Writing Workshops to Foster Written Communication Skills in Statistics Graduate Students

Abstract Being able to communicate effectively is an essential skill for statisticians and data scientists. Despite this, communication skills are not frequently taught or emphasized in statistics and data science courses. In this article, we describe a series of four workshops that were developed to enhance the written communication skills of statistics graduate students. We also present results of a survey administered prior to the first workshop and after the last workshop to assess changes in attitudes towards writing and obtain student feedback to improve the workshops. Lastly, we discuss potential modifications of these workshops including ways to use these workshops or elements from them for undergraduates or in statistics and data science courses. Supplementary materials for this article are available online.


Introduction
Effective communication, both written and oral, is often emphasized in statistics education. The 2014 Curriculum Guidelines for Undergraduate Programs in Statistical Sciences by the American Statistical Association recommended that students "should be able to communicate complex statistical methods in basic terms to managers and other audiences and visualize results in an accessible manner" (p.10). The reasons for this emphasis on communications have been two-fold. First, emphasizing communication enhances the learning of statistics (Khachatryan and Karst 2017;Radke-Sharpe 1991) and second, effective communication skills are critical for statisticians to possess in the workplace (Gardenier 2012;Radke-Sharpe 1991).
In his report of recommendations for teaching statistics, Cobb (1992) advocated that due to advances in computing, less emphasis needs to be placed on formulas and the mechanical aspects of statistics and more emphasis needs to be placed on written and oral communication. Oral communication skills can be developed using various strategies such as the use of small group cooperative learning (Garfield 1993), oral exams (Theobold 2021), and oral presentations. Numerous strategies to develop written communication skills also exist and range from small-scale writing such as interpreting results on homework or exams to more large-scale projects including writing an editorial or a research paper. Previous research has found that emphasizing writing improves students' interpretative and computational skills (Beins 1993;Holcomb and Ruffer 2000). Further, being able to communicate effectively Sharpe 1991, p.292). Because statisticians frequently work and interact with nonstatisticians such as politicians, journalists, and the general public, it is critical to be able to disseminate the information of statistical studies without statistical jargon as " [u]ltimately, statistical studies cannot offer any benefits in the real world unless they can be understood by nonstatisticians" (Gardenier 2012, p. 652).
Despite the evidence to indicate the benefits of including communication in statistics courses, it is often not taught or emphasized (Gardenier 2012). Limited work (e.g., Batova and Ruediger 2019;Stromberg and Ramanathan 1996;Khachatryan and Karst 2017;Woodard, Lee, and Woodard 2020) exists on how to enhance communication skills in statistics courses. The purpose of this article is to describe a series of four workshops designed to develop communication skills in statistics graduate students and to provide advice on how these workshops or elements of these workshops could be used at the undergraduate and graduate levels. Through a collaboration between the statistics faculty and the Writing Center at Montana State University (MSU), a series of four workshops were developed targeted at improving communications skills in statistics graduate students, primarily master's students. Prior to the first workshop and after the fourth workshop, students were given the opportunity to respond to a survey regarding their attitudes toward writing and their writing goals. This article is organized as follows: Section 2 describes the workshops developed and their implementation. In the Section 3 we provide information on the writing survey and present the results from the pre-and post-surveys. Section 4 concludes and provides recommendations.

