Address correspondence to Jane E. Miller, Ph.D., Research Professor, Institute for Health, Health Care Policy and Aging Research, and Edward J. Bloustein School of Planning and Public Policy, Rutgers University, New Brunswick, NJ 08901.
Watching: Creating a poster
APPENDIX A.1. Comparison of Research Papers, Presentations, and Posters—Materials and Audience Interaction.
APPENDIX A.2. Comparison of Research Papers, Presentations, and Posters—Contents.
Objectives
Posters are a common way to present results of a statistical analysis, program evaluation, or other project at professional conferences. Often, researchers fail to recognize the unique nature of the format, which is a hybrid of a published paper and an oral presentation. This methods note demonstrates how to design research posters to convey study objectives, methods, findings, and implications effectively to varied professional audiences.
Methods
A review of existing literature on research communication and poster design is used to identify and demonstrate important considerations for poster content and layout. Guidelines on how to write about statistical methods, results, and statistical significance are illustrated with samples of ineffective writing annotated to point out weaknesses, accompanied by concrete examples and explanations of improved presentation. A comparison of the content and format of papers, speeches, and posters is also provided.
Findings
Each component of a research poster about a quantitative analysis should be adapted to the audience and format, with complex statistical results translated into simplified charts, tables, and bulleted text to convey findings as part of a clear, focused story line.
Conclusions
Effective research posters should be designed around two or three key findings with accompanying handouts and narrative description to supply additional technical detail and encourage dialog with poster viewers.
Keywords: Communication, poster, conference presentation
An assortment of posters is a common way to present research results to viewers at a professional conference. Too often, however, researchers treat posters as poor cousins to oral presentations or published papers, failing to recognize the opportunity to convey their findings while interacting with individual viewers. By neglecting to adapt detailed paragraphs and statistical tables into text bullets and charts, they make it harder for their audience to quickly grasp the key points of the poster. By simply posting pages from the paper, they risk having people merely skim their work while standing in the conference hall. By failing to devise narrative descriptions of their poster, they overlook the chance to learn from conversations with their audience.
Even researchers who adapt their paper into a well-designed poster often forget to address the range of substantive and statistical training of their viewers. This step is essential for those presenting to nonresearchers but also pertains when addressing interdisciplinary research audiences. Studies of policymakers (DiFranza and the Staff of the Advocacy Institute 1996; Sorian and Baugh 2002) have demonstrated the importance of making it readily apparent how research findings apply to real-world issues rather than imposing on readers to translate statistical findings themselves.
This methods note is intended to help researchers avoid such pitfalls as they create posters for professional conferences. The first section describes objectives of research posters. The second shows how to describe statistical results to viewers with varied levels of statistical training, and the third provides guidelines on the contents and organization of the poster. Later sections address how to prepare a narrative and handouts to accompany a research poster. Because researchers often present the same results as published research papers, spoken conference presentations, and posters, Appendix A compares similarities and differences in the content, format, and audience interaction of these three modes of presenting research results. Although the focus of this note is on presentation of quantitative research results, many of the guidelines about how to prepare and present posters apply equally well to qualitative studies.
WHAT IS A RESEARCH POSTER?
Preparing a poster involves not only creating pages to be mounted in a conference hall, but also writing an associated narrative and handouts, and anticipating the questions you are likely to encounter during the session. Each of these elements should be adapted to the audience, which may include people with different levels of familiarity with your topic and methods (Nelson et al. 2002; Beilenson 2004). For example, the annual meeting of the American Public Health Association draws academics who conduct complex statistical analyses along with practitioners, program planners, policymakers, and journalists who typically do not.
Posters are a hybrid form—more detailed than a speech but less than a paper, more interactive than either (Appendix A). In a speech, you (the presenter) determine the focus of the presentation, but in a poster session, the viewers drive that focus. Different people will ask about different facets of your research. Some might do policy work or research on a similar topic or with related data or methods. Others will have ideas about how to apply or extend your work, raising new questions or suggesting different contrasts, ways of classifying data, or presenting results. Beilenson (2004) describes the experience of giving a poster as a dialogue between you and your viewers.
By the end of an active poster session, you may have learned as much from your viewers as they have from you, especially if the topic, methods, or audience are new to you. For instance, at David Snowdon”s first poster presentation on educational attainment and longevity using data from The Nun Study, another researcher returned several times to talk with Snowdon, eventually suggesting that he extend his research to focus on Alzheimer”s disease, which led to an important new direction in his research (Snowdon 2001). In addition, presenting a poster provides excellent practice in explaining quickly and clearly why your project is important and what your findings mean—a useful skill to apply when revising a speech or paper on the same topic.
