commute time (timej1 - timej16)

Dear fellow AHS users,

 

This question is in regards to the national AHS survey.  The 1997-2003 codebook describes the range of timej variables as 001-998.  In those years, there is a clumping of values at 996, which encoded “Does not work” and “Works at home” in previous survey periods.  I suspect that the clumping is either a miscoding or there is an error in the data dictionary.  Has this problem been found before?  Is there an explanation?  If this is a data dictionary problem, is there an updated dictionary?  If this is a data problem, are there updated timej values? 

 

Grateful for any advice,

Ilya Beylin