Friday, October 12, 2018

Statistics, we still have a problem and we need your help

One year ago, the #MeToo movement to increase awareness about sexual harassment and sexual assault began. The field of Statistics was no exception. Eleven months ago, the American Statistical Association board of directors approved the formation of a Task Force on Sexual Harassment and Assault. Ten months ago, Kristian Lum published a blog post titled Statistics, we have a problem where she called for our community to stop tolerating a culture of harassment with the hope that her story will help other women come forward who have been affected by sexual harassment or assault (and she’s not the only one). The day after I read Kristian’s blog post, I decided to organize a panel session at the largest statistics meeting (annual) in North America (Joint Statistical Meetings) on Addressing Sexual Misconduct in Statistics, which took place in Vancouver, Canada in August 2018.

A common theme that was echoed by both the panelists and audience was that it would be great if we could identify effective strategies to promote an inclusive, equitable culture, free of gender bias and sexual harassment. While we, as a community up until now, have primarily focused on the negative, downstream effects of what happens in a culture that does not value women, I would argue that what we should be focused on is how to identify, discuss, and encourage strategies that promote a positive culture. 

Why this is extremely important? 

Last week, the New England Journal of Medicine published an article titled Men’s Fear of Mentoring in the #MeToo Era — What’s at Stake for Academic Medicine? The article stated: 
“In response [to the #MeToo movement], some men in positions of power now say they are afraid to participate in mentoring relationships with women. In a study focused on engaging men in gender-equity initiatives, 74% of male senior business managers cited fear as a barrier to men’s support for gender equity. A 2018 survey of nearly 3000 employed U.S. adults found that some men have stopped meeting alone with women, and others will not meet with women they do not know well or who are considered to be their subordinates. Men say they fear false allegations of sexual misconduct that could compromise their reputations and end their careers, even if they were found to be innocent.
This has serious repercussions and consequences for women who are looking to advance their careers:
Being denied mentorship relationships deprives young women of career-enhancing experiences during critical periods of their professional development… In medicine, where female leaders are few — women represent nearly half of medical school graduates yet only 16% of deans — denying women access to mentoring relationships will perpetuate this gender gap.”
The field of academic statistics is similar. I have heard first-hand from many of my male peers that they too worry about false allegations of sexual misconduct, to the point that they have now altered the way they interact with their female students (e.g. no longer taking meetings outside of the office, such as meeting for coffee), but continue to provide these opportunities to their male students. My concern is not with the idea of meeting outside the office (that is a healthy discussion to have and by no means a straightforward discussion, for example, it can vary across cultural and religious norms), but rather the concern is with the discrepancy in mentorship between female and male students. 

This is extremely distressing to me because I wholly understand the loss of mentorship to women in the statistics community that will now happen if we don’t start discussing effective strategies to promote a positive culture and training on how implement those strategies. As a student and trainee, I was fortunate enough to have wonderful mentors (both male and female). I saw first hand that many, insightful conversations about career development often happens outside of the office (e.g. meeting a coffee shop or a conference). Had my mentors thought similarly to the 74% male managers above, I’m not sure I would have made it through the 14 years of post-high school education and training to get where I am today. We need young, female students and trainees to have access to good mentors (male or female) and male mentors to feel like they can successfully mentor women without feeling threatened of false accusations. While my male peers are sympathetic and would like to encourage further progress, I frequently hear it can be difficult to figure out where to start. 

What can you do to help? 

My call to action is that we need to identify effective strategies for promoting an inclusive and equitable culture for women and provide education and training to both men and women on how to successfully implement those strategies. This includes senior leadership in our community dedicating protected time for these discussions and training in our own work environments and at professional events.

This August, I submitted an invited panel session proposal to JSM for 2019 to “highlight innovative efforts by statisticians who have actively sought to positively change culture in their work environment and through local, national, and international platforms in the fields of statistics and data science. The panelists’ major objective is to discuss specific strategies on how to make a positive culture in our community. The ultimate goal for this session is that audience members will be able to implement strategies described by the panelists in their own work environment to change the culture in a tangible way.” The panelists stemmed from academia, government, non-profit and industry:
  • Wendy Martinez -- ASA President-Elect in 2019 and Director of the Mathematical Statistics Research Center at the Bureau of Labor Statistics
  • Emma Benn -- Assistant Professor of Biostatistics and Director of Academic Programs at the Icahn School of Medicine at Mount Sinai, and a member of the ASA Task Force on Sexual Harassment and Sexual Assault.
  • Debashis Ghosh -- Professor and Chair of the Department of Biostatistics and Informatics at the Colorado School of Public Health
  • Karthik Ram -- Senior Scientist at the Berkeley Institute for Data Science at UC Berkeley and co-founder of rOpenSci, a non-profit initiative to make scientific data retrieval reproducible
  • Jennifer Hecht -- Vice-President People Operations at RStudio with extensive experience in human resources in industry 
  • Gabriela de Queiroz -- Sr. Developer Advocate at IBM and Founder of R-Ladies, a non-profit, international organization to increase the gender diversity in the R community
While the proposal was not selected, I plan to submit it as a topic-contributed session. In the meantime, I decided it was important to start discussing this now (hence this blog post). 

