Wednesday, May 31, 2017

On Patents, Printers, and Consumer Behavior

Yesterday, the U.S. Supreme Court ruled on a very interesting case. Lexmark argued that 3rd parties refilling their ink cartridges was a patent violation. The Supreme Court took this case as an opportunity to further define patent exhaustion - the point at which a patent holder can longer control what happens to an individual instance of their patented product. According to the court, refilling ink cartridges is not a patent violation because once an individual purchases a product, what happens to it after is no longer in the patent holders control:
Lexmark’s rights to control the use of its patented refillable print cartridges would be “exhausted” when it sells those cartridges to retail buyers, even if Lexmark conditions the sale on the promise that the buyer will not refill the cartridge. That, at any rate, is the argument of Impression Products, which makes a business out of refilling Lexmark cartridges in violation of those agreements. Lexmark’s argument, by contrast, supported by a quarter-century of Federal Circuit precedent, is that modern commerce requires that innovators have the flexibility to devise contracting structures that segment the market into separate sectors, each of which gets a different price commensurate with the uses to which products will be put in that sector.

[T]he court concluded that “extending the patent rights beyond the first sale would clog the channels of commerce, with little benefit from the extra control that the patentees retain.” The court pointedly noted that “increasingly complex supply chains [well might] magnify the problem,” offering a citation to an amicus brief suggesting that a “generic smartphone … could practice an estimated 250,000 patents.”
In a sense, the case is one about control - do companies have control over what happens to their products after someone has purchased them? Specifically, can companies control the behavior of consumers. You can understand where Lexmark is coming from: they're missing out on extra sales if people can simply buy their cartridge once and refill it. But printer cartridges can be expensive, so you can also understand the consumer's behavior here. This ruling will obviously have an impact on the behavior of companies as well as consumers. Third party companies are liable to test the boundaries of this ruling. And I would imagine Lexmark (and other companies manufacturing printers) are going to look for ways to redesign printer cartridges that can't be refilled.

Tuesday, May 30, 2017

B.F. Skinner: The Last Behaviorist

Via Advances in the History of Psychology, I learned today about an upcoming film, The Last Behaviorist, which is "an audio-visual portrait" of Skinner and his theories:
“If I am right about human behavior, I have written the autobiography of a nonperson.”
- B.F. SKINNER, A Matter of Consequences: Part Three of An Autobiography

The Last Behaviorist takes Skinner’s proposition as a conceptual point of departure - it is an audio-visual portrait, examining the biographical history, ideas, words, and representations of a non-person through raw footage of the subject and their environment.
No information is provided on release of the film, but you can sign up to be on their mailing list. As a recovering radical behaviorist, I'll definitely check out this film.

Gender Bias in Political Science

This morning, the Washington Post published a summary (written by the study authors) of a study examining gender bias in publications in the top 10 political science journals.
Our data collection efforts began by acquiring the meta-data on all articles published in these 10 journals from 2000 to 2015. Web-scraping techniques allowed us to gather information on nearly 8,000 articles (7,915), including approximately 6,000 research articles (5,970). The journals vary in terms of the level of information they provide about the nature of each article, but we were generally able to determine the type of article (whether a research article, book review, or symposium contribution), the names of all authors—from which we could calculate the number of authors—and often the institutional rank of each author (for example, assistant professor, full professor, etc.). In what follows, we describe the variable generation process for all types of articles in the dataset, but note that the findings we report stem from an analysis of authorship for research articles only, and not reviews or symposia.

Using an intelligent guessing technique (compared against a hand-coding method) we used authors’ first names to code author gender for all articles in the database. We also hand-coded the dominant research method employed by each research article. We were further able to generate women among authors (%) which is the share of women among all authors published in each journal, as well as other variables related to the gender composition for each article, which include information about whether each article was written by a man working alone, a woman working alone, an all-male team, an all-female team, or a co-ed team of authors. Because the convention in political science is generally to display author names alphabetically, we have not coded categories like “first author” or “last author” which are important in the natural sciences.
As you can see from the table below, there were low percentages of women among authors across all 10 journals:


One explanation people offer for underrepresentation of women is that there are simply fewer women in the field. But that's not the case here:
Women make up 31 percent of the membership of the American Political Science Association and 40 percent of newly minted doctorates. Within the 20 largest political science PhD programs in the United States, women make up 39 percent of assistant professors and 27 percent of tenure track faculty.
Instead, they offer 2 explanations:

1) Women aren't being offered as many opportunities for coauthorship:
The most common byline across all the journals we surveyed remains a single male author (41.1 percent); the second most common form of publication is an all-male “team” of more than one author (24 percent). Cross-gender collaborations account for only 15.4 percent of publications. Women working alone byline about 17.1 percent of publications, and all-female teams take a mere 2.4 percent of all journal articles.
2) The research methods most often used by women political sciences (qualitative methods) are less likely to be published in these top journals than studies using quantitative methods. As a mixed methods researcher, I frequently use qualitative methods - this was especially true in my work for the Department of Veterans Affairs, where we studied topics that were not only complex and nuanced, but poorly studied and sometimes occurring in a small subset of the population. These are the perfect conditions for a well-done qualitative study to establish some concepts that can be studied quantitatively. But it's difficult to write a survey or create a measure without that basic knowledge. (That doesn't stop people from doing it, leading to bad research. But hey, it uses numbers, so it must be good, right? </sarcasm>) I frequently received snide remarks from other researchers and consumers of research, who didn't believe qualitative methods were rigorous or even scientific. And, as I've blogged about before, I received similar comments in some of my peer reviews.

The authors recognize that perhaps the reason for low representation of women may be because they simply aren't submitting to these journals. But:
[I]f women are not submitting to certain journals in numbers that represent the profession, this is the beginning and not the end of the story. Why not?

Political scientists have helped forge crucial insights into the “second” and “third faces” of power — ideas that help explain that the effects of power can be largely invisible.

The second face of power refers to a conscious decision not to contest an outcome in light of limited prospects for success, as when congressional seats go uncontested in districts that are solidly red or blue.

The third face of power is more subtle and refers to the internalization of biases that operate at a subconscious level, as when many people assume, without thinking, that wives — and not husbands — will adjust their careers and even their expectations to accommodate family and spouse.

Let’s apply those insights to the findings from our study. If women aren’t submitting in proportional numbers to prestigious journals, that may result from conscious decisions based on the second face of power: They don’t expect their work to be accepted because they don’t see their type of scholarship being published by those journals. Or they may refrain from submitting because of a more internalized, third-face logic, taking it for granted that scholars like “me” don’t submit to journals like that.

Either way, publication patterns are self-enforcing over time, as authors come to see it as a waste of time to submit to venues whose past publications do not include the kind of work they do or work by scholars like them.

Monday, May 29, 2017

Sara's Week in Psychological Science: Conference Wrap-Up

I'm back from Boston and collecting my thoughts from the conference. I had a great time and made lots of great connections. While I didn't have a lot of visitors to my poster, I had some wonderful conversations with a few visitors and other presenters - quality over quantity. I'm also making some plans for the near future. Stay tuned: there are some big changes on the horizon I'll be announcing, starting later in the week.

