20 August 2017

My Weekend, by Cecily

by Cecily
What did you do over the weekend? I went to my dad and stepmom's house by Georgetown Lake. Remember this? In the olden days of Summer 2012, they had some land by Georgetown Lake. Now, in modern-day times, there is a house on the land. The back yard still looks the same, though. I took this picture:

Looking west, I think? From whatever hill or mountain the house is on.
You may notice a difference in clarity and how many far distant mountains you can see, comparing the 2012 pictures to the modern ones. (There is also a difference in how much snow is on the mountains, but the old pictures are from June and these are from August so don't read too much into it.) Anyway the state is on fire and even in places where the air is relatively clear, like Georgetown, it is still not as clear as it might be.

Meanwhile, in Missoula, somebody else took this picture from the main street downtown:* 

Looking south down Higgins*

There is a really big, scary fire on Lolo Peak. It was at 30,000 acres when I looked it up this morning. (Update 8/22: now it's 32,300.) (Update 9/4: 45,012 acres) (update 9/14: 52,745)

Missoula is basically surrounded by fires (but not on fire itself) so that whichever way the wind blows, it is still blowing smoke on me. Air in Missoula is intermittently unbreathable, which is the only way it is affecting me personally, but the fires are also jumping lines and burning up houses and killing people so I'm trying not to complain about how everything smells like camping.

This fire, and most of the other ones, will keep burning until it freezes. Is what everyone says. Mid-September, probably.

I've had Johnny Cash stuck in my head a lot lately. Here you go!



*I stole this picture off Facebook but I can't figure out where it came from originally. If you know, tell me and I'll add credit. (I didn't try very hard. I might try harder some other time.)
 PHOTO CREDIT: Ross M Perkins https://www.instagram.com/wanderlustnotless/

18 August 2017

Anthropology for Beginners

by Cecily
As we all know, one of my hobbies is making up really, really elaborate metaphors and pushing them as far as they will go. This week's episode is:

Human Toddlers are 60-s Era Anthropologists on an Unknown Planet

People, when they are born, show up in the world as tiny naked anthropologists stranded on an unknown planet with no tools and no instructions and no way to record anything so they have to memorize all the data.  It's a stressful situation! You are in an unfamilar environment and you don't know how anything works here. There are inhabitants, who seem friendly overall, but they are much more powerful than you and you don't seem to have any control over anything. An overwhelming prospect! But there is a strong innate instinct in humans to Figure Stuff Out, and after a couple of days of jet lag, you get to work.

(The first step is obviously to figure out how to work your body, but that doesn't really fit in my metaphor so I'm mostly ignoring it. That happens in parallel, but it is less like anthropology.)

Tiny Anthropologist is an expert gatherer of data, and an expert noticer of patterns. She collects an enormous amount of data. Some combinations of sound waves occur very frequently, and others barely ever. Certain people correllate with certain smells. When she screams, the most common result is that an adult comes over to check on her.  She collects data on everything, and on how often it occurs with everything else. She keeps extensive databases full of detailed information, and does sophisticated things to the data with statistics and probabilities. Pattern after pattern emerges. The patterns are kept in a separate database to analyze which ones are significant and which ones aren't.

The adults produce certain sounds and/or gestures far more often than chance would predict. They seem to be able to communicate with each other this way. Tiny Anthropologist needs to figure out how to mimic those sounds and gestures, in order to test her theories about what they might mean. She practices controlling her body; getting better and better at making sounds and gestures that are similar to the adults'. Eventually she is able to mimic them in a way that gets enthusiastic approval from observers. She keeps track of which things are crowd-pleasers, which attempts at communication are successful, and so on.  She refines her theories about contrastive elements and phonotactic constraints.

By toddlerhood, the Tiny Anthropologist has a reasonably reliable phonological inventory for the language, and a functional ability to make sounds and gestures that nearby adults and older children recognize. She develops a lexicon, frequently running quality check tests by repeatedly pronouncing a word and noting the results. The adult interlocutors' reactions begin to show signs of impatience. Once upon a time, they were thrilled with her whenever she correctly identified an object. Now they seem less impressed. "Yes, it's a dog," they say, but they display traces of negative affect while saying it. It's time to move on from lexical inventory to more complex aspects of communication.