The Development and Implementation of Writing Workshops
The Department of Mathematical Sciences at MSU offers degrees in statistics at all levels including a BS in mathematics with a statistics concentration, a MS and PhD in statistics, and a graduate certificate in applied statistics. As part of the requirements for the MS degree, students complete either a thesis (at least 10 thesis credits) and give an oral defense or complete a writing project (2 to 4 credits) and give a 15-minute seminar, typically completed during the second and final year in the program. Nearly all students choose to complete a writing project. To help improve the quality of writing projects and to raise awareness of writing resources available on campus, the statistics faculty collaborated with the Writing Center at MSU to develop four 50-minute workshops that were delivered every other week during the first half of the Spring 2018 semester. Students working on a writing project were required to attend these seminars. In addition, statistics faculty, first year master's students, and statistics PhD students attended these workshops. The slides used in these workshops are available in the supplemental materials. Though there are many writing and communication skills that are important for statisticians to possess, it was not possible to cover all topics. Because of this, writing and communication skills frequently cited as being important for statisticians to possess were identified. Most frequently mentioned was the ability to communicate statistical concepts and results in a nontechnical manner to the intended audience (see Derr 2000;Gardenier 2012;Hoeting and Givens 2002;Love 2016). Based on this, the overarching goal for the workshops was to develop communication skills needed to effectively communicate with nonstatisticians. Four skills frequently cited as crucial for statisticians to possess to effectively communicate with nonstatisticians were identified with each skill forming the basis of one of the workshops: • Using an appropriate format and structure to present work ( The first three workshops focused on written communication while the last workshop focused on verbal communication. The concept of written communication was broadly defined to mean any type of written work ranging from formal academic papers and theses to slides for presentations to documents prepared for clients and collaborators outside of statistics. Because statisticians often present written work verbally to clients and collaborators the final workshop focused on the verbal and nonverbal aspects of effective oral communication. While these workshops were developed to develop communication skills identified as critical for statisticians to possess, these workshops are also applicable to data science students as these same communication skills are required in that field.

Workshop 1: Recognize Good Writing
The main purpose of the first workshop was how to correctly format and structure written work. To accomplish this, two learning objectives were used: (1) learn how to read aesthetically and (2) learn how to write and structure a paper in a way that enhances aesthetics for the reader. Aesthetic reading is a concept developed by Rosenblatt (1989). According to Rosenblatt (1989) there are two ways a reader approaches reading text: efferent reading where the reader's focus is on acquiring information (e.g., reading an instruction manual) and aesthetic reading where the reader's focus is engaged in the process of reading such as for enjoyment or entertainment (e.g., reading poetry). Though most work done by statisticians is intended for efferent reading (e.g., academic journal articles and technical reports), it is important to consider aesthetic reading as this ensures that the work is engaging to nonstatisticians.
To motivate what reading aesthetically entails, workshop attendees formed small groups and were asked to discuss the prompt: "Think about writing in your discipline. How do you know when it's good?" After several minutes of discussion, groups shared their ideas. Common themes included having ideas flowing together and being easy to understand.
To help develop the skill of reading aesthetically, six guiding questions were presented: 1. What is the structure of the text? (The genre) 2. What do the sentences look like? Sound like? 3. What voice does the author take? Tone? 4. What evidence does the author use? 5. How does he or she make his or her point? 6. What does the audience value?
Specific guidelines on how to write to enhance aesthetics for the reader, aimed at statistics, were then presented using the advice of Cummins (2009) and Love (2016). These guidelines included: (1) use the common structure of a paper (introduction, methods, results, and conclusions); (2) use graphs to enhance understanding as graphs are often the most concise and impactful way to present patterns in data; (3) use the past tense; and (4) describe general patterns without relying on values and then use values to support those statements.
The first guideline was discussed by presenting the common structure of a paper (introduction, methods, results, and conclusions), discussing what elements typically go in each section, and noting that there is flexibility in this structure. For the second guideline, which was briefly touched on as presenting tables and graphs was the focus of the second workshop, it was noted that text should be understandable without the graphics. Though not done in this workshop, incorporating examples from Cummins (2009) would be helpful here, specifically by using the examples from the section "Integrate Nontextual Results with the Text" (p. 40). A specific example was not provided for the third guideline as this was part of the final small group activity described below. To illustrate the fourth guideline, an example from the Purdue Online Writing Lab was used (Purdue Online Writing Lab n.d.). An example of a poor way to communicate results is: "A t-test (t = 3.59) showed that the two groups were significantly different (p < 0.01)" while a better way to present the results is: "Women scored higher than men on the aptitude test (t = 3.89, p < 0.01). " After introducing these guidelines small groups were formed again. Each group was asked to evaluate the paper The effects of run-of-river hydroelectric power schemes on invertebrate community composition in temperate streams and rivers by Bilotta et al. (2016) using the guidelines of reading aesthetically. This article was chosen as it followed the format recommended for aesthetic reading, incorporated graphics, and extensively used the results of statistical analyses familiar to our students. Though any paper could be used, we recommend choosing a paper that is fairly well-written in terms of aesthetics. Groups discussed what elements were well-written in terms of aesthetics and what elements could use improvement. After small group discussion, groups again came together to share what their groups had discussed and how we, as statisticians, could use the information in the workshop to help improve our writing. In general, most participants struggled with focusing on the aesthetics of the paper as opposed to the statistical content and wanted to critique the statistical methods used by the authors.

