A statistical model for abstract submission patterns to the International ...

5 downloads 7077 Views 88KB Size Report
information is of great importance for conference organizers as attendance, space ... Note. These statistics are based on information from the Vision '99 abstract book ... The number of abstracts in the submission system for the Open Call as a ...
Proceedings of the 9th International Conference on Low Vision, Vision 2008, Montreal, Quebec, Canada July 7 – 11, 2008

A statistical model for abstract submission patterns to the International Conference on Low Vision: Vision 2008 Walter Wittich a

a, b *

, Olga Overbury

b, c c , Jacques Gresset

Department of Neurology & Neurosurgery – Neuroscience, McGill University, b Department of Ophthalmology, SMBD Jewish General Hospital, c École d’optométrie, Université de Montréal, Montréal, Québec, Canada …

Abstract. Every conference relies on the number and quality of abstract submissions in order to assure the success of its scientific content. It remains, however, difficult to predict the final number of submissions that can be expected. This information is of great importance for conference organizers as attendance, space requirements and registration fees heavily depend on this estimate. Previous conferences in the same area are often used to estimate future attendance and submission numbers. There is general consensus that abstract submission follows an exponential function, whereby most abstracts arrive within the final few days before deadline. Vision 2008 followed its own submission process and used an exponential model to predict the number of submissions for the scientific program, which is presented here. Keywords: abstract submission, statistics

1. Introduction The organization of an international conference is challenging and requires the organizers to take certain risks. For example, until the day the conference opens, it is difficult to estimate how many attendees will indeed register and appear at the event. There are, however, certain clues to the eventual size of a conference of this type. Previous conferences in the same area are often used to gage the size of subsequent events. The International Low Vision Conference has been under the auspices of the International Society for Low Vision Research and Rehabilitation (ISLRR) for two decades. The conference is now held every three years and has continuously grown from the initial meeting in Pacific Grove, USA (1986), to subsequent conferences in Waterloo, Canada (1987), Beverly Hills, USA (1988), Melbourne, Australia (1990), Groningen, The Netherlands (1993), Madrid, Spain (1996), New York, USA (1999), Göteborg, Sweden (2002), London, UK (2005), to now Montreal, Canada (2008), and its next meeting scheduled in Kuala Lumpur, Malaysia, in 2011. One key component of estimating the number of attendees is the number of abstracts submitted for presentation in the scientific program. The conference relies on this number and quality of abstract submissions in order to assure the success of its scientific content. In addition, this program is intended to attract attendees that are not presenting, for example, for the purpose of obtaining Continuing Education (CE) credits. It remains, however, difficult to predict the final number of submissions that can be expected. This information is of great importance for conference organizers as attendance, space requirements and registration fees heavily depend on this estimate. Attendance and submission statistics are not always readily available for future conferences; however, in the case of Vision 1999 (New York), Vision 2002 (Göteborg) and Vision 2005 (London), the number of presented abstracts can be extracted from published abstracts and proceedings. Table 1 displays summary statistics for these conferences. Table 1. Number of abstracts published at the last 4 conference. Conference # of Posters # of Oral Presentations Total # of Abstracts Vision 1999 613 Vision 2002 186 231 417 Vision 2005 268 457 725 Vision 2008 (as of May 23) 287 531 818 Note. These statistics are based on information from the Vision ’99 abstract book (Lighthouse International, 1999) and the on-line programs of Vision 2002 (ISLRR, 2006) and Vision 2005 (RNIB, 2005).

Even though these statistics already give a first indication about an increase in submissions across consecutive conferences, what is not reflected is the actual number of abstracts that were submitted. The scientific committee of each event rejected a certain (unknown) percentage of abstracts that were considered unsuitable; in

* Corresponding author: e-mail address: [email protected] , Department of Ophthalmology, SMBD Jewish General Hospital, 3755 Cote-Ste-Catherine, Pavilion E-008, Montreal, Quebec, H3T 1E2, Canada

addition, presentations were withdrawn by the presenters for personal reasons (lack of funding, change in health status, conflicting dates, etc.). At this point, it needs to be mentioned that Vision 2002 most likely experienced a drop in attendance due to the aftermath of the events following September 11, 2001, in New York. In order to understand and describe abstract submission behaviour and to predict the final number of submissions for Vision 2008, the Program Committee decided to monitor their submission site and build a statistical model estimating the final number of posters and oral presentations at the conference.

2. Method and Results Vision 2008 decided to utilize the commercially available submission software provided by pharma service, a business unit of documediaS GmbH (pharma service, 2008). This company was chosen because of their decision to make their product accessible to screen reader programs for the visually impaired and blind. The abstract submission process for Vision 2008 was divided into three steps. First, the Open Call for abstracts began on October 5th 2007 and closed on January 15th 2008 (103 days). Second, the site was reopened on February 29th until March 31st 2008 for the submission of Invited Abstracts in R&R (Research & Rehabilitation) sessions, workshops, keynote speakers and CE credit courses. Third, and in parallel, another site was available from March 15 to May 23 2008 for submissions to the Late-Breaker Poster Session. For the Open Call, every 7 days the total number of new abstracts in “draft” and “submitted” status was added and entered in the model, using the graphing program and trend-line option within Excel. The function had the following form: y = 3.6968e0.2872x, accounting for 97.13% of the variance in the total number of submissions (see Figure 1). However, upon visual inspection of the data, it seemed more appropriate to divide the submission process into 2 time periods, each of which appear more linear in their increase over time: Period 1 from day 0 to day 94, and period 2 from day 94 to day 106. Using linear regression, a line was fit to each period. The function for period 1 was defined by y = 1.65x - 21.48 whereby r2 = .88, and for period 2 was defined by y = 47.44x - 4307.6, whereby r2 = .86.

