Dr. Tsiatis' work is widely cited as creative and path-breaking and has had considerable impact on statistical theory and practice. His productivity and prowess as an influential researcher is reflected in the almost 30 years of continual support (since 1983) for his methodological research through grants from NIH. Methods he has developed alone and with colleagues and his 40 PhD students are widely used and are now considered standard techniques for data analysis or have served as pioneering innovations that have established new methodological areas and motivated new advances by others; space considerations allow noting only a few highlights. His paper entitled "A nonidentifiability aspect of the problem of competing risks" (Tsiatis, 1975) has been cited over 500 times in the statistical, operations research, and engineering literature. His Annals of Statistics article presenting the first study of the asymptotic properties of the partial likelihood estimator for the proportional hazards model (Tsiatis, 1981) has received close to 400 citations, while his 1990 Annals paper "Estimating regression parameters using linear rank tests for censored data" (Tsiatis, 1990) has been cited nearly 300 times. Dr. Tsiatis is responsible for groundbreaking advances in the theory of group sequential testing that revolutionized the conduct of large clinical trials (e.g., Tsiatis, Rosner, and Mehta, 1984; Scharfstein, Tsiatis, and Robins, 1997) that are now standard in software routinely used in practice, and he is broadly recognized as the "founder" of what is now widely-used and cited methodology for joint statistical modeling of censored time-to-event outcomes and longitudinal processes (e.g., Tsiatis, DeGruttola, and Wulfsohn, 1995; Tsiatis and Davidian, 2004). His highly cited work with James Robins (Robins and Tsiatis, 1991, 1992) laid the groundwork for numerous advances in semiparametric method and causal inference, and his work on new methods for consistent estimation of quality adjusted survival time (Zhao and Tsiatis, 1997) broke new ground in this area. Methods he developed the for the first time allowed consistent estimation of costs with censored data (Bang and Tsiatis, 2000) are now considered standard among health policy economists. Dr. Tsiatis is responsible for important and extensive work on the use of semiparametric theory to derive methods for numerous data-analytic challenges (e.g., Tsiatis and Ma, 2005; Davidian, Tsiatis, and Leon, 2005; Zhang, Tsiatis, and Davidian, 2008; Cao, Tsiatis, and Davidian, 2009), and his 2006 book (Tsiatis, 2006), Semiparametric Theory and Missing Data, is a best-seller and considered the definitive text on the subject. Dr. Tsiatis continues to make notable contributions to statistical methodology; his more recent work in press on estimating optimal dynamic treatment regimes and causal inference (e.g., Zhang et al., 2012), for example, has already garnered citations and requests.
Dr. Tsiatis is also renowned as an exceptional educator and mentor. His reputation as an outstanding instructor is reflected in his receipt of the Excellence in Continuing Education Award from the ASA, which he was awarded for his one-day short course on Semiparametric Theory and Missing Data presented at the 2011 Joint Statistical Meetings. He has guided 40 PhD students and several post-docs at Harvard and NCSU, many of whom have gone on to distinguished research careers in their own rights, including as department chairs and leaders of research organizations.
