Abstract: This paper reports on a novel approach to providing undergraduates with feedback on STEM writing assignments via an otherwise untapped educational resource: university alumni and employees who normally play no role in the institution's educational mission. In the Volunteer Expert Reader (VER) approach, students are paired with volunteers whose backgrounds make them suitable readers for specific writing assignments. Given the realities of labor in STEM undergraduate teaching contexts, VER may be particularly valuable there, facilitating student interactions with experienced STEM professionals who have the time and inclination to give them substantive feedback on their writing based on real-world experience. Results from this 3-year study suggest that (1) VER can be an effective means of increasing student engagement in STEM writing assignments and may also improve the quality of student writing and increase learning of course content; and (2) how well VER works for any STEM writing assignment is dependent on a number of factors including whether student participation is required, whether students write alone or as a team, how well readers' backgrounds fit the assignment, and reader engagement and availability to students. VER is also an effective faculty development tool: it offers faculty in STEM disciplines a compelling reason to collaborate with writing program administrators and necessitates the inclusion of best practices in writing assignment design, such as making explicit the rhetorical context and aims of the writing task and setting a reasonable pace for drafting/revision cycles.
For it is time—or lack of it—that so often manages to redeem us from the admission that we have corrected, circled, checked, and assigned points to our students' writing but forgotten, in the arduous and painful process, to listen to what they have been saying.
—Chris Anson, Writing and Response: Theory, Practice and Research
When engineering employers were asked recently to rank the relative importance of various learning outcomes for prospective hires, communicating effectively came out on top—above engineering problem solving (Lattuca et al., 2006). This would surely surprise (and probably distress) those many undergraduate engineering students who assume that having chosen a major in a Science, Technology, Engineering or Math (STEM) discipline, they were safe from a future of writing.
While suggested approaches for improving written communication skills in STEM vary markedly, there does appear to be broad consensus about this: students need feedback on their writing that is timely and that sponsors serious rethinking and revision (see for example Harris & Schaible, 1997; Jerde & Taper, 2004; Gottschalk & Hjortshoj, 2004; McLaren & Webber, 2009). To date, research on innovation in STEM writing pedagogy has developed under the assumption that such feedback would necessarily come from the course instructor, a teaching assistant, classroom peers, or writing center tutors. Here I propose another option: members of the university community who are not on the teaching staff but who have the scientific, technical, or STEM management background needed to respond to STEM student writing as members of the target audience. In what I am calling the "Volunteer Expert Reader" (VER) approach, alumni and employees of the institution are solicited as "readers" for a course and then matched with students, giving them feedback on their work-in-progress. Depending on each reader's physical location and preferences, readers interact with the students in person or by webcam or phone, as well as in writing shared via email.
The broad goals of the current research were to learn whether VER could be an effective addition to STEM curricula and to uncover the factors associated with more or less favorable outcomes of the approach in various STEM course settings. This essay is intended to provide a big-picture view of this approach and this research to date. My objectives here are (1) to provide a description of the VER approach broadly conceived, in concept and in practice; (2) to present results, primarily from assessment surveys of students and volunteer readers, showing the major trends across a range of STEM courses; (3) to identify lessons learned to date from this research.
The main findings from this research are as follows:
I begin with a brief discussion of response to student writing, showing the importance of personalized, reader-centered feedback and the barriers that have historically limited students' ability to receive such feedback, especially in STEM. I then discuss VER in relation to these barriers and explain that why it offers a valuable type of feedback that many students are unlikely to get elsewhere followed by an overview of the VER approach. Next is an overview of assessment data from my research of VER in ten STEM courses followed by a discussion of lessons learned. I conclude with suggestions for future research. I this paper I use VER to refer to the pedagogical approach itself, in which students are matched with VE readers. Duke Reader Project refers to the particular enactment of VER at Duke University.
While formal research on feedback, or "response" to student writing dates back at least to the middle of the twentieth century, it has been a particular focus within writing studies research since the early 1980's resulting in a large and varied literature. Like any aspect of teaching, many strategies for giving feedback have been identified that might be productive in one educational setting or another. Yet is there anything we might identify as essential?
Horvath (1984) provides a summary and synthesis of the literature on response as "formative evaluation" to that date. Of the conclusions that Horvath draws, three are particularly relevant here and are broadly accepted within writing studies today: First, students benefit most from feedback that is neither "so tied to specific texts that students cannot discover in them general guidelines" (138) nor so generic that it fails to engage with the students' actual work on the page. Second, building on Sommers (1982), feedback is generally more effective when offered as a response to work-in-progress rather than to a finished product. Finally, and perhaps most important for VER, feedback should consider "the essay's full rhetorical situation" and "respond to it as an integrated work intent on accomplishing a certain aim—the student's intended aim—in the world" (140). (It does not go without saying, unfortunately, that providing such a response requires first that the student have an intended rhetorical aim beyond the obvious instrumentalist one of securing the desired grade. The matter of providing effective feedback thus circles back to the problem of designing rhetorically meaningful assignments.)
Writing and Response (Anson, 1989), a diverse collection of essays published just a few years later, presents response as an activity too complex to be reduced to a single technique or approach. Yet the importance of offering students the chance to be read is reiterated in many of these essays. In "Transactional Theory and Response," Robert E. Probst (1989) emphasizes the need for feedback that eschews the "narrow focus implicit in the roles of editor or diagnostician" (p. 74). Russell A. Hunt (1989) writes that "the most important change" we might make in how we give feedback is "finding ways to...make it real," and that "the most effective way to do this is to create situations in which student writing ... is read for its meaning, for what it has to say..." (p. 95). Taken as a whole, these essays confirm that whatever assistance and guidance students might receive, constructive response from an engaged reader is essential. Such detailed, personalized feedback continues to be identified as an essential component of effective writing pedagogy into the twenty first century (Gibbs & Simpson, 2004; Higgins, Hartley, & Skelton, 2002; Lipnevich & Smith, 2009).
We might also approach the question of what is essential in providing feedback by analogy. The jazz pianist must indeed master her "chops," but to really learn how to backup a singer or soloist, or to get "in the pocket" with a drummer and bass player, requires interacting with other musicians. Without the feedback of hearing the musical choices she makes in the context of what others play, how they respond with their own choices, and what they have to say about her playing, she can only imagine (hope) that her approach "works." Similarly, the aspiring basketball player learning to play a zone defense may understand it conceptually, but would we expect her to develop even the slightest proficiency on the court without the frequent opportunity to test those moves against real opponents? Finding out what happens when one does one's best in the context where it matters provides the feedback most essential for true advancement. For writers, such feedback comes from testing out our drafts with members of the target audience, finding out how they react to our texts.
Yet we rarely provide such opportunities for STEM undergraduates. Instead, we ask them to write as if they were making a pitch for a design to a potential investor, as if they were writing a research report for a journal, as if they were writing a policy memo on environmental policy to government officials. But how are they to judge whether these hypothetical attempts at writing are successful? Therein lies the challenge inherent in learning to write in school: as if. Without such feedback, students well know that they are not "engaged in real writing but only an exercise, an empty simulation of writing" (Probst, 75). The realities of undergraduate STEM education must necessarily limit such opportunities, but if we acknowledge that these are indeed limitations, we will provide students with such feedback when we can.
Readers of this essay will not be surprised that instructor labor is often identified as a primary barrier to students receiving personalized feedback from engaged readers on their written work (Carless, 2006; Gibbs & Simpson, 2004; National Commission, 2003). Anson, who notes in his introductory essay to Writing and Response that ample instructor time is not sufficient to guarantee high-quality feedback, nevertheless acknowledges it as a problem. (See the epigraph to this essay.) Within STEM, providing personalized feedback for undergraduates can be especially challenging. STEM units tend to have sizable "service" roles—providing prerequisite courses for students in other STEM departments, pre-med majors, and so on. These courses tend to have large enrollments, and, as prerequisites for other STEM courses, are often required to cover a great deal of specific content.