Figure 1. The number of abstracts in the submission system for the Open Call as a function of time. The submission period was divided into two time spans: the initial 94 days and the final 12 days (9 before and 3 after submission deadline).

It should be pointed out that human behaviour rarely closely follows mathematical predictions or deadlines without a certain amount of error. In the case of the Open Call, 684 abstracts had been submitted by the deadline date; however, an additional 28 abstracts were “submitted” after the deadline before the review process began, bringing the final number of Open Call submission to 712. For the Invited Abstracts, the number of submissions was more easily predicted because each speaker had been approached individually. However, even in this scenario, the expected number of 125 was exceeded to 153 because additional CE courses and changes in the composition of R&R panels increased the number of submissions. The total number of Late-Breaker Poster submissions was 37 (as of May 23, 2008) and followed a similar pattern as the Open Call. In total, 809 abstracts were placed in the schedule by the scientific program committee, creating a challenge in accommodating all speakers and presenters at Vision 2008. 3. Discussion The accuracy of the projected number of abstracts changed substantially during the submission period, based on the number of additional data points. Using the exponential model, final prediction ranged from 1400 abstracts after 8 weeks to 700 during the final 2 weeks. In addition, several choices by the scientific committee challenged the prediction. Authors who submitted abstracts during the initial 60 days were offered to undergo pre-review for language and suitability for the conference, possibly resulting in an increased number of submissions early on and causing the initial overestimate. In addition, the R&R sessions and Late-Breaker posters consisted of 169 abstracts that were submitted after the deadline, which were not included in the prediction. Overall, the model was able to predict the final number of 712 submissions with reasonable accuracy 2 weeks before the deadline; however, it generally underestimated submissions during the last 2 weeks. For comparison, Vision 2005 in London had received 256 abstracts one day before submission deadline and the number rose to 600 at the deadline, with a final of 725 abstracts in the conference, including late submissions (Gary Rubin, personal communication). In terms of its predictive utility, the presented statistical model remains to be tested once the conference itself opens. The Organizing Committee was advised by Opus3, their professional conference organizers, that conference attendance can generally be estimated by multiplying the submitted abstracts by a factor of 3. However, in the case of Vision 2008, this is unlikely to be true because presenters were allowed to submit up to three abstracts. In addition, the international nature of the conference and the connected financial investment required to attend may not allow every individual interested in the event to attend. 4. General Conclusions The importance of abstract submission for the progress of scientific discovery and the need to communicate research findings in written form has often been emphasized. Specifically, for the process of abstract composition and acceptance, previous authors have published guidelines and advice across several academic domains (Cole & Koziol-McLain, 1997; Lindquist, 1993; McNamara, Grannell, Watson, & Bouchier-Hayes, 2001). The Scientific Program Committee of Vision 2008 received inquiries about abstract submission that ranged from “What is an abstract?” to a request for extension of the deadline from Nepal, stating that “electricity and internet access are not always available long enough to complete the on-line form”. The Committee was very pleased with both the number and general quality of submissions and encourages authors to continue their work, specifically to develop their materials into manuscripts for publication in peer-reviewed journals – a key research tool in communicating recent research and rehabilitation progress. References Cole, F. L., & Koziol-McLain, J. (1997). Writing a research abstract. Journal of Emergency Nursing, 23(5), 487-490. ISLRR. (2006). International Society for Low Vision Research and Rehabilitation - Home Page. Retrieved April 1st, 2008, from http://www.islrr.org/ Lighthouse International. (1999). Vision '99 International Conference on Low Vision - Vision Rehabilitation for the 21st Century. In Lighthouse International (Ed.), Vision '99. New York: Lighthouse International. Lindquist, R. A. (1993). Strategies for writing a competitive research abstract. Dimensions of Critical Care Nursing, 12(1), 46-53. McNamara, Grannell, M., Watson, R. G., & Bouchier-Hayes, D. J. (2001). The research abstract: worth getting it right. Irish Journal of Medical Science 170(1), 38-40. pharmaservice. (2008). pharmaservice - Home Page. Retrieved April 1st, 2008, from http://www.pharmaservice.de/ RNIB. (2005). Vision 2005 - Home Page. Retrieved April 1st, 2008, from http://www.rnib.org.uk/xpedio/groups/public/documents/code/public_rnib004044.hcsp

Financial Disclosure and Funding The authors do not have any financial interest in the software product mentioned within the article. Graduate student funding for WW was provided by the Canadian Institutes of Health Research (CIHR, STP # 53875; CGS DRA # 168341), The Réseau Vision of the Fonds de la Recherche en santé du Québec (FRSQ, # 9745), as well as the McGill Tomlinson Foundation and the Canadian National Institute for the Blind (CNIB).