Blog Guilt and a Categorical Data Course

I read a blog post entitled On Not Writing and it felt a little close to home. The author, an academic who is in a non-tenure position, writes,

If you have the luxury to have time to write, do you write scholarship with the hope of forwarding an academic career, or do you write something you might find more fun, and hope to publish it another way?*

The footnote read, “Of course, all of this writing presupposes that the stacks of papers get graded.” Ouch. Too close to home. I sent this on to some of my non-tenure track peers and Rob responded that I had tapped into his blog guilt. My blog guilt had been at an all time high already, and so I vowed that I would immediately post something to Citizen Statistician. Well, that was several weeks ago, but I am finally posting.

Fall semester I taught a PhD seminar on categorical data analysis that I had proposed the previous spring. As with many first-time offerings, the amount of work was staggering and intellectually wonderful. The course notes, assignments, etc. are all available at the course website (which also doubled as the syllabus).

The course, like so many advanced seminars, had very few students actually take the course for a grade, but had quite a few auditors. The course projects were a blast to read and resulted in at least two pre-dissertation papers, a written prelim paper, and so far, two articles that have been submitted to journals!

After some reflection, there are some things I will do differently when I teach this again (likely an every-other-year offering):

  • I would like to spend more time on the classification methods. Although we talked about them a little, the beginning modeling took waaaay more time than I anticipated and I need to re-think that a bit.
  • I would like to cover mixed-effects models for binary outcomes in the future. This wasn’t possible this semester since we only had a regression course as the pre-requisite. Now, there is a new pre-requisite which includes linear mixed-effects models with continuous outcomes, so at least students will have been exposed to those types of models. This course also includes a much more in-depth introduction to likelihood, so that should also open up some time.
  • I will not teach the ordinal models in the future. Yuck. Disaster.
  • I probably won’t use the Agresti book in the future. While it is quite technical and comprehensive, it is expensive and the students did not like it for the course. I don’t know what I will use instead. Agresti will remain on a resources list.
  • The propensity score methods (PSM) were a hit with the students and those will be included again. I will also probably put together an assignment based on those.
  • I would like to add in survival analysis.

There are a ton of other topics that could be cool, but with limited time they probably aren’t feasible. I think in general my thought was to spend the first half of the course on introducing and using the logistic and multinomial models and the second half of the course on advanced applications (PSM, classification, etc.)

If anyone has any great ideas or suggestions, please leave comments. Also, I am always on the lookout for some datasets that were used in journal articles or are particularly relevant.

 

 

Data Analysis and Statistical Inference starts tomorrow on Coursera

It has been (and still is) lots of work putting this course together, but I’m incredibly excited about the opportunity to teach (and learn from) the masses! Course starts tomorrow (Feb 17, 2014) at noon EST.

coursera_dasi

A huge thanks also goes out to my student collaborators who helped develop, review, and revise much of the course materials (and who will be taking the role of Community TAs on the course discussion forums) and to Duke’s Center for Instructional Technology who pretty much runs the show.

This course is also part of the Reasoning, Data Analysis and Writing Specialization, along with Think Again: How to Reason and Argue and English Composition 1: Achieving Expertise. This interdisciplinary specialization is designed to strengthen students’ ability to engage with others’ ideas and communicate productively with them by analyzing their arguments, identifying the inferences they are drawing, and understanding the reasons that inform their beliefs. After taking all three courses, students complete an in-depth capstone project where they choose a controversial topic and write an article-length essay in which they use their analysis of the data to argue for their own position about that topic.

Let’s get this party started!

JMM 2014

Two weeks ago I traveled to Baltimore to the Joint Mathematics Meetings. These meetings are very much like the Joint Statistics Meetings except for mathematicians. “Now, um, usually I don’t do this but uh….Go head’ on and break em off wit a lil’ preview of the remix….” (Kelly, 2003).

The JMM are a great place to educate and work with mathematics teachers at the collegiate level who are teaching introductory statistics courses. One group that is quite active in this community is the Statistics Education Special Interest Group of the Mathematical Association of America (SIGMAA). If you are a member of the MAA, let me put in a plug to join this SIGMAA. Each year they sponsor at least one contributed paper session and often several minicourses.