Conventional alternatives to instructor labor are often inadequate for this task. Graduate-student teaching assistants who work most closely with student writing in many STEM courses are usually novice writers of science themselves, tend to have little pedagogical training or teaching experience—especially in relation to student writing, and often have the additional challenge of being non-native speakers. Tutors and peer feedback can both be used to some advantage, although perhaps less so than for non-STEM courses. Tutors especially can offer a "safe" space for students to discuss their work away from the evaluative gaze of the professor. Tutors, especially if trained through a campus writing center program, may have more knowledge of some aspects of writing pedagogy than the faculty member teaching the course. But while writing centers tend to be well staffed with tutors trained in the humanities, they are rarely able to hire enough tutors with the variety of STEM backgrounds needed to provide STEM student writers with the kind of feedback I have discussed.
With proper guidance, peers can provide useful feedback on those aspects of each other's writing that do not require "insider" knowledge of the field or genre, and giving feedback to peers is beneficial for the giver as well as the receiver. But for the forms of professional discourse STEM faculty increasingly assign in WID contexts, inexperience limits the value of peer input. Peer feedback may also be a poor fit in lab-based courses which are not "inquiry-based" since the students are, in effect, all attempting to write the same paper (Moskovitz & Kellogg, 2005).
Even when conditions make it practical for faculty to provide personalized feedback for their students, their responses can be limited in other ways. First, giving useful reader-based feedback requires significant experience with the assigned genre—if not as a producer of such texts than as a consumer. A tenet of Writing in the Disciplines philosophy is that while faculty in the various disciplines may need training and guidance in writing pedagogy, they are uniquely positioned as disciplinary experts to instruct their students in the writing practices of their field. Carter and colleagues expressed this idea recently (2011):
[W]hat is most important about communication, i.e., its effectiveness, lies squarely within the disciplinary boundaries. ...There is no such thing, we argue, as effective communication in general, only effective communication in particular situations, including those defined by disciplines. Consequently, the people best able to teach effective communication in any discipline ... are the discipline's practitioners and faculty. This does not mean that students cannot learn the general communication skills of writing, speaking, and reading typically taught by specialists in those areas. Rather, it means that those general skills need to be shaped and honed for use in a discipline by specialists in that discipline. (p. 21)
One might, then, reasonably assume that faculty in a given discipline are also experts in the various disciplinary genres their students most need to learn. Yet my experience as a writing program administrator and consultant has taught me that this assumption is not always accurate in STEM. While professors in STEM typically are expert in genres such as the journal article and research grant proposal, many lack experience with workplace genres common in industry or government settings.
Other challenges are inherent in the school context itself, which can make it difficult for both students and their teachers to treat student writing as "a communication activity between audience and author" (Cohen & Riel, 1989, as quoted in Cho & MacArthur, 2010, p. 329). Even as instructors attempt to switch roles between reader and evaluator, students are well aware that the giver of feedback is also the giver of grades. Also, instructors often know too much about the students' work to stand in meaningfully for an audience who would not have this familiarity. This is especially problematic in lab courses where students often end up parroting information about an experiment back to their lab instructor. In these situations, students may perceive the writing task as "a pointless exercise in error-avoidance" (Probst, 1989, p. 75). Even after required courses in first-year writing, students often have difficulty shifting their concept of good writing from the school-situated "what the professor wants" perspective to the more sophisticated understanding of "what is expected in a discourse community" (Fraizer, 2010, p. 51).
While instructors, peers, and writing center tutors can all play useful roles in providing feedback on student writing in STEM, their feedback is necessarily limited by institutional and structural constraints. Because VE readers can be selected specifically for each course, their professional expertise, genre familiarity, and topic knowledge can be more closely aligned with student writing assignments. Consider for example the research grant proposal, an increasingly common undergraduate assignment in STEM. VE readers can be solicited whose background closely matches the target audience: they are familiar with grant writing and have a working knowledge of the field but know little if anything about the particulars of the research being proposed.
A second benefit is that VE readers generally work with only one student or student group during the semester. As volunteers, they can devote considerable time to their interactions with individual students. And because responding to student writing isn't part of their normal grind, VE readers tend to have more energy and enthusiasm than even the most devoted faculty members are likely to muster once they've reached the middle of a stack of papers. Third, because VE readers are recognized by students as working professionals, their involvement may encourage students to approach the writing task as an apprentice activity rather than just a school assignment. Finally, many students are motivated by the networking opportunity VE readers might provide. While such opportunities cannot be expected, the possibility does seem to encourage some students to participate and perhaps put more effort into their writing.
The VER approach can also affect teaching. The prospect of securing 3 to 4 hours of expert writing consultation per student is a strong motivator for faculty to participate in the needed work of revising their assignments and syllabi to make them suitable for VER. Faced with the task of specifying what constitutes "appropriate" readers, many instructors recognize that they have not adequately articulated the rhetorical dimensions of the writing they are asking their students to attempt.
Finally, there are benefits for the institution itself. Offering alumni and staff from the various parts of the institution a meaningful role in the educational mission of the school could be an effective tool for community building. Alumni affairs units are always looking for new ways to engage alums and VER offers opportunities for alums to give back to the institution in ways other than monetary gifts or event planning. All of this hinges, of course, on whether members of the institutional community will in fact volunteer. At Duke, there has been strong interest, especially among alumni. While there have indeed been challenges in soliciting volunteers for some courses; response to solicitations for volunteers has far surpassed my initial expectations, as I explain below.
It is worth noting that in some STEM courses, notably engineering capstone projects, it has become fairly common for students to work with/for industry clients. The goal of these partnerships is give students experience designing for a real-world setting. Such clients may give students feedback on written documents produced as part of the design work. In spite of the similarities, differences between this arrangement and VER are important. For one, VE readers' interactions with students are off the record. Since these are private exchanges between student and reader, students are freer to make mistakes, ask what they might consider foolish questions, and interact informally in ways that are likely to be educationally useful. Another important distinction is that the timing of interactions in VER is designed to maximize learning, while for clients it is driven by the needs of the project and the availability of the clients. Finally, clients may have neither the time nor inclination to give the type of extended feedback we expect in VER, nor could most clients be expected to spend time learning how to give appropriate feedback. From my conversations with instructors of such courses, it is frequently the case that if students get any substantive feedback from clients on written materials, it arrives after the project has been completed and revisions are no longer possible. That said, I believe that VER could serve as a useful model to help students get more benefit from their interactions with such clients.
I began the current research project with NSF support after three years of implementing VER in over forty different courses across the undergraduate academic landscape at Duke. I had figured out many of the important questions, but I also understood that few of these could be answered outright given the complexities described above. My goals for this research, then, have been more modest: to better understand the practicalities of implementation in various undergraduate STEM contexts and to get a first cut in identifying factors that make VER more or less likely to be successful in those contexts. By experimenting with VER in a broad range of STEM undergraduate settings—different disciplines, different sizes and types of courses, different types of writing and so on—I hoped that this research could provide some guidance for those who might wish to implement VER at other institutions.
A main limitation of this research is the prevalence of confounding factors. A pedagogical intervention this intertwined with writing and research projects of multiple courses cannot be studied in a controlled manner varying one parameter at a time; there are simply too many variables at play to sort out their effects with any certainty. Nevertheless, if future scholars are to build on this preliminary research, it makes sense that I share whatever insights I have obtained knowing that others will understand the limitations of those insights and thus the need to investigate them further.
