David Hensher "How do Respondents Process Stated Choice Experiments? -
Attribute Consideration under Varying Information Load", Journal of Applied
Econometrics, Vol. 21, No. 6, 2006, pp. 861-878.

Data are in the file hensherdat.txt, an ASCII file in DOS format. It is
zipped in the file hensherdat.zip. Unix users should use "unzip -a".

There are sixteen variables, in the following order:

altz, naig, nlvls, ntb, nalts1, fftd, addtim, congt1d, congt2d, timtd, 
chset, pinc, naign6, naign5, naign4, naign3

Definitions of Variables:

Altz		Selection rule to eliminate subsets of data
Naig		Selection rule to eliminate subsets of data
Nlvls 		number of levels defining an attribute
Ntb		narrow range in design
Nalts1		number of alternatives in choice set
Fftd		freeflow time of base alt minus freeflow time in stated 
                choice alternative
Congt1d		congestion time of base alt minus congestion time in stated 
                choice alternative
Congt2d		congestion time of base alt minus congestion time in stated 
                choice alternative
Timtd		total time of base minus total time in stated choice 
                alternative
AddTim		In SC expt 1=added times up in response, 0= did not add 
                them up, 3=N/A
Chest		Number of choice sets evaluated
Pinc		Annual Personal Income coded in natural units (using top of 
                range): 0,10,15,20,30,40,50,60,80,100,120,140 
Naign6		dependent variable for 6 attribute alternatives
Naign5		dependent variable for 5 attribute alternatives
Naign4		dependent variable for 4 attribute alternatives
Naign3		dependent variable for 3 attribute alternatives

(Note: congt1d is sum of slowed down and stop/start time; congt2d is the
single attribute slowed down/stop/start time)

Descriptive Statistics
All results based on nonmissing observations.
===============================================================================
Variable        Mean         Std.Dev.        Minimum         Maximum      Cases
===============================================================================
-------------------------------------------------------------------------------
All observations in current sample
-------------------------------------------------------------------------------
ALTZ      10.5909306      4.16463486      5.00000000      20.0000000      12702
NAIG      8.06046292      4.20883819      1.00000000      15.0000000      12702
NLVLS     3.19059991      .816314517      2.00000000      4.00000000      12702
NTB       .273500236      .445772926      .000000000      1.00000000      12702
NALTS1    4.09470949      .825784407      3.00000000      5.00000000      12702
FFTD      5.31024220      9.30488266     -42.0000000      52.0000000       9744
ADDTIM    1.32805857      .980640088      .000000000      3.00000000      12702
CONGT1D  -7.55954186      11.1816296     -98.0000000      32.0000000       5937
CONGT2D  -6.51063830      11.1957376     -54.0000000      37.0000000       3807
TIMTD    -2.25693036      11.8722591     -54.0000000      32.0000000       2958
CHSET     11.7080775      3.16356650      6.00000000      15.0000000      12702
PINC      58.8320737      32.5226645      10.0000000      140.000000      12702
NAIGN6    .355928200      .830575114      .000000000      4.00000000      12702
NAIGN5    .954180444E-01  .310987330      .000000000      2.00000000      12702
NAIGN4    .305384979      .605489653      .000000000      3.00000000      12702
NAIGN3    .102975909      .316129094      .000000000      2.00000000      12702

Note: -999 means not applicable to that data subset.

Model Specification used in Limdep8/Nlogit3

*********natts=6*******************
sample;all$
reject;altz<5$ To reject base alt
reject;naig=0$
reject;naig>5$
Ordered;lhs=naign6
;rhs=one,nlvls,ntb,nalts1,fftd,congt1d,addtim 
;het;rh2=  congt1d,chset 
;logit ;marginal effects$

************natts=5*******************
sample;all$
reject;altz<5$ To reject base alt
reject;naig=0$
reject;naig<6$
reject;naig>8$
Ordered;lhs=naign5
;rhs=one,nlvls,ntb,nalts1,fftd,congt1d,addtim 
;het;rh2= congt1d,chset ,pinc  
;logit ;marginal effects$ 

*********natts=4*******************
sample;all$
reject;altz<5$ To reject base alt
reject;naig=0$
reject;naig<9$
reject;naig>12$
Ordered;lhs=naign4
;rhs=one,nlvls,ntb,nalts1,fftd,congt2d,addtim 
;het;rh2=congt2d,chset,pinc 
;logit ;marginal effects$ 

*********natts=3*******************
sample;all$
reject;altz<5$ To reject base alt
reject;naig=0$
reject;naig<13$
crosstab;lhs=naign3;rhs=chset$
dstats;rhs=naign3$
ordered;lhs=naign3
;rhs=one,nlvls,ntb,nalts1,timtd,pinc 
;het;rh2= chset,pinc
;logit ;marginal effects$ 

Background Source of Data

The data are drawn from a larger study, reported in Hensher (JAE), in which
16 stated choice sub-designs have been embedded in one overall design, with
each sub-design being used in surveying a sample of car commuter trips in
Sydney in 2002. Each commuter evaluated one randomly assigned sub-design;
however, across the full set of stated choice experiments, the designs
differed in terms of the number, range and levels of attributes, the number
of alternatives and the number of choice sets. The combination of these
dimensions of each design is often seen as the source of design
'complexity', and it is within this setting that we have varied the number
of attributes that each respondent is asked to evaluate. 

For further details, please see the file Hensher-Data.pdf.