Workshop 2: Don't Forget the Figures
The focus of the second workshop was how to incorporate figures, such as tables and graphs, in written documents. There were three learning objectives: (1) understand the reason why figures are used and their importance; (2) choose the correct figure (e.g., table, type of plot) to display information; and (3) appropriately format figures.
To begin the workshop, small groups were formed, and two prompts were discussed: "Why do we use figures?" and "As a statistician, how do you understand the use of figures in your discipline?" Groups typically reported figures to be important as they are a visual way to convey complex information.
Because of the importance of figures in statistics, the remainder of the workshop was devoted to understanding the importance of developing good figures and what it means for a figure to be "good". To better understand this, the field of sociology of scientific knowledge was introduced, which asserts that the construction of scientific knowledge is a social process. Consequently, it is crucial for statisticians to use and develop figures that help nonstatisticians make sense of our work. As a large group, it was discussed what decisions need to be made when constructing figures.
The first decision discussed was deciding if a figure is, in the words of Tufte (2006), explanatory or exploratory. Explanatory figures are used to communicate results, such as pointing out important comparisons, whereas exploratory figures are used to better understand or describe the data. It was discussed that, in practice, figures are both explanatory and exploratory and that the writer needs to be conscious of whether the figure is intended to be more explanatory or exploratory as this affects how a reader interprets the figure.
The second decision discussed was how to best display data which consisted of three main considerations. The first consideration discussed was whether a table or graph should be used. Two recommendations by Tufte (2006) were provided: (1) use graphs instead of tables when looking at data collected over time as tables are more prone to a regency bias and (2) use an appropriate aspect ratio in graphs so that the information is not distorted. The second consideration was labeling figures including titles, axis labels, keys/legends, and captions. This consideration was illustrated using plots from Reese (2017). The third consideration was the use of color where it was noted that color can make plots easier to read though too much color creates clutter and confusion. A figure from Reese (2017) was also used to illustrate this consideration.
Using this information, participants then viewed several figures from Follett, Genschel, and Hofmann (2014), a paper based on the winning poster from the 2011 ASA Data Expo. One figure discussed in detail was Figure 1, a map of the Gulf of Mexico that displays where dead animals were sighted. Specifically, participants discussed the authors' use of color, readability of the plot, and potential concerns with overplotting. Alternative ways to display the data were also discussed.

Workshop 3: Let Your Data Speak
The third workshop focused on how to present written work simply and without statistical jargon to nonstatisticians. There were three learning objectives: (1) understand why nonstatisticians may have problems interpreting statistical results; (2) how to use elements of storytelling so that writing is understandable to nonstatisticians with an emphasis on the SUCCES Model by Heath and Heath (2007); and (3) appropriately structure writing to tell a story.
This workshop used ideas from Writing Science by Schimel (2012) who advocates scientists are writers who need to learn how to write compellingly. The workshop began with a prompt which was discussed in small groups: "Agree or disagree? Statistical results are often misunderstood, misrepresented, or ignored by the public. " Not too surprisingly, all participants agreed with this statement. This then led to a discussion regarding why participants agreed with this statement. One reason highlighted was from Schimel (2012) who contends that scientists are often ineffective at communicating outside of their specializations. Based on this, the workshop then explored ways to write that are understandable to nonstatisticians by using ideas from storytelling and fiction writing.