While conducting this research on courses just adopting VER, I have simultaneously overseen the implementation of VER in dozens of other courses—some STEM, some not. My understandings and insights about VER in STEM are informed, then, not only by this formal research but also by internal assessment data regularly collected for many other courses along with many informal conversations and correspondences with our paid project coordinators, instructors, students, and readers. Some of my early hunches about implementing VER have been supported; others have not. In order to make this essay as useful as possible, I will try to articulate the reasons for decisions I made along the way and what I think can be learned from those experiences.
VER is simple in concept: students are matched with appropriate volunteers who give them feedback on drafts of their writing. Yet once we move from concept to implementation the complexities soon become evident, for each combination of instructor, course, and assignment poses its own implementation issues. This is especially true in STEM, given the wide variety of writing contexts: students write alone or in teams; they produce one major multi-section paper turned in a section or two at a time, or they write half a dozen smaller papers turned in complete; they write research reports intended for experts in the field or design proposals intended for potential investors without advanced technical training; and so on. In a real sense, each course has its own teaching/learning context and culture requiring adaptation for VER. Here are a few of the questions that regularly arise when preparing to add VER to a course: What kind of writing will students be doing and what kinds of readers can be expected to give appropriate feedback? Are students writing alone or in teams? Can we recruit enough volunteers with the right kinds of backgrounds? Is the written product only a tool for learning within the course, or are their stakes beyond the course—perhaps actual clients or a real grant application—that impose additional constraints and complications? Will written products be short enough that readers can respond cold in real time or will readers need a significant turnaround time to read and comment on the draft?
Given these complexities, implementing VER involves a sequence of steps, each of which includes a number of distinct tasks: (1) preparatory instructor consultation; (2) reader recruiting and matching; and (3) guiding reader-student interactions. Here I sketch out the tasks and rationale for each step for a typical course. As will be discussed below, particulars often varied from this template due to the needs of different course contexts as well as intentional variation for purposes of experimentation.
To be a good fit for VER, a course should meet two conditions: There should be at least one substantive writing project in a genre that exists beyond the classroom setting; we must be able to imagine potential volunteers who have experience as writers or readers of such texts. In addition, there must be adequate time for a meaningful writing-response-revision sequence that can accommodate the schedule of volunteer labor. The first step, then, is a meeting with the course instructor to decide whether the proposed course is in fact suitable for VER. If these conditions can indeed be met we then consult with the instructor to develop or revise the writing assignment(s) as needed, articulate the writing and feedback tasks so the goals of VER are clear to students and readers, and map out a plan for reader-student interactions.
For students to get useful feedback, the context—whether real or imagined—must be clear to both student and reader and in line with instructor expectations. If the student believes she is writing for, say, an audience of disciplinary specialists with a high level of technical knowledge while the reader gives feedback from the perspective of a non-specialist with policy interest in the work, much of the reader's response will be off-base. VER personnel, then, need an accurate understanding of the rhetorical context to solicit appropriate readers and make good student-reader matches. As noted above, many instructors will not have articulated a specific rhetorical context for their writing assignments prior to our first conversation, and even among those who have, the contexts they have specified may not be in alignment with other aspects of their assignment. Most instructors will need some help thinking through their options regarding audience and genre and figuring out how to articulate the context in a way that will give both students and readers a clear understanding of the writing task.
Instructors will also need assistance working out a set of student-reader interactions that fit the nature and pacing of the course and the requirements for VER. While readers often get back to their students with feedback within a day or two of receiving a draft, we try to build in a one-week window for turning around drafts. In practical terms, this means students will typically need at least two weeks between when they send the draft to the reader and when they are expected to have completed revisions. The amount of time instructors allot for major assignments is often insufficient for the scope of the task—even without VER. In those cases I help the instructor figure out how to get students started on the first stages of the assignment earlier in the term.
For the Duke Reader Project, I have developed two preparatory documents to facilitate the consulting process. One asks for details such as the number of students in the course, whether students will be writing solo or co-authoring, how the assignment is staged and paced during the term, and so on. This information is used to determine how many readers will likely be needed and when matches need to be in place. The other articulates the type of writing students will be doing and the audience for that writing in a way that makes the task clear to volunteer readers and students as well as for VER personnel. For some courses, an existing writing assignment will work for VER with only minor adjustments. More often I meet two or three times with the instructor to guide them in revising (or reimagining) an assignment and to figure out how to pace the writing process in line with the other needs of the course.
In recruiting readers, we direct our solicitations to members of the institutional community who do not otherwise play a direct role in the education of our undergraduates. In spite of their good intentions, instructional staff who act as VE readers tend to be less involved and enthusiastic readers than others. This is not surprising, since VER involves tasks similar to their routine work. In contrast, helping students with writing assignments is a novel experience for most of our volunteers so they generally bring a higher level of attention and enthusiasm to the task. In fact, many readers report that participating in such intellectually oriented activities is one of the pleasures of their involvement.
We recruit readers in two ways: general and targeted. General recruitment involves solicitation of volunteers with any background. This is done through notices in newsletters for alumni of Arts and Sciences and our professional schools. Those who respond to these announcements complete an online survey that adds them to the pool of volunteers who receive invitations to participate each semester. Targeted recruitment, conducted in cooperation with the Office of Alumni Affairs, various professional school alumni offices, and other on-campus contacts, is used to locate volunteers with specific backgrounds whenever the current pool is insufficient to meet demand. This is usually done by email blasts or postings to list-serves.
A few weeks before the start of each term, the volunteer pool is sent a link to an online list of upcoming courses. For each course there is a description of the writing assignment(s) and a statement about the kind of background readers should have, along with the standard course description. Volunteers who feel they would be a good reader for one or more of these courses respond with their preference through an online survey. The project manager then compiles this information for the match process and additional readers are solicited as needed. If there are still not enough readers for all interested students in a given class, students are matched with readers on a first-come basis. For most courses, readers who are a good fit for a course are randomly matched with students in that course. In those courses in which students would benefit from readers with specialized knowledge related to their individual projects (especially capstone courses and independent research projects), the project coordinator collaborates with the instructor to match specific readers with individual students.
Because instructors are assisted in designing a protocol that fits the context and pacing of their specific courses, the actual plan for student-reader interactions for any given class will vary from that described in the "standard" version here—which is the template from which we develop course-specific plans.
At the appropriate point in the semester, the Project Manager notifies students and their assigned readers of the match by email, providing each with the other's contact information, and asking them to schedule an introductory meeting (in person or via webcam or phone). In this first meeting participants get to meet each other and the student learns about the reader's background relevant to the project. In some cases students have a short written document already prepared—perhaps a prospectus or statement of a research question—and readers give input on that document after the introductions. When students have a reasonably coherent draft ready for feedback, they e-mail a copy of the draft to their reader who is asked to provide feedback in a timely fashion, typically within one week. Students then revise the draft taking into account the reader's feedback (along with any other feedback they may have received). They then meet with their reader again to discuss the revised draft in advance of preparing the final version of the paper. Once that project is completed, the student is expected to share the final version with the reader, and both student and reader participants are asked to complete an online assessment survey.
Giving feedback for VER involves special considerations. Because feedback from experienced professionals comes with considerable authority, students can be tempted to accept the input without considering for themselves whether particular changes are warranted. VE readers, then, need to understand that whereas the aim of providing feedback on writing in the workplace is to improve the text, the goal for VER is to help students grow as writers and thinkers. We want volunteers to recognize that, beyond concerns relating to plagiarism, students will learn more if readers help them recognize what might need changing and why, rather than making the changes for them. In our guidelines for giving feedback (Appendix B), we ask readers to give mainly "reader-based" feedback—addressing the text as a sincere attempt to accomplish its rhetorical aims, expressing their honest reactions and struggles with patience and good will. When readers encounter a passage or feature of the text they see as problematic, we ask them to explain their concern (they got lost, bored, confused, and so on), rather than telling the student what to do. That said, there are many occasions when students can benefit from direct advice given by an experienced practitioner. When advice is warranted, we ask readers to give it in relation to principles students can apply in the future, rather than as only as fixes to specific problems.