The Tiny Anthropologist begins to study what it is that these beings do, exactly, with their time. She watches, and she attempts to participate. Sometimes an adult holds a broom and moves it around on the floor. Tiny Anthropologist requests to have a turn. Everyone sits on the couch. Tiny Anthropologist sits on the couch, too. An adult tells a long story, including a number of evocative gestures. The Tiny Anthropologist attempts to emulate the scenario. Eventually, patterns of behavior emerge, and the Tiny Anthropologist is thrilled when she begins to correctly predict strings of events. Equally, she is very disappointed when her predictions fail. She thought she had identified a pattern, and now all her work has to be thrown out! The Tiny Anthropologist is unable to contain her distress. Nearby adults are dismayed at her visible disappointment.

For the Tiny Anthropologist, there is also much to learn about how basic conversations work. How do you get someone to be in a conversation with you? Who goes first? How do you know whose turn it is? This is a daunting project. Tiny Anthropologist digs in. She starts by finding out about ways to get attention from other people. Screaming, which up til now has been a failsafe option, has lately been becoming less effective. Producing other sounds works sometimes, but not very reliably. Eye contact seems to get very good results. The Tiny Anthropologist notices that often, when an adult makes eye contact with her, the next thing that happens is that the adult says something, or moves something, or gives something to her. Many fancy statistical tests indicate that P is less than 0.05! The hypothesis is confirmed! The Tiny Anthropologist uses this information to initiate her own conversations: she makes eye contact with adults, and shows them things that she is holding. The adults say something to her! Tiny Anthropologist gets busy initiating conversations with whoever's eye gaze is around. She needs to practice.

Some of Tiny Anthropologist's best work is in the area of Quasilinguistic Discourse. She has come to believe that when the adults say "no" (and/or yell, and/or shake their heads, and/or furrow their eyebrows, and/or pronounce her name with a specific, ominous tone contour), this indicates that they would like her not to do whatever it is that she is doing, or is about to do. This is very useful information, but Tiny Anthropologist knows she needs to make absolutely certain that her understanding is correct. She spends hours, weeks, months, locking eyes with nearby adults and beginning to do things she has (hypothetically) been told not to do. Every time, the adult demonstrates some form of negative signalling until the Tiny Anthropologist sits back down, or stops shrieking, or backs away from the fire pit. She tests again and again, always making sure to lock eyes with an adult first. She starts to spit out her food. No! Food back in the mouth. Okay. She feints with a cup of milk. Scowl. Cup upright, the scowl relaxes. These tests, too, are successful. Achievement unlocked! (The adults seem to be less ecstatic than they should be at all this evidence that the Tiny Anthropologist now clearly understands what "no" means.)

One particular pattern jumps out. A pattern where an adult, while making eye contact with the Tiny Anthropologist, says a word from the known lexicon, while pointing to or holding some object related to the meaning of the word. The new H1 is that if someone, while holding or pointing at something, makes eye contact, the participants are expected to say words that are related to that thing. After many, countless, exhausting hours of strenuous testing (during which the adult subjects often become restless and impatient), the Tiny Anthropologist has enough data to support the hypothesis.

This is a huge breakthrough, because now the Tiny Anthropologist can find out what everything is called, and begin a number of concurrent studies related to sequences of words, and tones, and various co-occurring gestures. The Tiny Anthropologist is an extremely talented researcher, and her project now makes very quick progress. The Discourse notebook gets more and more notes and lists of different conversations. The Lexicon is expanding hourly. She figures out the pronoun system. She observes that (in English), questions are formed by wh-movement and prosody. The Tiny Anthropologist is more and more successful at participating in social events and traditions, and able to understand and comply with cultural and behavioral expectations. Everyone spends less and less time screaming.

