As I mentioned, I did an analysis of ASA submissions and sessions for 2006-9. This is “regular” sessions only, not section sessions. The data source is a handout given to program committee members to help them see the patterns and plan future ASA meetings. It is not “official” data that is carefully cleaned, and contains some possible errors. After a little bit of backstage brouhaha about whether it is appropriate to share this data and some re-analysis in light of our exchanges, here are some graphs that leave out the topic areas. I may “publish” the findings with the areas named in a later post. There are some important findings:
(1) The number of sessions is relatively constant across years, but the number of submissions varies greatly. Thus the ratio of sessions to submissions (and thus, by implication, the probability of having a paper accepted) varies greatly from year to year.We had a backstage debate about possible explanations for the year-to-year variation in submissions (with some hypotheses being clearly rejected with data) and about whether these variations in submissions are an artifact (with my conclusion being they are not, base both on detailed patterns in the data and conversations with ASA about possible sources of artifacts). It is harder to get into an ASA regular session in a year when a lot of other people are also submitting papers. Per ASA, one trend they have seen is a rising percentage in the proportion of convention attendees who are on the program. I assume this is because most people cannot get money for convention travel unless they are on the program.
(2) Session submission topics get a widely varying number of submissions. Further, the same number of submissions yields varying numbers of sessions.Each data point in the graphs below is an open submission regular session topic.
Regular session organizers can ask for more sessions if they get a many good submissions, but they have to know this is possible (many inexperienced session organizers do not know this) and they have to hustle and ask early, because the total number of sessions is relatively fixed so earlier requests are more likely to be honored than later requests. The plot of number of sessions by number of submissions shows that the same number of submissions can yield different numbers of sessions, most often with the number of sessions varying by 1, but in a few cases by more. Further, the correlation between number of sessions and number of submissions is lower in years with more submissions (and was especially low in 2008).
Plotting the the ratio of sessions to submissions by the number of submissions, you can see that in more competitive years, the session/submission ratio is generally higher for the less-popular topics (which are guaranteed at least one session and sometimes get two), although there are a few topics that seem to have yielded zero sessions. The correlation between the number of submissions and the ratio of sessions to submissions is negative and is more strongly negative in years with more submissions. If we restrict analysis to only those topics that appeared in all four years (the graph on the right), the negative correlation is weaker although still seemed to be operating in 2008.
Although there is a broad year-to-year consistency in which areas are more popular than others, and broad consistency in the between-year trends (so that most areas had substantially more submissions in the higher-volume years), there are also substantial year-to-year fluctuations in the number of submissions to particular topics within these broad trends, with some areas being much more volatile in submission numbers than others. I speculate that some of these fluctuations are random and that others are the indirect effects of the mix of session offerings available in a given year’s program. I also see in the data evidence that some subareas are consistently better than others at getting more sessions, which I believe correlates with the extent to which a subarea is a well-organized community in ASA. More about that in a follow-up post.
What are the take-home messages? First, submitters need to understand that getting a paper accepted to a regular session at ASA is not some kind of pure merit standard. The competition for slots in regular sessions is not one big competition, but many smaller competitions with different numbers of competitors. Second, the number of other competitors for a slot has a large random component that has nothing to do with you. Third, if your paper qualifies for a more specialized or narrow topic area that is not a standard offering, your chances of acceptance would seem to be higher in that pool than in the more standard topics, especially in years when the overall level of submissions is higher. Fourth, there is a second-level competition between area organizers to be first into the ring with requests to add extra sessions. Session organizers with more savvy and hustle do better in this second-level competition.
In this analysis, I have not considered the possibility that differences in the session/submission ratio may be due to the session organizer’s assessment of the quality of the pool of submissions in a given year.
It should also be noted that the names of some areas evolve over time and get grouped and ungrouped, and this can affect apparent session/submission ratios.