In addition to our standard guidelines, we ask each instructor to note two to four issues that they would particularly like readers to address in their comments for each writing project. These course-specific feedback notes are included on the VER webpage for the course. We also provide models of written feedback. Initially we used only a small set of examples representing different response methods and disciplines. As we collect models from more courses we are able to provide course-specific examples for readers in those courses.
The research described here involved data collection over a three-year period, starting in the fall of 2011. Courses included in this research were intentionally selected to include a broad range of course types and writing assignments, and diversity of disciplines. Table 1 lists participating courses and the modal educational year of the students. Basic information about writing assignments in each course is given in Table 2. Whether student participation would initially be optional or mandatory was decided on a course-by-course basis in consultation with each instructor. As shown in Table 2, participation was required in three of seven courses for 2011/2012: COMPSCI, ENGINEER, and CHEM PHYS. For the 2012/2013 year, student participation for ENGINEER was changed to optional.
None of the courses included in this research were new and all had included substantial writing assignments prior to involvement with VER. Using the information forms in Appendix A, I worked with instructors individually to revise writing assignments as needed and to make decisions about the kind and pacing of student-reader interactions as described above. Readers were provided with guidelines as shown in Appendix B.
All assessment surveys were administered using Qualtrics online survey software.
For student surveys, instructors were asked to administer the survey in class if possible, and to ask any students who were not in class that day to complete the survey out of class. Readers received the survey link by email. The student survey had two major sections: the first was for all students in the course, whether or not they participated in VER; there was no mention of VER prior to or within this section. The survey then bifurcated, asking different questions depending on whether students had participated in VER. Numbers of student responses by course are shown in Table 3; each entry shows the number of student respondents who participated in VER and the total number of responses for the course. Assessment methods including informed consent processes were approved by the Duke University Office of Human Subjects Research.
For this exploratory study, assessment data for each course from one semester was used to guide changes implemented the next. Given the substantive differences in reader-student interaction plans, matching arrangements, and so on between courses and even between subsequent offerings of the same course, detailed comparative analysis of pooled research data is excessively complicated and potentially misleading. Detailed data analysis focusing on specific subsets of these data most relevant to faculty and administrators in particular fields will be presented in forthcoming publications.
Here I discuss results highlighting general trends across course settings. I first present data for student participants, including their assessment of the quality and quantity of interactions with their readers and their perceptions of how they benefited from participation. This is followed by comparisons of responses from student participants and non-participants regarding their experiences with writing in the course. The final set of student results reports on problems and challenges students faced. The second major subsection includes results from reader surveys followed by a short section on instructors.
In planning this research, I posed for myself two primary questions: Can VER increase student engagement in STEM writing assignments, and if so, what makes it more or less likely to work well? The first question is now easy to answer—at least according to student survey data. As I will show, these data suggest that VER participation does enhance engagement for most students who participate and that most would participate again if offered the chance or would recommend VER to other students. That said, results very considerably across courses, which makes answering the second question more challenging. Given the limited data for each course, sorting out the factors that might account for these differences requires finding patterns across different measures. Because this is difficult to present sequentially, I state my general interpretations of these data here and then discuss trends related to these interpretations as I present specific results.
The most important trend I find in these data is that fewer students reported benefits or satisfaction with the experience when their participation was required or when the writing was done as part of a team. Of the many questions on the survey, I see the one asking whether students would recommend participation to other students as the best measure of overall satisfaction with the experience. For the first year of this study, 19% of students who wrote papers as the sole author said they would not recommend participation to other students, compared with 46% of those who coauthored papers with classmates. While data from the following semesters shows improvement in this measure for both groups, the trend continued. In the most recent data, 6% of students writing on their own would not recommend participation vs. 20% of those who coauthored.
Comparison of specific results from courses with collaboratively written papers is revealing. The two courses with the weakest student results for this question in the 2011-2012 year were the engineering and computer science courses; both had coauthored papers with a single VE reader per team and student participation was required. An additional course with team projects, Restoration Ecology, was added the following year. In response to the first round of assessment data from the engineering and computer science course, students in Restoration Ecology and the second offering of the engineering course were given individual readers and participation was made optional. That semester student satisfaction with VER in both courses was strong. In the final year of this study, the course Global Health was added, in which students also coauthored papers. Following on prior success in the ecology and engineering courses, participation was optional and students were assigned individual readers. Student assessment was also quite positive for this course. These results suggest that students can indeed have positive experiences with VER in collaborative writing contexts as long as students have their own readers.
Student surveys began with a series of questions about their experience with writing in the course without reference to VER, which provided comparison group data. I note that though these comparisons are useful, there was no true control: students self selected into VER (for courses not requiring it) rather than being randomly assigned to VER or the comparison group. After the first set of questions the survey bifurcated, with different questions for VER participants and non-participants. VER participants were asked about the VER experience; non-participants were asked why they chose not to participate, what they heard about VER from those who did, and whether they would have chosen differently given what they knew about VER now.
Figure 1 presents a comparison of responses from student participants and non-participant to a set of prompts regarding their experience with writing in the course from fall 2012 through fall 2013 for courses in which student participation was optional. There are two notable differences between the VER participant and comparison groups. Of VER participants, 50% strongly agreed that they were engaged in the writing process, compared to 21% for non-participants. Of participants, 39% strongly agreed that the feedback they received helped them improve their papers and 34% strongly agreed that the feedback will help them write more effectively in the future; for non-participants the values were 25% and 23% respectively. Participants were also more likely than non-participants to report that they were motivated to do their best work. While these data reflect favorably on VER they should be interpreted with caution since students self-selection into the VER or comparison groups could have resulted in a sampling bias, particularly regarding issues of engagement and motivation.
Students who participated in VER were also asked to indicate agreement with a series of statements regarding perceived benefits (Figure 2). Even though enhanced subject matter learning was not an explicit goal of VER, the top graph (a) shows that over half of participating students felt their participation improved their understanding of the subject they were writing about. Nearly 90% of student participants believed that their work was better as a result of the experience (b). Over two thirds reported that the experience made the writing process more engaging, with over one quarter strongly agreeing with that statement. The bottom graph (d) shows 80% of students reporting that VER was a "valuable addition to the course." Results shown in (d) are particularly instructive since their response should take into account both perceived benefits such as those shown in (a) - (c) and any perceived costs—particularly the additional time required.
To better gauge students' overall satisfaction they were also asked, "If you had the opportunity to participate in the Reader Project again, would you?" Of the 50 responses to this question from the 2011/2012-year, 76% answered yes. Given that this was the first attempt at including VER in any of these courses, this was encouraging. However, as shown in Figure 3, results varied markedly by course. All of the students in the chemistry courses and nearly all participating students in the Genomics Perspectives course indicated they would do so again; in contrast, the proportion of favorable responses for CompSci and Engineering were 67% and 52%, respectively.
Figure 4 shows the results for this prompt for the most recent year. Courses with co-authored papers are COMPSCI, ENGINEER, ENV REST ECO and GLOBAL HEALTH. Note that the course with the lowest proportion of "yes" responses is the computer science course—still under the required participation/one-reader-per-team model. In 2011 (Figure 3) participation in the engineering course was required and each group had one reader. In 2013 it was optional and students were assigned their own readers. (Note: 19 of 25 students opted in.)
