The all_data file is a tab-delimited data file containing all subject decisions and responses from Studies 1 and 2 of the paper Sequential Search and Learning from Rank Feedback: Theory and Experimental Evidence by Asa B. Palley and Mirko Kremer to appear in Management Science (2014).

The variables in the file are as follows (note we use the words option and apartment interchangeably here as they denote the same thing):

gsid: A unique subject identifier.
treatment: Indicates the treatment the subject was assigned to. Treatments 1-4 are from Study 1 and treatment 5 was from Study 2 of the paper. Treatment 1 is the search condition with no discounting and full information, treatment 2 is the search condition with no discounting and partial information, treatment 3 is the search condition with a discount rate of 10% and full information, and treatment 4 is the search condition with a discount rate of 10% and partial information. Treatment 5 was the apartment value estimation task from Study 2 in the paper.
trial: Indicates which of the 50 trials the subject was playing (a trial is a set of 20 apartments that the subject could consider). Trials were indexed as t=1...T (T=50 in Study 1, and T=25 in Study 2), with trials -1 and 0 indicating the 2 initial practice rounds. Note that for four subjects (22-25) in our sample, the data from the last round (i.e., trial 50) is missing, due to a software issue.
discount: The additional discount factor 1/(1+r) that was applied to the payoff in that condition for each additional option considered (in other words, the payoff from selecting apartment m would be discounted by a factor of (1/(1+r))^m).
trial_id: Indicates which set of trials the subject was presented with. (We pre-generated 7 sets of 50 trials (with 20 apartment values each), resulting in a total of 50*7=350 independently sampled trials.)
option_number: Indicates which number apartment was being considered (m=1...20).
option_value: The true value of the apartment being considered (a real number between 0 and 20).
option_rank_ascending: The rank of the apartment being considered (option_number=m) among all of the m apartments observed so far in that trial. An apartment with a rank of 1 indicates that the apartment was the best out of the m and an apartment with a rank of m indicates that the apartment was the worst out of the m. This can be translated to the order statistic k used in the mathematics of the paper according to the equation k=m+1-option_rank_ascending. This was the information given to subjects in the partial information treatments 2,4, and 5.
option_expected_value: The true (objective) expected value of the option given the relative rank information (option_rank_ascending). This was calculated according to the equation 20*k/(m+1).
continuation_value: The optimal continuation value V_m (or V_m^FI) at that point in the search.
decision_to_stop: An indicator variable for whether the subject chose to stop (ending the search) and select that apartment. decision_to_stop=1 if the subject stopped and selected that apartment. decision_to_stop=0 if the subject did not select that apartment (either they observed it and decided to continue or they did not observe it because they stopped already).
decision_to_stop_opt: An indicator variable for whether that apartment would have been selected by a decision maker following the optimal policy throughout that trial. decision_to_stop=1 if the value of the current apartment (either the exact value in the full information conditions or the objective expected value in the partial information conditions) exceeds the optimal continuation value and the optimal decision maker would not have stopped previously. decision_to_stop_opt=0 if that apartment would not have been selected by following the optimal policy throughout the search (either they would have observed it and decided to continue or they would not have observed that apartment because it was already optimal to stop earlier in the trial).
decision_to_stop_cond_opt: An indicator variable for whether it would be optimal to select that apartment if it were observed. This is determined by simply comparing the value (either the exact value in the full information conditions or the objective expected value in the partial information conditions) of the apartment to the optimal continuation value. decision_to_stop=1 if the value of the current apartment exceeded the optimal continuation value. decision_to_stop=0 if the value of the current apartment was not greater than the optimal continuation value.
estimate: The subjects estimated value of the apartment in Study 2 (treatment 5). estimate=0 if the apartment was not observed because the search had ended. estimate=0 for treatments 1-4 because these subjects were not asked to provide estimates.
option_observed: The  variable indicates whether, in a given trial, the subject observed this option (1) or did not observe the option (0).
