Thursday, October 27, 2016

Girls, Math, and Grade School Stereotype Threat

Stereotype threat, especially with regard to women and math ability, has been one of my interests for a long time (see past posts here and here). To recap, stereotype threat occurs when a stereotype about a group (e.g., "women are bad at math") affects a group members' performance (e.g., a woman encountering math test). Though some studies have tested this by specifically stating stereotype prior to the test phase, other research suggests that this isn't necessary. Simply being aware of the stereotype is enough to impact performance.

Recent research suggests this stereotype is still alive and well, and begins rearing its ugly head around 1st grade:
A new study shows that first-grade teachers consistently rate girls’ math ability below boys’ — even when they have the same achievement level and learning style. The study out today in the journal AERA Open from researchers at New York University and the University of Illinois at Urbana-Champaign seems to represent a setback for gender equity in math. A widely reported 2008 study found that girls score as well as boys do on standardized state math tests. But the latest study suggests that early in their math education, many girls run into a teacher who perceives them as being worse at the subject than they are — which could discourage some of them from heading down a path that could lead to a career in math, science or engineering.
For me, I think I became consciously aware of the gender math stereotype in middle school. And they were especially obvious to me in junior high and high school, when I took algebra courses for the first time. Believe it or not, I struggled big-time with algebra. But I excelled at geometry. Still, I started to buy into the stereotype, if not about all women, but definitely about myself. I used to be involved with science and math clubs, but around 7th or 8th grade, I stopped going because I believed I couldn't do math.

My Algebra 2 teacher definitely bought into gender stereotypes. I approached him early in the course to tell him I was having trouble. He invited me to stop by during study hall. I showed up, he gave me some problems to work on, and walked away. A couple of boys from my class showed up and he spent the whole time showing them how to do the problems on the board, correcting them when they gave a wrong answer, and just being really hands-on with them. Meanwhile, I was working through problems I had no idea how to do with no attention from him. I did this one more time before I got the hint and stopped showing up. That was my one and only C in high school. That was also the last math class I took in high school.

Now that I've been working with higher-level statistics for so long, algebra makes perfect sense to me. I wonder how different things would have been in my career choices if I'd had more help.

Lying Liars Who Lie

In the words of one of my favorite characters, Gregory House:

One lie can easily turn into two, and small lies can easily become big lies. And today I encountered some recent research that suggests why.

Whenever you tell a lie, you experience a little twinge of emotion - usually guilt. That guilt may not be enough to keep you from lying, especially if the lie benefits you and does not necessarily hurt someone else. And past research has shown that increased exposure decreases our emotional response over time. So just like your first break-up is likely to hurt a lot more than your fifth break-up (cue "The First Cut is the Deepest"), the guilt you feel from your first lie is going to be much greater than the guilt you feel after lie #793.

To test this hypothesis, and get at the specific brain response to lies, researchers had people participate in a game while undergoing functional Magnetic Resonance Imaging (fMRI):
Specifically, we adapted a two-party task used previously to elicit and measure dishonesty. Participants advised a second participant, played by a confederate, about the amount of money in a glass jar filled with pennies. We changed the incentive structure over the course of two experiments such that dishonesty about the amount of money in the jar would either benefit the participant at the expense of their partner (Self-serving–Other-harming, Experiment 1), benefit both (Self-serving–Other-serving, Experiment 1), benefit the partner at the expense of the participant (Self-harming–Other-serving, Experiment 1), benefit the participant only without affecting the partner (Self-serving, Experiment 2) or benefit the partner only without affecting the participant (Other-serving, Experiment 2). Importantly, the participants believed that their partner was not aware of this incentive structure but thought that they were working together at all times to provide the most accurate estimate, which would benefit them both equally. A baseline condition enabled us to infer the amount of dishonesty on each trial without the participant being instructed to act dishonestly or required to admit to dishonesty.
The researchers observed dishonesty escalation - that people became more dishonest over time - when it was self-serving. This was the case even when the lie hurt the other person, though people were more likely to be dishonest when served the other as well. (So they still lied if it hurt the other, but not as much as if it helped the other.) Results from the brain scan showed reduced amygdala activity over time. As I've blogged about previously, the amygdala, part of the mid-brain, is involved in emotional response.