This year, aside from the perennial Teaching introductory statistics (for instructors new to teaching intro stats minicourse, the SIGMAA also endorsed two minicourses aimed at using randomization/bootstrapping in the introductory course, CATALST: Introductory statistics using randomization and bootstrap methods and Using randomization methods to build conceptual understanding of statistical inference. Both mini courses were well attended and will likely be offered again next January.

JMM-2014-Minicourse-Nicola

Nicola during the CATALST minicourse.

The SIGMAA also sponsored a Contributed Paper Session entitled, Data, Modeling, and Computing in the Introductory Statistics Course. The marathon session, running from 1:00pm–6:00pm, was very well attended and included 15 presentations.

Nick-Horton

Nick Horton gives the paper, Big Data in the Intro Stats Class: Use of the Airline Delays Dataset to Expose Students to a Real-World, Complex Dataset by himself, Ben Baumer, and Hadley Wickham.

One of my favorite things at JMM is attending the SIGMAA Stat-Ed Business Meeting. This took place immediately following the CPS, so we were able to capitalize on inviting many of the attendees to join us. After eating what might have been the best spread of food I have encountered at one of these meetings, we had our meeting.

The SIGMAA presents two awards during these meetings.

The Dex Whittinghill Award is presented to the first author of the paper that receives the highest evaluations during the CPS session from the previous JMM. This year, it was presented to Kari Lock-Morgan of Duke University (who was unable to be there, but sent her heartfelt thanks via her parents).

The Robert V. Hogg Award for excellence in teaching introductory statistics was presented to Johanna Hardin of Pomona College. Johanna’s colleague, Gizem Karaali, gave a heartwarming talk when presenting Johanna the award.

IMG_3772

Scott Albers, SIGMAA chair, congratulates Johanna Hardin on winning the Robert V. Hogg Award

IMG_3776

Gizem Karaali reads a heartwarming note from the Johanna’s colleagues.

 

References

Kelly, R. (2003). Ignition (remix). On Chocolate factory. Chicago: Jive, Sony.

Conditional probabilities and kitties

I was at the vet yesterday, and just like with any doctor’s visit experience, there was a bit of waiting around — time for re-reading all the posters in the room.

vodka

And this is what caught my eye on the information sheet about feline heartworm (I’ll spare you the images):

cond

The question asks: “My cat is indoor only. Is it still at risk?”

The way I read it, this question is asking about the risk of an indoor only cat being heartworm positive. To answer this question we would want to know P(heartworm positive | indoor only).

However the answer says: “A recent study found that 27% of heartworm positive cats were identified as exclusively indoor by their owners”, which is P(indoor only | heartworm positive) = 0.27.

Sure, this gives us some information, but it doesn’t actually answer the original question. The original question is asking about the reverse of this conditional probability.

When we talk about Bayes’ theorem in my class and work through examples about sensitivity and specificity of medical tests, I always tell my students that doctors are actually pretty bad at these, looks like I’ll need to add vets to my list too!

R Syntax for Ranked Choice Voting

I have gotten several requests for the R syntax I used to analyze the ranked-choice voting data and create the animated GIF. Rather than just posting the syntax, I thought I might write a detailed post describing the process.

Reading in the Data

The data is available on the Twin Cities R User Group’s GitHub page. The file we are interested in is 2013-mayor-cvr.csv. Clicking this link gets you the “Display” version of the data. We actually want the “Raw” data, which is viewable by clicking View Raw. The link is using a secure connection (https://) which R does not handle well without some workaround.

One option is to use the getURL() function from the RCurl library. The text= argument in the read.csv() function reads the data in using a text connection, and is necessary to not receive an error.

library(RCurl)
url = getURL("https://raw.github.com/tcrug/ranked-choice-vote-data/master/2013-mayor-cvr.csv")
vote = read.csv(text = url)

A quick look at the data reveal that the three ranked choices for the 80,101 voters are in columns 2, 3, and 4. The values “undervote” and “overvote” are ballot also need to be converted to “NA” (missing). The syntax below reduces the data frame to the second, third and fourth columns and replaces “undervote” and “over vote’ with NAs.

