00:00:08 Mae Macfarlane
Hi, I'm Mae Macfarlane here with the Professor Podcast, the podcast of your professors, their research, and their academic lives here at St. Thomas.
00:00:16 Mae Macfarlane
This week we are delighted to have with us Dr. Amelia McNamara, a professor in the St. Thomas Department of Computer and Information Sciences. Welcome Dr. McNamara.
00:00:26 Amelia Mcnamara
Thanks for having me.
00:00:28 Mae Macfarlane
So, first up, can you briefly explain what your primary research is about.
00:00:34 Amelia Mcnamara
Sure! So, I'm in the Department of Computer and Information Sciences, but my PhD is actually in statistics, so you might know that the stat program at St. Thomas is joint between the Math Department and the computer science department. And I'm on the computer science side and I think that it's a great fit because my research intersects with computer science.
00:00:53 Amelia Mcnamara
So, I'm really interested in making it easier for everyone to do and to understand statistics. And to me, if you're going to do statistics, you need to be using a computer and working with data.
So, a lot of my research has to do with how people use computers, either to do data analysis or how they present their data analysis to a wider audience.
00:01:18 Mae Macfarlane
Now that's really cool.
And I, you know, the kind of purpose of this podcast is to make research that is kind of information heavy, more accessible.
00:01:25 Mae Macfarlane
Can you kind of talk about how your research in statistics and your background plays into everyday life?
00:01:31 Amelia Mcnamara
Ah, yes, yes.
So, I think, you know, that kind of relates to why I like statistics. I think that statistics and data get used in almost every realm of life. Like that's one of the reasons why I think it's really cool and interesting, and I want people to see how it can be relevant to...You know, if you're a biology major, I think, you know, that biologists collect a lot of data and they analyze it and they work with it.
But economists also do that. Every major corporation has a data or a statistics department that is working behind the scenes. If you're an English major, there is data... you can use text as data. I just taught a full semester class about that and you can do analysis that way.
So, I really think that there's data about everything in our lives. And so then that makes statistics really relevant to a lot of different areas.
00:02:27 Mae Macfarlane
I just know from speaking to, like, my peers at St. Thomas, that, like statistics is used in psychology. Yeah, psychology. And that, you know, that's very useful in a lot of different areas.
And you've spoken about, in our pre-session interview about how journalists reach out to you and how it's part of our news. We hear about statistics all the time and how it kind of comes around. Do you wanna talk about that a little bit?
00:02:52 Amelia Mcnamara
And sure, yeah, I mean, I'm really interested in communication as well. So, I think there's a kind of meta connection with the podcast here.
And so, I think one of the ways that you can communicate is with your spoken words. Another way would be with something visual that you could look at. And in terms of statistics, I think about data visualization there. And then, of course, there's the written word as well. And I teach a class on data communication where we cover all three of those modes of communication.
00:03:22 Amelia Mcnamara
But the journalistic written communication is near and dear to my heart. My graduate advisor, or one of my graduate advisors, is involved in data journalism. And so, I go to a fair number of conferences for journalists who are interested in working with data.
And I think that journalism school is changing a little bit now and giving students more context in statistics and data analysis, but that historically has not been the case.
And so, people are often kind of reskilling. And they are sometimes a little bit nervous about their stats skills. And so, because I know people from these conferences, journalists will reach out to me. I don't know If it's really technically on background, because I'm not like a source. But they'll say, “hey, we've done this data analysis and here is our results and we just wanna kind of like run it past you and make sure that everything is watertight.” Because you know they need to check their work and be able to back up their stories.
So, I do that a couple times a year, I'd say a journalist reaches out to me for that so And this is kind of a question for me.
00:04:22 Mae Macfarlane
So, I'm a journalism major. I don't have a huge math background necessarily. What is statistics as a field?
00:04:30 Amelia Mcnamara
That's a great question. Obviously, it's a hard question, as with, you know, defining any field. But the way that I usually put it is that statisticians are really obsessed with two big questions. And one question is, is this number different from zero? And another question is, what are some other values that we might have observed?
