Chapter

Design and instrument for hidden bias

Myoung-Jae Lee

in Micro-Econometrics for Policy, Program and Treatment Effects

Published in print April 2005 | ISBN: 9780199267699
Published online February 2006 | e-ISBN: 9780191603044 | DOI: http://dx.doi.org/10.1093/0199267693.003.0005

Series: Advanced Texts in Econometrics

 						Design and instrument for hidden bias

More Like This

Show all results sharing this subject:

  • Econometrics and Mathematical Economics

GO

Show Summary Details

Preview

If the treatment (T) and control (C) groups are different in unobserved variables e as well as in observed variables x, and if e affects both the treatment and response, then the difference in outcome y cannot be attributed to the difference in the treatment d. The difference in x causing overt bias can be removed with one of the methods discussed in the preceding chapters, but the difference in e causing hidden bias is hard to deal with. In this and the following chapter, hidden bias due to the difference in e is dealt with. An econometrician’s first reaction to hidden bias (or endogeneity problem) is to ‘use an instrument’. But good instruments are hard to come by. Much easier, but less conclusive, is exploring ways to detect the presence of hidden bias; this is done in the name of ‘coherence’ (or consistency, if one does not mind the abuse of this term), whether the main scenario of the treatment effect is coherent with other auxiliary findings. This task can be done with multiple treatment groups, multiple responses, or multiple control groups, which are easier to find than instruments; checking coherence leads to an emphasis on study design rather than estimation techniques. The treatment-effect literature sheds new light on instrumental variables, i.e., the instrumental variable estimator is shown to be for the effect on those whose treatment selection is affected by the instrument.

Keywords: hidden bias; partial treatment; reverse treatment; multiple responses; multiple comparison groups; instrumental variable estimator; compliers

Chapter.  17301 words. 

Subjects: Econometrics and Mathematical Economics

Full text: subscription required

How to subscribe Recommend to my Librarian

Buy this work at Oxford University Press »

Users without a subscription are not able to see the full content. Please, subscribe or login to access all content.