FEM inc Research and Why It Matters | Rachel Payne & Natash Mohanty

So when we were starting out— this is really early, by the way. What I'm going to show you is over a year old.

Yes.

But I want to show you this because it's part of that process. So when we were starting out, we asked ourselves, okay, what is the problem we're trying to solve? Because this is huge. When you think about systemic issues like the representation of certain segments of the population in popular entertainment, that's like massive, right? And is there even a problem? And if there were a problem, what would you do about it? And how would you actually solve that? So I'm taking you in sort of a time warp back to a period where we were trying to quantify the problem and really think about what, what is that angle that we might want to pursue to really have an impact, build an amazing product, something that the market wants and that we can make profitable? Those are a lot of questions to answer. So one of the things we did is really look at data.

Yeah, I mean, I think, I don't know, I'm guessing all of you are also big data fans, right? So we started with quantifying the problem and collecting data, doing a lot of our own analytics. There's a lot of research out there in whatever field you do. The internet's big, right? So there's research on everything, and there'll be a lot of stuff out there. And what we wanted to make sure is we kind of supplemented all of that with analytics of our own, research of our own, so that we really, really understood the problem in depth. Especially as we're trying to understand our market and our positioning in the market, all the analytics legwork we did early on, which was painful in parts, was worth every penny. Because now whenever we go to meetings, whenever we're presenting, we're always armed with all the analytics, all the data, a lot of research of our own, so we can answer pretty much any question in the space.

And it actually opened up the prospect of us helping redefine metrics in an industry, which is We just had a meeting this morning with a major media conglomerate who is very interested in redefining engagement. And so when we've been at the forefront of not just articulating and understanding current patterns, but also collaborating with major players to think about how might you redefine that next set of metrics, it's interesting because it allows for It allows for a lot of collaboration and innovation in a space that isn't always, I guess, receptive to change. So what we did is— this is, again, it's early research— but we were asking ourselves the question of, OK, let's see, what is the impact that media has on some of the trends that we're seeing today? And there was a big study that was done by the Geena Davis Institute on gender and media. who, by the way, is now an advisor of ours. And they did this research around how leadership positions are represented from a gender standpoint. So when you look at primetime television, it actually comes pretty close to reality. It's not too far off. But what was interesting was family films had the lowest representation of women in any form of leadership position. And this is interesting because it's actually precisely at those ages when girls are most susceptible to messages from media around the appropriateness of certain roles and their future in those roles. So that was an aha moment. I'm like, oh, maybe unwittingly there is a lack of representation in women in leadership roles in family films. And maybe with better data we can actually help change that to at least— Catch up with reality. Another question we asked ourselves was, well, why would this be? Why would there be this pattern of representation that's so outdated? I mean, we're talking 20, 25 years out of date, okay? Well, we did our own research and we studied 125 popular TV shows and we looked at primary and secondary characters and we quantified the results on different dimensions and one of the dimensions was How is passion about your work shown from a gender perspective in TV? So I'm going to ask the question, pretend you don't see this. How many people in this room are passionate about what they do? Right, because if you weren't, why would you be here? That's 100%, by the way. And it's fully gender equal. Okay? So this discrepancy didn't make sense to me, and it didn't make sense to a lot of people that I know who are accomplished and passionate, and it could be any kind of work. But the reality is there was a big difference between the passion that was shown versus the day-to-day reality that we live in modern times. And also to that end, more likely men are shown as leaders by a lot.

By a lot, yeah.

So this seems to me be a way of perpetuating outdated leadership models and styles as well. So we wrote a paper about that and it's available online. It's free. So we publish what we learn, by the way, and we make that available. But what we found is that the leadership positions could be creating this sort of self-perpetuating cycle where if messages are being sent at really young ages, then decisions are made at younger ages about whether or not to be a leader. And so then the cycle continues. Television characters are also much whiter and more male than the US, and I think that's probably not a surprise, but it's just data. We're just looking at every— we're slicing this from a lot of perspectives. And female television characters are often quite young, much younger than their male counterparts. And if you read some of the other studies that we didn't do, but for example, the Geena Davis Institute has done They're also wearing less clothing generally, and especially animated characters for kids. Don't ask me why. Okay, and then women are shown to be much more sexual than men. That's often the case in a lot of the TV shows, which surprised me, but not really.

Excuse me. Yes.

I was wondering how you define sexual in the previous slide. Yes, and it has to do with behaviors as well as identity. So it could be And we broke this out. We have a number of different things that we use to create a composite score. And so part of it is clothing and skin, the appearance of bare skin, that sort of thing. Part of it is behaviors, initiating behaviors versus receiving behaviors. And there are a few other factors like that. So it was both behavioral as well as appearance-oriented.