So, to answer Liz Phair's question, "Why I Lie?" the answer is: it just keeps getting easier.

Wednesday, October 26, 2016

Polling the Polls: Meta-Analysis in Action

I blogged recently about a new course I'm taking on the topic of meta-analysis, which is a set of techniques for aggregating results from multiple studies. I realized this morning that you've probably encountered meta-analysis recently, in the form of election polls - specifically, sites that aggregate data from multiple polls. A couple examples are Talking Points Memo's Poll Tracker and FiveThirtyEight's Election Forecast.

Since we can't poll everyone in the population, we use samples as a stand-in. (Terminology side note: When we poll a sample, we call it a survey; when we poll an entire population, we call it a census.) There are techniques to use in surveying, to ensure the sample is representative, but of course, weird things can happen. We may get bias in who responds, or bias because of how a question was worded, or any number of issues. These polls of polls are great examples of the usefulness of meta-analysis. While an individual study, even when done very well, has limitations, aggregating across studies and weighting each study's contribution by its sample size allows us to uncover relationships that may be too small to detect in a single group. And as sample sizes get bigger, we can get a closer and closer estimate to the true population value. In the case of presidential elections, that population value is of course the proportion of voters who will vote a particular way.

See? You've been watching meta-analysis in action all along!

Tuesday, October 25, 2016

Everything You Always Wanted to Know About Early Voting (But Were Afraid to Ask)

Though Election Day is still a couple of weeks away, you may have already heard friends and family say they've already voted. Thanks to early voting, which is already underway in many states, people can go cast their ballot today. Christina Silva at FiveThirtyEight offered a post today that explains early voting and differences by state, as well as analyses of early voting patterns (in terms of who votes early and its potential impact on outcome):
Well-organized campaigns do have opportunities to capitalize on early voting, however, and this year that could benefit Hillary Clinton, who has a stronger ground game than Donald Trump.

It "opens up more possibilities for voting, boosting turnout in the long run," said Mark Stephenson, the CEO of Red Oak Strategic, a political consulting firm in Arlington, Virginia. "But it also gives the campaign tacticians the opportunity to analyze and see what is happening over a longer period of time and be efficient with where spending is going as a result. Both, when done successfully by either party, can provide a real tactical and strategic advantage."

The Clinton campaign uses a variety of techniques for reaching out to early voters, including door knocks, phone calls, emails and text messages, said Lily Adams, a Clinton campaign spokeswoman.

"Hillary Clinton was in Iowa on the first day of early vote in person and suggested to all of the attendees of the rally that they go vote," Adams said. Similarly, President Obama held a rally for Clinton in Ohio just before early voting began in that state.

"The DNC is dominating early voting [outreach]" in Nevada, [Jon] Ralston said. And it seems to be paying off: So far, the proportion of Nevada early voters who are Democrats is higher than the proportion of registered voters who are Democrats, which suggests Clinton’s lead in the polls there may be mirrored in the results.
And in related news, the results of Channel One News OneVote 2016 (which gets America's youth involved in the political process by having them vote in a mock election) are in: Hillary Clinton was the winner. OneVote has accurately predicted the next president since 1992 - which coincidentally is when I participated in OneVote and voted for Bill Clinton.

With a Little Help From His Friends

Thanks to research, Christopher Marlowe will now be listed as an official coauthor on three of William Shakespeare's plays, specifically the three dealing with Henry VI:
The new Oxford edition, which will be available in November, was edited by four Shakespeare scholars: Gary Taylor of Florida State University, John Jowett of the University of Birmingham, Terri Bourus of Indiana University and Gabriel Egan of De Montfort University.