In the meantime, I'm revisiting notes from talks I attended. One in particular presented a flip side of a concept I've blogged about a lot - the Dunning-Kruger effect. To refresh your memory, the Dunning-Kruger effect describes the relationship between actual and perceived competence. People who are actually low or high in competence tend to rate themselves more highly on perceived competence than people with a moderate level of competence - and this effect has been observed for a variety of skills.

The reason for this effect has to do with knowing what competence looks like. You need a certain level of knowledge about a subject to know what true competence looks like. People with moderate competence know quite a bit but also know how much more there is to learn. But people with low competence don't know enough to understand what competence looks like - in short, they don't know what they don't know. (In fact, you can read a summary of some of this research here, which I co-authored several years ago with my dissertation director, Linda Heath, and a fellow graduate student, Adam DeHoek.)

The way to counteract this effect is to show people what competence looks like. But one presentation at APS this year showed a negative side effect of this tactic. Todd Rogers from the Harvard Kennedy School presented data collected through Massively Open Online Courses (MOOCs - such as those you'd find listed on Coursera). These courses have high enrollment but also high attrition - for instance, it isn't unusual for a course to have an enrollment of 15,000 but only 5,000 who complete all assignments.

Even with 66.7% attrition, that's a lot of grading. So MOOCs deal with high enrollment using peer assessment. Students are randomly assigned to grade other students' assignments. In his study, Dr. Rogers looked at the effect of quality of randomly assigned essays on course completion.

He found that when students received high quality essays, they were significantly less likely to finish, than if they received low quality essays. A follow-up experiment, where participants were randomly assigned to receive multiple high quality or low quality essays, confirmed these results. When people are exposed to competence, their self-appraisals go down, mitigating the Dunning-Kruger effect. But now they're also less likely to try. Depending on the skill, this might be the desired outcome, but not always. Usually when you try to get people to make more accurate self-assessments, you aren't trying to make them give up entirely, but perhaps accept that they have more to learn.

So how can you counteract the Dunning-Kruger effect without also potentially reducing a person's self-efficacy? I'll need to revisit this question sometime, but share any thoughts you might have in the comments below!

In the meantime, I leave you with a photo I took while sightseeing in Boston:

Sunday, May 28, 2017

Statistics Sunday: Getting Started with R

For today's post, I'm going to get you started with using R. This will include installing, importing data from an external file, and running basic descriptives (and a t-test, because we're fancy).

But first, especially for statistics newbies, you may be asking - what the heck is R?

R is an open source statistical package, as well as the name of the programming language used to run analysis (and do some other fancy-schmancy programming stuff we won't get into now - but I highly recommend David Robinson's blog, Variance Explained, to see some of the cool stuff you can do with R). R comes with many statistical and programming commands by default, part of what's called the 'base' package. You can add to R's statistical capabilities by installing different libraries. Everything, including new libraries and documentation about these libraries, is open source, making R an excellent choice for independent scholars, students, and anyone else who can't blow lots of money on software.

R will run on multiple operating systems, so whether you're using Windows, Mac OS, or a distro of Linux, you'll be able to install and run R. To install R, navigate over to the Comprehensive R Archive Network (CRAN). Links to install are available at the top of the page. I just reinstalled R on my Mac, with the newest version (at the time of this writing) called "You Stupid Darkness" (aka: 3.4.0). If and when you write up any statistical analysis you did on R, you'll want to report which version you used (this is true anytime you use software to run analysis, not just when you use R).

After you install R, you'll also want to install R Studio. It's an excellent resource regardless of whether you're new to R or an advanced user.


R Studio organizes itself into four quadrants:
  1.  Upper left - Any R scripts or markdown (for LaTex lovers, like myself - future post!) files are displayed here. Code you write here can be saved for future use. Add comments (starting the line with #) to include notes with your code. This is great if you (or someone else) will revisit code later, and it's also helpful to remind yourself what you did if and when you write up your results. Highlight the code you want to run and click 'Run' to send it to the console.
  2. Lower left is the console. This is where active commands go. If you run code from a script above, it will appear here along with any output. You can also type code directly here but note that you can't save that code for later use/editing.
  3. Upper right records any variables, lists, or data frames that exist in the R workspace (that is, anything you've run that creates an object). There's also a history tab that displays any code you ran during your current session.
  4. Lower right is the viewer. You can view (and navigate through) folders and files on your computer, packages (libraries) installed, any plots you've created, and help/documentation.
The great thing about R Studio is that you can access many things by clicking instead of typing into the console, which is all you get if you were to directly open R instead of R Studio. For some things, you'll find typing code is faster - such as to change your working directly, or load or install libraries. In fact, when I first started using R regularly, I was installing 4-5 libraries a day, which I briefly considered (half-jokingly) using as a measure of productivity. Now that I've reintalled R on my Mac (because I completely wiped the hard drive and reinstalled - long story), I could actually collect these data instead of just joking about doing so.

But when you have to go through multiple steps for certain things - such as viewing the help for a specific command within a specific library - you'll find R Studio makes it much easier.

R Studio will also do some auto-complete and pop-help when you type things into the script window, which is great if you can't quite remember what a command looks like. It can also tell when you're typing the name of a dataset or variable and will pop up a list of active data and variables. Super. Helpful.

Hopefully you were able to install these two programs (and if you haven't done so yet because you've been distracted by this love letter to R Studio fantastically written post, do that now). Now, open R Studio - R will automatically load too, so you don't need to open both. By default, the whole left side of the screen will be console. Create a new script (by clicking the icon that looks like a white page with a green plus and selecting R Script or by clicking File -> New File -> R Script) and the console will move down to make room.

The first thing I always do in a new script is change the working directory. Change it to whatever folder you'll be working with for your project - which can vary depending on what data you're working with. For now, start by downloading the Caffeine study file (our fictional study about the effect of caffeine on test performance, first introduced here), save it wherever you want, then change the working directory to that folder by typing setwd("directory") into the script (replacing directory with wherever the file is saved - keep the quotes and change any \ to /).  (If you really don't want to type that code, in the lower right viewer, navigate to where you saved the file, then click More -> Set As Working Directory. The code you want will appear in the console, so you can copy and paste it into the script for future use.