vote = vote[ , 2:4]
vote[vote == "undervote"] = NA
vote[vote == "overvote"] = NA

The syntax below is the main idea of the vote counting algorithm. (You will need to load the ggplot library.) I will try to explain each line in turn.

nonMissing = which(vote[ , 1] != "")
candidates = vote[nonMissing, 1]
#print(table(candidates))

vote[ , 1] =  factor(vote[ , 1], levels = rev(names(sort(table(vote[ , 1]), decreasing=TRUE))))
mayor = levels(vote[ , 1])
candidates = vote[nonMissing, 1]

p = ggplot(data = data.frame(candidates), aes(x = factor(candidates, levels = mayor))) +
	geom_bar() +
	theme_bw() +
	ggtitle("Round 1") +
	scale_x_discrete(name = "", drop = FALSE) +
	ylab("Votes") +
	ylim(0, 40000) +
	coord_flip()

ggsave(p, file = "~/Desktop/round1.png", width = 8, height = 6)
  • Line 1: Examine the first column of the vote data frame to determine which rows are not missing.
  • Line 2: Take the candidates from the first column and put them in an object
  • Line 3: Count the votes for each candidate
  • Line 5: Coerce the first column into a factor (it is currently a character vector) and create the levels of that factor so that they display in reverse order based on the number of votes. This is important in the plot so that the candidates display in the same order every time the plot is created.
  • Line 6: Store the levels we just created from Line #5 in an object
  • Line 7: Recreate the candidates object (same as Line #2) but this time they are a factor. This is so we can plot them.
  • Line 8–16: Create the bar plot
  • Line 18: Save the plot onto your computer as a PNG file. In my case, I saved it to the desktop.

Now, we will create an object to hold the round of counting (we just plotted the first round, so the next round is Round 2). We will also coerce the first column back to characters.

j = 2
vote[ , 1] = as.character(vote[ , 1])

The next part of the syntax is looped so that it repeats the remainder of the algorithm, which essentially is to determine the candidate with the fewest votes, remove him/her from all columns, take the second and third choices of anyone who voted for the removed candidate and make them the ‘new’ first and second choices, recount and continue.

while( any(table(candidates) >= 0.5 * length(candidates) + 1) == FALSE ){
	leastVotes = attr(sort(table(candidates))[1], "names")
	vote[vote == leastVotes] = NA
	rowNum = which(is.na(vote[ , 1]))
	vote[rowNum, 1] = vote[rowNum, 2]
	vote[rowNum, 2] = vote[rowNum, 3]
	vote[rowNum, 3] = NA
	nonMissing = which(vote[ , 1] != "")
	candidates = vote[nonMissing, 1]
	p = ggplot(data = data.frame(candidates), aes(x = factor(candidates, levels = mayor))) +
		geom_bar() +
		theme_bw() +
		ggtitle(paste("Round", j, sep =" ")) +
		scale_x_discrete(name = "", drop = FALSE) +
		ylab("Votes") +
		ylim(0, 40000) +
		coord_flip()
	ggsave(p, file = paste("~/Desktop/round", j, ".png", sep = ""), width = 8, height = 6)
	j = j + 1
	candidates = as.character(candidates)
	print(sort(table(candidates)))
	}

The while{} loop continues to iterate until the criterion for winning the election is met. Within the loop:

  • Line 2: Determines the candidate with the fewest votes
  • Line 3: Replaces the candidate with the fewest votes with NA (missing)
  • Line 4: Stores the row numbers with any NA in column 1
  • Line 5: Takes the second choice for the rows identified in Line #4 and stores them in column 1 (new first choice)
  • Line 6: Takes the third choice for the rows identified in Line #4 and stores them in column 2 (new second choice)
  • Line 7: Makes the third choice for the rows identified in Line #4 an NA
  • Line 8–18: Are equivalent to what we did before (but this time they are in the while loop). The biggest difference is in the ggsave() function, the filename is created on the fly using the object we created called j.
  • Line 19: Augment j by 1
  • Line 20: Print the results

Creating the Animated GIF

There should now be 35 PNG files on your desktop (or wherever you saved them in the ggsave() function). These should be called round1.png, round2.png, etc. The first thing I did was rename all of the single digit names so that they were round01.pnground02.png, …, round09.png.

