A Review of: Canuel, R., Hervieux, S., Bergsten, V., Brault, A., & Burke, R. (2019). Developing and assessing a graduate student reference service. Reference Services Review, 47(4), 527–543. https://doi.org/10.1108/RSR-06-2019-0041 Abstract Objective – To evaluate the effectiveness of a reference training program for graduate student employees that seeks to encourage use of reference interview and instruction techniques in virtual and in-person reference interactions. Design – Naturalistic observation with qualitative content analysis. Setting – A large, public research university in Montreal, Canada. Subjects – Three graduate students in Library and Information Science employed by the university library to provide virtual and in-person reference services. Methods – After completing a training program, the three participants provided virtual and in-person reference training for two consecutive semesters. They self-recorded their desk interactions in a Google form. These self-reports, along with their online chat transcripts from QuestionPoint, were the subject of this study’s analysis. Focusing on the QuestionPoint data, the authors coded the transcripts from these participants’ online reference interactions to reflect the presence or absence of a reference interview and various instructional techniques in their responses to patrons. Also, all in-person and virtual questions were examined and categorized as being either transactional or reference questions. Reference questions were further categorized as basic, intermediate, or advanced questions. Main Results – Of the chat transcripts analyzed, 49% were classified as containing reference questions rather than transactional questions. At the desk, 21.9% of interactions were coded as reference questions. Taking the two semesters together, 232 of 282 virtual reference questions were considered basic, while 41 were labelled intermediate, and 9 classified as advanced. Similarly, of 136 desk reference questions, 120 were classified as basic, 14 as intermediate, and 2 as advanced. In their coding of chat transcripts, researchers indicated whether the interaction contained no reference interview, a partial reference interview, or a complete reference interview. Virtual chat transcripts from both fall and winter semesters showed that no reference interview took place in 77.3% of interactions. Authors noted evidence of partial reference interviews in 19.3% of fall transcripts and 21.5% of winter transcripts. Complete reference interviews took place in 3.4% of fall and 1.2% of winter transcripts. Additionally, authors found that 65.5% of chat transcripts contained elements of instruction, with Modelling and Resource Suggestion being the most prevalent forms. Conclusion – Because the graduate students used complete or partial reference interviews in a small number of their virtual reference questions, the authors of this study determined that more emphasis ought to be placed on reference interviews, particularly virtual reference interactions, in future training programs. Graduate students employed instructional strategies in observed virtual reference interactions, a promising trend.