Let's read this file into R to play with. The file is saved as a tab-delimited file. R base has a really easy command for importing a delimited file. You'll want to give the dataset a name so you can access it later. Here's what I typed (but you can name the first part whatever you'd like):

caffeine<-read.delim("caffeine_study.txt", header=TRUE, sep="\t")

You've now created your first object, which is a dataframe called "caffeine." The command that follows the object name tells R that the file has variable names in the first row (header=TRUE) and that the delimiter is a tab (sep="\t"). Now, let's get fancy and run some descriptive statistics and a t-test, recreating what you saw here. But let's make it easy on ourselves by installing our first package: the psych package*. Type this (either into your script, then highlight and Run, or directly into the console): install.packages("psych"). You just installed the psych package, which, among other things, lets you run descriptive statistics very easily. So type and Run this: 

library("psych") (which loads the library you need for...)
describe(caffeine) (or whatever you named your data)

You'll get output that lists the two variables in the caffeine dataset (group and score), plus descriptive statistics, including mean and standard deviation. This is for the sample overall. You can get group means like this:

describeBy(caffeine, group="group")

Now you'll get descriptives first for the control group (coded as 0) and then the experimental (coded as 1). It will still give you descriptives for the group variable, which is now actually a constant, because the describeBy function is separating by that variable. So the mean will be equal to the group code (0 or 1) and standard deviation will be 0. You should have group 0 M = 79.27 (SD = 6.4) and group 1 M = 83.2 (SD = 6.21). Now, let's run a t-test. R's base package can run a t-test: t.test(DV ~ IV, data=nameofdata). So with the caffeine dataset it would be:

t.test(score ~ group, data=caffeine)

If your data are normally distributed, you'll get a standard Student t. Otherwise, you'll get a Welch's t, which shifts your degrees of freedom slightly to account for lack of normally distributed y variable (future post!). Apparently, my data weren't normal, so I got Welch's for my output. That's what I get for fabricating a dataset - oh yeah, as I said previously, these data are fake, so don't try to publish or present any results.

R can read in many different types of data, including fixed width files, and files created by different software (such as SPSS files). Look for future posts on that. And R can go both ways - not only can it read a tab-delimited file, it can write one too. For instance, if you're doing a lot of different transformations or computing new variables, you might want to save your new datafile for later use. I've also used this command to write results tables to a tab-delimited file I can then import into Excel for formatting. You'll need to reference whatever you named the command, so if you wanted to write your descriptives to a tab-delimited file, you'd need to name the object:

desc<-describe(caffeine)

Note that above, we just typed the describe command in directly, so you'll want to rerun it with a name and the arrow (<-). (This is, in my opinion, the easiest way for a new R user, but there is a way to do all of this in one step that we can get into later.) Now, write the descriptives to a tab-delimited file:

write.table(desc, "desc.txt", row.names=FALSE, sep="\t")

Without the row.names command, R will add numbers to each row. This might be helpful when writing data to a tab-delimited file (it basically gives you case numbers) but I tend to suppress this, mostly because I almost always give my cases some kind of ID number from the beginning. 

One note for any R-savvy readers of this post - the sep command technically isn't needed in either the read.delim or write.table commands, because tab ("\t") is actually the default, but I include just to be clear what delimiter I'm using, and so you get used to specifying. After all, you might need to use a comma delimiter or something else in the future.

Hopefully this has given you enough to get started. You can view help files for different packages by going to the packages tab in the lower right, then clicking on the package name. Scroll through the different commands available in that package and click on one to see more info about it, including sample code. I hope to post some new R tutorials soon! And let me know in the comments if you have any questions about anything.

*Check out William Revelle's page for great resources about the psych package (which he created) and R in general.

Saturday, May 27, 2017

Different Distributions

As I was logging some recently watched movies in Letterboxd, I found something interesting: the ratings for Alien: Covenant are normally distributed.


Get Out, on the other hand, is negatively skewed:


I'm still at the conference. More later. 

Friday, May 26, 2017

Sara's Week in Psychological Science: Conference Day #1

Today was my first full day at the conference - the annual meeting of the Association for Psychological Science. (Last night's post was hastily written on my phone while enjoying a beer and dessert, hence the lack of links.)

I'll be presenting a poster tomorrow afternoon. In the meantime, I've been sitting in interesting presentations today.

First up this morning was a panel on psychometric approaches. There was a lot of attention given to Bayesian approaches, and this just signals to me something I've suspected for a while - I should learn Bayesian statistics. I'll probably write more about this approach in a future Statistics Sunday post, but to briefly summarize, Bayesian statistics deal with probability differently than traditional statistics, mostly in the use of "priors" - prior information we have about the thing we're studying (such as results from previous studies) or educated guesses on what the distribution might look like (for very new areas of study). This information is combined with the data from the present study to form a "posterior" distribution. There are some really interesting combinations of Bayesian inference with item response theory (a psychometric approach, which I've blogged about before and should probably discuss in more detail at some point). One great thing about Bayesian approaches is that they don't require normally distributed data.

The panel was devoted to the benefits and drawbacks of different kinds of psychometric models and the research situations in which you should use special models - here's one of my favorite slides of the panel:


I also attended a presentation for a brand new journal, Advances in Methods and Practices in Psychological Science, which will be publishing its first issue early next year:
The journal publishes a range of article types, including empirical articles that exemplify best practices, articles that discuss current research methods and practices in an accessible manner, and tutorials that teach researchers how to use new tools in their own research programs. An explicit part of the journal’s mission is to encourage discussion of methodological and analytical questions across multiple branches of psychological science and related disciplines. Because AMPPS is a general audience journal, all articles should be accessible and understandable to the broad membership of APS—not just to methodologists and statisticians. The journal particularly encourages articles that bring useful advances from within a specialized area to a broader audience.
I already have an idea for a paper I'd like to submit.

The last session of the day I attended was on implicit bias, and how they impact real-world interactions between police and community members, doctors and patients, and employers and employees.

All that's left is a reception tonight. At the moment, I'm relaxing in my hotel room before heading out to try a German restaurant with an excellent beer selection.

Thursday, May 25, 2017

Psychological Science in Boston

I just arrived in Boston earlier this afternoon to attend the annual meeting of the Association for Psychological Science. I'll have details on the conference tomorrow. While they had events and workshops earlier today, these were mostly pre-conference activities. I attended the official opening reception earlier this evening. Here's some photo highlights of the day:

The view from my hotel

Some great psychology buttons I found at the opening reception 

A beautiful old church I walked by on my way back to the hotel

The hotel bar

And dessert

In fact I'm enjoying dessert right now. After this, I'll head back up to my room for some reading and/or Netflix before heading to bed.

Wednesday, May 24, 2017

Science, Uncertainty, and "The Hunt for Vulcan"

Today, I'm listening to a science podcast from earlier this month, "How the Planet Vulcan Changed Science Forever":
In the podcast, which runs in FiveThirtyEight’s What’s The Point feed, senior science writer Maggie Koerth-Baker, lead science writer Christie Aschwanden and senior editor Blythe Terrell talk through how science ideas evolve over time — and how challenging that process can be.

The second part of this month’s podcast features Christie interviewing [Thomas] Levenson about [his] book, [The Hunt for Vulcan].
I'll have to add Levenson's book to my reading list. And if you want to read ahead for next month's podcast, they'll be talking about "Flavor" by Bob Holmes.