Yes, Statistics, we still have a problem, but we will soon have an even bigger problem if we lose young, incredibly talented, female students and trainees because of discrepancies in training or completely insufficient mentorship. 

Questions, comments? Feel free to reach out to me by email or twitter.

Friday, August 17, 2018

How to successfully submit a conference session proposal

Conferences are a wonderful place to learn about exciting research happening in your field, to meet new people and/or potential collaborators and to catch up with old friends. Recently, I attended the annual Joint Statistical Meetings (JSM) in Vancouver, Canada, which is the largest conference for statisticians in North America. One of the features of this conference is the program committee invites anyone to submit a proposal for an Invited Session. In terms of JSM, the due date is almost a year in advance of the conference. For example, Invited Session proposals for JSM 2019 (July 27-Aug 1, 2019) are due September 6, 2018. There are also other types of sessions called Topic-contributed Sessions and Contributed Sessions with due dates a bit later in the year. The problem is because this conference is so large, it requires a long time to read through the proposals and to organize all the sessions.

In previous years, I helped organize and participated in an invited session at JSM. This year, I submitted a proposal for a topic-contributed session, but it was elevated to a late-breaking session. Most recently, I organized an Invited Session, which was just accepted, for the 2019 Eastern North American Region (ENAR) International Biometric Society conference. Finally, I am on the program committee for the 2019 Symposium on Data Science and Statistics (SDSS).

Given my recent experiences organizing sessions for conferences and the JSM 2019 Invited Session proposals are due in a few weeks, I thought I would it would be relevant to write a blog post on strategies I use when putting together a proposal.

1. Come up with a interesting, timely and relevant topic. You might bounce ideas off of your colleagues to see if the topic would have wide enough interest and might be of interest or be relevant to a particular conference. This is often the most difficult part of organizing a session. If your topic is not really of interest to a wide enough audience, it is highly unlikely that it will be selected. However, you want it to be focused enough that you can reasonably talk about the topic in 1-2 hours.

2. Create a title and abstract for your session proposal. If you have thought carefully about the topic, this should be an easier step. The title should be succinct and representative of what you want your session to be about. The abstract should contain (1) why this topic is important and relevant, (2) the focus and goal of the session, (3) what the speakers will discuss. Ask a colleague to review the title and abstract to give you feedback.

3. Decide on speakers and send out invitation emails. First, think carefully about who is in your audience and who you want to invite. Some things to think about when coming up a list of potential speakers: their backgrounds, their expertise, their perspective, their ability to give good presentations and the diversity of the speakers. The last one is most commonly overlooked, but can bring such rich and valuable discussions if you have a diverse set of speakers.

Here is a suggested format to send the invitation emails:

Hi ____, 

I am organizing a <add name of session> session for the <add name of conference> conference in <add location> next year. The session will be focused on <add topic of session>. 

My goal with this session is for <add goal of session>. Instead of focusing on <a previously discussed topic>, I want to focus on <a new topic>. My hope for the session is that audience members will be able to <add what you want the audience to get out of the session>. 

As the <add the person's title, etc>, I would like to invite you to speak in the session <(or) join as a panel member to share your insight and perspectives (if a panel)>. Your expertise in <all the reasons why this person would be a good speaker> would be highly valuable and a great contribution to the session. 

I hope you will join the session if you plan to attend and aren’t otherwise committed. Could you let me know by <fill in the date> if you would be willing to speak? I plan to include 3-4 speakers <(or) panel members> and welcome suggestions for additional speakers. 

I am happy to answer any other questions that you may have. 

All the best, 
<add your name here>
<add your affiliation here>

Some will say yes, some will say no. If needed, send out more email requests. The main things are to explain (1) the details of the conference and session, (2) the focus and goal of the session, (3) why you are inviting them or why you think they would be a good contribution to the session, (4) and the date you need for them to respond to you by.