The research program is enormous, but efficient, constantly generating statistically significant results. The Tiny Anthropologist moves from one subfield to another, refining and correcting and adding new information, until at last all of the necessary conversational norms have been strenuously tested. The work is complete. The Tiny Anthropologist has learned enough to participate in conversations, rather than study them. She is accepted as a member of the group, and can communicate successfully with most interlocutors. She understands the language and the social norms (mostly) of these strange, huge, people. No further research is needed. The Tiny Anthropologist is ready to move on to a new project. She begins to study her older siblings. It is time to learn how to bicker, squabble, tease, and tattle.

06 August 2017

inequality

by Cecily
Everybody* is always** talking about how much less women and various minority groups make than white men, and often they do it by saying something like "Women in this group earned 90 cents for every dollar a man in the same age group earned."  And then I am indignant but not surprised, and brood over the inequality that is still pervasive in our world, and consider history, and look up things on Wikipedia. But the I find the how-much-of-a-dollar thing boring and irritating. It is overly simplistic. Yes, the nation and the world continue to be in a state of inexcusable inequity and iniquity. That is not new information. (Also, getting women up to the full dollar would not rid the world of iniquity and inequity. I dislike the dollar thing slightly less when it includes minorities.)

But I'm just bored by the white-man's-dollar thing, not the topic in general. I like thinking about the statistics that are behind that, and what it might look like, and wondering how the data*** was organized, and where it came from, and what would happen if you binned it in different ways and did more statistics to it.

This particular white-man's-dollar thing made the Facebook rounds recently, so I started thinking about where those numbers came from, and what kind of average does it represent, and how much variation is being obscured by the average, and does that archetype white guy who earned the dollar include the super rich white guys? What does the top of the income scale look like, compared to the bottom?

(citation needed)

And also I started thinking about what I would do with the data if I had it.

Here's the study I want to see: a huge, huge sample size, collected nationwide (we'll do the rest of the world later) coded for a ton of things. Income is still the dependent variable, but I want way more independent variables than just gender and race. The things I have thought of so far are
  • number of years experience (0, 1, 2, 5, 10+)
  • degrees held (none, BA, MA, etc)
  • specific job type (e. g. nurse, lawyer, tour guide) 
  • general field (medicine, technology, sportsball)
  • geographic location (city, state)
  • type of area (urban/rural)
  • race
  • disability
  • parental income
  • (these independent variables are getting a little out of hand. We might need to save some of for another project using the same data set.)
  • with/without children
  • single/married
  • height? weight?
  • plus probably some more
I would send out my army of minions (or maybe just use facebook for most of it?) to find out all these things about lots and lots of people. And after the minions had visited thousands and thousands of people,I would take all of that data and put it into a huge, beautiful spreadsheet, and I'd do statistics to it. I'd create a bunch of study sets, ranging from extremely narrow (black nurses with RNs and 5 years experience, in California) to very broad (everyone with 2 years experience at any job). I'd add a column for [modal income minus actual income], and one for [male income minus female income]. Then I'd make another spreadsheet, with all the numbers from the first one, with subgroups as tokens, instead of people. And I'd calculate some things about the subgroups, like their modes and means and mediums.

(Isn't this a realistic project? Among other things, we're going to need a huge travel budget for all the places we have to go and then find sufficiently large numbers of participants for each most-specific subgroup: at least 30 male and 30 female nurses who have RNs and 5 years experience, in Chicago, from each minority category, and also 30 of each with 1 year of experience, and also with whatever other nursing degrees there are. And all the variations of all the other jobs (mechanics, high school teachers, circus performers, etc.)**** Just the search for minions may take a while.)

And then I'd get out my trusty, rusty, dusty old R program on my computer and do some statistics to my statistics and create some visual representations! I have been thinking about what kind of graph would fit best for various questions. Lines? Scatter plot? Chopped up dollars? The ones I've been thinking about would have groups of people on the x axis and plot the income disparities on y. (I also want to know what would happen if we put  modal incomes on x and looked at race and gender, and if we put actual income on x and looked at disparity. But I haven't been thinking about how to make graphs of those, yet).