Tuesday, May 23, 2017

Trump's Budget

Trump released his first budget, which FiveThirtyEight observes is built on fantasy:
President Trump’s first budget, released Tuesday, is not going to become law. First, because presidents’ budgets never become law, not the way they’re initially proposed. And second, because the specifics of Trump’s fiscal 2018 budget — enormous cuts to nearly every significant government program other than defense, Social Security and Medicare in order to pay for huge tax cuts that would go disproportionately to the wealthy — seem designed to alienate not just Democrats (at least a few of whom Trump needs to get his budget through the Senate) but also moderate Republicans and the public at large. Trump likely knows this; the White House released the budget while he is thousands of miles away on his first foreign trip as president.
Hmmm...

The fantastical part of his budget? He's basing it accelerating economic growth, up to 3% by 2021. This is much higher than estimates from the Congressional Budget Office (1.9%), the Federal Reserve (1.8%), and what was observed last year (1.6%). There are also countless threats to economic growth, including limits on immigration and the retirements of baby boomers. Productivity is also slowing down and no one knows why, making it difficult to predict what the economy will look like.

The response from the White House is basically chiding the Obama administration for being so pessimistic about the nation's economic growth, and faith that we can obtain 3% growth. Hope and faith is important, but not to build a budget upon.

Monday, May 22, 2017

Would You Like Fryes With That?: A Psychological Analysis of Fraud Victims

What started as an over-the-top music festival in the Bahamas ended up as a social media joke. The Frye Festival, which was supposed to take place in late April, was canceled - after guests had already started arriving:
On social media, where Fyre Festival had been sold as a selfie-taker’s paradise, accounts showed none of the aspirational A-lister excesses, with only sad sandwiches and free alcohol to placate the restless crowds. General disappointment soon turned to near-panic as the festival was canceled and attendees attempted to flee back to the mainland of Florida.

“Not one thing that was promised on the website was delivered,” said Shivi Kumar, 33, who works in technology sales in New York, and came with a handful of friends expecting the deluxe “lodge” package for which they had paid $3,500: four king size beds and a chic living room lounge. Instead Ms. Kumar and her crew were directed to a tent encampment. Some tents had beds, but some were still unfurnished. Directed by a festival employee to “grab a tent,” attendees started running, she said.

Now, they're under federal investigation for fraud. In hindsight, the whole thing is clearly a scam. Websites disappeared because designers weren't getting paid. Past customers of previous services complained that special offers never materialized. Not to mention hearing from disgruntled past employees and contractors. In fact, it's so clearly a scam, it's surprising anyone fell for it.

It's very easy for us to look at all of this information now, and come to the conclusion that it was a scam. The problem with hindsight is that it's always 20/20. The same cannot be said for foresight but that doesn't stop people from saying they would have known all along. This is called hindsight bias.

There's probably also some victim blaming going on here. How could these people not know any better? Had we been in the same situation, of course we would have known. We distance ourselves from the victims of this fraud, because it helps us feel more safe, more in control of our world. The same thing could never happen to use because we wouldn't let it.

It's easy to understand reactions after-the-fact. What's more interesting, I think, is to try to figure out what got the attendees and contractors to buy into this fraud to begin with. We ask incredulously, "What were they thinking?" But seriously - what were they thinking?

Human beings are social creatures. We have to be. In order for our species to survive in a hostile environment, it was necessary for us to band together. We formed groups, which became tribes, which become whole societies. And in order to survive in these social structures, it was necessary for to have some trust in the people around us. You could argue that trust is an evolutionarily selected trait in humans. Let's face it, if you don't trust anyone else, it's really unlikely that you're going to reproduce. You have to at least trust one person to do that (at least, if you're reproducing on purpose).

So now we have a species pre-disposed toward trusting others. But we don't give our trust to just anyone - rather, to people we perceive as having certain traits. The more charismatic the leader, the more likely we are to trust them. And if everyone else in our social group trusts a certain person, we're more likely to trust that person too, at least externally.

Internally we may be more skeptical. If we look at the results of the Milgram study, we find that many people reported after the fact feeling very uncomfortable with what they were doing. They even had doubts as to whether they were doing the right thing. But they continued shocking the learner nonetheless. Why? Because somebody in the lab coat, somebody they perceived as having expertise, told them to. This person knows better than me, so I'm just going to keep doing what they say. It doesn't matter whether they actually have any expertise. It's the perception of expertise. And that is something charismatic leaders can do. They can convince you that they know more than they actually do, that they are an expert in something that you are not an expert in. Mc Farland had people believing that he was an expert in entertainment, technology, and rubbing elbows with celebrities. He had people convinced that he could help them to do the same thing.

I'm sure there are some people who didn't trust him. But they went along with him anyway, because there were people who did believe him, who believed that he could do exactly what he said he was going to do, despite instances in the past where he had simply wasted other people's money. But that's the nature of conformity. At the very least, if everyone else is doing it, that makes us more likely to question why we aren't doing it too. Maybe the rest of the group knows something that we don't. Maybe we're misreading the situation.

In the 1950s, Solomon Asch conducted what he said was a study on perception, that was actually a study of conformity.  Actors who pretended to be fellow participants publicly selected what was clearly the wrong answer, to see if the true participant would do the same; 32% of participants conformed with the wrong answer every time across multiple trials, and 75% conformed at least once.

Obviously, there are some other cognitive fallacies occurring here and in similar scams. The sunk cost fallacy, for instance, would explain why people held onto the idea of the festival, especially if they kept paying into it over time. It's the same principle that keeps people pumping money into slot machines or staying in bad relationships - if I keep this up, eventually it will be worth it, and I've put in too much time, money, and/or effort to walk away now. That's what happens when something has a variable schedule of reinforcement. We learn from variable schedules that if you just keep it up, the reward will eventually come.

Combine the sunk cost fallacy with a charismatic leader, the promise of rubbing elbows with people we admire, and other members of our social group going along with it, and it's not surprising at all that people fell for this scam. The problem is that people are going to keep falling for it. The people who were hurt in this particular scam will probably learn their lesson and stay far away from McFarland and his endeavors. But there will always be others will fall for it. And they're unlikely to learn anything from the negative experience of their peers - they'll blame the victims, they'll insist they would have known all along, and they'll distance themselves from those who have been hurt. They'll think of them as the outgroup - people who aren't like them in the ways that matter - and ascribe negative characteristics to them.

There will always be people like McFarland. And there will always be people who fall for his song and dance.

Sunday, May 21, 2017

Statistics Sunday: The Analysis of Variance

In the t-test post and bonus post, I talked about how to use the t-test to compare two sample means and see if they are more different than we would expect by chance alone. This statistic is great when you have two means, but what if you have more than 2?

Here's a more concrete example. Imagine you're going to a movie with three friends. You buy your tickets, get your popcorn and sodas, and go into the theatre. You turn to your friends to ask where they'd like to sit.

The first says, "The back. You don't have anyone behind you kicking your seat and you can see the whole screen no matter what."

"No," the second friend says, "that's way too far away. I want to sit in the front row. No one tall in front of you to block your view, and you can look up and see the actors larger-than-life."
And you can be like these guys

"You're kidding, right?" asks the third friend. "And deal with the pain in neck from looking up the whole time? No, thank you. I want to sit farther back, but not all the way in the back row. The middle is the best place to sit."

