Kim Love is the owner and lead collaborator at K. R. Love Quantitative Consulting and Collaboration. She earned her PhD from Virginia Tech in 2007 and, prior to starting her practice, served as the associate director of the University of Georgia Statistical Consulting Center. Visit her blog about starting her private practice.
One of the biggest challenges faced by any collaborative statistician is communicating statistical information to those with less knowledge of statistics. Many of us with a formal education in statistics receive extensive training in theory, methods, and application; however, even with a PhD in statistics, it is not uncommon to have taken one or no courses that focus on communicating this knowledge to those who can benefit from it. In other words, many of us leave school with little understanding of how to put our skills into effective practice.
What do our nonstatistician colleagues need from us to get the most out of our interactions? As posited by Janice Derr in her 1999 textbook Statistical Consulting: A Guide to Effective Communication, there are five dimensions of quality that nonstatisticians evaluate when collaborating with statisticians:
- Availability of support
- Responsiveness of support
- Timeliness of support
- Completeness of support
- Pleasantness of support
Note that none of these dimensions directly incorporates correctness, technical savvy, or methodological awareness. It’s not that those are unimportant; they are extremely important, and a collaborative statistician will not last long without solid abilities in those areas. It’s simply that most nonstatisticians are unable to evaluate those aspects of a collaboration and have to make the assumption that their statistical collaborator possesses those skills.
The question then becomes, how can we improve our communication skills when working with nonstatisticians so they will understand and appreciate our expertise? The following recommendations are based on my personal experience and the advice of other statisticians in the consulting and collaboration community. I refer to nonstatisticians as “clients” in these recommendations, but that term is not limited to what one might view as a traditional consulting client; it could be a boss, a coworker, or even a friend who asks for help with a quantitative problem.
Focus on the client and project at hand, rather than general statistical concepts.
During my time at the University of Georgia Statistical Consulting Center, I supervised many students who were just learning to become collaborative statisticians. Across the board, when these students were initially challenged to explain a statistical method to a client, they provided equations full of Greek letters and other mathematical notation. While each client’s needs should be evaluated individually, for many clients, this tends to add to their confusion about a method rather than mitigate it.
Here is my personal hierarchy, from greatest to least chance of success, of techniques to explain statistical methods to most clients:
- Plain English
- Equations with your client’s variables written out in words
- Equations with mathematical notation
Example: “Across your group of students, for every additional point a student scores on the entrance exam, the final achievement score increases by an average of about half a point.”
Example: “Estimated Average Achievement Score = 0.32 + (0.54 x Entrance Score)”
Examples: “ŷ= 0.32 + 0.54x₁”; “y = β₀+β₁X₁+ε”
Well-labeled figures are always helpful when it comes to understanding, and should be used in tandem with these techniques when possible. Don’t label figures using the cryptic, abbreviated variable names we often use as programming shortcuts; this is a barrier to a client who does not think like a programmer and who would need to continuously remind him or herself of the meaning of those labels.
Some addendums: Some clients require mathematical notation for their eventual research output, including those who are publishing in quantitative academic journals in their fields. Many of my clients require a combination of these methods to both understand the concepts and be able to provide a final product that meets the requirements of their stakeholders.
Some clients do have an interest in learning more general information about statistics. When a client asks a general question (e.g., What is power?), it is still helpful to explain it in a way that is specific to that client and his or her research. For example, for a client in agriculture, “Power is the probability that you will choose a sample of lettuce plants for your study that will result in a statistically significant difference between your two lettuce strains, assuming a difference exists.” The less appropriate explanation would be “Power is the probability of rejecting the null hypothesis when the alternative hypothesis is true.”
Actively improve your communication; communication skills can be learned.
Why do so many new statisticians respond to clients’ difficulties in understanding with Greek letters and nonproject-specific explanations? The short answer is this is how most of us learned statistics. This notation and general conceptual discussion is useful in a classroom—it’s a shorthand language we have in common so we can learn advanced concepts quickly. However, its usefulness diminishes greatly when working with someone who does not share our background in quantitative sciences. The good news is, contrary to what many of us believe (at least in practice), communication abilities can be practiced and learned.
One of the best ways to improve communication in client interactions is through the use of video review. Think of it as collecting data on client interactions (with the client’s approval, of course). Set the recording device in a location where both the client and statistical collaborator are visible on the video, record the interaction, and then choose some or all of the video to watch—preferably with a trusted colleague. A video by Eric Vance from the Laboratory for Interdisciplinary Statistical Analysis at is a helpful resource for implementing video coaching and feedback sessions.
Focus on the following during video review:
- Was your communication effective? Did you provide explanations that were appropriate for your client, and did he/she seem to respond to them? Did you ask the necessary questions when you didn’t understand something? Listen to the verbal exchange, but also examine body language.