To begin this discussion, participants were prompted with: "Agree or disagree? When you write in statistics you are telling a story. " There was less agreement with this statement. Faculty tended to agree with this statement while students were more split on their agreement. A discussion then occurred where participants discussed why and how statistical writing may be like telling a story and how elements from storytelling can inform statistical writing. To explore how storytelling can be used in statistical writing, the SUCCES Model by Heath and Heath (2007) was introduced. Under this model, there are six basic principles to use to make an idea stick: Simple, Unexpected, Concrete, Credible, Emotions, and Stories, which are defined in Table 1. Each principle was briefly explained while the last principle, Stories, was emphasized as the focus of this workshop was on how to tell a story. For more information on the SUCCES Model, we refer the reader to the initial work of Heath and Heath (2007) and, for examples, we direct the reader to Le (2017) and Readingraphics (2018).
The workshop then focused on how to tell a story by using an appropriate structure. The basic OCAR structure of a paper (Schimel 2012) was highlighted as it is commonly used in scientific writing. Under this structure, a paper contains four elements: Opening, Challenge, Action, and Resolution. Each of these four elements were then further explained. Specifically, in the opening the problem is presented while the challenge is the question or hypothesis of interest. In the action portion, the work done to meet the challenge, including the methods and results, is explained and the resolution provides what was learned from this work. The OCAR structure was then related to the paper structure from Workshop 1. Specifically, the introduction section would contain an opening, background information (e.g., literature review), and a challenge (e.g., research question(s) of interest). The methods section begins describing the action by describing the data and what methods will be used to analyze the data while the results section continues the action by describing results. The conclusion section then is the climax of the story by addressing the challenge from the introduction (e.g., answering the research question of interest) and providing resolution such as limitations and areas of future work. At the end of this workshop, participants then discussed how to use these ideas from storytelling in their writing. The recommendations of focusing on sentence and paragraph structure, word choice, and being concise from Schimel (2012) were highlighted.

Workshop 4: Communicate Your Work
The fourth and final workshop focused on how to effectively communicate verbally and nonverbally with nonstatisticians. This topic was chosen as all graduate students are required to complete a consulting course for their degrees which requires them to communicate a statistical analysis and results with nonstatisticians. The three learning objectives for this workshop were: (1) identify client and collaborator barriers to communication; (2) develop strategies to break down barriers to communication; and (3) identify appropriate communication strategies based on how the content of the message will be perceived. The book Statistical Consulting: A Guide to Effective Communication by Derr (2000) was used as a reference throughout the workshop.
To begin the workshop, participants were given the prompt "During the course of your work as a statistician, what situations might you find yourself in where you have to communicate your work to people outside of your field?" Participants individually brainstormed and then got together in small groups to share situations where they have had to communicate our work to nonstatisticians. Many participants, including both faculty and students, noted that teaching and consulting were situations where they have had to frequently work with students and researchers in other fields.
The first topic discussed was barriers to effective communication from the perspective of a client in consulting. This included barriers such as being intimidated by statistics, not understanding statistics, and feeling vulnerable (e.g., feeling "dumb" for not knowing statistics and being able to analyze data). This segued into how statisticians could help break down these barriers by creating the conditions necessary for good communication. The conditions discussed were being accountable and dependable and building trust and rapport with the client. Specifics were also discussed such as asking good questions, replying to phone calls and emails in a timely and professional manner, providing accurate results, and being willing to admit when you (the statistician) made a mistake or do not understand something the client is telling you. Using the advice of Derr (2000), a strategy for effective communication during consulting was provided which involves setting an agenda that satisfies the client's needs, paying attention to cues and being willing to change course if needed, and being able to work with a client based on the client's abilities and explaining concepts multiple times and in multiple ways.
To better understand how to explain concepts to nonstatisticians in multiple ways, two small group activities were used. In the first activity, participants were given a list of common statistical concepts and methods such as dichotomous variable, autocorrelation, multicollinearity, and odds ratio and told to choose one. They were given one minute to prepare an oral explanation and then explained the concept in small groups without using statistical jargon. In the second activity participants were instructed to visually explain the concept that was chosen in the first activity using tables, charts, pictures, or any other visuals. Overall, participants enjoyed figuring out ways to visually explain statistical concepts and found the activity to be valuable as it required thinking creatively.