How do you solve this dilemma? With research, of course! (Why? How do you solve arguments between friends?) You could pass out a survey to movie goers to see who has the best experience of the movie based on where they sit - front, middle, or back. But now you want to see which group, on average, has the best experience. You know a t-test will let you compare two groups, but how do you compare three groups?

Yes, you could do three t-tests: front v. middle, front v. back, and middle v. back. But remember that you inflate your Type I error with each statistical test you conduct. You could correct your alpha for multiple comparisons, but you also increase your probability of Type II error doing that. As with so many issues in statistics, there's a better way.

Enter the analysis of variance, also known as ANOVA. This lets you test more than two means. And it does it, much like the t-test, by examining deviation from the mean. In any statistical situation, the expected value is the mean - in this case, it's what we call the grand mean, the mean across all 3+ groups. If seating location makes no difference, we would expect all three groups to share the same mean; that is, the grand mean would be the best descriptor for everyone. We're testing the statistical hypothesis that the grand mean is not the best descriptor for everyone. So we need to see how far these groups are from the grand mean and if it's more than we expect by chance alone.

But the mean is a balancing point; some groups will be above the grand mean, and some below it. If I took my grand mean, and subtracted each group mean from it, then added those deviations together, they would add up to 0 or close to it. What do we do when we want to add together deviations and not have them cancel each other out? We square them! Remember - this is how we get variance: the average squared deviation from the mean. So, to conduct an ANOVA, we look at the squared deviations from the grand mean. Analysis of variance - get it? Good.

Once you have your squared deviations from the grand mean - your between group variance - you compare those values to the pooled variance across the three groups - your within group variance, or how much variance you expect by chance alone. If your between group variance is a lot more than your within group variance, the result will be significant. Just like the t-test, there's a table of critical values, based on sample size as well as the number of comparisons you're making; if your ANOVA (also known as a F test - here's why) is that large or larger, you conclude that at least two of the group means are different from each other.

You would need to probe further to find out exactly which comparison is different - it could be only two are significantly different or it could be all three. You have to do what's called post hoc tests to find out for certain. Except now, you're not fishing - like you would be with multiple t-tests. You know there's a significant difference in there somewhere; you're just hunting to find out which one it is. (Look for a future post about post hoc tests.)

The cool thing about ANOVA is you can use it with more than one variable. Remember there is a difference between levels and variables. A level is one of the "settings" of a variable. For our caffeine study, the levels are "experimental: receives caffeine" and "control: no caffeine." In the movie theatre example, the variable is seating location, and the levels are front, middle, and back. But what if you wanted to throw in another variable you think might effect the outcome? For instance, you might think gender also has an impact on movie enjoyment.

There's an ANOVA for that, called factorial ANOVA. You need to have a mean for each combination of the two variables: male gender-front seat, female gender-front seat, male gender-middle seat, female gender-middle seat, male gender-back seat, and female gender-back seat. Your ANOVA does the same kind of comparison as above, but it also looks at each variable separately (male v. female collapsed across seating location, and front v. middle v. back collapsed across gender) to tell you the effect of each (what's called a main effect). Then, it can also tell you if the combination of gender and seating location changes the relationship. That is, maybe the effect of seating location differs depending on whether you are a man or a woman. This is called an interaction effect.

On one of these Statistics Sundays, I might have to show an ANOVA in action. Stay tuned!

Saturday, May 20, 2017

Long Road Ahead

So a special counsel has been appointed to continue the Russia investigation. People on both sides of the political continuum are pretty happy about this (after all, 78% of Americans in a recent poll were in support of this) - people who believe Trump is innocent of any wrongdoing can depend on the investigation to exonerate him, while people who believe Trump is guilty can finally move one step closer to removal from office.

But it's liable to be years before anything definitive comes out of this investigation. Clare Malone at FiveThirtyEight sat down with political scientist, Brandon Rottinghaus, to discuss the history of political scandals and when this particular investigation might end:
Number one, the president is insulated politically so that it’s hard to get the president’s staff and counsels to turn on the president.

Number two, presidents are often insulated legally; they have the ability to do a lot of things that staff or Cabinet members aren’t able to do.

The third thing is that independent counsels, special counsels and any other investigatory bodies are reluctant to challenge the president in a way that might lead to impeachment for fear that it looks like a non-democratic outcome to the legal process. Although obviously these things run into partisanship very quickly, people are less willing to remove a president unless the crisis is severe and the implications are egregious.

[Y]our standard investigation, even of a person who’s a Cabinet member or staff member, is probably between two and three years. For a president in particular, it tends to be longer because the amount of care to be taken is greater.
Regardless of how long the process will take, the fact that an investigation is underway will still have an impact on the current administration, long before any results are shared:
[T]hese kind of events often lead to legislative paralysis, and if you’re not producing legislation, the public tends to take it out on the incumbent party, especially the president. So it’s a kind of double whammy for presidents looking to keep those approval ratings above water.
It's a long road ahead:

This gorgeous photo is by photographer Glenn Nagel

Friday, May 19, 2017

Meanwhile, In the Mind of AI

Researcher and lover of neural networks, Janelle Shane, was thinking about the strange and usual paint names and wondered what a neural network would name different paint shades. So she fed in 7,500 Sherwin Williams paint names and RGB values, and let the network go to work. The results is not only an interesting insight into neural networks and a higher-level view of how they work; it's pretty hilarious what the network came up with:
One way I have of checking on the neural network’s progress during training is to ask it to produce some output using the lowest-creativity setting. Then the neural network plays it safe, and we can get an idea of what it has learned for sure.

By the first checkpoint, the neural network has learned to produce valid RGB values - these are colors, all right, and you could technically paint your walls with them. It’s a little farther behind the curve on the names, although it does seem to be attempting a combination of the colors brown, blue, and gray.
Later in the training process, the neural network is about as well-trained as it’s going to be (perhaps with different parameters, it could have done a bit better - a lot of neural network training involves choosing the right training parameters). By this point, it’s able to figure out some of the basic colors, like white, red, and grey, although not reliably.

In fact, looking at the neural network’s output as a whole, it is evident that:
  1. The neural network really likes brown, beige, and grey.
  2. The neural network has really really bad ideas for paint names.

Stoner Blue is pretty nice. Think I'll skip Sindis Poop.

In Your Mind, The Future is Now

In an article published today by the New York Times, journalist John Tierney teams up with psychologist and founder of the positive psychology movement, Marty Seligman, for a great article about anticipation and the importance of the future in our approach to the present:
But it is increasingly clear that the mind is mainly drawn to the future, not driven by the past. Behavior, memory and perception can’t be understood without appreciating the central role of prospection. We learn not by storing static records but by continually retouching memories and imagining future possibilities. Our brain sees the world not by processing every pixel in a scene but by focusing on the unexpected.

