Upon receipt of the reports on the census mid-year, the Committee undertook an analysis of the data in order to address a series of questions pertaining to the representation and advancement of women and minority groups. We were assisted by an undergraduate at Princeton with training in statistics, who was given access to the raw data (the anonymity of members was duly protected) from 2014 as well as from 2004.
One of the major aims of the analysis was to compare the numbers from 2004 with those from 2014 in order to assess whether disparities in gender representation and the representation of minority groups were being corrected. We here highlight several important points based on these charts.
- For a number of questions concerning underrepresented minorities, the sample size is too small to yield statistically meaningful conclusions, particularly as concerns representation in the profession (hiring and promotion). This makes it important to identify other forms of information gathering that can help the SCS track the representation and advancement of minorities and monitor the status of problems in equity and non-discrimination.
- What does appear to be the case, however, is greater representation of minorities (and women) at the BA level than at the graduate level and in full-time positions. The data suggests that one area that the SCS and CSWMG might together address is the pipeline: what factors contribute to the disappearance of underrepresented groups along the track from BA major to Full Professor, with particular attention in the case of underrepresented minorities to the gap between BA and PhD.
- The representation, in terms of percentage, of women at the level of Full and Associate has improved over the ten-year period surveyed. In the 2014 census 61% of full-time tenured professors were male and 39% female vs. 72% vs. 28% in the 2004 data. In 2004, 79% of full professors were men and only 21% women; whereas in 2014, the numbers are 68%/32%. At the associate level, 62% of associate professors in 2004 were men and 38% women vs. 56%/44% in 2014. We therefore have seen improvement in representation of women at the levels of associate and full. But the discrepancy in representation remains statistically significant.
Moreover, it is not entirely clear what to make of the numbers at the assistant professor level. In 2004, 56% of assistant professors were men and 44% were women. These numbers are the same in 2014. We see a similar situation in the category “hired on the tenure track”: in 2004, 54% men, 46% women; in 2014, 54% men, 46% women. Finally, we can compare numbers in the “tenure-track” category. The representation of women actually goes down here: in 2004, 55% of professors on the tenure track were men, 45% women; in 2014 those numbers are 57%/43%.
It is difficult for a non-expert to draw the appropriate conclusion here. It does seem likely, though, that we won’t see as significant gains in equity at the tenured levels if the representation of men and women at the lower levels of the ladder does not change. More research is needed to know what it is reasonable to expect in the next census (2017) based on time to tenure and to full-time promotion. Anecdotally, there is a problem across the university in terms of women advancing from associate to full. How might our data identify in a more fine-grained way problems with our profession, in particular?
Addressing these latter questions leads to the second set of conclusions.
- The membership of CSWMG does not have the skills to perform analysis of the raw data or draw conclusions that are statistically sound. It should be noted further that given that the SCS protects members’ identities in the census, no member of the committee could access the raw data. That is, even if a member did have statistical training, they couldn’t perform the analysis that the undergraduate research assistant undertook this year. This puts the onus on the Chair and members of the committee to outsource and finance the analysis undertaken to identify problems of representation and advancement of minority groups. The situation is too contingent and unstable for a task as important as this one.
- There is a second point of clarification worth stressing here: CSWMG does not strictly speaking provide data analysis on the basis of the census. Rather, the committee tries to analyze the data in a way that identifies and tracks problems of representation and advancement. The census information posted on the website for the membership of the SCS to review is not raw data: it has itself been analyzed according to rubrics presumably given to the company that undertook the census. This means that if the SCS were to recognize that the representation and advancement of minority groups was a priority for the membership as a whole, as this committee believes it should, the kind of analysis just undertaken by this committee could be seen as part of the professional process of converting the raw data of the census into meaningful information for the organization that solicits the data collection.
- Especially for those of us without training in data science, the idea of “the data” can be misleading, even mystifying. But the kind of data that comes in will depend on the questions asked; the value of the data depends on what the community gathering it believes needs to be accomplished through data gathering and analysis. It is worth repeating that the professionalization and regularization of the census has been a major step forward for CSWMG and the SCS in the past five years and brings the SCS into line with other major professional academic bodies. It is now worth thinking harder about what we want to know and how to get at that information. Given the complexity of data science, it is important that the SCS be able to consult more extensively on the design of the survey with the professionals undertaking it in view of the aims identified for data collection. Dialogue with other professional academic organizations would be of great help here.
- In the process of reviewing the census data, the committee became aware of other data streams within the SCS—in this case, information collected by the Placement Service on candidates. Such data used to be analyzed by CSWMG but the last report was conducted in 2008/2009, and it is clear that it required some expertise in data processing. It is not clear what happened to the data being collected by the placement service after 2009.