Then I opened Terminal and used ImageMagick to create the animated GIF. Note that in Line #1 I move into the folder where I saved the PNG files. In my case, the desktop.

cd ~/Desktop
convert -delay 50 round**.png animation.gif

The actual animated GIF appears on the previous Citizen Statistician post.

Ranked Choice Voting

The city of Minneapolis recently elected a new mayor. This is not newsworthy in and of itself, however the method they used was—ranked choice voting. Ranked choice voting is a method of voting allowing voters to rank multiple candidates in order of preference. In the Minneapolis mayoral election, voters ranked up to three candidates.

The interesting part of this whole thing was that it took over two days for the election officials to declare a winner. It turns out that the official procedure for calculating the winner of the ranked-choice vote involved cutting and pasting spreadsheets in Excel.

The technology coordinator at E-Democracy, Bill Bushey, posted the challenge of writing a program to calculate the winner of a ranked-choice election to the Twin Cities Javascript and Python meetup groups. Winston Chang also posted it to the Twin Cities R Meetup group. While not a super difficult problem, it is complicated enough that it can make for a nice project—especially for new R programmers. (In fact, our student R group is doing this.)

The algorithm, described by Bill Bushey, is

  1. Create a data structure that represents a ballot with voters’ 1st, 2nd, and 3rd choices
  2. Count up the number of 1st choice votes for each candidate. If a candidate has 50% + 1 votes, declare that candidate the winner.
  3. Else, select the candidate with the lowest number of 1st choice votes, remove that candidate completely from the data structure, make the 2nd choice of any voter who voted for the removed candidate the new 1st choice (and the old 3rd choice the new 2nd choice).
  4. Goto 2

As an example consider the following sample data:

Voter  Choice1  Choice2  Choice3
    1    James     Fred    Frank
    2    Frank     Fred    James
    3    James    James    James
    4    Laura 
    5    David 
    6    James              Fred
    7    Laura
    8    James
    9    David    Arnie
   10    David

In this data, James has the most 1st choice votes (4) but it is not enough to win the election (a candidate needs 6 votes = 50% of 10 votes cast + 1 to win). So at this point we determine the least voted for candidate…Frank, and delete him from the entire structure:

Voter  Choice1  Choice2  Choice3
    1    James     Fred    <del>Frank</del>
    2    <del>Frank</del>     Fred    James
    3    James    James    James
    4    Laura 
    5    David 
    6    James              Fred
    7    Laura
    8    James
    9    David    Arnie
   10    David

Then, the 2nd choice of any voter who voted for Frank now become the new “1st” choice. This is only Voter #2 in the sample data. Thus Fred would become Voter #2′s 1st choice and James would become Voter #2′s 2nd choice:

Voter  Choice1  Choice2  Choice3
    1    James     Fred
    2     Fred    James
    3    James    James    James
    4    Laura 
    5    David 
    6    James              Fred
    7    Laura
    8    James
    9    David    Laura
   10    David

James still has the most 1st choice votes, but not enough to win (he still needs 6 votes!). Fred has the fewest 1st choice votes, so he is eliminated, and his voter’s 2nd and 3rd choices are moved up:

Voter  Choice1  Choice2  Choice3
    1    James
    2    James
    3    James    James    James
    4    Laura 
    5    David 
    6    James              
    7    Laura
    8    James
    9    David    Laura
   10    David

James now has five 1st choice votes, but still not enough to win. Laura has the fewest 1st choice votes, so she is eliminated, and her voter’s 2nd and 3rd choices are moved up:

Voter  Choice1  Choice2  Choice3
    1    James
    2    James
    3    James    James    James
    4     
    5    David 
    6    James              
    7    
    8    James
    9    David    Laura
   10    David

James retains his lead with five first place votes…but now he is declared the winner. Since Voter #4 and #7 do not have a 2nd or 3rd choice vote, they no longer count in the number of voters. Thus to win, a candidate only needs 5 votes = 50% of the 8 1st choice votes + 1.

The actual data from Minneapolis includes over 80,000 votes for 36 different candidates. There are also ballot issues such as undervoting and overvoting. This occurs when voters give multiple candidates the same ranking (overvoting) or do not select a candidate (undervoting).