WRITING FOR A VARIED PROFESSIONAL AUDIENCE
Audiences at professional conferences vary considerably in their substantive and methodological backgrounds. Some will be experts on your topic but not your methods, some will be experts on your methods but not your topic, and most will fall somewhere in between. In addition, advances in research methods imply that even researchers who received cutting-edge methodological training 10 or 20 years ago might not be conversant with the latest approaches. As you design your poster, provide enough background on both the topic and the methods to convey the purpose, findings, and implications of your research to the expected range of readers.
Telling a Simple, Clear Story
Write so your audience can understand why your work is of interest to them, providing them with a clear take-home message that they can grasp in the few minutes they will spend at your poster. Experts in communications and poster design recommend planning your poster around two to three key points that you want your audience to walk away with, then designing the title, charts, and text to emphasize those points (Briscoe 1996; Nelson et al. 2002; Beilenson 2004). Start by introducing the two or three key questions you have decided will be the focus of your poster, and then provide a brief overview of data and methods before presenting the evidence to answer those questions. Close with a summary of your findings and their implications for research and policy.
A 2001 survey of government policymakers showed that they prefer summaries of research to be written so they can immediately see how the findings relate to issues currently facing their constituencies, without wading through a formal research paper (Sorian and Baugh 2002). Complaints that surfaced about many research reports included that they were “too long, dense, or detailed,” or “too theoretical, technical, or jargony.” On average, respondents said they read only about a quarter of the research material they receive for detail, skim about half of it, and never get to the rest.
Readmore: 9780030672781 – Your Search For Holt Elements Of Literature
To ensure that your poster is one viewers will read, understand, and remember, present your analyses to match the issues and questions of concern to them, rather than making readers translate your statistical results to fit their interests (DiFranza and the Staff of the Advocacy Institute 1996; Nelson et al. 2002). Often, their questions will affect how you code your data, specify your model, or design your intervention and evaluation, so plan ahead by familiarizing yourself with your audience”s interests and likely applications of your study findings. In an academic journal article, you might report parameter estimates and standard errors for each independent variable in your regression model. In the poster version, emphasize findings for specific program design features, demographic, or geographic groups, using straightforward means of presenting effect size and statistical significance; see “Describing Numeric Patterns and Contrasts” and “Presenting Statistical Test Results” below.
The following sections offer guidelines on how to present statistical findings on posters, accompanied by examples of “poor” and “better” descriptions—samples of ineffective writing annotated to point out weaknesses, accompanied by concrete examples and explanations of improved presentation. These ideas are illustrated with results from a multilevel analysis of disenrollment from the State Children”s Health Insurance Program (SCHIP; Phillips et al. 2004). I chose that paper to show how to prepare a poster about a sophisticated quantitative analysis of a topic of interest to HSR readers, and because I was a collaborator in that study, which was presented in the three formats compared here—as a paper, a speech, and a poster.
Explaining Statistical Methods
Beilenson (2004) and Briscoe (1996) suggest keeping your description of data and methods brief, providing enough information for viewers to follow the story line and evaluate your approach. Avoid cluttering the poster with too much technical detail or obscuring key findings with excessive jargon. For readers interested in additional methodological information, provide a handout and a citation to the pertinent research paper.
As you write about statistical methods or other technical issues, relate them to the specific concepts you study. Provide synonyms for technical and statistical terminology, remembering that many conferences of interest to policy researchers draw people from a range of disciplines. Even with a quantitatively sophisticated audience, don”t assume that people will know the equivalent vocabulary used in other fields. A few years ago, the journal Medical Care published an article whose sole purpose was to compare statistical terminology across various disciplines involved in health services research so that people could understand one another (Maciejewski et al. 2002). After you define the term you plan to use, mention the synonyms from the various fields represented in your audience.
Consider whether acronyms are necessary on your poster. Avoid them if they are not familiar to the field or would be used only once or twice on your poster. If you use acronyms, spell them out at first usage, even those that are common in health services research such as “HEDIS®”(Health Plan Employer Data and Information Set) or “HLM”(hierarchical linear model).
Poor: “We use logistic regression and a discrete-time hazards specification to assess relative hazards of SCHIP disenrollment, with plan level as our key independent variable.”
Comment: Terms like “discrete-time hazards specification” may be confusing to readers without training in those methods, which are relatively new on the scene. Also the meaning of “SCHIP” or “plan level” may be unfamiliar to some readers unless defined earlier on the poster.