4. Submit a proposal to the conference by the due date. This usually includes at minium a title and an abstract. You also want to include the name of all the speakers who have agreed to participate in your session, their affiliations (departments / institutions / company name, etc), and usually an email address. This helps the conference organizers get a better idea of what will be discussed. This last one is often required, so check out what is needed for your specific conference.

5. Wait for a response from the conference organizers. This can be a quick or very long process, depending on how large the conference is. Typically the larger the conference, the longer it takes to go through all the proposals. JSM's Invited Session proposals are notoriously long:

And my favorite response from the amazing Shannon Ellis:

If your proposal is accepted, CONGRATULATIONS! Send an email to all the speakers to share the good news. They will need to register and may need to submit individual abstracts for each of their respective talks. Remind the speakers of any upcoming deadlines. If not, consider trying again next year!

Most importantly, if you are a student or postdoc, organizing a conference session can be a fantastic way to meet people in your field! You have the opportunity to craft a session on a topic that excites you the most and chair that session. Conferences need fresh perspectives and new ideas, and I find students and postdocs actively working on a research topic have some of the most insightful suggestions.

Thursday, June 21, 2018

Using Slack for Academic Departmental Communication

This is a joint blog post joint between Leonardo Collado-Torres and Stephanie Hicks. It has been cross-posted on Leo's blog too! We want to share with you our experience using Slack and why you should join us. This post is in an interview style.
What is Slack?

[SH] Slack is a communication tool for teams. The main idea is you have individual chat rooms (referred to as channels that always begin with the # symbol), which are organized by topics. Traditionally, if an email is sent to everyone on your team, each person must decide how tag or organize emails in their account. In contrast, Slack provides the structure and organization for all users on the team. Each user clicks on a specific topic, or channel, and all the messages related to that topic are already there, thereby reducing the organizational burden. In a slack team used by academic department, topics can be anything from announcements of #conferences, #workshops, #jobs, #seminars, #working_groups, #good_reads, #food_alerts, #payroll, and #IT_help. You can write public messages in the individual channels or send private direct messages to anyone on your team. Everything in Slack is searchable (e.g. files, conversations, people). Finally, Slack includes a lot of functionality and integration with other useful tools such as Google Drive.


[LCT] Slack is a website that provides access to a group of message channels bundled together in what they call a workspace. These channels can have specific purposes and different sets of members. Slack also provides many integrations with other software that we commonly use, like integration with GitHub, todo lists, Google Calendar, Google Docs, etc. The idea of using Slack is to keep your communications with close colleagues organized and make them more fluid than email. You might use the general channel for making broad announcements, such as an interesting talk from a guest. Then switch to a private channel with your 4 colleagues for that new secret project you are working on and discuss some ideas that would be useful to explore in more detail. And finally, switch to a one-on-one channel with your advisor and ask if you can meet for a few minutes that later today. Like many websites, Slack is also available as its own desktop and phone applications.


How do you use Slack?

[SH] While I communicate with colleagues everyday on Slack, I think it’s important to distinguish between Slack and email. Both are important, I just use them for different purposes. I view communication on Slack to be for instant, real-time messaging and discussion to increase collaboration. These are usually short messages compassing a single idea. In contrast, I view email to be a slower, more deliberate medium. There is less expectation for immediate response. In both cases, you can always turn them off completely and enjoy some peace too!

versus

[LCT] I’m part of multiple Slack workplaces (collections of channels and users) but I mostly use one with my genomics collaborators. There, we have a channel per project, some of which can be silent for weeks/months. But that’s ok, because I can always jump back to them and revise what we were talking about. We have a channel for R questions, one for the team that I’m a part of (Andrew Jaffe’s lab), one for organizing when and where to eat lunch, another one with the recount2 team, a diversity channel, one for our computing cluster JHPCE, among others. I also frequently ask and get asked questions on one-on-one channels with Andrew’s lab members and people elsewhere from our genomics workplace. I love using Twitter for networking in academia and keeping myself updated. So I also tend to share relevant news to specific channels I’m a part of. For the project specific channels I also use the integration with GitHub to keep everything in one place and reduce the chance of git merge issues :P


What do you like about Slack?

[SH] I recently transitioned from being a postdoctoral research fellow to an assistant professor. Many people warned me about the increase in the number of emails that I would receive. While that is certainly true, I would attribute Slack to being the biggest mitigator in reducing the number of emails that I have received. It reduces my email burden and increases communication between collaborators and colleagues. When an email sometimes seems too formal, Slack is perfect for sending out quick messages.

