NotesFAQContact Us
Search Tips
Peer reviewed Peer reviewed
PDF on ERIC Download full text
ERIC Number: ED562332
Record Type: Non-Journal
Publication Date: 2015
Pages: 6
Abstractor: ERIC
Reference Count: 3
Identifying the Average Causal Mediation Effects with Multiple Mediators in the Presence of Treatment Non-Compliance
Park, Soojin
Society for Research on Educational Effectiveness
Identifying the causal mechanisms is becoming more essential in social and medical sciences. In the presence of treatment non-compliance, the Intent-To-Treated effect (hereafter, ITT effect) is identified as long as the treatment is randomized (Angrist et al., 1996). However, the mediated portion of effect is not identified without additional unverifiable assumptions (Yamamoto, 2013; Robins, 2003). The inability to identify the mediated portion of ITT effect presents a serious problem in conducting causal mediation analysis in the presence of treatment non-compliance, and a researcher often uses the naïve approach for estimating the mediated ITT effect where 1) the treatment assigned is used as if it was the treatment received and 2) a standard causal mediation analysis is conducted. Yamamoto (2013) proposed an alternative approach to identifying the mediated portion of the ITT effect using the instrumental variable approach, and compares his proposed approach to the naïve approach when a single mediator exists. Considering the complexity of the causal mechanisms in reality, it may be natural to assume multiple variables that mediate through the causal path. Thus, this paper extends Yamamoto (2013)'s approach to the multiple-mediators case where mediators do not or do influence one another. One table is appended.
Society for Research on Educational Effectiveness. 2040 Sheridan Road, Evanston, IL 60208. Tel: 202-495-0920; Fax: 202-640-4401; e-mail:; Web site:
Publication Type: Reports - Research
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: Society for Research on Educational Effectiveness (SREE)