Experimental Design Details
Participants are matched into pairs and informed that they’re playing with another person in the
same room. Endowments will be randomly distributed as 15 or 35 ECUs. All the players will be
informed that there’s an equal chance for both endowment levels for them and their partners.
In the first treatment, Hidden Information treatment from DWK(2006) will be implemented to
the trust game setup. Players won’t be aware of their partners’ endowment levels. Player 1 will
be informed that Player 2 can reveal information on Player 1’s initial endowment level. If Player 1
sends 10ECU to Player 2, Player 2 will make a decision on whether to learn Player 1’s endowment or
not. Following that decision a BDM auction will take place to implement the revealing information
In the other two treatment games, the information on partner’s endowment won’t be given to the
players and again Player 2s make decision on revealing Player 1’s endowment level or not. . The
treatments will be focused on the trustee’s behavior on revealing the information about trustor’s endowment
and will compare the willingness to get the information and to stay ignorant. . Design of the
game is very close to Hidden Information in (Dana et al., 2007) with the exception of manipulation
on revealing the information ex-ante and ex-post as in (Grossman and Van Der Weele, 2017).
The timing of the experiment is as follows. At the beginning of each treatment t = 0, subjects
are randomly paired and assigned to the roles as trustors and trustees and to the endowment levels as
high and low. In t = 1 trustors decide to send 10ECUs or nothing to the trustees without knowing
the endowment level of their partner. In t = 2 of the hidden information and ex-ante information
treatments trustees decide to get the information I = 0;1 about the trustors’ endowment and if they
don’t want to get that information they will attend in a BDM auction(b = 1). The trustors who want
to get information won’t take place in the auction(b = 0). The bids in the BDM auction will give the
continuous measure of willingnes to pay of to avoid information for each individual, k. That enables
me to see how they value the avoidance. In t = 3 the auction will be resolved and the randomly
chosen price will be announced to trustees. If they win information won’t be revealed and the price of
avoidance k will be charged from trustees. If they lose information will be revealed and nothing will
be charged. In t = 4, trustees make the distribution decision which is “a” in the self-signalling model
and changes the xixj. This decision is effected by the altruism level of the trustee q. Following the
proposed models with that decision the trustee get a signal about her own type and the psychological
utility of the signaled type is mq. Therefore, avoiding the information helps trustee to maintain her
belief on her own type. On the other hand without the self signalling tool the trustee only decides by
the altruism level of her. Since she knows the possible endowments in the first place even without the
information she suffers from that possibility as in Eq. 2. In t = 5 final payoffs are realized and in t=6
players take place in the post-experiment questionnaire. In the ex-post information treatment, t = 4
change place in the order and goes before t = 2, after making the distribution decision without any
information trustees are asked if they want to get information or not. In addition to the time order in
the ex-ante game, after the auction a revision part will take place in ex-post game. Therefore, in t = 5
after the auction is resolved the player who receive information are asked if they want to revise their
distribution decision. Trustors and trustees will not be informed about the revision part in advance.
And as in the ex-ante treatment in t = 6 the final payoffs are realized and in t = 7 a questionnaire
takes place (see fig.1). The main stages will be explained in a more detailed way in the following
paragraph. Information & Auction In the first treatment, trustees who choose to get the information
will attend an auction to realize that decision. The same mechanism takes place in other treatments
as well: in the ex-ante /ex-post information treatments before/after the distribution decision, trustees
will face a BDM auction to stay ignorant. Trustees will be told that they’re given additional 1ECU
and over that amount they can submit their reservation prices if they want to get the information on
their partner’s endowment, in the second and third treatment they submit reservation prices if they
want to stay ignorant. After trustees submit their reservation prices, a price will be randomly chosen
between 0-1ECU. If the bid is lower than the chosen price that subject loses the auction and don’t
receive/receive the information, if it’s higher then the price will be taken from the subject and will be
informed / ignorance situation will be kept. Not getting the information keeps the trustor’s endowment
and final payoff hidden from the trustee, in the self image model decreases the psychological cost of
that choice and protects the self-image in case of making an unfair decision. In the outcome based
models the information avoidance has no lowering effects on the psychological cost of the decision
since p and 1 p are known.
Sending Back After receiving the sent amount from the trustor, trustee will make the distribution
decision with/without knowing his partner’s initial endowment.To get a more detailed idea on the
insights of the decisions first order and second order beliefs about the other party’s endowment and
sending/sending back decision will be elicitated.
Revision In all treatments subjects makes conditional decisions on the possible endowment levels
of the trustor. Then, if they get the information in all treatments (if they win the auction in treatment
1 and loose in treatments 2 and 3) that conditional decision will be implemented. If they don’t get the
information they will be asked to make another sending back decision.
Other covariates like demographics, risk behavior, reciprocity etc. will be controlled in the questionairre
by using the survey from Falk et al. (2016).
Depending on the results of this experiment a second part of the idea will be conducted on social
image effects. The self-image model seems to provide a very useful tool to understand social decisions.
If it can be generalized to all social decisions the next step of the experiment will be to see the
different effects of self and social image. In the model multiple self concept is used and it enables
one to interpret the model in social image concept as well. In the self image there are two selves: the
observer and the decision maker. Both are not aware of the type of the player. So the observer self can
easily be interpreted as an other person. Although in the model self and social image seem to have
the same effect, as B´enabou and Tirole (2006) they might be weighted differently in subjects utility.
In short, even when there is no observer other than oneself, individuals engage in altruistic actions
like voting or donating to charities, or avoid information which can possibly hurt their self-image like
feedback on a homework or critics on the supported party etc. The present research idea with the
second part of it intends to see the effects of these image concerns(if any) in social decisions.