The remainder of the workshop was devoted to the idea that "the medium is in the message" such that the form of communication (the medium) affects how the content (the message) will be perceived. This suggests that different forms of communication (e.g., email, oral presentations, face-to-face meetings, reports, and phone calls) require different approaches. Because of this, careful consideration is needed on how to present information using different forms of communication. In a large group, it was discussed how to present content in a constructive manner. For example, instead of telling a client that "your study was poorly designed" say that "I'm concerned about how the study was designed. I believe there will be limitations about the conclusions we can make. " To better understand the role in how content matters, two prompts were discussed in small groups: "What are your strengths and weaknesses as a communicator?" and "Think about past interactions. When have you experienced breakdowns? Successes? What were the conditions that may have influenced each situation?" This allowed participants to reflect on what has and has not been successful previously in terms of communication and how future situations could be improved.

Survey Instrument and Results
Prior to the first workshop and after the completion of the fourth workshop, students were sent links to surveys, referred to as the pre-and post-surveys. The surveys were administered using Google forms and were approved by the Institutional Review Board at Montana State University. Students were first directed to a page where they were provided with the goals of the study, the inherent risks, and contact information of the IRB director and the principal investigator of the study and were asked to provide consent. Students providing consent were then directed to a second page containing the survey questions. Students were asked to provide their email address so that responses could be matched and were informed that after matching responses identifying information would be deleted.
In the pre-survey, found in the supplemental materials, students were first asked to provide their year in the graduate program and to identify previous writing services they have utilized. Students were then asked to respond to five sets of Likert scale items to assess attitudes toward writing, discussed in more detail shortly. At the end of the pre-survey students were asked to provide their goals for the workshops and areas in which they would like additional writing support.
In the post survey, also found in the supplemental, students were first asked to provide how many workshops they attended and what they found to be most helpful. Students were then asked to provide ways to improve the workshops and how they will incorporate the information from the workshops. Lastly, students responded to the same five sets of items pertaining to writing attitudes that they responded to in the pre-survey.
The survey instrument containing the five sets of Likert scale items to assess writing attitudes was developed by the third author based on a thorough literature review of instruments to assess writing attitudes and motivation and was designed to be applicable to a variety of fields. Each set of items addressed a different facet of writing attitudes: (1) attitudes toward writing in general; (2) attitudes toward planning a graduate writing project; (3) attitudes toward drafting a graduate writing project; (4) attitudes toward using conventions from statistics in writing; and (5) attitudes toward the writing process and time management. These five facets were chosen to be consistent with identified facets on commonly used and validated writing instruments, such as the Writing Activity and Motivation Scales (Troia et al. 2013) and the Self-Beliefs, Writing-Beliefs, and Attitude Survey (Wright, Hodges, and McTigue 2019).
Previous writing instruments indicate that writing attitudes and motivation is multi-dimensional with two major dimensions: (1) beliefs about the ability to complete the writing task and (2) beliefs about the writing task itself. The first dimension, beliefs about ability, has two components: self-efficacy, the belief that one possesses the competence to complete a given writing task, and self-concept, the beliefs one has about their writing abilities. The items in facets 2 and 3, drafting and planning a graduate writing project, were used to assess self-efficacy as both drafting and planning are parts of completing a writing task while self-concept was assessed using items from facet 5, attitudes toward the writing process and time management. Beliefs that are specific to the task of writing were assessed using the items from facets 1 and 4, attitudes toward writing in general and attitudes toward using conventions from statistics. Two sets of items were chosen to assess attitudes towards the task of writing as students may have different attitudes regarding writing in general versus writing in a specific discipline.
The items used in the survey were either written by the third author to be similar to items on existing instruments or modified from previous instruments to reflect the target audience of graduate students as most writing instruments have been designed to assess K-12 students. Consistent with other writing instruments, items were measured on a Likert scale (with choices of strongly disagree, disagree, neutral, agree, and strongly agree) as the goal of the surveys was to assess attitudes and to ease the burden of respondents. Though the survey instrument used in this study has not been psychometrically validated, it was developed based on psychometrically validated instruments.
For each set of items, respondents were first provided the prompt "Please rate your level of agreement with the following statements regarding ________" where the blank was replaced to reflect the facet of interest (e.g., writing in general). The respondents then responded to each item. The number of items varied by facet with six items for the first and fourth facets of writing in general and using conventions from statistics, four items for the second facet of planning a graduate writing project, seven items for the third facet of drafting a graduate writing project), and nine items for the fifth facet of writing process and time management.