Beginning in 2011, the student survey asked, "Was your reader engaged and responsive?" (Table 4). Slightly over half of students responded "yes" in the first year of the study, increasing to about 75% in the following two years. Students' overall assessment of their experience was associated with their responses on this question: 77% of students who responded "yes" indicated that they would participate again if offered the chance, vs. 33% who answered "somewhat." (Very few responded "not at all.")
If student reports of reader enthusiasm are accurate, this may well reflect the readers' perception of their student's commitment to VER process. Increases in reported reader enthusiasm, then, might be indirectly influenced by improvements in the matching process and measures to improve student follow-through and accountability. Figure 5 shows responses to this prompt from our last full year of data collection, by course. The computer science course again gives the most negative responses, with about 40% of respondents indicating a less-than-ideal perception of reader engagement. Notably, this was the only course that year in which students both coauthored papers and had a single reader per team.
Students were also asked about the quantity of their interaction with readers. In response to the prompt, "What are your thoughts about the amount of interaction you had with your reader?" 67% of students reported that this was "just right," while 29% "would have liked more." Only 4% (8 students) "would have preferred less." These results were fairly stable over time, suggesting that we are generally planning the right amount of interaction between students and readers—with the assumption that it is better to leave participants wanting more interaction than to expect them to interact more than they would like.
Figure 6 shows student responses to this question over the last full year of data collection by course. While students in most courses were generally satisfied with the amount of interaction they had with readers, the exceptions are interesting. 80% of students in the computer science course wanted more interaction. Given the relatively low rate of student satisfaction described above, these data suggest at least one possible reason: in student teams with a single reader, students may have too little interaction with the reader. Yet the responses were quite similar for the physical chemistry course for which students reported favorably on tehir experience with VER. Student desire for more interaction might imply that these students would have liked yet more of what they felt was a good thing.
Starting in fall 2012 we began asking students about the quality of the reader match with the prompt, "Considering your reader's background and experience, how good a fit was this reader for giving feedback as a member of the target audience for the assignment(s)?" Of 139 responses, 62% felt their reader was a "good" fit, 30% a "satisfactory" fit, and the remaining 8% "not a good fit." Examining these data across time suggests that improvements in soliciting and matching procedures have improved reader fit: In 2012/13, 47% of students rated their reader's fit as good, 39% as satisfactory, and 15% as not good, whereas for the last semester of this study, the ratings were 80% good and 20% satisfactory, with no students reporting poor fits.
Students were asked about challenges they faced as a VER participant in two ways: via an open-ended prompt about whether they encountered any problems and a question asking specifically about conflicting feedback they received from their VE reader and their instructor or teaching assistant. The proportion of students reporting some type of problem was highest the first year of the study at 24.1%. The proportion has since decreased: 21.3% for the ‘12/'13 year, down to 13.9% for the most recent term.
On the open-ended prompt, the most common themes were a lack of reader engagement and not getting feedback from the reader quickly enough to meet the timeline of the assignment. One student, for example, commented that her reader "was most often busy and the response time too slow to be valuable in doing drafts of a report for a college class." The pacing of student writing tasks, combined with the multiple assignment deadlines and exams students must juggle, does pose a challenge for VER. This may be why assessment data are more positive for courses with a single large writing project than for multiple shorter assignments. A number of students also commented on reader fit, noting that their reader had insufficient subject-matter expertise or didn't know enough about the genre of the assignment.
In response to the prompt, "If you received feedback both from your reader and your instructor (or T.A.), did you ever receive conflicting feedback?" 24% students answered in the affirmative. Those students were then asked how they dealt with the conflict. The majority (56%) reported that they deferred to the instructor or teaching assistant:
I inevitably had to default to the feedback of my TA. She is the person who grades my assignments and thus I need to follow her instructions and improvements.
It was mostly feedback from the reader that didn't take into consideration small quirks required by the instructor for the paper, so the instructor's wants won out always.
The TA grades my paper, so I took his advice. But I know that in real life that my writing will be different and make more sense than the stuff we do for grades in this class, so hearing my reader's advice is good for when I do actual writing in real life.
Those who deferred to their reader (14%) generally noted the reader's greater experience, particularly in comparison to the teaching assistant:
Our writing assignments were policy memos about impact assessing and recommended policy action. My reader is a professional working in environmental impact analysis and regulation, while my TA was a student in environmental finance. So, I deferred to my reader because her feedback was more informed and pointed.
I used the feedback from my reader, as she was an expert in the field that my memo addressed.
As my reader was an expert in her field, I trusted her feedback over that of the TAs, who are essentially my peers.
The remaining 25% attempted to synthesize input from both sources:
Most of these conflicts were related to personal opinion, so I just went with the feedback I was most comfortable with.
It really dealt with the topics of the writing project and how I as a student should go about writing them. [My reader] …was giving advice more as if in a real world situation whereas the TA talked mostly about grading and grading policies which differed in terms of content and style.
I tried to incorporate both types of feedback in a way that would satisfy both the professor and the reader.
Because qualified and willing volunteers are the essential element in VER, I focused data gathering for readers on questions that would provide insight on their motivations for participating, their approaches to providing feedback, and their likelihood of participating again. In response to the question "Why did you participate in the Reader Project?" common responses related to three themes: the desire to interact with Duke students, the opportunity to give back to the institution, and the importance of writing in the field (Figure 7).
Table 5 shows the ways in which readers provided feedback and the proportion of readers who used each. Readers were asked to select all of the modes they used. Email was the most common, followed by Word "comments" and then "track changes." Since most readers do not reside near the university, it is not surprising that only 15% of readers met with their students in person.
To understand the kinds of feedback readers provide for students, we began collecting feedback samples in the second year of the study. Collecting such data is challenging for a number of reasons: The variety of response modes (email comments separate from the student text, comments made within documents, and those made orally in person, or by phone or webcam), participants reluctance to share these data, and because IRB protocol requires securing both student and reader permissions for each document. While I have not analyzed the data collected to date, I have worked with instructors to select model responses from those submitted for their courses to use in future reader training. A sample of these models is presented in Appendix C.
Readers were asked whether the type of feedback they gave changed during the semester, to which 35% responded affirmatively. Comments tended to remark on changes made either as a result of better understanding the role of the VE reader and the educational context or recognizing that student needs changed as the work developed:
I pulled back from editor mode to more like a senior engineer reviewing the work of a less experienced engineer.
It became more generalized and less direct which gave the students a chance to think for themselves.
From helping her brainstorm on sources for information and perspectives to the best way to organize your arguments and information in the paper.
We started out just talking about ideas, and gradually became more targeted.
The first set of feedback was a little more generic about organization, purpose, and clarity. The second feedback spent more time discussing specific points.
The first draft was much more basic, pointing out the initial issues about organization and the content working together. The second draft was more about asking questions about the subject and content that I would have anticipated the intended reader to ask.
Readers were also asked what, if anything, they learned in the process of being a reader. Interestingly, many of the responses were in line with current best practices in teaching writing. (How much of this was learned from the experience itself or from reading the materials we provided to guide them in giving feedback is not clear.) Many commented on the importance of not appropriating the student's text or doing their work for them:
Leaving work in students' hands.
I have learned how to suggest changes without making the changes. It was definitely different as all of my previous editing experience has been making changes. It was also challenging to suggest changes in a way that suggests learning.
I've learned how to comment by giving suggestions, rather than telling the student what to change. It is the student's paper, and therefore his or her decision to accept or reject any of my comments or suggestions.
It is in the best interests of the students to give them less direct feedback—they learn more and think for themselves.
Try not to give students answers but ask questions.
Others commented on the importance of including positive comments and having an encouraging attitude:
... to give positive feedback as well as critical feedback.
Feedback should always be constructive and never negative.