Our emotions are less reactions to the present than guides to future behavior. Therapists are exploring new ways to treat depression now that they see it as primarily not because of past traumas and present stresses but because of skewed visions of what lies ahead.

Prospection enables us to become wise not just from our own experiences but also by learning from others. We are social animals like no others, living and working in very large groups of strangers, because we have jointly constructed the future. Human culture — our language, our division of labor, our knowledge, our laws and technology — is possible only because we can anticipate what fellow humans will do in the distant future. We make sacrifices today to earn rewards tomorrow, whether in this life or in the afterlife promised by so many religions.

The central role of prospection has emerged in recent studies of both conscious and unconscious mental processes, like one in Chicago that pinged nearly 500 adults during the day to record their immediate thoughts and moods. If traditional psychological theory had been correct, these people would have spent a lot of time ruminating. But they actually thought about the future three times more often than the past, and even those few thoughts about a past event typically involved consideration of its future implications.

When making plans, they reported higher levels of happiness and lower levels of stress than at other times, presumably because planning turns a chaotic mass of concerns into an organized sequence. Although they sometimes feared what might go wrong, on average there were twice as many thoughts of what they hoped would happen.
The article is based on a book Marty wrote with fellow psychologist Roy Baumeister, philosopher Peter Railton, and psychiatrist Chandra Sripada, called Homo Prospectus.

Thursday, May 18, 2017

Can I Get A "Amen"?

A special counsel has been named to continue the investigation into Russia's potential involvement in the election: former FBI Director Robert S. Mueller III. I'm pleasantly surprised that this happened - the White house was also surprised, though I don't think pleasantly.

Perry Bacon Jr., of FiveThirtyEight, analyzes this development from all sides, including pros and cons for Trump himself, Republicans, Democrats, and the American people:
Trump can now say there is an independent investigation going on, by someone he did not personally appoint and who is not beholden to his party. [But] Mueller’s appointment ensures that the Russia controversy won’t just go away — at least not anytime soon. And he could gravely threaten Trump’s presidency if he finds clear, improper connections between the president’s campaign and Russian officials.

Republican members were being repeatedly asked about the Trump investigation. Like Trump, they can now defer to Mueller’s probe. The one problem? Mueller is only investigating the Russia issue. It’s likely Trump will do something else controversial — in the past two weeks alone, he allegedly shared highly classified intelligence with the Russians, and he fired Comey in a clumsy way that created all kinds of political problems. Republicans will still have to answer for Trump’s other controversial moves.

Democrats strongly disagree with Republicans like Mitch McConnell and Paul Ryan on policy, including on sweeping issues currently on the congressional docket such as health care and taxes. In the eyes of many Democrats, Trump and the potential laws he might sign could damage the country for years to come. A process that could (in the long run) lead to Trump’s removal from office is a major step for liberals.

In the short term, they may have lost an issue. Democrats could have pounded Trump and Republicans on their lack of accountability every day till next year’s midterms. Make no mistake: If Democrats had won control of Congress next year and Trump had blocked a special counsel up until then, impeachment would have been on the table. Now, Democrats have to wait and see what Mueller concludes.

Wednesday, May 17, 2017

Out of Touch, Out of Time

Hard to believe it's only been a week since FBI Director James Comey was fired. There are still debates about whether this firing was for the official reason given by the White House (poor handling of the Clinton email investigation) or to halt the investigation into the Trump administration's connections with Russia. There are reports that Trump asked Comey to cease his investigation into Michael Flynn's connections with Russia before the firing, as well as reports that Trump shared classified information with Russia.

Obviously, some investigation is necessary, and this situation - in which the administration has a conflict of interest in conducting or being involved in the investigation - a special prosecutor can be brought in to conduct an independent investigation. The majority of Republicans aren't in support of this. Concerning, considering that a recent poll by NBC and the Wall Street Journal shows that 78% of Americans are in favor of an independent investigation.


And before you shout "fake news!" and insinuate most respondents were Clinton supporters, the sample was pretty evenly split between Clinton and Trump votes; 37% voted for Trump and 40% voted for Clinton. The remainder voted for another candidate or did not vote in the election.

Some other golden nuggets from the survey: 57% had little to no confidence in Trump as president, 48% think the new health care legislation is a bad idea, 38% disapprove of Comey's firing, and 46% agree or strongly agree Trump fired Comey to slow down the investigation. And Trump's actions of late have increased many respondent's doubts about him as president:
  • Donald Trump’s decision to fire FBI Director James Comey - 42% somewhat or a lot more doubts
  • The steps that Donald Trump has taken to separate his business dealings from his official duties - 44% somewhat or a lot more doubts
  • The Trump administration’s handling of allegations of campaign and administration’s officials’ contacts with Russia - 48% somewhat or a lot more doubts

Tuesday, May 16, 2017

History of the World

For your afternoon distraction, this great video about world history starting from the beginning of time:

Something Happens and I'm Head Over Heels

I finally got to hear one of my favorite songs performed by the band live. Last night, I went to a double-header concert of Tears for Fears and Hall & Oates. It was an awesome show, though there were a few technical difficulties and I have a couple things I was hoping to have happen that didn't. Everyone seems to have taken great care of their voices, because they sound exactly the same, and John Oates can still rock out on the guitar. He's also growing a mustache again - keep it up, John! Everyone loves the mustache.

The show was at the Allstate Arena in Rosemont. Tears for Fears had a lot of fun joking with the crowd about Rosemont versus Chicago. "Is everyone here from Rosemont? Or Chicago? Or is the same thing?" They did some witty banter throughout.

Hall & Oates, on the other hand, mostly stuck to making music, with Daryl being the only one really talking to the crowd - mostly to tell us how awesome we are and how much they enjoy performing for us, the usual script for a concert performance. Tears for Fears was much more likely to go off book.

Musically, it was about the opposite. Tears for Fears did faithful versions of their music, as well as a cover of "Creep" (that had the sweet older couple next to us Googling Radiohead during the break between acts). They closed out their set with one of my favorite songs ever, Head Over Heels, then encored with Shout.
  • Everybody Wants To Rule The World 
  • Secret World 
  • Sowing The Seeds Of Love 
  • Advice For The Young At Heart 
  • Everybody Loves A Happy Ending 
  • Change 
  • Mad World 
  • Memories Fade 
  • Creep 
  • Pale Shelter 
  • Break It Down Again 
  • Head Over Heels 
  • Shout
Hall & Oates, on the other hand, did a lot of riffing, improvisation, and extended solos for guitar and especially saxophone (played by Charles DeChant, who's been with the group a long time). A few of these extended solos went on a little long for my taste, especially close to the end of the night, when Hall & Oates had gone way over their set time. I still very much enjoyed their part of the show.