- What was the relationship between you and the client like? Was it collaborative or hierarchical? Was it constructive or combative?
- Did you follow a proper structure for the interaction (see the next recommendation)? Make a checklist.
- Most importantly, leave your review session with one or two goals for future client interactions.
Statisticians also can improve communication outside of live client sessions. New collaborators can practice communication skills using role play scenarios, in which one person acts as a client and the other acts as a statistical collaborator (and these can be video reviewed as well). Workshops and continuing education opportunities to help improve communication throughout a statistician’s career also are available at ASA conferences (such as the Conference on Statistical Practice or the Joint Statistical Meetings) and through other professional organizations.
Structure your interactions and their outcomes.
One of the keys to practicing communication effectively is to have a well-thought-out plan for your interactions. Several prominent collaborative statisticians have presented structures for interactions, including Derr (again in her textbook, Statistical Consulting: A Guide to Effective Communication) and Doug Zahn, who developed the POWER process for client interactions. While it is probably more important to have a structure in the first place than to adopt a specific meeting structure, these structures do have a number of elements in common. There is an initial period to prepare for the meeting; terms for the interaction and a mutual agenda are agreed upon between the statistical collaborator and the client; there is a work session in which information is exchanged productively, with opportunities to question and enhance understanding on both sides of the table; and time is allowed at the end to review the interaction and agree on steps going forward.
While it will take some time and experience to implement a meeting structure smoothly and with appropriate flexibility, the benefits to the participants are well worth it. When there is a structure in place and the statistician no longer needs to concentrate on the meeting logistics, it becomes much easier to focus on communication and gauge effectiveness. Note that this idea can be extended to written reports and other outcomes outside of immediate interactions—having a general, flexible structure in place to organize statistical information is immensely helpful to making that information understandable.
Gauge your clients’ knowledge and communication needs on an individual basis.
Not all nonstatisticians who need statistical expertise are the same. That may sound like common sense, but it is easy to begin treating all clients as if they are the same over time. While I have seen many attempts to categorize clients—some serious and some humorous—they fall short of describing the variety and nuances of nonstatisticians who seek statistical collaboration.
With respect to background knowledge, the simplest approach is to ask clients what kind of experience they have had with statistics. It’s important to do this in a respectful manner (see my next recommendation), as immediately firing off a barrage of questions related to specific statistical techniques and courses can be intimidating. Instead, clarify that you are asking because you want to make sure you use appropriate vocabulary and provide proper explanations. Also, invite them to ask questions any time you are not being clear.
Communication needs are a bit subtler, and I’ve emphasized just a few points in the following bullets:
- Atmosphere: Some clients prefer an atmosphere that includes friendly conversation, while others prefer a more polished, down-to-business environment. I usually try to gauge this based on my client’s demeanor after the introduction—is the discussion moving to the parking situation outside? Or is the client already placing material on the table and starting to tell me about some of the issues involved in his or her project?
- Attitudes: Unfortunately, some clients have preconceived negative perceptions about statistics; they have had little exposure and believe they have poor abilities. They may be nervous or even fearful about meeting with a statistician. Make an effort to be particularly patient when you recognize a client has an emotional reaction to working with statistics, and that client will be more receptive to what is being communicated.
- Directness: Some clients are very good about asking questions and guiding the direction of the meeting. Others are quieter and prefer to be invited to contribute. Never be afraid to ask a client if he or she understands something or is happy with the pacing and direction of the meeting.
Respect takes many forms during client interactions. General politeness (such as showing up on time, remaining focused on your interaction, and not interrupting your client) is one way to demonstrate respect. Some of what I have already discussed in my other recommendations also go toward demonstrating respect and are more specific to collaboration: acknowledge your client as an individual; structure your interactions so they are efficient; and be patient with your clients’ hang-ups when needed.
There is also an issue of professional respect, which cannot be understated—clients collaborate with us because they acknowledge we have statistical expertise they do not. It is important to recognize in return that our clients have expertise in their areas that we do not have. Just like clients do not become experts in statistics after an hour-long meeting, we do not become experts in their fields or areas of expertise during that meeting. Allow and encourage clients to contribute to the interaction. This not only results in open lines of communication and positive relationships, but even improves the technical aspects of statistical work, as it provides a more complete view of clients’ research problems and objectives.
One of the greatest contributions experienced statisticians can make to the field is to share their experiences. How do you explain complicated statistical procedures to nonstatisticians (think random effects, ordinal responses, and computationally intensive techniques)? How did you learn to expand your communication abilities?
Please feel free to add your own comments and thoughts about communication with nonstatisticians below.