Importance of giving positive feedback along with critical feedback.
The best results and experiences have come from remaining positive, encouraging, and if there are issues to address, be sure to be clear and focused if you have specific concerns.
Students are used to criticism, but I'm not sure they are used to the idea that it is always intended to make the product better. In other words, this reader needs to make clear that their comments are in the spirit of business / enterprise collaborative writing - it's not about critiquing the performance, but rather about improving the performance.
One has to be very careful not to do any of the student's actual research, but to suggest areas where he or she might find relevant material, new material, challenging perspectives. In face to face (via web cam in my case), I always find asking questions produces more interest and follow up work from the student. I often disagree with a student's assumptions or conclusions, but I see my job as helping the student present his or her thesis or analysis as effectively as possible. Often I find I have learned something new along the way as I read and listen.
A number of readers mentioned keeping comments focused:
I'm used to giving feedback to younger professionals, but I found that with college students without real world experience you have to be a little more targeted.
Be specific about what you're trying to help them achieve. It's more than "good writing"; it's about developing a tight package about a problem and describing the issues succinctly.
I think giving too many comments can be overwhelming. Focusing on a few critical points can be more effective.
Finally, many readers gave responses that are interesting for the level of engagement they show for the VER process:
I think I have a pretty cut-to-the-bone, possibly harsh style with my colleagues as we turn around materials very urgently at times and also we tend to write comments rather than talk. We work from various parts of the country so are not in the same offices. I do not think that style would have been productive with a student. I certainly gave my opinion, but the chance to talk rather than just mark up a paper was very helpful. At the same time, it allowed me a chance to reflect on what my well-informed but generally gut reactions to policy writing are based on. I could explain them, with this reflection, and also discern what was actually a personal preference.
I've learned to really pay attention to how the student is reacting to the conversation to understand how/if feedback is being understood or valued. I've participated for several years and each student is different.
How to separate comments on content from those on delivery. It seems to be easier to assimilate the information when compartmentalized this way.
Don't assume that what seems obvious to you is obvious to others.
Sometimes it is necessary to correct/provide feedback for a student for the same thing in multiple situations/multiple times—can't just use one example and expect the student to recognize and change similar situations, but if you point out the same thing two or three times then they might recognize it themselves in the fourth time. This isn't because students are slow or don't understand, it's because the feedback and material is new to them. I think that professionals with a lot of experience in a certain field or skill have a tendency to forget that at one point we had to learn that material, skill, way of doing things—it isn't completely intuitive or natural. Working to teach a student through feedback on a piece of writing provided a valuable experience which I think will help me work with junior staff and new employees in the future.
Readers were also asked about the challenges they faced in providing feedback, the amount of interaction they had with students, and whether they would participate again. As shown in Figure 8, a slight majority of respondents indicated no difficulties in providing feedback. Of those who reported challenges, the most common response was refraining from correcting students' work (30%).
Table 6 shows how readers felt about the amount of interaction they had with students—overall and as a function of team vs. group projects.
Overall, readers were almost evenly split between those who thought the amount of interaction as "about right" and those who wanted more. Breaking these data down by type of authorship suggest that readers tended to be less satisfied with the amount of interaction for collaborative writing projects than for individually authored papers, consistent with results for students. For team-based writing projects, most readers (60%) would have preferred more interaction, whereas for solo-authored papers most readers (67%) felt the amount of interaction was about right. No respondents indicated that they had too much interaction. For the team projects it was theoretically possible for readers to have more opportunities to interact with students, since each reader was matched with three to five students (depending on the course). These data suggest, however, that the dynamics of one-reader-multiple-student matching were less favorable than the one-to-one matching in the solo-authored contexts.
A crucial matter for sustaining VER is whether readers will volunteer repeatedly—both because of the need for a sufficient number of readers but also because repeaters bring the benefit of prior VER experience (which includes the kinds of things readers reported learning from their experience discussed above). When asked whether they would be interested in participating again, 79% of 113 respondents chose "Yes, as soon as possible," while 19% responded "Yes, but not next semester." Only 2 readers indicated that they did not want to participate again. In fact, over one third of readers taking the survey indicated they had been a reader at least one time before.
While formal assessment offers important data on the experiences of volunteers, readers occasionally communicate with the program informally by email. While these data are anecdotal, they provide insight into the motivations of some readers. Here are two excerpts from unsolicited notes describing the reader's interest in participating:
As an engineering manager, I have had the opportunity to recruit and develop many bright young engineers over the past couple decades. It seems that only a very small fraction have the writing skills to communicate their ideas in an organized, coherent fashion.
Oftentimes, engineers seem to think that the calculations are all they need to be good at. My response has always been that it doesn't matter how good your idea is if you can't communicate it, and that much of the time, the people deciding whether to approve a project expenditure only understand the text, not the technical data.
My research on VER to date has focused primarily on students and secondarily on readers. Nonetheless, we have collected a limited amount of data from instructors. These data suggest that for some participating instructors, their involvement in VER has influenced their teaching in ways we would consider positive from a WID perspective:
It has helped me to think about how to structure the assignments so that they are not merely outputs, but… part of a process.
I am using additional interim deadlines and assignments for the students. I have utilized more class time for writing workshops, including peer and model editing...
Previous semesters, we have left documentation tasks until near the end of the semester. Being part of the Reader Project has facilitated our more effectively integrating documentation issues into the core of the course.
These responses suggest that aside from the direct benefits of students interacting with readers, VER may result in enhancements to the writing pedagogy of the course more broadly. Finally, one instructor offered a comment on the value of VER that aligns well with the broad aims of the approach:
Without a doubt, [my students'] writing has improved dramatically [...] through their interactions with their reader. This is not because of feedback on sentence structure and grammar, but because of the real-life experience these readers bring with them. [...] As a result of their interactions with their readers (e.g., via e-mail or Skype), their papers have been transformed from just another term paper into a "living document" that has real meaning beyond the classroom. These students are no longer just writing for the "professor", but know that there is a larger audience out there that may find their research and ideas of value.
While such comments have limited value as evidence of the effectiveness of VER, they do suggest that faculty can understand what VER offers and value the experience for their students. The other and perhaps most telling data regarding instructors is indirect: Of the ten courses involved in this study, instructors of all but one—the computer science course—have asked to continue VER in their courses after the study.
My research to date suggests that VER can indeed be an effective tool for STEM writing projects. But it is also clear that a number of factors may affect the success fo VER. Here I present my current thinking on issues that appear to play a role in the success of VER and how VER can best be implemented.
One of the most rewarding aspects of directing the Reader Project has been seeing the variety and caliber of volunteers. When I first experimented with VER in my own health-science-oriented first-year writing courses in 2006, I expected that if were I fortunate enough to get many volunteers, they would be primarily younger professionals with limited experience. I found the opposite. Most of those who volunteered had considerable experience and some were at the top of their fields. In fact, readers for this course over two semesters included the directors of the Duke Diet and Fitness Center and the Rice Diet Center, both nationally renowned experts. This trend has continued. To provide a sense of the caliber of professionals who are willing to share their expertise with students in this way, I provide a sample of reader biographies in Appendix D.
One of the earliest questions I faced in implementing VER was whether student participation could be required. Some instructors have had initially a strong preference that all their students participate—either because they wanted all of their students to get the benefits or because they wanted to avoid the complications resulting from partial participation. While sympathetic to these concerns, I preferred from the outset to make participation optional. For one reason, I was not certain we would be able to solicit enough qualified volunteers to make good matches for every student. I have also found that while our volunteers have been remarkably reliable overall, it occasionally happens that readers don't work out, either because personal issues arise and they have to withdraw or, more commonly, because they take too long to respond to their student's work.