The animated GIF below shows the results after each round of elimination for the Minneapolis mayoral election.

2013 Mayoral Race

The Minneapolis mayoral data is available on GitHub as a CSV file (along with some other smaller sample files to hone your programming algorithm). There is also a frequently asked questions webpage available from the City of Minneapolis regarding ranked choice voting.

In addition you can also listen to the Minnesota Public Radio broadcast in which they discussed the problems with the vote counting. The folks at the  R Users Group Meeting were featured and Winston brought the house down when commuting on the R program that computed the winner within a few seconds said, “it took me about an hour and a half to get something usable, but I was watching TV at the time”.

See the R syntax I used here.

 

Personal Data Apps

Fitbit, you know I love you and you’ll always have a special place in my pocket.  But now I have to make room for the Moves app to play a special role in my capture-the-moment-with-data existence.

Moves is an ios7 app that is free.  It eats up some extra battery power and in exchange records your location and merges this with various databases and syncs it up to other databases and produces some very nice “story lines” that remind you about the day you had and, as a bonus, can motivate you to improved your activity levels.  I’ve attached two example storylines that do not make it too embarrassingly clear how little exercise I have been getting. (I have what I can consider legitimate excuses, and once I get the dataset downloaded, maybe I’ll add them as covariates.)  One of the timelines is from a day that included an evening trip to Disneyland. The other is a Saturday spent running errands and capped with dinner at a friend’s.  Its pretty easy to tell which day is which.

movings1movings2

But there’s more.  Moves has an API, thus allowing developers to tap into their datastream to create apps.  There’s an app that exports the data for you (although I haven’t really had success with it yet) and several that create journals based on your Moves data.  You can also merge Foursquare, Twitter, and all the usual suspects.

I think it might be fun to have students discuss how one could go from the data Moves collects to creating the storylines it makes.  For instance, how does it know I’m in a car, and not just a very fast runner?  Actually, given LA traffic, a better question is how it knows I’m stuck in traffic and not just strolling down the freeway at a leisurely pace? (Answering these questions requires another type of inference than what we normally teach in statistics. )  Besides journals, what apps might they create with these data and what additional data would they need?

The Future of Inference

We had an interesting departmental seminar last week, thanks to our post-doc Joakim Ekstrom, that I thought would be fun to share.  The topic was The Future of Statistics discussed by a panel of three statisticians.  From left to right in the room: Songchun Zhu (UCLA Statistics), Susan Paddock (RAND), and Jan DeLeeuw (UCLA Statistics).  The panel was asked about the future of inference: waxing or waning.

The answers spanned the spectrum from “More” to “Less” and did so, interestingly enough, as one moved left to right in order of seating.  Songchun staked a claim for waxing, in part because  he knows of groups that are hiring statisticians instead of computer scientists because statisticians’ inclination to cast problems in an inferential context makes them more capable of finding conclusions in data, and not simply presenting summaries and visualizations.  Susan felt that it was neither waxing nor waning, and pointed out that she and many of the statisticians she knows spend much of their time doing inference.  Jan said that inference as an activity belongs in the substantive field that raised the problem.  Statisticians should not do inference.  Statisticians might, he said, design tools to help specialists have an easier time doing inference. But the inferential act itself requires intimate substantive knowledge, and so the statistician can assist, but not do.

I think one reason that many stats educators might object to this because its hard to think of how else to fill the curriculum.  That might have been an issue when most students took a single Introductory course in their early twenties and then never saw statistics again.  But now we must think of the long game, and realize that students begin learning statistics early.  The Common Core stakes out one learning pathway, but we should be looking ahead, and thinking of future curricula, since the importance of statistics will grow.

If statistics is the science of data, I suggest we spend more time thinking about how to teach students to behave more like scientists.  And this means thinking seriously about how we can  develop their sense of curiosity.  The Common Core introduces the notion of a ‘statistical question’– a question that recognizes variability.  To the statisticians reading this, this needs no more explanation.  But I’ve found it surprisingly difficult to teach this practice to math teachers teaching statistics.  I’m not sure, yet, why this is.  Part of the reason might be that in order to answer a statistical question such as “What is the most popular favorite color in this class” we must ask the non-statistical question “What is your favorite color.”  But there’s more to it than that.  A good statistical question isn’t as simple as the one I mentioned, and leads to discovery beyond the mere satisfaction of curiosity.  I’m reminded of the Census at Schools program that encouraged students to become Data Detectives.

