Daniela Witten writes:

This Bulletin issue announces the Class of 2023 IMS Fellows. First and foremost: congratulations to each and every new Fellow! This is an incredible career accomplishment. You make me optimistic about the future of our field.

However, today I’d like to discuss the fact that only three of the 27 new Fellows are women, within the context of a broader conversation about gender among IMS members. These two topics are inextricably linked, because: (i) to be eligible for IMS fellowship, a candidate must have been a member of IMS for the past two years; and (ii) IMS members comprise the set of active participants in the IMS community, and thus the set under consideration for IMS fellowship.

Over the years, I have talked with a number of accomplished statisticians about some of the ideas in this column. Almost without exception, they have been open to what I have to say (and, in many cases, in enthusiastic agreement). So, if you already agree with the points I make in this column, then I’m so glad to have you on board. If you do not currently agree with me on these issues, but are curious about my perspective, then thank you for reading my column, and I hope to give you some food for thought!

I also want to emphasize that the statisticians who I know individually (and almost certainly those who I do not) are wonderful people. Nobody “wants” IMS to have a gender problem. However, there are systemic issues at play that go far beyond individual action or intent. The goal of this column is to point out some of these, and to start a conversation about how IMS can address them.


Q1: Why should I care about the representation of women within IMS?

A1: There are a number of reasons. Here are two:

(i) Creating an inclusive professional society is the “right” thing to do. I believe that women are no less talented (or interested) in statistics/probability than men; however, they often face societal/cultural barriers. By breaking down these barriers, we can enable women to have fair access to the fulfilling career opportunities available in statistics/probability.

(ii) A substantial body of research shows that diverse teams lead to better results. Thus, the field of statistics/probability — and the quality of the field’s research — will directly benefit from greater diversity.

Why does this column focus exclusively on women, as opposed to other marginalized genders, or other types of diversity unrelated to gender? See Q11.


Q2: Before we talk about IMS awards, we should talk about IMS membership. Does the IMS membership have a gender problem?

A2: According to the 2021 IMS Membership Survey, the overall membership of IMS is around 20% women.

What “should” the number be? We can try to triangulate it. In the US in 2014, according to the 2017 NSF Women, Minorities, and Persons with Disabilities in Science and Engineering Digest, 42% of bachelor’s degree recipients in mathematics or statistics were women. For master’s degrees, the number was 43%. In the US in 2021, 38% of doctoral degrees in statistics were awarded to women. This number was well above 40% (and as high as 45%!) in 2005–2011, the period when many of this year’s IMS fellows completed their PhDs.

The American Statistical Association (ASA), which has worked hard to increase diversity in statistics for a number of years, has a far greater proportion of women Fellows than does IMS: approximately 35% of ASA members are women (see https://magazine.amstat.org/blog/2020/10/01/asa-fellows-analysis/ and https://magazine.amstat.org/blog/2016/02/01/genderupdate16/).

Now, you might argue that these numbers aren’t the right baseline: for instance, perhaps you believe that women are less likely than men to pursue mathematical statistics and probability, as opposed to (say) applied statistics. That might be part of the story, but it’s also a cop out. A lot of factors contribute to the fact that few women become mathematical statisticians and probabilists, and to the fact that few women join IMS. Instead of shifting the blame from one factor to another, let’s instead work together to address the entirety of factors.

I think that part of the reason that there are so few women in IMS is because, well, there are so few women in IMS. Representation matters, and it’s hard to be what you can’t see. If we want more women to become involved in IMS (and I hope we do!!) then we need to play an active role in making that a reality. Why would we expect someone to pay a $105 USD annual membership fee to support an organization of people that don’t look like them and that don’t make them feel included?

There are a number of ways to make IMS a more appealing professional organization for women; one way is by having more women award recipients. More on this later.


Q3: Won’t the demographics of IMS naturally change over time, without our intervention?

A3: Nope. Problems like this get solved through years of hard work and tireless advocacy, not by twiddling our thumbs while waiting for time to pass. The proportion of women who earned bachelor’s and master’s degrees in math or stat in the US declined substantially from 2004 to 2014, as did the proportion of women earning doctorates in statistics in the past 15 years. Among IMS members below age 30 in 2021, the proportion of women remains very low, at ~25%.


Q4: This year, only 3 of 27 IMS Fellows are women. Is this a fluke?

A4: Table 1 displays the gender breakdown of IMS awards over the past few years. The percentage of women recipients hovers around 20%. So 2023 was a particularly “bad” year in terms of women receiving IMS fellowships; but the “typical” percentage is still quite low. As a point of comparison, typically around 30% of new ASA fellows are women; this number was 48% in 2021.

Table 1. Major IMS awards with more than four recipients to date, as well as total number of recipients, and number of women recipients.


Total Recipients

# Women

% Women

Larry Brown PhD Student Award




Tweedie New Researcher Award




Wald Lecture [since 2010]




Blackwell Lecture




Le Cam Lecture




Neyman Lecture [since 2000]




Rietz Lecture [since 2000]




Harry Carver Medal




Medallion Lecture




IMS Fellow [since 2007]




[Many thanks to IMS staff for assembling this data; however, I take full responsibility for any errors due to failure to double-check.]


Q5: So, circling back: given that 20% of IMS members are women, is it actually a problem that only 20% of IMS award recipients are women?

A5: Yes, it is. Representation matters, and it’s hard to be what you can’t see. If we want more women in the field, then we should make sure women are represented in IMS: awards are a good place to start.

Furthermore, the total number of women members in IMS is smaller than it should be (see Q2). The women who do choose to join IMS (as opposed to self-selecting out of IMS) are particularly qualified for IMS awards, and thus I believe that they should win awards at a higher rate.


