Sanghamitra Das, Charles F. Manski, and Mark D. Manuszak, "Walk or Wait? An 
Empirical Analysis of Street Crossing Decisions", Journal of Applied
Econometrics, Vol. 20. No. 4, 2005, pp. 529-548.

The data used in this article were gathered from four intersections in 
New Delhi, India. Vehicle and pedestrian flows were videotaped at the 
four intersections, coded separately, and then combined into a final 
dataset. Support for this data collection was generously provided by 
a grant from the Society of Indian Automobile Manufacturers to the 
Transportation Research and Injury Prevention Program of the Indian 
Institute of Technology, New Delhi.

The observations in the final data set gaps.txt are gaps in traffic 
associated with each pedestrian making a particular crossing at a 
particular intersection. Thus, an observation in the data file is 
a gap in a sequence for pedestrian i making a crossing from point 
j at intersection s. An (i,j,s) combination may have multiple gap 
observations associated with i.  A pedestrian-intersection 
combination (i,s) may make crossings from two points j,the origin 
and median. In total, there are 6,838 gap observations associated 
with 1,339 pedestrians. Each gap observation contains 13 variables 
relevant for the gap and (i,j,s) combination. (Note that the data 
also include observations for a fourth intersection that was not 
used in the estimation due to a different intersection design.)

The analysis in the paper concerns a pedestrian facing a sequence 
of gaps at a given crossing point. Hence, for each (i,j,s) combination, 
the sequence of gaps relevant for that crossing need to be extracted to
perform the estimation in the paper.

For some estimation routines, additional demographic information was
used. This demographic information is contained in the file peds.txt
This file contains age and gender for each of the 1,340 pedestrians
identified by their (i,s) combination.

There are two data files, gaps.txt and peds.txt. Both files are ASCII files
in DOS format. They are zipped in the file dmm-data.zip. Unix users should
use "unzip -a".

Documentation:

File Name: gaps.txt

Identification variables:

1) Intsctn - intersection number
   1 = Motibagh
   2 = Africa Ave.
   3 = AIIMS
   4 = Panchsheel
   Note: Data on Panchsheel not used in estimation (intersection is 
	different design and has small number of observations)

2) Drctn - direction of crossing
   Four main crossing patterns: {1-2, 2-1, 2-3, 3-2}

3) PedId - pedestrian id number


Gap characteristic variables:

4) GapNo - gap number in gap sequence

5) Gaps - total number of gaps (G) in pedestrian's gap sequence

6) GapSize - gap size (seconds)
   = -1 if gap dne (gap is not observed)
   Note: In the empirical analysis, unobserved initial gaps are included
	in the estimation and are set to have a size of 20 seconds.

7) Platoon - platoon gap indicator
   = number of vehicles in platoon if platoon gap
   = 0 if not platoon gap

8) Veh - vehicle length (meters)
   = -1 if veh dne (when gap dne)
   Note: when gap is a platoon gap, Veh = max(length of vehicles in platoon)

9) Veh2 - vehicle length 2 (meters)
   = mean(length of vehicles in platoon) if platoon gap
   = 0 if not relevant (i.e. Platoon=0)

10) Safe - safety status upon arrival at decision point
   = 1 if safe signal
   = 2 if quasi-safe signal (red/yellow in non-threatening direction)
   = 3 if unsafe signal
   Note: All estimates based only on observations not facing a safe signal
	(i.e. Safe ~= 1)

11) Time - time until safe change at decision time
   = -1 if safe (i.e. Safe = 1)

12) Accept - gap acceptance variable
   = time between gap arrival and pedestrian departure if gap is accepted
   = -1 if gap is rejected
Note: If gap is accepted then Accept >= 0

13) Wait - cumulative wait time (seconds) until gap arrived


File Name: peds.txt

Identification variables:

1) Intsctn -- intersection id
   1 = Motibagh
   2 = Africa Avenue
   3 = AIIMS
   4 = Panchsheel
2) Pedid -- pedestrian id number

Pedestrian info variables:

3) Gender	1 = M 		2 = F

4) Age	1 = child	2 = youth	3 = middle age		4 = old

5) Pedtype -- pedestrian type
   1 = normal adult
   2 = handicapped adult
   3 = adult w/ children
   4 = adult w/ heavy packs
   5 = child

For answers to questions regarding the data or data collection, 
contact:

Mark Manuszak
Assistant Professor of Economics
Graduate School of Industrial Administration
Carnegie Mellon University
5000 Forbes Ave.
Pittsburgh, PA 15213-3890
USA
manuszak [AT] andrew.cmu.edu