Because the surveys were designed to be applicable to a wide variety of disciplines, not every item was directly applicable to the goals of the workshops. We primarily chose items that are relevant to the two dimensions of writing attitudes and motivation and their applicable components and that align with the goals of the writing workshops. Two additional items, items 2 and 9, though not directly tied to the goals of the workshops, were chosen based on personal experience-it had been noticed by many faculty in the department that students often procrastinated when it came to writing and/or had anxiety regarding writing. Items chosen included: • Items that assess beliefs about writing ability: self-efficacy (dimension 1, component 1) • Items that assess beliefs about writing ability: self-concept (dimension 1, component 2) (a) I can communicate effectively with advisors/professors about expectations for the project. (b) I can overcome procrastination, fear of failure, or writing blocks.
• Items that assess beliefs about the writing task (dimension 2) (a) I can write a strong paper in a graduate course.
(b) I can use appropriate conventions from statistics. (c) I can adjust my writing to meet the expectations of different disciplines when writing for different courses.

Pre-Survey Results
Fifteen students responded to the pre-survey. The year in the graduate program of the 15 respondents was relatively equally spread with three students in their first year, four in their second year, five in their third year, and the remaining three in their fourth or later year. The most used writing service was participation in peer review with seven students indicating using this service. The remaining services included formal and informal writing groups (2 students), workshop attendance (3 students), review by staff at the Writing Center (2 students) and visiting the Center for Student Success (1 student) which is a center on campus that provides programs and resources to students that promote student success. Two students reported that they had never utilized any writing services and two other students did not respond. Table 2 provides the results of chosen items with complete results found in Table S1 in the supplemental material. These results indicate that, in general, students reported high levels of self-efficacy as a majority of students either reported agreeing or strongly agreeing with 5 of the 7 items. These included item 1 (taking notes and organizing information; 11 out of 15), item 3 (coming up with a strong argument or claim; 12 out of 15), item 5 (developing a clear organizational structure; 10 out of 15), and item 7 (maintaining a sense of who the audience is; 10 out of 15). Students also reported relatively positive attitudes regarding the ability to complete a writing task as a majority of students either agreed or strongly agreed with 2 out of the 3 items including item 10 (writing a strong paper for a graduate course; 10 out of 15) and item 11 (using conventions from statistics; 11 out of 15). In contrast, students tended to report lower levels of self-concept as the majority of students either agreed or strongly agreed with 1 of the 2 items, item 8 (communicating effectively with advisors/professors; 9 out of 15). Notably, the items with the most responses of disagree or strongly disagree are items 2 (writing a paper without experiencing overwhelming feelings of fear or distress; 5 out of 15) and item 9 (can overcome procrastination, fear of failure, or writing blocks; 6 out of 15) both of which concern emotions students feel around writing. Overall, these results suggest that it may not be writing or communication where students struggle but rather the emotions that arise while writing or communicating. Goals expressed by students fell into one of two categories: writing and research. Writing goals included enhancing creative and technical writing skills, writing more strongly worded essays, learning how to write a refereed paper including proper formatting, understanding how to communicate thoughts especially to nonstatisticians, and improving confidence in writing. Research goals included learning ways and resources available to conduct research, becoming better at organizing time and ideas to write, and setting writing goals.
The areas students identified as where they would like additional support tended to be similar to their goals for the workshops. For example, several students noted that they would like additional support in terms of how to write and structure a paper and of the resources available to them which mirror the goals of learning how to write a refereed-paper and learning about the writing resources available. Students also responded that they would like additional support in understanding the conventions of writing in statistics while also being able to clearly explain results to nonstatisticians.

Post-Survey Results
Seven students responded to the post-survey. Six of these students attended all four workshops while the remaining student attended two. Four students indicated that working with other students and faculty was most valuable as it allowed for more opinions and different perspectives to be explored. Students also found working with Writing Center staff to be helpful as this allowed for a different disciplinary perspective that they had not been exposed to previously. Students also noted that they became aware of the services offered by the Writing Center and several students noted that they either were using these services or planned to in the future. Additionally, students enjoyed the interactive nature of the workshops especially the fourth workshop where they participated in activities that mimic situations that statisticians often are in.