Taylor tells NPR the conclusion that Marlowe should be credited as co-author is partly based on a combination of new and old research. In particular, Taylor cited 2009 research by Hugh Craig and Arthur Kinney that analyzed vocabulary from the Henry VI plays and compared it to plays known to have been written by Marlowe, and a 2015 article by John Nance analyzing the prose of Part 2 of Henry VI.

Taylor himself has published scholarly work on Marlowe and Shakespeare, including work from last year titled Imitation or Collaboration? Marlowe and the Early Shakespeare Canon, on which he collaborated with Nance.
This research is a great demonstration of the power of computational linguistics. It's similar to some of the past research I've blogged about, such as this study using sentiment analysis of Trump's tweets. In fact, see this recent post on Natural Language Processing for a more in-depth look.

To put it very simply, every writer, no matter how they were taught or who influenced them, has their own unique style. Some of the unique characteristics are very subtle - simple word choice or sentence length. Others are more noticeable. And of course, writers from the same period and genre will be more similar to each other. So based on the fact that both playwrights were active at the same time in the same location, and dealing with similar content, we would expect Shakespeare's writing to be more similar to Marlowe than, say, Isaac Asimov.

But even after taking that into account, there will be some slight differences. So if the writing of Shakespeare is more similar to Marlowe than we would expect based on their similarity in approach and paradigm, we can start to speculate that the two may have been collaborators, rather than simply rivals. This is hinted at in the film Shakespeare in Love, where Shakespeare discusses his writer's block with Marlowe, and gets some advice and direction for the play that would become Romeo and Juliet. This short scene isn't surprising considering that people have been speculating about Marlowe's role in Shakespeare's plays for a very long time. But while a person may read the work of two writers and notice similar patterns, a computer can really dig into the data. Just as statisticians test against a null hypothesis (that a relationship does not exist) and estimate the probability that if there is no relationship we would observe a difference, these analyses can estimate the probability that, if the two authors did not work together, we would see such similarity in vocabulary and style.

Obviously, there could be alternative explanations for the similarities. One strong possibility is offered in the article:
Carol Rutter, a professor of Shakespeare and performance studies at the University of Warwick, told the BBC, "It will still be open for people to make up their own minds. I don't think [Oxford University Press] putting their brand mark on an attribution settles the issue for most people."

Rutter told the BBC, "I believe Shakespeare collaborated with all kinds of people ... but I would be very surprised if Marlowe was one of them."

As for how Marlowe's vocabulary and style could have made it into Shakespeare's work without direct collaboration, Rutter said: "It's much more likely that he started his career working for a company where he was already an actor, and collaborated not with another playwright but with the actors — who will have had Marlowe very much in their heads, on the stage, in their voices. ... They were the ones putting Marlowe's influence into the plays."

Monday, October 24, 2016

With Liberty and Justice For All

One of my favorite topics in social psychology is justice, or rather, how individuals determine whether something is just or fair. I did one of my candidacy exams in grad school on justice, citing work on, for instance, belief in a just world and Norman Finkel's contribution of commonsense justice. The various justice frameworks certainly influence how I perceive interactions with others, and I often remind myself of these different approaches anytime the phrase "not fair" enters my mind.

As I mentioned in a previous post, there are 2-3 overarching frameworks (depending on who you ask): distributive justice (which deals with the different types of distributions of outcomes believed to be fair), procedural justice (which states that, as long as the process is fair, the outcome is believed to be fair), and (the potential third) interactional justice (which is sort of an extension of procedural justice, with some attention to distribution of outcomes). Of these, procedural justice is arguably what most political processes are based upon. By clearly delineating the process through which decisions (such as elections) are made, and by ensuring those procedures are followed, we can be confident that the outcome of that process is just.

