7

For a paper, I need an example of a dataset $(x_i,y_i)$ where the residuals are $iid$ (the $x$ do not represent time) with a discontinuity on the effect of $x$ on $y$.

  • I have already found a dataset in Berger and Pope 2010 but the jump there is not clearly visible
  • The 'LifeCycleSavings' dataset has been suggested, but it has only 50 observations (ideally, I would be looking at 100).

Does anyone know of other datasets?

dimitriy
  • 31,081
  • 5
  • 63
  • 138
user603
  • 21,225
  • 3
  • 71
  • 135
  • The variates cannot possibly be iid if there is any effect at all. Presumably you mean there is a function $f$ such that $y_i - f(x_i)$ are iid conditional on the $x_i$ (or some such thing) and where $f$ has a discontinuity. – whuber Mar 02 '11 at 19:49
  • yes. Corrected. – user603 Mar 02 '11 at 19:53
  • 5
    See the genetic sequences at http://anson.ucdavis.edu/~mueller/changetrees.pdf (at the end). A dataset of 56 cases where a slope exhibits a discontinuous jump is given at http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_mcmc_sect048.htm . Googling "multivariate changepoint problem dataset -time" yields promising hits. – whuber Mar 02 '11 at 21:51
  • 1
    Thank you @WHuber, the google keyword you suggested lead me to examples from a problem i did not know of. Can you post your comment as an answer (Snijders and al 2001). – user603 Mar 03 '11 at 04:53

1 Answers1

2

In economics, this is called "regression discontinuity." For one example, check out

David Card & Carlos Dobkin & Nicole Maestas, 2008. "The Impact of Nearly Universal Insurance Coverage on Health Care Utilization: Evidence from Medicare," American Economic Review, American Economic Association, vol. 98(5), pages 2242-58. [data]

Here's a summary piece:

Lee, David S., and Thomas Lemieux. 2010. "Regression Discontinuity Designs in Economics." Journal of Economic Literature, 48(2): 281–355. [article page]

Hope that helps.

Charlie
  • 13,124
  • 5
  • 38
  • 68