There were a few sound issues, and even Daryl seemed to be a bit annoyed by them. Our seats were pretty close to the stage, so we could see Darryl turn to the sound guy mixing for the stage monitors and making gestures. And at one point, he also had a couple choice words for the sound guys mixing for the hall itself. In fact, during Maneater, DeChant's sax line got completely lost when guitar came in. Pretty sad, considering that, rather than having the sound of 2 saxophones (as they do in the original, with the line echoing back), they had a duet between saxophone and Oates's guitar. This was a cool effect, for the parts I could hear, and I would have liked to have heard more. They seemed to fix the saxophone sound issues after that, though, because he was clear for the rest of the night.

As with Tears for Fears, Hall & Oates also stuck to their hits:
  • Adult Education  
  • Maneater 
  • Out Of Touch 
  • Say It Isn't So 
  • You've Lost That Lovin' Feelin' 
  • One on One 
  • Possession Obsession 
  • She's Gone 
  • Sara Smile 
  • Wait For Me 
  • Is It A Star 
  • Method of Modern Love 
  • I Can't Go For That 
  • You Make My Dreams 
  • Rich Girl 
  • Kiss On My List 
  • Private Eyes
Other than the sound, here are my only complaints.

1. Hall & Oates didn't perform one of my favorites of their songs:


The video is hella cheesy, but it's a great song. It might be considered one of their deep tracks, so I don't know if they really perform it, especially these days.

2. Considering that Daryl Hall has a show called Live from Daryl's House (which they had playing during the break between sets), where he performs with other artists and groups, I would have loved for Hall & Oates to play one or two songs with Tears for Fears.

Minor complaints, really. It was an awesome show and I'm thrilled I got to be there! And now, here's the video for Head Over Heels, for no other reason than I love this song:

Monday, May 15, 2017

Science Fiction Meets Science Fact: Machine Learning and Precognition

Many science fiction and fantasy stories feature characters who can predict events before they occur. In Minority Report (both the short story and film),"precogs" are able to foresee crime, allowing the police to arrest the perpetrator before the event actually occurs. As of yet, we have no evidence of people with precognition abilities. But the real-life equivalent may already be here.

A study published just over a month ago in Clinical Psychological Science used emergency room data and machine learning to predict suicide attempts:
Due to convention and the limitations of most traditional statistical approaches, clinical psychological science has often attempted to use simple algorithms to solve complex classification problems. This approach can produce statistical significance but has a limited ability to produce clinical significance. For example, as noted earlier, recent meta-analyses on hundreds of studies from the past 50 years indicate that the ability to predict future suicide attempts has always been at near chance levels. The primary reason for this lack of progress is that researchers have almost always used a single factor (i.e., a simple algorithm) to predict future suicide attempts.
They used claims data to identify people who had attempted (but not succeeded in committing) suicide, by specifically looking for self-harm diagnostic codes. Two experts reviewed the resulting 5,543 records to identify actual suicide attempts. Two control groups were used: a random sample of 12,695 patients from the same database who had no history of suicide attempts, and 1,917 individuals from the 5,543 self-harm records that did not appear to have a suicide attempt. They used Python and R to conduct their analyses.

Machine learning outperformed traditional prediction methods (such as regression), and though predictive ability was better closer to the actual attempt (e.g., 7 days before rather than 720 days before), ML still performed far better than chance at all time points. In the table below, you can see the frequencies of true positives (correctly identified as a suicide attempt), false positives (incorrectly identified a control case as a suicide attempt), true negatives (correctly identified as control case), and false negatives (incorrectly identified suicide attempt as control case):



ML correctly classified (as either suicide attempt or control) between 82% and 86% of cases overall, and between 94% and 98% of suicide attempts specifically. And keep in mind, this algorithm used information available in patients' charts, rather than the more in-depth information one could get by speaking to a patient one-on-one. This information could be useful for health care systems, who could routinely use machine learning on patient records to identify patients who may need follow-up or intervention.

War on Open Data?

Open data - that is, making datasets freely available for other researchers and even the general public to inspect and reanalyze - seems to be a natural progression of the scientific method, particularly with regard to dissemination as well as replication. That is, in order to call a particular inquiry 'science,' we have to follow specific rules and steps. And one of these steps is to share our methods and the results with others so that they may examine what we did and come to their own conclusions. Given that researchers can game the system by inflating their Type I error rate, misreporting or fabricating results, or even just making simple mistakes, it's essential that others have access to methods, results, and even the original data where possible.

That's why it is incredibly disturbing that the Trump administration is removing publicly available datasets - a lot of them:
Across the vast breadth of the government, agencies have traditionally provided the public with massive data sets, which can be of great value to companies, researchers and advocacy groups, among others. Three months ago, there were 195,245 public data sets available on www.data.gov, according to Nathan Cortez, the associate dean of research at Southern Methodist University’s Dedman School of Law, who studies the handling of public data. This week it stood at just under 156,000.

Data experts say the decrease, at least in part, may reflect the consolidation of data sets or the culling of outdated ones, rather than a strategic move to keep information from the public. But the reduction was clearly a conscious decision.

Cortez said the Obama administration increased the amount of government data offered to the public, although the information was at times incomplete or inaccurate and sometimes used as a “regulatory cudgel.” Under Trump, the government is taking transparency “in the opposite direction.”

In some cases, federal Web pages are being routinely maintained. In other cases, information that was once easily accessible to the public has moved to locations that are harder to find, access and interpret. Yet other data has entirely vanished.

Sunday, May 14, 2017

Statistics Sunday: What's Normal Anyway?

I've already done one bonus statistics post, which was published on a Thursday. But I wanted something that alliterated well, and Statistics Sunday (with Sara!) seemed perfect and just cheesy enough. I'll try to post something about statistics every Sunday. Once again, feel free to contact me with questions and I might cover them here.

The normal distribution is very important in statistics. Because statistics is about determining the probability of certain outcomes - and inferring that when an outcome (the result of a statistical test) is unlikely, it has an explanation beyond random chance - it's important that we know what the distribution of scores looks like, or should look like in the population, which we are always referring and generalizing back to. We use sample data as a stand-in for the population and to infer what the population distribution might look like if we had that data.

The normal distribution is well-understood, and we can easily determine probabilities of certain results using area under the curve.


So when we use samples - as we almost always do - to study something we often need those data to also be normally distributed, so we can determine those probabilities. Many statistics are based on the assumption (the rule) that the data have a known distribution, and usually that distribution is normal. The distribution from the sample data may not look exactly like the standard normal distribution, but how normal does it need to be? Or more specifically, how far can we depart from normal before we are unable to use probabilities from a normal distribution?

We should first look at the distribution of scores using a histogram, but this doesn't tell you if your data are normal enough. Remember, in statistics, we don't eyeball things. We don't use subjectivity with our results. We let the math do the talking. But it's still important to do this step, because looking at the histogram tells us whether there is only one most frequent score (mode), making the distribution what we call 'unimodal.'

