Talk:Average treatment effect

Estimand? WTF is "casual estimand"??

Dr. Huber's comment on this article
Dr. Huber has reviewed this Wikipedia page, and provided us with the following comments to improve its quality:

"For the following equation, some parameters are not defined in the text: "y = \Beta_{0} + \delta_{0}d2 + \Beta_{1}dT + \delta_{1}d2 \cdot dT , where y is the response variable and \delta_{1} measures the effects of the policy change on the population." This makes it hard for non-experts to understand the aim of the equation.

Furthermore, this regression equation is in fact a way to represent the diff-in-diff approach. This is not obvious from the discussion of diff-in-diff: "The difference in differences equation would be \hat \delta_{1} = (\bar y_{2,T} - \bar y_{1,T}) - (\bar y_{2,C} - \bar y_{1,C}), where T is the treatment group and C is the control group. In this case the \delta_{1} measures the effects of the treatment on the average outcome and is the average treatment effect. "

In fact, an ESTIMATE of the regression equation (which is however defined in terms of population parameters) would also correspond to the "diff in diff equation"

In general I find the diff in diff discussion sloppy because it only works under particular assumptions (common trends) which are not mentioned, so the discussion is not well connected to the rest of the article. Furthermore, I think the following claim is somewhat sloppy: "In this case the \delta_{1} measures the effects of the treatment on the average outcome and is the average treatment effect." Strictly speaking it only estimâtes the ATE among the treated population (which is different to the total population)

Also, Diff in Diff represents only one way to estimate treatment effects and maybe now deserves too much attention vis a vis of IV, natural experiments, regression discontinuity etc

A further remark concerns the references: At least under "Further reading", the following article should occur (as it appears to be the most important one, at least in economics):

Imbens, Guido W. and Jeffrey M. Wooldridge. 2009. Recent developments in the econometrics of program evaluation. Journal of Economic Literature 47, no. 1: 5-86. Published Version doi:10.1257/jel.47.1.5"

We hope Wikipedians on this talk page can take advantage of these comments and improve the quality of the article accordingly.

Dr. Huber has published scholarly research which seems to be relevant to this Wikipedia article:


 * Reference : Frolich, Markus & Huber, Martin & Wiesenfarth, Manuel, 2015. "The Finite Sample Performance of Semi- and Nonparametric Estimators for Treatment Effects and Policy Evaluation," IZA Discussion Papers 8756, Institute for the Study of Labor (IZA).

ExpertIdeas (talk) 17:46, 21 July 2015 (UTC)

Dr. Mellace's comment on this article
Dr. Mellace has reviewed this Wikipedia page, and provided us with the following comments to improve its quality:

"The introduction could be improved. The sentence "the ATE measures the difference in mean (average) outcomes between units assigned to the treatment and units assigned to the control." is in my opinion misleading. In fact, it should be clear to the reader that the ATE is a causal (unknown) parameter which is not necessarily equivalent to the average of the difference in the (observed) outcomes between treated and control units. In my opinion the estimation section is very confusing. Neither Natural experiment nor quasi-experiment are estimation methods. As the related articles describes those two terms actually include all of the methods listed below them. Strictly speaking with difference in differences it is only possible to identify the ATE on the treated population known as the ATET or ATT and not the ATE  (unless the two are the same of course). Similarly regression discontinuity designs allow to estimate the ATE only locally at the threshold and not the ATE for the entire population. I'm not sure what "methods based on the theory of local IVs" means. Similarly this sentence doesn't make much sense to me: "Once a policy change occurs on a population, a regression can be run controlling for the treatment. The resulting equation would be". Moreover, I find the Diff-in-diffs example right below this sentence not necessary and potentially misleading.

It is also important to mention IV an identification strategy for the LATE (see Imbens and Angrist, 1994)."

We hope Wikipedians on this talk page can take advantage of these comments and improve the quality of the article accordingly.

We believe Dr. Mellace has expertise on the topic of this article, since he has published relevant scholarly research:


 * Reference : Huber, Martin & Mellace, Giovanni, 2012. "Relaxing monotonicity in the identification of local average treatment effects," Economics Working Paper Series 1212, University of St. Gallen, School of Economics and Political Science.

ExpertIdeasBot (talk) 16:41, 27 July 2016 (UTC)

Dr. Koch's comment on this article
Dr. Koch has reviewed this Wikipedia page, and provided us with the following comments to improve its quality:

"This is fairly well-done. The "Estimation" section focuses a bit too much on one type of approach, although four additional approaches are mentioned. In my view, each of those should receive a small paragraph, while any additional links can also be included to more specific pages. It also appears that "local IVs" are not linked anywhere, thus, I would suggest a linking page related to local IVs.

More generally, the references are a bit short, as well. I would suggest at least including Angrist and Pischke's "Mostly Harmless Economics" textbook"

We hope Wikipedians on this talk page can take advantage of these comments and improve the quality of the article accordingly.

We believe Dr. Koch has expertise on the topic of this article, since he has published relevant scholarly research:


 * Reference : Dambala Gelo & Steven F. Koch, 2011. "The Welfare Effect of Common Property Forestry Rights:Evidence from Ethiopian Villages," Working Papers 201123, University of Pretoria, Department of Economics.

ExpertIdeasBot (talk) 16:11, 24 August 2016 (UTC)