The current election cycle has had an underlying subtheme of justice since the very beginning, especially with regard to a certain political candidate. Earlier this year, a New York Times article discussed Trump as the Anti-PC Vote; that is, he appeals to people who believe that political correctness is hurting America, forcing the majority to submit to the will of the minority. Trump gives his supporters (which polls suggest are predominantly men) an outlet for their "mad as hell and not going to take it anymore" feelings and a perceived opportunity to take back their country. Trump heightens these feelings, through in-group/out-group tactics - othering and scapegoating - paradoxically blaming minority groups for their own suffering while also blaming them for making the majority group suffer. I say paradoxically because in order to do both, they have to be both powerless and powerful.

But one part of Trump's rhetoric is especially related to the concept of justice: he has stated on multiple occasions that he will only trust the outcome of the election if he wins:
As Trump has fallen in the polls, he has said that the electoral system is rigged against him and that rampant voter fraud could rob him of votes, even though documented cases of such fraud are rare. Trump said Thursday that undocumented immigrants are illegally voting in elections, even though only U.S. citizens are allowed to register to vote, and that Democrats are voting on behalf of people who have died, even though most jurisdictions regularly update their voter rolls.

“This is having nothing to do with me but having to do with the future of our country,” Trump said. “We have to have fairness.”
This is a frightening notion - not just because the thought of a Trump presidency scares the sh*t out of me. He is making a mockery of our system, inciting further division and uproar among his supporters, that will carry long past Election Day. He is giving his followers license to distrust any process that doesn't deliver the outcome they want, promoting circular reasoning over critical thinking skills. And he is furthering sexism by insinuating that the only way a woman could win the presidency (or really any contest against a man) is by cheating.

Of the different justice frameworks, Trump's seems most like distributive justice. In distributive justice, there are three distribution rules that can be applied: equality, in which each party receives an equal share; equity (also sometimes known as merit), where an individual party’s share is based on the amount of input from that party; and need, where share is based on whether the party has a deficit or has been slighted in some other distribution. Trump's approach, however, does not fall cleanly in one of these rules. It's sort of equity/merit, where his estimation of input is subjective and egocentric, but also a perversion of need, because he's convinced himself (and his followers) that the system is rigged and his win will reverse some of that. And who has rigged the system?
"I think that the media and the Clintons and Obama have all rigged the system and they're trying to make us all believe that she's the winner."
So, among others, they are Obama (his stand-in for scapegoating the entire African-American community) and Clinton (his stand-in for scapegoating women). In fact, the news that Clinton would win if only women voted led many of Trump's followers to proclaim we should "repeal the 19th amendment." Yes, it seems his idea (and his followers' idea) of making America great again is taking it back to a time when only white men could vote.

This is not fairness or justice; this is a mockery of the concept. This is not how someone worthy of being our leader determines justice; this is how a child does it.

Animals Are Funny

In case you need a laugh this Monday morning, let me introduce you to the Comedy Wildlife Photography competition:
The Comedy Wildlife Photography Awards, ingeniously titled to avoid any confusion, was the result of two factors: Firstly, a need for a photography competition that was light hearted, upbeat, possibly unpretentious and mainly about wildlife doing funny things. After its first year, these objectives seem to have been met. Secondly, and way more importantly, this competition is about conservation.

None of us are perfect, all of us at some point will fly somewhere, drive somewhere, cook something, burn something and probably provide some direct input into the general warming of the globe. Indirectly, we will also have some impact on the animals that share this planet with us. So the end result?

By entering this competition it gives both Paul and Tom and the rest of you talented photographers a chance to do a little bit for conservation. How? Well… you are now obviously going to go to your office, home, pub, club or wherever and talk about the dire need for us all to be conservationists in our own little way. Also, perhaps you will go to Born Free’s website and have a look at the work they do and spread that word as well.
The competition is now closed and they are down to the finalists. Winners from the 6 categories will receive a trophy, and the overall winner will receive an all-expenses paid photographer-led safari in Kenya and a new Nikon camera (a D810 body and 24-120mm f/4G ED VR AF-S NIKKOR for the camera nerds out there).

Here are some of my favorites.

The whole gallery can be viewed here.