So, we usually have data from a sample and so you could think about taking a random sample -- calling people out of the phone book, or if you had a list of all the St. Thomas students you know and going through that roster and just calling people randomly -- we use samples because they're cheaper and easier to access than accessing the whole population.
But we get some sample data. We run some, you know, number crunching and figure out something about it.
But we want to make a conclusion about the greater population -- all the St. Thomas students -- and we often want to know, is there an effect here or not or, you know, what other kinds of things could you have observed?
00:05:29 Amelia Mcnamara
And let me see if I can think of a cool example. I'm working on grading some final projects because final grades are due soon and one of my students this semester was trying to predict how much ski resort passes would cost based on how many, I think It was kilometers. It was in Europe. How many kilometers of intermediate, advanced and beginner slopes they had. So, you know, here's what we observed that relationship was in the sample data.
But what are some other things that we could have observed?
00:06:02 Mae Macfarlane
Thank you for clarifying that. It's one of those things that you hear a lot about might not know the technical vernacular for it.
So, sure, what kind of drew you to statistics and then kind of on to your academic interest and your research with it?
00:06:17 Amelia Mcnamara
I think it's again a kind of a complicated story. I've had an interesting educational path in my life. My first year of college I started out at design school at the University of Cincinnati and, I did a full year of Art Foundation, learned how to draw and paint, and about color theory and 3D design. And you know, digital design. All that kind of stuff
I realized that it didn't actually want to be a designer, so I transferred to Macalester College with the goal of being an English major, and you can probably tell that some of my English major tendencies are still shining through.
And I did end up graduating with that English major, but I also tacked on a math major while I was there, and I had a really influential professor who I did research with and who encouraged me to think of myself as a scholar and someone who could go to graduate school. So, then I decided I wanted to go to graduate school.
00:07:10 Amelia Mcnamara
I think I had the idea that I wanted to be a professor because again, I'm kind of all about communicating things, whether it's visually or with words or with numbers. And it seems like a job in which you get to communicate a lot of stuff. So, I went to grad school with that goal.
And when I applied to graduate school, I applied to programs that were in pure mathematics, applied mathematics, and statistics which I think if you are outside of stats, those sound like they're all the same thing, but if you are a statistician It's like, “whoa! those fields are completely different!” And I ended up going with statistics, mostly because I saw the most applicability in the real world and everyday life --more so than the mathematical fields.
00:07:54 Mae Macfarlane
Can you describe for the average Joe, what is statistics in our lives?
00:08:00 Amelia Mcnamara
Here I mean, I think again one of the things that statisticians care a lot about is variability.
So, I think you experience variability in your life every day, like, “OK, I'm gonna go to the gas pump and the prices are a little bit higher than they were last time.” “Are they kind of trending upward” as I think they have been lately? Or is this just like, “ooh, it's going to be higher for one week and then it's gonna drop back down.”
That might influence your decision making. Statistics and probability are pretty related, so I think when you check the weather and it says there's a 60% chance of rain you're using some probabilistic thinking to say “I'm going to bring my umbrella with me.” Like “I'd rather have my umbrella along if…even if it doesn't rain, than then get caught out without one.”
00:08:43 Amelia Mcnamara
And then I think a lot of people, the way that they interact with statistics the most is they read statistics in journalistic outlets. So, you know, The FiveThirtyEight is saying we have run these statistical models and here are our predictions for the presidential race.
Or, we are talking about a basketball team and why they're on this, like, amazing winning streak.
So, I think that a lot of statistics does get filtered through the news media.
00:09:11 Mae Macfarlane
So, Dr. McNamara, how does your research and like your interest in stats and visualization and everything play into your teaching and in your classes?
00:09:22 Amelia Mcnamara
I mean, I think my research has overlap with the field of statistics education. So, I'm interested in -- again because I want everyone to be able to do and to understand statistics really well -- I am interested in how to make it easier to, you know, to kind of learn statistics and this…
So one thing that I'm really interested in is the programming language R which gets used by a lot of professional statisticians. But some folks think that it is too complicated for college students to learn, and it is really complicated, I'm not going to lie about that. But, I don't think that it's too complicated for college students. And so, some of my research is about making sure that I have prepared materials that are as concise as possible so that we just use a really small vocabulary of functions in that programming language.