When asked how to improve the workshops, four of the students did not provide a response. The other three students provided similar feedback in that they would like the workshops to be more focused on statistics. Students noted that they found the information presented helpful but that it was general in nature. The students recommended tailoring the workshops to include more specific advice on how to write with statistics by providing more examples and writing samples and by having students bring in their own writing for critique similar to what would occur in writing groups. Table 3 contains the results of the post-survey for the 12 chosen items. These results are interpreted cautiously due to the small sample size and because the respondents to the postsurvey tended to be respondents that completed the pre-survey with responses of agree and strongly agree. Complete results are found in Table S3 in the supplemental materials.
In contrast to the pre-survey, a majority of students either agreed or strongly agreed to all 12 items. Overall, this may suggest more positive attitudes toward writing though it may also be an artifact of the attitudes of the students choosing to Items Assessing Beliefs about Writing Ability: Self-Efficacy (Dimension 1, Component 1) I can take notes and organize information. 0 1 1 1 4 I can write a paper without experiencing overwhelming feelings of fear or distress.  complete the post-survey. Though this is a concern, the shift in those being able to write a paper without experiencing overwhelming feelings of fear or distress is considerable and may indicate that the workshops helped students with their emotions toward writing.

Pre-and Post-Survey Results
Six students responded to both the pre-and post-surveys. Table 4 provides the results for the items directly related to the goals of the workshops. For ease of comparison, the mode(s) for the pre-and post-surveys are provided as are the number of students that show less, the same, and more agreement from the pre-to post-survey. This allows for a more holistic interpretation of results as opposed to just focusing on modes or converting the categories to numerical values (e.g., strongly disagree is coded as a 1 and strongly agree is coded as a 5). Complete results for all survey items are found in Table S3 in the supplemental materials. When interpreting these results, it is necessary to note that they must be interpreted cautiously due to the small sample size. Unfortunately, it was not possible to increase the number of students responding to both surveys, in particular the postsurvey. Further, because the most students responding to both surveys tended to have responses of agree and strongly agree in the pre-survey, care needs to be taken to not misrepresent the results as they should be considered exploratory in nature. Results indicate that the majority of students either had a response that stayed the same or showed greater agreement for all 7 self-efficacy items. For items 1 (taking notes and organizing information) and 5 (develop a clear organizational structure) 3 students indicated stronger agreement while for items 2 (writing without overwhelming feelings of fear or distress), 6 (com-municate ideas clearly), and 7 (maintain a sense of who the audience is) 4 students indicated stronger agreement. For both items regarding self-concept, items 8 (being able to communicate effectively with professors/advisers) and 9 (being able to overcome procrastination, fear of failure, and writing blocks) all students either had the same level of agreement (4 students) or a greater level of agreement (2 students). Encouragingly, items 2 (writing without overwhelming feelings of fear or distress) and 9 (being able to overcome procrastination, fear of failure, and writing blocks), which both relate to emotions students may experience around writing showed improvement. Because fear and distress can be a major obstacle in writing, a decrease in students experiencing these emotions is encouraging as it indicates the workshops may have helped decrease some of the fears of students. These results are also encouraging as the remaining items indicate a positive shift in attitudes. Though it is not possible to say this is solely due to the workshops, these results, in conjunction with the qualitative information provided by respondents, indicate the workshops were beneficial in this area.
Results for attitudes toward the writing task are noticeably different in that for all three items, (item 10: writing a strong paper for a graduate course; item 11: using conventions from statistics; item 12: ability to adjust writing to meet expectations of different disciplines) there were two students reporting lower levels of agreement. Interestingly, though item 12 showed the largest decrease in agreement, there was also a large increase in agreement with three students indicating a greater level of agreement. Unfortunately, none of the open response questions provide insight as to why there were decreases in agreement for these items. One potential reason why there may have been a decrease in agreement for these items, is that the workshops introduced students to ideas and concepts they were unaware of which could lead to a decrease in their confidence regarding these items.