One of the clearest lessons from our ongoing assessment of the Duke Reader Project is that when students are required to participate, rates of student follow-through decrease markedly. In fact, changing student participation from required to voluntary has consistently improved outcomes according to our data. Three of the ten courses in this study began with students being required to participate, one of which—engineering—transitioned to optional after the first term. Of the two remaining, the computer science course consistently returned the weakest assessment results while physical chemistry has been strong. The computer science course enrolls nearly 20 students, has many moving parts (including external, real-world clients) and requires students to write in teams. The physical chemistry course has had unusually low student enrollment during the two semesters of this study with only 3-5 students per term (corresponding to a general dip in departmental enrollments). The chemistry instructor is also extremely engaged and committed to VER, to the extent of soliciting readers for his students himself—largely from his network of former graduate students. Every student in his course reported positively on the experience. Evidence from this study, together with informal assessment of many other VER courses has convinced me that requiring students to participate is unlikely to be effective except in unusual circumstances.
While most Reader Project assignments have been targeted to specialist readers, some have been specifically intended for non-specialists. When I started the Reader Project I assumed that we should choose volunteers who were the best fit as members of the target audience for each assignment. For non-specialist assignments this meant matching students with readers who were interested in the subject (either personally or professionally) but did not have expertise in the field. Assessment data has complicated this idea: it seems that in spite of the potential merits of the more "authentic" reading non-specialists might provide, students are more likely to fully engage with readers who are experts in their topic area. Readers with topical expertise can provide more input on the substance of the work—questioning content, suggesting resources, and so on. For students interested in a career related to the topic or discipline, these readers may be more interesting to interact with and may also provide potential networking opportunities. In these cases, getting feedback from both a VE reader and a tutor might be fruitful—the reader providing the insider textual and content knowledge and motivation and the tutor responding through the lens of the non-specialist.
In other cases, it may be advantageous for readers to have genre knowledge but not content knowledge. For example, students in the environmental science course Conserving the Variety of Life are assigned to write (and then video themselves delivering) a 5-minute congressional testimony on a species conservation case of current interest. The target audience for this assignment would include people familiar with the rhetorical context of congressional testimony but not necessarily expert in species conservation. In the most recent term of this course, one reader had experience giving just this sort of testimony on an environmental issue, another served as a staffer on a congressional committee, and yet another was the past chair of the House Republican Caucus. I assume that all of these readers could be good VE Readers for this course.
When we began matching students and readers in 2007, we solicited readers broadly to join a volunteer pool, using a detailed survey to collect information about education and work history, personal interests, and so on. We then used these data to select readers as needed for each course. This approach, however, had its drawbacks. First, and perhaps most important, having readers sign up to volunteer for the program doesn't guarantee they will be available for any given term. This meant we had to check in with the volunteers we selected and find replacements for some, adding extra labor and delaying the match process. Second, this approach required our personnel to make the judgments as to which readers would be the best fit for each course. Given the range of topics, genres, and fields, this was impractical at best. Now, before the start of each semester, we present the list of upcoming courses to readers in our pool and then let them decide when they want to volunteer and which courses they are suited for.
Also, when we began the Reader Project we didn't involve the instructors in the match process, trying to minimize what we asked of them so as not to discourage them from trying VER in their courses. We now know that most instructors are happy to provide input in selecting readers for their courses, and this input clearly results in better matches. Instructors often notice relevant aspects of a reader's background that VER personnel miss, especially for independent research projects since instructors know the particular topic of each student's paper. Here is are two such examples as described by the instructor of the course Environment and Conflict:
[O]ne of the students this year was writing about the marshlands shared between Iraq and Iran and the potential conflicts over water that could erupt as Iraq moves forward with its reconstruction efforts. This student [was] paired with a Duke alumnus [in the service] stationed in Iraq. Another student was interested in the relationship between conservation of sea turtles and the narcotics trade in Mexico. She was paired with a Duke alumnus who is one of the world's leading experts on the protection of sea turtles along the Mexican coastline.
In our current practice, once a roster of potential readers for a given course offering has been identified, we ask instructors to help in making the final selections for their students.
We have also learned that readers occasionally volunteer themselves for courses that interest them but for which they do not possess the necessary background. When readers volunteer themselves for a specific course, we now ask them to briefly explain their interest and fit for that course—information which usefully supplements the general biographic data we collect for each volunteer. Overall, then, we have found it best to let readers propose themselves for courses each term and then select among those readers based on available information with input from instructors.
Unlike courses in the humanities and interpretive social sciences, writing assignments in STEM often involve collaborative writing. Using VER for group projects involves additional considerations of student-reader matching. The two obvious matching options are one reader per group or one reader per student. Each has its theoretical advantages. Those acquainted with undergraduate student writing in STEM are aware that, unless required to do otherwise, teams will often distribute the labor of various tasks such that only one or two members end up doing most of the writing and editing. If each team has a single reader, most students in this scenario would have little if any substantive interaction with their reader, and thus get no benefit. One reader per team also introduces scheduling complications that pose a serious threat to the success of the project. To date, we have tried the one-reader-per-group model five times and each has resulted in weaker assessment data from both students and readers. On the other hand, pairing a single reader with each team requires far fewer qualified readers and less administrative labor in making matches and following up with volunteers. Having a reader for each student, however, is more likely to sponsor meaningful relationships. As described above, data from this study suggest this as the better option. Students matched one-on-one with readers have tended to be more engaged and responsible partners.
For coauthored papers, one reader per student means multiple readers per group, and this does make incorporating reader feedback more complicated for students. As novice writers, students often see these complications as problematic—especially if they have learned to succeed at school writing tasks by interpreting teacher feedback as a comprehensive set of instructions to be executed in exchange for a higher grade. But there is reason to believe that the choices forced on students through multiple, personalized readings is valuable for students growth as writers. Robert Probst (1989) puts the point this way:
Regardless of personal ambition, ... the writer must have integrity, which means not unthinkingly accepting the judgments of others. Nor does it mean rejecting them out of hand. Rather, the writer must assimilate these judgments and use them in making her own judgments. The responses we make to writers at all stages of their development must encourage them to begin to take that responsibility. (p. 76)
Exposing students to multiple readings of their coauthored texts may turn out to be an additional advantage of the one-reader-per-student arrangement for team projects.
From a faculty development perspective, it's fortuitous that implementing VER requires that instructors integrate it into the fabric of their courses and adopt accepted best practices in WID: a required revision process, an explicit rhetorical context, and feedback that focuses on the intentions of the author within that context. In order to secure the benefits of VER for their students—which are especially apparent to faculty in STEM—instructors must participate in what for many of them are their first collaborations with the writing program. Of the ten instructors participating in this study, only four had prior interactions with our writing program. As of this writing, I have worked with nearly fifty instructors to include courses in the Duke Reader Project; the great majority had no prior contact with our writing program or, to my knowledge, other formal training in writing pedagogy.
In "Rethinking the WAC/Writing Center Connection," Michael Pemberton (1995) asks "What benefits can the writing center and its tutors provide in conferences that would not be more fully realized in meetings with professors or other experts in the field [my emphasis]?" (373) Although addressing WID courses generally, Pemberton's question obtains here: Tutors may, in fact, serve a useful role as mediators for students, helping them interpret and decide how to act on VER feedback. Again, while Pemberton's essay predates VER, his insights are applicable: "[A]lthough writing center tutors may not be the best people to comment on papers produced for courses in WAC programs (in terms of their subject-area knowledge and familiarity with discipline-specific conventions), they may very well be the one quasi-authoritative source that students feel most comfortable with" (p. 374).
