Here is the introduction:
In his comment, Train (2015) claims to have uncovered “numerous errors and misstatements” in our recently published article (Alvarez et al., 2014) and accompanying corrigendum (Alvarez et al., 2015), yet close inspection reveals that all these errors and misstatements have been discussed at length in the article and corrigendum. This document outlines Train's claims and the corresponding section of the article and corrigendum that address each of his concerns. However, we first offer some background on the evolution of the research and the ensuing response from Train and BP plc.
The Alvarez et al. (2014) study relies on data continuously collected by a third party, the National Marine Fisheries Service, for purposes other than measuring the effects of oil spills and made publicly available online as the MRFSS/MRIP data. The study was conceived as an illustration of using the MRFSS/MRIP data to create a choice model capable of depicting angler behavior in response to a quality change brought about by an oil spill. In doing so, our study diverges from the previous literature on the effects of oil spills on recreational use of natural resources, which relied entirely on post-hoc surveys of recreational users designed for the sole purpose of fulfilling the data needs for the creation of these models. While our approach overcomes the post-hoc nature of data collection prevalent in the previous literature, it suffers from the necessary shortcoming of relying on data that was not collected for the purpose at hand. To get around these shortcomings, a series of assumptions must be made, and it is with these assumptions that Train appears to be taking issue. As will be discussed in this reply, these assumptions are clearly documented in the original article.
Furthermore, the Alvarez et al. (2014) study was conceived, designed, and carried out completely independent of the legal process surrounding the Deepwater Horizon spill. It did not rely on funding from any stakeholder group, and its objectives or content were never reviewed by attorneys involved in either side of the legal process. However, the research was presented at various conferences and other venues, and comments and suggestions received from participating scholars shaped the research process, which began in late 2010 and culminated with the publication of the article in late June 2014.
In August 2014, we were contacted by Train, who expressed concerns about the compensation estimate documented in Alvarez et al. (2014), in particular the use of the ‘short-cut’ formula for estimation of welfare changes, or equation (7) in Alvarez et al. (2014). He also suggested we retract the article. In response, we developed the corrigendum which was submitted to the journal and shared with Train as a courtesy during September 2014. The corrigendum focuses on the use, and misapplication, of the ‘short-cut’ formula, explains the relationship between the welfare estimate and the number of sites affected, and provides a number of alternate welfare estimates that would result under different scenarios of sites affected. Besides addressing Train's initial concern, the corrigendum adds to the literature by shedding light on the correct use of the short-cut formula, and as such, it should prove useful to the Journal of Environmental Management's readership.
In October 2014, we were again contacted by Train with additional concerns. As a professional courtesy we shared all of our raw data, computer code, and estimation datasets that offered complete disclosure of our approach and assumptions, which we had already fully disclosed in Alvarez et al. (2014). Disclosure of the data and computer code allowed Train to re-estimate or re-design the model using differing assumptions, and these results could be used for their scholarly merit or as evidence in the legal process.
In the following section we list the issues that Train claims to have uncovered and outline the sections of the article and corrigendum where they have been discussed. In summary, none of these topics add to the peer-reviewed literature with respect to either the theory or empirical application. ...