Conclusion
Being able to communicate effectively is a necessary skill for statisticians and previous studies have indicated that incorporating written and oral communication in statistics courses improves statistical understanding. Despite this, limited emphasis is often placed on communication in statistics courses. In this article, we presented the material covered in four writing workshops geared toward statistics graduate students and discussed the results from pre-and post-workshop surveys. Results indicate that prior to the workshops, students had high levels of self-efficacy and positive attitudes toward writing tasks and that after the workshops students reported high levels in both components of beliefs about ability to complete the writing task dimension (self-efficacy and self-concept) and the beliefs about the writing task itself. When examining students that completed both surveys, students responded with either the same or greater level of agreement for all 12 items. Most encouraging is that there were noticeable improvements for the items that examine the emotional responses students have regarding writing. Though most students had the same or greater agreement, for all three items regarding attitudes toward the writing task itself, 2 students responded with a lower level of agreement.
There were several benefits of these workshops we observed. First, students were able to devote time to reflect on their writing which, though important, is often not something that many faculty have noticed they typically do either in their courses or on their own. Second, students became more aware of the writing resources available to them through the Writing Center and more willing to use those resources. Many of our PhD students joined the student writing groups run by the Writing Center and/or were working with or planning on working with Writing Center staff. Third, over the course of the workshops, students reported feeling less fear and distress stemming from writing. This fear can be a major obstacle for students to begin or continue writing so being able to help reduce this fear is beneficial. Lastly, students found it beneficial to discuss writing with other students and faculty. These interactions allowed students to realize that other students and faculty also face writing challenges and that there are resources available to help them.

Incorporating Writing Workshops in Statistics Courses
These workshops or elements from them could easily be implemented by instructors in statistics courses. One element that would be easy to include and would help faculty is administering the survey instrument or items from it. As faculty we may be aware that students struggle with writing, but it can be difficult to identify why students struggle with writing and what would help them most. The results of the survey would help faculty in this regard. These workshops or elements from them can also help faculty identify whether students are struggling with writing, course material, or both. The faculty that participated in these workshops also noted that due to the discussions that occurred during the workshops they gained a better understanding of where their students are struggling.
Though these workshops were developed for statistics graduate students, they are also suitable for undergraduates, especially those in mid-to upper-level statistics courses. The skills emphasized in these workshops are recommended in the GAISE guidelines and consequently incorporating the workshops or elements from them would benefit undergraduates as well. For example, the activities in the fourth workshop where students are given one minute to prepare a brief presentation describing a statistical concept could easily be incorporated into courses. The first author has used this activity in several courses and has had several students remark that it has helped them realize that while they internally understand a concept it was very hard to explain it to someone else, especially those outside of statistics. The workshops and activities in them could be similarly modified to the level of the course and material being discussed to help benefit students and items in the survey instrument modified to reflect the course. Overall, these workshops are flexible in nature and can be used to help improve writing for both undergraduate and graduate students.

Limitations and Recommendations
As noted previously, there are several limitations of this study when it comes to the survey instrument. Specifically, this writing instrument has not been psychometrically validated nor was it written to be specific to statistics, sample sizes are small, and the students choosing to respond to the post-survey are not representative of all students attending the workshops.
To overcome some of these limitations, we provide several recommendations. First, strongly encourage students to complete both surveys. Because these workshops were not part of a course, it was difficult to incentivize students to complete both as we could not offer extra credit or require their completion as an assignment. Second, instructors may want to exclude certain items from the survey. For example, one item is regarding the use of library databases which are not covered in the workshops. This item could easily be removed. Third, items that are specific to statistics and the objectives of the workshops should be included. In our study, the writing instrument used was developed to be applicable to multiple disciplines as it was developed by Writing Center staff for this purpose. Items could be developed that are more directly applicable. For example, one item could be "I can make figures and tables that follow standard statistical conventions. " This would provide more detailed feedback for instructors. Lastly, we recommend including items that address the emotions students may experience with writing. As instructors, we do not "see" this aspect of writing in written work though we know it can, and often does, play a major role in writing. Being able to gauge students' feelings regarding writing provides valuable information to the instructor which allows the instructor to address these concerns and provide necessary support.