Serving as mediator could also benefit the tutors. The role reduces pressure on tutors to any pretense of insider knowledge they do not possess. Also, the opportunity to examine VER feedback along with the client educates the tutor about discursive norms in that field. I can easily imagine a professional tutor assigned to a VER course who, in reviewing many VER responses, develops rare and valuable insights that could be shared with the professor, future students in the course, other tutors. These insights might even be publication-worthy. Research into how tutors might be integrated into VER courses could reveal synergies between these two sources of support.
Like any research into a new area, this study raises more questions than it answers. I conclude with a list of those questions I feel are now ripe for investigation:
Anson, Chris M. (1989). Writing and response: Theory, practice, and research. Urbana, Illinois: National Council of Teachers of English.
Association of American Colleges and Universities. (2007). College learning for the new global century. Washington, DC: Association of American Colleges and Universities.
Burke, Virginia M. (1961). A candid opinion on lay readers. The English Journal, 50, 258-264.
Burke, Virginia M. (1960). The lay reader program in action. Special Bulletin, No.1, Wisconsin Council of Teachers of English.
Calibrated Peer Review, 2001. Website. Retrieved from http://cpr.molsci.ucla.edu/
Carless, David. (2006). Differing perceptions in the feedback process. Studies in Higher Education, 31, 219–233.
Carter, Michael, Gannod, Gerald, Burge, Janet, Anderson, Paul, Vouk, Mladen & Hoffmann, Mark. (2011). Communication Genres: Integrating communication into the software engineering curriculum. In Proceedings of the 24th IEEE-CS Conference on Software Engineering Education and Training, 22-24.
Cho, Kwangsu & MacArthur, Charles. (2010) . Student revision with peer and expert reviewing. Learning and Instruction, 20, 328-338.
Cohen, Moshe & Riel, Margaret. (1989). The effect of distant audiences on students' writing. American Educational Research Journal, 26, 143-159.
Ford, Paul M. (1961). Lay readers in the high school composition program: Some statistics. The English Journal, 50, 522-528.
Frazier, Dan. (2010). First steps beyond first year: Coaching transfer after FYC. WPA: Journal of the Council of Writing Program Administrators, 33, 34-57.
Gibbs, Graham & Simpson, Claire. (2004). Conditions under which assessment supports students' learning. Learning and Teaching in Higher Education, 1, 3-29.
Gottschalk, Katherine & Hjortshoj, Keith. (2004). The elements of teaching writing: A resource for instructors in all disciplines, Boston, Bedford/ St. Martin's.
Harris, David E. & Schaible, Robert. (1997). Writing across the curriculum can work. Thought and Action, 13, 31–40.
Higgins, Richard, Hartley, Peter & Skelton, Alan. (2002). The Conscientious consumer: Reconsidering the role of assessment feedback in student learning. Studies in Higher Education, 27, 53-64.
Horvath, Brooke K. (1984). The components of written response: A practical synthesis of current views. Rhetoric Review, 2, 136-156.
Hunt, Russell A. (1989). A horse named Hans, a boy named Shawn: the Herr von Osten theory of response to writing. In Anson, Chris (Ed.) Writing and response: Theory, practice and research. (pp. 80-100). Urbana, Illinois: National Council of Teachers of English.
Jerde, Christopher L. & Taper, Mark L. (2004). Preparing undergraduates for professional writing: Evidence supporting the benefits of scientific writing within the biology curriculum. Journal of College Science Teaching, 33, 29-33.
Lattuca, Lisa R., Terenzini, Patrick T. & Volkwein, J. Fredricks. (2006). Engineering change: A study of the impact of EC2000. Baltimore, ABET, Inc.
Lipnevich, Anastasiya A. & Jeffrey K., Smith. (2009). "I really need feedback to learn:" students' perspectives on the effectiveness of the differential feedback messages. Educational Assessment, Evaluation and Accountability, 21, 347-367.
McLaren, Ingrid & Webber, Dale. (2009). Writing right: Enhancing student engagement and performance in an ecology course. International Journal of Environmental & Science Education, 4, 365-380.
McLeod, Stephen, Brown, Gavin, McDaniels, Preselfannie & Sledge, Lawrence. (2009). Improving writing with a PAL: Harnessing the power of peer assisted learning with the reader's assessment rubrics. International Journal of Teaching and Learning in Higher Education, 20, 488-502.
Moskovitz, Cary and Kellogg, David. (2005). Primary science communication in the first-year writing course. College Composition and Communication, 57, 307-334.
National Commission on Writing. (2003). The neglected "r": The need for a writing revolution. New York: College Entrance Examination Board. Retrieved from http://www.collegeboard.com/prod_downloads/writingcom/neglectedr.pdf
Pemberton, Michael A. (1995). Rethinking the WAC/Writing center connection. In Zawacki, Terry & Rogers, Paul (Eds.) Writing across the curriculum: A critical sourcebook., Bedford/St. Martins, 2012. Originally published in the Writing Center Journal, 15(2): 116-33.
Probst, Robert E. (1989). Transactional theory and response to student writing. In Anson, Chris (Ed.) Writing and response: Theory, practice and research. (pp. 68-79). Urbana, Illinois: National Council of Teachers of English.
Reynolds, Julie, & Moskovitz, Cary. (2008). Calibrated peer review assignments in science courses—are they designed to promote critical thinking and writing skills? Journal of College Science Teaching, 37, 60-66.
Sauer, Edwin H. (1962). Contract correcting: The use of lay readers in the high school composition program. Report to the Cooperative Research Program of the United States Department of Health, Education, and Welfare. Rev. ed. Distributed by Harvard University Press, Cambridge, MA.
Sommers, Nancy. (1982). Responding to student writing. College Composition and Communication, 33, 148-56.
The Rutgers plan. (1961). College Composition and Communication, 12, 151-153.
Topping, Keith. (1988). Peer assessment between students in college and university. Review of Educational Research, 68, 249-276.
 I wish to express my sincere appreciation to Alfie Guy, David Malone, Dara Regaignon and Nick Carbone, who gave me thoughtful feedback on various drafts of this article. The Duke Reader Project is supported by the Duke Alumni Association and the Jean T. and Heyward G. Pelham Foundation. This research was supported by the Division of Undergraduate Education of the National Science Foundation (DUE-1043674).
 Although the approach proposed here—having experts who are alumni and employees of the institution provide feedback for undergraduate writing projects—is novel, a related concept, the "Lay Reader Program," was developed for secondary school students in the late 1950's and early 1960's. In that program, members of the local community, usually "housewives" with degrees in English or related fields, were partnered with middle school or high school teachers and were paid to identify errors, make comments, and—occasionall —to grade student papers. The issue was the subject of considerable attention and some debate in English pedagogy scholarship in the early 1960's (Burke, 1960; Burke 1961; Ford, 1961; Rutgers Plan, 1961; Sauer, 1962). Lay Reader Programs existed in some fashion as recently as the late 1990's (Richard Bullock, personal correspondence).
 If the purpose of feedback is to help students develop mature writerly skills and sensibilities, peer feedback using computerized scoring systems (Topping, 1998; McLeod, 2009) such as CPR (Calibrated Peer Review, 2001) may be even less effective. The nature of the reviewing tasks means that students may have little internal motivation to take the reviewing process seriously and often find ways to game the system. A study of CPR practices found that feedback tended to be highly reductive and that the system indirectly encouraged practices not in line with current best practices of teaching writing (Reynolds & Moskovitz, 2008).
 For a discussion of feedback on student laboratory reports, see Moskovitz & Kellogg, 2005.
 This is precisely what happened with the first engineering course in the Duke Reader Project a couple of years before the current research began.
Moskovitz, Cary. (2014, December 22). Volunteer expert readers for STEM student writers. Across the Disciplines, 11(2). Retrieved from https://wac.colostate.edu/atd/articles/moskovitz2014.cfm