost for acquiring information (between US$ 0 and 2) plus a bonus (US$ 5) if the electorate determined the correct event via majority vote. The matching of payment and participant takes place with the help of their alphanumeric id, such that the participants remain anonymous to the researchers.
Detailed protocol
During the HIT, a participant will first read and click through written online instructions. While progressing through the instructions (a sample can be found in Attachment 3),the participant will answer some questions testing the level of comprehension. A participant who correctly answers control questions can proceed, while wrong answers require the participant to return to the relevant sections in the instructions before proceeding.
The instructions are truthful and complete (i.e. they do not contain any deception). Hence, once participants have read and understood the instructions they will be fully informed about the task at hand and how the payment will be determined. Once a participant has proceeded through the instructions she will be credited with a reward of US$ 2 as a base payment.
Then the participant is asked if she wants to pay a randomly determined amount (between US $0 and 2) to open one of three boxes. Two of the boxes contain the correct outcome while the outcome contained in the remaining box is the wrong one. It has been made clear in the instructions, that outcome that opening a box reveals has a probability of 2/3 of being the correct outcome. Hence, buying information improves the likelihood that the participant can pick the correct event from 1/2 to 2/3.
Finally, the participant is asked to either vote for one of the events (heads or tails) or to abstain. The rational game-theoretic prediction is that the participant should never vote without acquiring information. Participants whose information cost lie below a certain threshold should acquire the information and vote, while those with cost above the threshold should abstain. The threshold depends on the size of the electorate.