Better: “Chances of disenrollment from the State Children”s Health Insurance Program (SCHIP) vary by amount of time enrolled, so we used hazards models (also known as event history analysis or survival analysis) to correct for those differences when estimating disenrollment patterns for SCHIP plans for different income levels.”
Comment: This version clarifies the terms and concepts, naming the statistical method and its synonyms, and providing a sense of why this type of analysis is needed.
To explain a statistical method or assumption, paraphrase technical terms and illustrate how the analytic approach applies to your particular research question and data:
Poor: “The data structure can be formulated as a two-level hierarchical linear model, with families (the level-1 unit of analysis) nested within counties (the level-2 unit of analysis).”
Comment: Although this description would be fine for readers used to working with this type of statistical model, those who aren”t conversant with those methods may be confused by terminology such as “level-1” and “unit of analysis.”
Better: “The data have a hierarchical (or multilevel) structure, with families clustered within counties.”
Comment: By replacing “nested” with the more familiar “clustered,” identifying the specific concepts for the two levels of analysis, and mentioning that “hierarchical” and “multilevel” refer to the same type of analytic structure, this description relates the generic class of statistical model to this particular study.
Presenting Results with Charts
Charts are often the preferred way to convey numeric patterns, quickly revealing the relative sizes of groups, comparative levels of some outcome, or directions of trends (Briscoe 1996; Tufte 2001; Nelson et al. 2002). As Beilenson puts it, “let your figures do the talking,” reducing the need for long text descriptions or complex tables with lots of tiny numbers. For example, create a pie chart to present sample composition, use a simple bar chart to show how the dependent variable varies across subgroups, or use line charts or clustered bar charts to illustrate the net effects of nonlinear specifications or interactions among independent variables (Miller 2005). Charts that include confidence intervals around point estimates are a quick and effective way to present effect size, direction, and statistical significance. For multivariate analyses, consider presenting only the results for the main variables of interest, listing the other variables in the model in a footnote and including complex statistical tables in a handout.
Provide each chart with a title (in large type) that explains the topic of that chart. A rhetorical question or summary of the main finding can be very effective. Accompany each chart with a few annotations that succinctly describe the patterns in that chart. Although each chart page should be self-explanatory, be judicious: Tufte (2001) cautions against encumbering your charts with too much “nondata ink”—excessive labeling or superfluous features such as arrows and labels on individual data points. Strive for a balance between guiding your readers through the findings and maintaining a clean, uncluttered poster. Use chart types that are familiar to your expected audience. Finally, remember that you can flesh out descriptions of charts and tables in your script rather than including all the details on the poster itself; see “Narrative to Accompany a Poster.”
Describing Numeric Patterns and Contrasts
As you describe patterns or numeric contrasts, whether from simple calculations or complex statistical models, explain both the direction and magnitude of the association. Incorporate the concepts under study and the units of measurement rather than simply reporting coefficients (β”s) (Friedman 1990; Miller 2005).
Poor: “Number of enrolled children in the family is correlated with disenrollment.”
Comment: Neither the direction nor the size of the association is apparent.
Poor : “The log-hazard of disenrollment for one-child families was 0.316.”
Comment: Most readers find it easier to assess the size and direction from hazards ratios (a form of relative risk) instead of log-hazards (log-relative risks, the β”s from a hazards model).
Better: “Families with only one child enrolled in the program were about 1.4 times as likely as larger families to disenroll.”
Comment: This version explains the association between number of children and disenrollment without requiring viewers to exponentiate the log-hazard in their heads to assess the size and direction of that association. It also explicitly identifies the group against which one-child families are compared in the model.
Presenting Statistical Test Results
On your poster, use an approach to presenting statistical significance that keeps the focus on your results, not on the arithmetic needed to conduct inferential statistical tests. Replace standard errors or test statistics with confidence intervals, p-values, or symbols, or use formatting such as boldface, italics, or a contrasting color to denote statistically significant findings (Davis 1997; Miller 2005). Include the detailed statistical results in handouts for later perusal.
Readmore: The Heart Of Wisdom Teaching Approach : Bible Based Homeschooling
To illustrate these recommendations, Figures 1 and and22 demonstrate how to divide results from a complex, multilevel model across several poster pages, using charts and bullets in lieu of the detailed statistical table from the scientific paper (Table 1; Phillips et al. 2004). Following experts” advice to focus on one or two key points, these charts emphasize the findings from the final model (Model 5) rather than also discussing each of the fixed- and random-effects specifications from the paper.
Categories: literature