For example: We want to know which makes more of an impact on a nurse's salary, experience or education, and if is it the same for men and women. We would look at the education and experience of nurses of all races and abilities, nationwide and sort them into groups twice, first by how many degrees (with each degree group split by gender), and again by years of experience (same). There will be 50 groups each time, for a total of 100 groups. Now we have a study set! Each line on our spreadsheet will represent one of those subgroups of nurses like this: [male, 0 years, RN];  [male, 1 year , RN]; [male, 2 years, RN]. And we'll get, from our other spreadsheet, the modes and medians and means (let's be very thorough) for each subgroup, and put them in the new spreadsheet. Now it's time to make a graph!

This super useful and precise example graph has all our subgroups on the x axis, and the mean difference between modal and actual incomes is plotted on y. Like this!

This graph bears no relation to reality; I made up the point placement out of thin air and was too lazy to make up what units to use- dollars? percentages? standard deviations?
Then I'd look at my graph and see if it seems interesting in any way, like something that might be a line or a shape or any pattern of any kind. If it does, figure out what and write a paper about it. If it doesn't, start over with different variable groups.

I would like to know much more about wage inequity.  Are there exceptions to the general rule? Is there a difference between how much a master's boosts salaries of black teachers' vs white teachers'? Do years of experience and/or degrees ever make up for being disabled?  Which fields of work are the worst? Which state is the best?

And so on. Basically my interest is in looking at the details of the big picture, rather than just the final, single, average. (Also I wish everyone would use more precise language when they talk about the "averages" of things.) "White men" is too heterogeneous to be the baseline. Are there any subgroups of white men where they are paid less than women? If there are, that would be very interesting. Looking at the big picture is great, but you have to check the details first to make sure the big picture isn't hiding anything.

I started thinking about this specifically because, while looking at the dollar picture above, I also thought about the extremely wealthy white men who have obscene salaries because they are CEOs, and how there are more men named John among them than there are women CEOs. That's a different kind of disparity, which is also very disheartening, but not within the scope of this project (don't get distracted!). Anyway the fact that all these CEOs and college presidents and whatever are being paid absurd amounts of money, the disparity might be much larger if you just look at the amount of money, but smaller as a percentage (the difference between $50 and $100 seems way more dramatic than the difference beween $500,000 and $500,050. I think the absurdly high salaries at the far end of the scale might screw up the data. So I was thinking about extreme outliers, and what stuff was accounted for and what wasn't, and where those parts of dollars came from. And for that you need more information about the data than the dollar picture gives you. And here we are!

During one of my many Adventures in Wikipedia, I looked at a lot of studies about this. (This disparity has been known, but not fixed, for a pretty long time, so there are quite a few studies.) Most of the ones I found were pretty careful and targeted specific populations. Many of them match jobs and years of experience. But none of them answer my details-of-the-big-picture questions from up above. And the Telephone game from actual research to media report to striking pictures on Facebook is a lossy, lossy transmission, resulting in the chopped-up dollar picture I am complaining about.

Maybe one of the silver linings of our impending transition to a dictatorship will be that the government will require everyone to report every detail of our lives anyway, and I can sweet-talk the dictator into letting me see the records to make some spreadsheets.

In conclusion, the things I am interested in knowing about income disparity in America are not adequately addressed in the chopped-up-dollar picture.

And now you know a new fact about me: one of the ways I entertain myself while bedridden is to invent unrealistic studies and think about how I would arrange the data in a spreadsheet, and what to do with the data, and what kind of visual representation of the results would work best.


*Some people, with whom I occasionally interact in some way

**Once in a while

***I refuse to treat data as a plural. We're speaking English, damn it! It's a singular mass noun and Latin can keep its stupid inflections to itself.

****Some of these subgroups may be empty sets. Disabled black lawyers with LLMs who live in Montana, for example.

It is really, really stupid that the white men at the very top make as much money as they do.