We also asked people Yeah, and we asked—

we validated all these, by the way, too, before we did the research. And we are continuing to build on this research, so I'll talk about that in a little bit. Features with significant gender differences. These are pretty stark, actually. And the one that might resonate most with you is science, technology, and engineering fields where, by a lot, men are overrepresented, and about 10x more than reality. of the females' participation. We looked at a lot of other aspects too. So we looked by race. We did a lot of work looking at other types of representations by ethnicity, age. And so, you know, it was striking, you know, and it was— there were patterns that were pretty persistent. Why does this matter? Well—

Go back a little step.

Oh, sorry.

Just to understand why we wanted to do it this way, one of the— we did a lot of surveys too because we were collecting a lot of user responses along with data analytics of our own. And one of the things we learned very early on is a lot of people choose to watch content because they identify with content, with the characters of the content. And therefore it became really important for us to really break down content by character representations. And that's why, you know, you can really break down content by many other things. And if you looked at classic content recommendation systems, they're mostly broken down by genre, and recommendations are very much, this is action and this is also action, and so we'll show it.

People don't resonate with genre, by the way. I mean, they may choose to watch a genre, right? But when you're talking about deep emotional connection— go ahead.

Yeah, so I think that's kind of why we chose this axis to do a very, like, fairly in-depth analysis, because we learned really quickly that characters were the predominant reason why people continued to watch a show or a video that they had gotten introduced to.

Well, and that insight around character identification is also some of the— it came out of some of the work that our co-founder, Meghana, did in some of the neuroimaging work around the parts of the brain that really light up when somebody is watching a character that they resonate with. Yes?

And then all this research was based on US culture and US media?

Right now, we were looking at US TV shows. That was our sample. And it was about 125 shows. And about 4 or 5 characters per show, roughly. Yeah? With people watching from the US?

So US watching?

Yeah, being narrated to the US audience. So you can imagine the differences, right? So as soon as you go out— and I think that's really important research. We just didn't have the scope and time budget to do. But it's really important to then ask that question, which is not just US content for US audiences or US content exported to other foreign audiences, but also locally produced content. For local audiences, vice versa. I think all of those are very interesting threads worth pursuing, which we just haven't done yet. How did you reach out to these people, and how did you do this research? So we can spend a lot of time on that, and I would like to, but there's a few more things I'd like to cover. So just simply what I'll say is there are ways that we crowdsourced information. We also did work with a lot of experts on the way that we framed the questions and how we defined the dimensions. and the attributes around the characters and the feature sets that we were looking at. And we did do some focus groups, but most of it was really— we used surveys and we used other types of crowdsourcing mechanisms. Yeah. This was a quick experiment, by the way. We didn't spend— this wasn't like building a product. This was just a way to validate, are we—

Basic research.

Yeah, basic research. And you're all going to do it, so I'm just giving you a flavor of So that's some of the ways that we thought about it.

Can I do a little dance though? So we did all this and then we got patents on this.

Oh yeah, and we got patents too. So we, yeah, we have several patents actually right now, and some of them are linked to the notion of character. Now let me just have you go with me on a thought experiment, right? Content recommendations to date have not really put characters at the center. What we wanted to create was a human-centric approach to content recommendations, and that's what we got a patent for.

Wow.

Right? Putting people at the center and having the attributes of what makes a human, a multidimensional human.

Awesome.

So why does this matter? Well, because negative stereotypes can actually have pretty significant consequences, particularly if these stereotypes are repeated So it's creating these patterns that are then internalized, and that internalization affects literal performance. In this study, this is a very robust study, it's actually cited in Sheryl's book Lean In, but it's a very, very robust study at Harvard University that basically primed people against different dimensions. One dimension was their race, another dimension was their gender, and then there was this sort of neutral. And this was an actual, this IQ test, so these are like—

2 groups.

real, you know, they, they're measurable. The impact is measurable. So what they did is they had people who identified with— I'll use in this example Asian because the difference is so big. So if you were an Asian female, right before you went in to take your test, they would prime you with the identification of being Asian. And separately, if you were a female, And they primed you on being female and didn't bring up your race, just said, you know, made a comment, something very subtle, by the way. These are really subtle effects, or I'm sorry, subtle primes, big effects. So when they primed the female piece, the performance was dramatically lower because we're internalizing bias and then we live that bias. And that was the findings that came out of the study. By the way, this study is Patinsky and Ambady at Harvard. I would definitely encourage you to look it up. It's an amazing study. It's very, very— it's been used a lot in research because of how direct and immediate those effects are. Yes? Is this where they put the different ethnic groups in the same room having to take the test next to each other? Yeah. Yeah, and that study had many— there were many parts to it.

Okay.

Yeah.

I don't know if it's exactly—

This just happened to be— I think that's the one. I mean, it sounds familiar, but this just happens to be one of the findings. They had several others in that study. Do you know an example of like what the subtleties would be that they would say before they would go in? Yeah, usually it's not something as overt as, you know, you're a woman, but they would ask a question, you know, like, what is your gender? Click female, and then they go take the test. It can be very subtle and indirect, and because it is subconscious processing that's happening, just priming that one element was enough to trigger all of these other internalized biases.