The Variance of Non-Parametric Treatment Effect Estimators in the Presence of Clustering

Samuel G. Hanson; Adi   Sunderam

}
//]]>
</script>
<input type="hidden" name="__VIEWSTATEGENERATOR" id="__VIEWSTATEGENERATOR" value="E39AC016" />
<script type="text/javascript">RegisterSod("sp.core.js", "\u002f_layouts\u002fsp.core.js?rev=7ByNlH\u00252BvcgRJg\u00252BRCctdC0w\u00253D\u00253D");</script><script type="text/javascript">RegisterSod("sp.res.resx", "\u002f_layouts\u002fScriptResx.ashx?culture=en\u00252Dus\u0026name=SP\u00252ERes\u0026rev=b6\u00252FcRx1a6orhAQ\u00252FcF\u00252B0ytQ\u00253D\u00253D");</script><script type="text/javascript">RegisterSod("sp.ui.dialog.js", "\u002f_layouts\u002fsp.ui.dialog.js?rev=Tpcmo1\u00252FSu6R0yewHowDl5g\u00253D\u00253D");RegisterSodDep("sp.ui.dialog.js", "sp.core.js");RegisterSodDep("sp.ui.dialog.js", "sp.res.resx");</script><script type="text/javascript">RegisterSod("core.js", "\u002f_layouts\u002f1033\u002fcore.js?rev=Emb5hNX1RZMNXk9oS82yvA\u00253D\u00253D");</script>
<script type="text/javascript">
//<![CDATA[
var _spFormDigestRefreshInterval = 1440000;window.g_updateFormDigestPageLoaded = new Date(); window.g_updateFormDigestPageLoaded.setDate(window.g_updateFormDigestPageLoaded.getDate() -5);function loadMDN2() { EnsureScript('MDN.js', typeof(loadFilterFn), null); }
function loadMDN1() { ExecuteOrDelayUntilScriptLoaded(loadMDN2, 'sp.ribbon.js'); }
_spBodyOnLoadFunctionNames.push('loadMDN1');//]]>
</script>
<script type="text/javascript">RegisterSod("sp.runtime.js", "\u002f_layouts\u002fsp.runtime.js?rev=IGffcZfunndj0247nOxKVg\u00253D\u00253D");RegisterSodDep("sp.runtime.js", "sp.core.js");RegisterSodDep("sp.runtime.js", "sp.res.resx");</script><script type="text/javascript">RegisterSod("sp.js", "\u002f_layouts\u002fsp.js?rev=\u00252B4ZEyA892P3T0504qi0paw\u00253D\u00253D");RegisterSodDep("sp.js", "sp.core.js");RegisterSodDep("sp.js", "sp.runtime.js");RegisterSodDep("sp.js", "sp.ui.dialog.js");RegisterSodDep("sp.js", "sp.res.resx");</script><script type="text/javascript">RegisterSod("cui.js", "\u002f_layouts\u002fcui.js?rev=OOyJv78CADNBeet\u00252FvTvniQ\u00253D\u00253D");</script><script type="text/javascript">RegisterSod("inplview", "\u002f_layouts\u002finplview.js?rev=WB6Gy8a027aeNCq7koVCUg\u00253D\u00253D");RegisterSodDep("inplview", "core.js");RegisterSodDep("inplview", "sp.js");</script><script type="text/javascript">RegisterSod("ribbon", "\u002f_layouts\u002fsp.ribbon.js?rev=F\u00252BUEJ66rbXzSvpf7nN69wQ\u00253D\u00253D");RegisterSodDep("ribbon", "core.js");RegisterSodDep("ribbon", "sp.core.js");RegisterSodDep("ribbon", "sp.js");RegisterSodDep("ribbon", "cui.js");RegisterSodDep("ribbon", "sp.res.resx");RegisterSodDep("ribbon", "sp.runtime.js");RegisterSodDep("ribbon", "inplview");</script><script type="text/javascript">RegisterSod("sp.ui.policy.resources.resx", "\u002f_layouts\u002fScriptResx.ashx?culture=en\u00252Dus\u0026name=SP\u00252EUI\u00252EPolicy\u00252EResources\u0026rev=YhBHGmUAGyJ3lAgSdE4V\u00252Fw\u00253D\u00253D");</script><script type="text/javascript">RegisterSod("mdn.js", "\u002f_layouts\u002fmdn.js?rev=gwmFFJ2\u00252FfFacqXWAqG\u00252FqKg\u00253D\u00253D");RegisterSodDep("mdn.js", "sp.core.js");RegisterSodDep("mdn.js", "sp.runtime.js");RegisterSodDep("mdn.js", "sp.js");RegisterSodDep("mdn.js", "cui.js");RegisterSodDep("mdn.js", "ribbon");RegisterSodDep("mdn.js", "sp.ui.policy.resources.resx");</script>
<script type="text/javascript">
//<![CDATA[

function _spNavigateHierarchyEx(nodeDiv, dataSourceId, dataPath, url, listInContext, type, additionalQString) {
    SetAdditionalNavigateHierarchyQString(additionalQString);
    g_originalSPNavigateFunc(nodeDiv, dataSourceId, dataPath, url, listInContext, type);
}

g_originalSPNavigateFunc = _spNavigateHierarchy;
_spNavigateHierarchy = _spNavigateHierarchyEx;
var _fV4UI = true;//]]>
</script>
<script type="text/javascript">RegisterSod("msstring.js", "\u002f_layouts\u002f1033\u002fmsstring.js?rev=QtiIcPH3HV7LgVSO7vONFg\u00253D\u00253D");</script><script type="text/javascript">RegisterSod("browserScript", "\u002f_layouts\u002f1033\u002fnon_ie.js?rev=EVTj1bu32\u00252FMla6SDN\u00252FsNTA\u00253D\u00253D");</script>
<script type="text/javascript">
//<![CDATA[
var __wpmExportWarning='This Web Part Page has been personalized. As a result, one or more Web Part properties may contain confidential information. Make sure the properties contain information that is safe for others to read. After exporting this Web Part, view properties in the Web Part description file (.WebPart) by using a text editor such as Microsoft Notepad.';var __wpmCloseProviderWarning='You are about to close this Web Part.  It is currently providing data to other Web Parts, and these connections will be deleted if this Web Part is closed.  To close this Web Part, click OK.  To keep this Web Part, click Cancel.';var __wpmDeleteWarning='You are about to permanently delete this Web Part.  Are you sure you want to do this?  To delete this Web Part, click OK.  To keep this Web Part, click Cancel.';//]]>
</script>
</form>

Go to main content

Article | Review of Economics and Statistics | November 2012

The Variance of Non-Parametric Treatment Effect Estimators in the Presence of Clustering

by Samuel G. Hanson and Adi Sunderam

Abstract

Non-parametric estimators of treatment effects are often applied in settings where clustering may be important. We provide a general methodology for consistently estimating the variance of a large class of non-parametric estimators, including the simple matching estimator, in the presence of clustering. Software for implementing our variance estimator is available in Stata.

Keywords: treatment effects; matching estimators; clustering; Software; Mathematical Methods;

Format: Print

Find at Harvard Read Now