But then we should examine two statistics: skewness and kurtosis. Skewness has to do with where the mode (the top of the distribution) falls. It should fall in the middle, rather than more on one side or the other. It does this by looking at the tail(s) of the distribution, the smallest part(s) of the distribution out to the side(s). A true normal distribution should have two tails, negative (below the mean) and positive (above the mean), that are approximately symmetrical. If the distribution is unskewed, the skewness statistic will be equal to 0, but the skewness statistic ranges (theoretically) from negative infinity to positive infinity. A negative skew means there is more of a tail on the negative end and less at the positive end of the distribution; a positive skew means there is more tail on the positive end.


But skew isn't all; we still need to look at kurtosis, which is a fancy term for how "peaked" the distribution is - the height of the mode. If the distribution is very peaked, that means there are far fewer scores in the tails (extreme scores are very rare); if the distribution is flat, that means there are many scores in the tails (extreme scores aren't very rare). There are three types of distributions with regard to kurtosis:
  • Mesokurtic (perfectly normal)
  • Leptokurtic (peaked)
  • Platykurtic (flattened peak)
Kurtosis is always positive and has a theoretically infinite range. A truly mesokurtic distribution will have kurtosis of 3, though some statistical analysis programs will subtract 3 from kurtosis (creating a measure often referred to as "excess" kurtosis), so that a mesokurtic distribution will have a value of 0.

There are conventions for both, though it can get more complicated than that. Often, programs will give you a standard error and you can conduct statistical tests using skewness and kurtosis. You divide your skewness by the standard error (like many statistics, the test formula is essentially a signal to noise ratio), and the resulting metric will be a Z-score. You would want that result to be less than 1.96, if you're using an alpha of 0.05 for that test. For kurtosis, the standard error is based on the standard error for skewness (it's the skewness standard error times 2 with a slight correction). Both standard errors are computed based on sample size; remember, as sample size increases, the more closely our data should resemble the population distribution.

But some people prefer to simply use conventions. In general, a skewness between -1.5 and +1.5 is considered acceptable. Kurtosis is a bit more disputed, in part because some analysis programs give excess kurtosis without clearly specifying. When using conventions, many people don't even worry about kurtosis and just focus on skewness.

This could be, in part, because skewness and kurtosis really aren't emphasized in statistics courses - at least not that I've seen. That data should be normally distributed is stated but then glossed over, and data provided for student exercises are often generated so that they do not violate assumptions. Real data is far messier. So by glossing over these concepts, courses aren't preparing students for situations they very likely will encounter. But once again, I digress.

My advice? Use the provided standard error and conduct a simple Z-tests. It really only adds one more step. If your data aren't normal, the results of the statistical tests could be wrong.

Saturday, May 13, 2017

Another Chicago Museum I Need to Visit

Next Tuesday, the American Writer's Museum will be opening on the 2nd floor of 180 N Michigan Ave. Chicago Magazine shares 5 cool things about it:
1. Lessons in Chicago history
The museum boasts a room dedicated to local literary lights, with recordings by Studs Terkel and Gwendolyn Brooks and a digital map of landmarks, including the birthplace of Ebony and Jet, the original public library network, and every bookstore within the city limits.

2. Audio tours by lit experts
In the American Voices exhibit, virtual docents, including NPR book critic Maureen Corrigan and Northwestern English professor Ivy Wilson, narrate the biographies and works of 100 emblematic writers, from Abraham Lincoln to James Baldwin.

3. Sensory immersion
Anchoring the museum are 100 wall-mounted boxes that pull out to reveal literary ephemera—items you can touch, hear, or smell.

4. Masterworks in progress
See the pillars of American lit deconstructed in Anatomy of a Masterwork, an exhibit that breaks down the creation and legacy of iconic titles like “Sonny’s Blues” and Huck Finn with digital renderings of early manuscripts, critical reviews, and close readings.

5. Rare treasures
One place you’ll find an absence of screens: the Writer’s Room Gallery, which features a rotating collection of relics. First up? Jack Kerouac’s 120-foot On the Road scroll.
The museum promises a multimedia, interactive experience. As Carey Cranston, president and one of the curators, says, "It's not about books under glass."

Friday, May 12, 2017

Reading Comey's Farewell Letter

Wednesday, CNN shared James Comey's brief farewell letter. The Mary Sue dissects the letter. It is hilarious and spot-on. A few favorite sections:
To all:

I’M OFFICIALLY NAILING THIS SHIT TO THE CHURCH DOOR LIKE MARTIN LUTHER

I have long believed that a President can fire an FBI Director for any reason, or for no reason at all.

IN CASE IT’S NOT PERFECTLY CLEAR I’M THE LATTER CASE Y’ALL OR ELSE I WOULDN’T HAVE MENTIONED IT

It is done, and I will be fine, although I will miss you and the mission deeply.

I’M NOT FINE I’M NOT FINE I’M BARELY KEEPING IT TOGETHER DID Y’ALL HEAR HOW THAT SONNOVABITCH IN CHIEF WAITED UNTIL I WAS OUT OF THE STATE TO FIRE ME AND SENT HIS PERSONAL BODYGUARD WITH A NOTE TO FBI HEADQUARTERS EVEN THOUGH I WASN’T THERE LIKE THAT IS SOME DOWNTON ABBEY SHIT YOU’RE GONNA MAKE THE BUTLER DO YOUR DIRTYWORK AND SEND A MCFRICKEN NOTE AS THOUGH ALEXANDER GRAHAM BELL DIDN’T SOLVE THIS COMMUNICATION PROBLEM IN 1876 FUCK

Working with you has been one of the great joys of my life. Thank you for that gift.

IF I EVER MEANT ANYTHING TO YOU MY G-MEN AND G-WOMEN YOU WILL BURN THIS MOTHER DOWN

What Do Americans Think About Science Protests

On April 22, 2017, Marches for Science occurred in Washington D.C. and other cities around the world (including Chicago!). Yesterday, the Pew Research Center published results of a recent survey, asking Americans what they thought about the marches and specifically, whether they thought it would help or hurt the cause:
Overall, 44% of adults think the protests, marches and demonstrations will boost public support for science, while an equal share believe the protests will make no difference and 7% believe the demonstrations will actually hurt the cause.
As is the case so often these days, responses were divided along party lines:
61% of Democrats and Democratic-leaning independents believe the marches will increase public support for science, while only 22% of Republicans and those who lean Republican say the same. Instead, 60% of these Republican backers think the protests will make no difference, compared with just 32% of Democratic partisans who think that.
Overall, 48% of Americans say they “support” or “strongly support” the goals of the pro-science marches, while 26% “oppose” or “strongly oppose” the goals of the demonstrations and 26% say they don’t know how they feel about them.

It's not too surprising how partisan this issue has become, considering that Trump and his ilk have politicized pretty much every issue. By making things that should be objective into partisan issues, you can ensure that your supporters will fall in line with your way of thinking, even if it doesn't make a whole lot of sense. People on the other side of the political spectrum are simply seen as being partisan themselves (and let's face it, with some issues, they are), which makes it much easier to discount their arguments.
I tire so of hearing people say
Let things take their course. Tomorrow is another day.
I do not need my freedom when I'm dead
I cannot live on tomorrow's bread.
-Langston Hughes