In short, its time to think seriously about teaching students why they should want to do data analysis.  And if we’re successful, they’ll want to learn how to do inference.

So what role does inference play in your Ideal Statistics Curriculum?

Should Programming Count as a “Foreign Language”?

I re-hashed this blog post title from the Edutopia article, Should Coding be the “New Foreign Language” Requirement? Texas legislators just answered this question with “Yes”. I hope Minnesota doesn’t follow suit.

Now, in all fairness, I need to disclose that when I taught high school, the Math department played a practical joke on the Languages department by faking a document that claimed that mathematics would be accepted as a foreign language requirement and then conveniently dropping the document outside the classroom door of the Spanish teacher. The ensuing result had the faculty laughing for weeks.

But, I would have no more stood up for mathematics fulfilling a foreign language requirement than computer science fulfilling the same requirement. I think a better substitution however is that computer science should count as fulfilling a mathematics requirement!

The authors of the Edutopia blog write,

In terms of cognitive advantages, learning a system of signs, symbols and rules used to communicate — that is, language study — improves thinking by challenging the brain to recognize, negotiate meaning and master different language patterns. Coding does the same thing.

Substitute the word “mathematics” for “language study” in the previous paragraph and in my mind, it is an even better sell.

While I hope coding does not replace foreign language, I am glad that it is receiving its time in the spotlight. And, I hope the statistics community can use this to its advantage. This is perhaps the perfect route for building on the success of AP statistics…statistical computing. The combined sexiness (sorry Mr. Varian!) of statistics and coding would be amazing (p < .000001) and would be beneficial to both disciplines.

City Hall and Data Hunting

The L.A. Times had a nice editorial on Thursday (Oct 30) encouraging City Hall to make its data available to the public.  As you know, fellow Citizens, we’re all in favor of making data public, particularly if the public has already picked up the bill and if no individual’s dignity will be compromised.  For me this editorial comes at a time when I’ve been feeling particularly down about the quality of public data.  As I’ve been looking around for data to update my book and for the Mobilize project, I’m convinced that data are getting harder, and not easier. to find.

More data sources are drying up, or selling their data, or using incredibly awkward means for displaying their public data.  A basic example is to consider how much more difficult it is to get, say, a sample of household incomes from various states for 2010 compared to the 2000 census.

Another example is gasbuddy.com, which has been one of my favorite classroom examples.  (We compare the participatory data in gasbuddy.com, which lists prices for individual stations across the U.S., with the randomly sampled data the federal government provides, which gives mean values for urban districts. One data set gives you detailed data, but data that might not always be trustworthy or up-to-date. The other is highly trustworthy, but only useful for general trends and not for, say, finding the nearest cheapest gas. )  Used to be you could type in a zip code and have access to a nice data set that showed current prices, names and locations of gas stations, dates of the last reported price, and the username of the person who reported the price.  Now, you can scroll through an unsorted list of cities and states and get the same information only for the 15 cheapest and most expensive stations.

About 2 years ago I downloaded a very nice, albeit large, data set that included annual particulate matter ratings for 333 major cities in the US.  I’ve looked and looked, but the data.gov AirData site now requires that I enter the name of each city in one at a time, and download very raw data for each city separately.  Now raw data are good things, and I’m glad to see it offered. But is it really so difficult to provide some common sensically aggregated data sets?

One last example:  I stumbled across this lovely website, wildlife crossing, which uses participatory sensing to maintain a database of animals killed at road crossings.  Alas, this apparently very clean data set is spread across 479 separate screens.  All it needs is a “download data” button to drop the entire file onto your hard disk, and they could benefit from many eager statisticians and wildlife fans examining their data.  (I contacted them and suggested this, and they do seem interested in sharing the data in its entirety. But it is taking some time.)

I hope Los Angeles, and all governments, make their public data public. But I hope they have the budget and the motivation to take some time to think about making it accessible and meaningful, too.