[LCT] It greatly reduces the amount of emails and email chains. Also, I like having the option of more communication with all my colleagues than would be typically allowed via email. Like no one would like an email about some interesting R package tweet or emails about lunch. Also, it was very useful this past year when I advised two students. I could get them involved in different projects, get them observe how we do some things, but also provide them a space to ask me questions during the week outside of our weekly meeting slot. Ultimately, I like to be organized and Slack helps me stay organized.


Any more tips?

[SH] I have found that the more I engage and write messages in channels on a Slack team, the more I get in return. As conversations flow in a given channel on a particular topic, I may or may not have anything useful to contribute. If it’s the former, I try to pass along links, advice, knowledge or whatever is appropriate. If it’s the latter, I still gain knowledge by being able to read other people’s responses and learning how others would have approached or solved a problem. Either way it’s a win-win!

[LCT] Regardless of what channel you use, it’s important to keep your messages polite. Basically, following a code of conduct similar to this one. This is not the anonymous internet, here you are interacting with your colleagues and you have to be aware of biases you might have, otherwise you might perpetrate some problems like sexism. Remember, you share the responsibility of making sure that your Slack workspace is a welcoming environment where anyone is allowed to ask questions. I also think that it’s very important to keep your boundaries clear between work and home. I highly recommend going to your preferences and automatically disabling notifications outside work hours as shown below. That’s something you can’t do with Google Chat: yes, I might be available after work hours, but I want to choose whether to respond or not and I don’t want others to expect an immediate answer if they see me online.
If you want more, check this great presentation about Slack by Stephanie Hicks.

Wednesday, June 6, 2018

Addressing Sexual Misconduct in Statistics

The cultural revolution from the recent #metoo movement has demonstrated that sexual harassment and sexual assault are far too common and often go unreported in professional settings, including in the field of Statistics. In response to this, I have organized a late breaking session at the Joint Statistical Meetings on Addressing Sexual Misconduct in Statistics on Monday July 30, 2018 2-3:50pm in Vancouver, Canada.

The focus of this session is to bring together seven panelists to discuss how sexual misconduct can negatively impact the careers of the victims and how we as a community can make positive changes. Our goal is to open the dialogue on recognizing and condemning predatory sexual behavior, and provide support and inclusion to all members of the statistics community.

This session will be a positive discussion to address this topic and will offer perspectives from conference participants, elected leaders of professional societies, academic journal editors, academic department chairs, and program committee members. In addition, this session will also include the perspective of individuals who have been directly impacted by sexual harassment or assault, which is particularly relevant to other individuals in the audience who have also been impacted by sexual harassment or assault, but may not have discussed it publicly. Keegan Korthauer will chair the session and our panelists will include:

If you are planning on attending #JSM2018, I invite you to join us.  

Friday, February 9, 2018

Lessons learned from applying for scholarships

As a graduate student (many moons ago), I applied for the Gertrude M. Cox Scholarship from the American Statistical Association. The purpose is "to encourage more women to enter statistically oriented professions". I was a woman pursing a PhD in Statistics and wanted to enter a statistically oriented profession, so I wrote an essay and submitted an application. In the essay, I explained how I got interested in statistics and why I decided to pursue a PhD. At the time, I did not have many examples of "acts of leadership", but I had done some community service and/or mentoring as an undergraduate student. Afterwards, I was sad to find out I was not selected. However, forcing myself to go through that process taught me a few general lessons when applying for scholarships (and more general things e.g. grants, papers) that have stuck with me over the years. I figured it might be worth sharing them here.