Q6: I think the IMS award process is objective and unbiased.
Are you saying it’s not?!

A6: I think that everyone’s intentions are good! However, there is extensive evidence that, across fields, “the scientific efforts and achievements of women do not receive the same recognition as do those of men”. This phenomenon is known as the “Matilda Effect”. It means that despite our best efforts at objectivity, in reality we are not that objective.

Furthermore, many women face gender-specific obstacles to careers in statistics and probability (such as cultural/societal expectations, subtle and overt sexism, difficulty finding senior faculty mentors/sponsors, etc.). And so I believe that the “average” woman who has achieved a given seemingly “objective” marker of success (e.g., publication in a top journal) has, in fact, accomplished more than the “average” man who has achieved this same marker. [Obviously, there are (many!) exceptions to any rule, and furthermore these comments absolutely apply to other groups besides women (see Q11).] So, if we want IMS awards to go to the most meritorious people, then we should give awards to more women.

To make my meaning crystal clear: I believe that women deserve to win more than 20% of IMS awards, and I hope that women will win more than 20% of IMS awards in 2024 and beyond.


Q7: Don’t look at me!! I always do my best to support women!!

A7: Yes! You do! And I really love that about you! You invite women to give talks, write glowing tenure letters for them, treat them extremely fairly whenever you serve on a hiring committee or award committee, etc etc. On behalf of all of the women: thank you!

But, the problems that I’m describing are systemic. Again, see my answers Q2 and Q6. This is not about anyone having ill intent (again, I think almost all of us operate with the best of intent); it’s about a well-documented sociological phenomenon.


Q8: I invited five women to speak at event XYZ, and they all said “no”. What gives??!

A8: First of all, thank you for being an ally! I can see why this is frustrating. Your experience is not unique, I’m afraid. I think of this as the “collective Rolodex” issue.

Here’s a little game: think of the “top” 10 women in statistics/probability (whatever the word “top” means to you). Then ask a colleague to do the same. I bet that there’s a lot of overlap between your two lists!

We have a tendency to always think of the same very small set of women when issuing invitations to give talks, apply for jobs, serve on advisory panels, etc. Then we feel frustrated when those women decline (because they have already received countless such invitations!).

The solution is not to stop trying: it’s to dig a bit deeper into our collective Rolodex. More on this in my answer to Q9 below.


Q9: So, what can we do to fix this?

A9: Thank you for asking! Of course, our individual actions are important. But we also need a top-down approach. If the IMS leadership views (a lack of) gender diversity as a problem, then it can work to solve it. Here’s one concrete suggestion.

For most (if not all) IMS awards, nominations are submitted to an award committee, which then selects among the nominees. If the list of nominees does not contain (enough) women, then women will not win the award.

So my suggestion is as follows: IMS should encourage (or even require) award committees to ensure—before review of nominations—that the nomination pool contains 50% women. This might require award committee members to actively solicit nominations of women: this is okay, and in fact this is often how award committees work already! (I have multiple times served on award committees where the committee members sought out additional nominations.) Then, once a diverse pool of nominees has been obtained, the committee can do its work, without explicit regard to gender. This will guarantee that the most deserving candidates win the award — while also ensuring that women are given a fair shake.

(In fact, the system that I’m describing is already in use at many universities in the US: often universities will not allow a faculty search to proceed without a sufficient level of diversity on the “long list” or “short list” of candidates. Once a diverse slate of candidates has been selected, the search proceeds without explicit regard to diversity considerations.)

I emphasize that this suggestion will not reduce the quality of IMS award recipients, nor will it change the rubric by which nominees are evaluated: it will simply increase the pool of nominees to include more women.

This suggestion will also help address the collective Rolodex issue described in my answer to Q8.


Q10: Can you close this column with a touching anecdote about the late Ingram Olkin?

A10: I sure can! During a lovely dinner in around 2012, Ingram shared an interesting perspective on why qualified women often are not hired into faculty positions. He said that a typical faculty search might shortlist six candidates, including one woman. However, when the committee meets to make a final decision on hiring, the discussion typically proceeds like this:

Committee member 1: The woman is the best at X, but she is not as good at A as man #1.

Committee member 2: Yes, she is the best at X, but she is not as good at B as man #2.

Committee member 3: I agree. Also, she is not as good at C as man #3.

Committee member 4: Also, man #4 is better at D and man #5 is better at E.

Committee member 1: OK, so it sounds like we shouldn’t hire her, because while she is the best at X, she is not the best at A, B, C, D, or E.

What happened here? Each individual committee member’s statements were true, and well-intentioned: but the woman was compared to an imaginary man who combined the best qualities of five real men.

How can we fix this problem? By having more women on the short list (see Q9). This will make it easier to avoid the (very natural) tendency to make a “man-versus-woman” decision before deciding who to hire.

It meant a lot to me to hear Ingram—who must have been almost 90 at the time—articulate these points so clearly.


Q11: We’re almost done here, but there’s time for one final question. Why does this column discuss only gender?

A11: Great question. There absolutely are crucially important conversations to be had about other marginalized groups within IMS (e.g. race, ethnicity, other genders, sexual orientation). However, as far as I know, IMS does not collect the relevant data: for instance, its annual membership survey does not ask about race. I understand that the situation is delicate: there are constraints on the questions that IMS, as an international organization, can ask.

Nonetheless, I believe that we cannot improve that which we cannot measure. In the future, I hope that the IMS membership survey will include questions on race, and other demographics of interest, in countries where this is allowed. Perhaps the exact set of questions to be asked should vary by geographic region. Of course, all country-specific laws should be followed. (And, as is the case for other questions in the survey, members should be given the option not to answer.)


I look forward to hearing how IMS will address these important issues, and I hope that they will put together a plan in time for the 2024 award cycle.