  1. As a student, it is easy to feel like your resume/CV is sparse or you do not have much to write about. You may have started a graduate program right after completing your undergraduate degree without any work experience. You may still be taking a lot of classes or not have much teaching experience. You may be early in your graduate program without any research experience or peer-reviewed publications. What is most important is your ability to communicate why you are interested in your area, why it is exciting, and why others should be excited about it. If you have done work or made contributions in the area, great. Write about it. However, as someone who now reviews applications for a full-tuition scholarship for women interested in STEM funded by Cards Against Humanity I believe it is much more important to say why you are passionate by whatever you are interested in, irrespective of how much you may or may not have accomplished. 
  2. Recommendation letters matter. You want to ask individuals who really know you. If you ask a professor who only knows that you made an A in their class and can't say much more, that comes across loud and clear in the recommendation letter. You want to ask individuals who will actively advocate for you. 
  3. Rejection is part of life and do not it deter you from pursing your passions or dreams. I will be honest: I am still struggling with this one even to this day. However, it has gotten easier over the years because I have learned to not take it personally. But seriously, some days I want to scream at the top of my lungs 'WHY DIDN'T YOU PICK ME? WHY DIDN'T YOU SELECT MY GRANT? WHY DIDN'T YOU ACCEPT MY PAPER?', etc. During those moments, I just remind myself I only have control over the next award I apply for, the next paper I submit, the next grant I write. 
Anyways, if you are woman in a full-time graduate Statistics program (MS or PhD) and a citizen or permanent resident of the US or Canada, I would encourage you to apply for the Gertrude M. Cox Scholarship. Applications are due two weeks from today (February 23, 2018). For what it's worth, I posted my (unsuccessful) application essay for the scholarship. Maybe it will help someone else write their essay! 


Saturday, October 14, 2017

Rainbow Chocolate Cake



My husband's birthday was two months ago and we really didn't get a chance to celebrate it at the time. As we were walking through the grocery store today, I asked him if he would like to make a birthday cake / dessert this weekend to properly celebrate it! I left him in the baking aisle to ponder all the possibilities. A few minutes later, he rounded the corner with a box of Rainbow "Tie-Dye" cake mix and chocolate funfetti icing!


When we got home, I started reading and following the instructions. The beginning was pretty straight forward. Mix cake mix, water, oil and, interestingly, only egg whites (no yolks). I can only assume that is to not tint the colors? Not sure though. Then, I split the cake batter into six bowls with approximately 2/3 cup cake batter in each bowl.

The box came with three packets of dye (yellow, red, and blue) and instructions on how many drops of each color to fill the six bowls of cake mix. I really liked the brightness of the red ad blue colors.


I don't have two 8" round cake pans, but the instructions said a greased 9x13 pan would work. For this pan, I poured all the cake batter from each individual bowl into the 9x13 pan in the following order: red, orange, yellow, green, blue, and finally purple. The trick is to pour the next color directly in the middle of where the last color was poured. Each layer will slowly spread out as you add an additional layer. The last step is to lightly jiggle the pan to even out any cake batter that may have been uneven. This was the cake after I jiggled it a bit.


I baked the cake in a 350F oven for 32 minutes and tested with a knife to see if it was done. Looked good to me! Then I let it cool an hour before trying to decorate it with icing.


After letting it cool, I removed it from the pan and covered it in chocolate funfetti icing! We sang Happy (belated) Birthday to him and enjoyed it after our dinner tonight. Hopefully, this post will inspire a little color creativity in your next baking adventure!


Wednesday, February 8, 2017

Sampling Points Proportional to Intensity in an Image

A few months ago Aaron Lun shared some code with me to sample points based on intensity of pixels in an image. With his permission, I thought it would be fun to share it here!


The idea is you start with some image (either in a JPEG or PNG format). Here I'm using Rafa's twitter profile picture in a JPEG format.


I'm using the readJPEG() function from the R jpeg package to read in the JPEG image. Alternatively, you can use the readPNG() function from the R png package to read in a PNG image. If you read the help file of readJPEG(), it says it "reads an image from a JPEG file/content into a raster array". OK, first off what do we mean by raster? You can think of a raster image as a rectangular grid of pixels. In the help file for the readJPEG() function, it states "most common files decompress into RGB channels (3 channels)" where RGB stands for red, green and blue.

So, this means for each pixel in the original image, we should expect 3 values (one for red, green and blue). These values range between 0 and 1 and can be thought of the intensity of each color in the image. Below, I'm showing the intensity values from the three channels colored on the same scale (top row) and colored in the red, green and blue scales (bottom row).

To get the intensity of how light a particular pixel is, we can just average the values across the three channels. Next, we flip the intensity scale by subtracting the intensity scale from 1 to see how dark/black a particular pixel is. This makes the previously dark areas now light and vice versa (top left figure below). To control the contrast of intensity, you can just take powers of the intensity values. The lighter the background, the better the contrast will be between the individual and background on the original scale. Depending on the background in the original image, the skin tone, etc you may need to tinker with the contrast. I've shown the original scale and powers of 2, 4, and 8 below.

Next, we sample points proportional to the (powered) intensity level, add a bit of random noise and then plot the sample points along the x and y axis. In the picture below, you can really see how the power is very important to control the contrast levels.


Happy sampling!