You can kind of think of them like verbs. So just a few verbs that we're going to use over and over again throughout this semester.
There's also a lot of different ways to kind of say the same thing in R different…they're kind of like dialects in human language, where if you were trying to learn English you probably wouldn't try to learn with a Southern accent and also a British accent, and also like in New Zealand accent. You would focus on learning one particular accent. You'd be able to listen to and understand other accents, but you would kind of focus on one.
00:10:46 Amelia Mcnamara
So that's my best metaphor for there's like these different syntaxes in R and people have had really strong opinions about this is the one that you should teach in intro stat
But no one had any data about it. So, a few semesters ago I was teaching 2 sections of the intro stats labs for Stat 220, and I thought, “I'm gonna do one in this formula syntax and one in the tidyverse syntax.” And I don't like, I genuinely didn't know which one was going to be better, but I wanted to compare them.
00:11:16 Mae Macfarlane
Was there one that was a little bit more successful?
00:11:17 Amelia Mcnamara
So I would say that there wasn't a ton, there wasn't a ton of difference when I looked at the statistics after the fact. So in terms of a lot of different variables that I looked at, there was no statistically significant difference.
It did look like there was a difference in the amount of time that students were spending on the computing platform depending on which of the syntaxes that they were using. And so in you know since then, in subsequent semesters I've been using the one that had less computing time, thinking like maybe that meant it was easier for students.
But I would love to do some follow-up because I didn't actually ask students like “hey, I noticed that this group was spending more time on the computer working.”
That could be because they were like, struggling and like, this is really hard and I hate it but it could have also been like “this is so cool and I wanna play around,” right?
So, I'm like kind of making up these stories in my head and I didn't actually ask the students to know.
00:12:15 Mae Macfarlane
Yeah. No, that's really interesting. And this is kind of an aside. My partner is a stats student at the U. He's in his master's program right now. And he uses R and his TA students are like “this sucks!” You know, they hate it. They're in the intro and the ones that are computer science majors get it a lot more than the other people.
And it's very interesting. But my roommate is an econ major at St. Thomas, and she uses R all the time and. She's always like this kid in class is not getting it. And I'm always doing all the work and you know.
So, it's really interesting to get and see the different fields using it, but people's reactions to it. And, the only coding kind of computational language I know is HTML, which I know is very different.
It is like using verbs and you just apply them in different ways. it's that's a really good, simplified example of it. It's like conjugating when you're learning…
00:13:11 Amelia Mcnamara
Yeah, I think computer programming is way more picky about syntax than human speech. So I always say, like if you are learning a human language like my second language is Spanish. And I say something in Spanish where I haven't conjugated the verbs quite right. A person is probably going to be like, well, I knew what that noun was and that verb. So, I have a guess as to what she's trying to ask me.
But, like computers, are not that smart. And so, if you have misspelled something, even by one letter or is capitalization specific. So, if you've capitalized a letter that's supposed to be lowercase. And it doesn't work and I know that can be really frustrating for people as they try to learn it.
00:13:51 Mae Macfarlane
In journalism, we always say if everyone needs an editor because when you're writing your own stuff, you don't catch the wrong capitalization or the wrong “then” or “than.”
So you brought, you talked about your art background and kind of using visual communication and your interests in that. Can you talk a little bit more how that plays into your data research and your interest in that?
00:14:13 Amelia Mcnamara
That so I have this paper that I wrote a few years ago now and it's called “Key Attributes of a Modern Statistical Computing Tool” or something like that.
And so, it's like 10 things that I believe you need to have in a computing tool that's going to be used for doing statistics and data science/data analysis. One of the things is that it needs to be easy for people to get started using. If it's like really high threshold to entry, then no one's going to do it, for example. But then on the other end, it needs to be flexible enough that you can kind of keep using it.
00:14:50 Amelia Mcnamara
Like sometimes people teach something in intro stat that can only be used for that one single course, and then if you want to go on, you have to learn something else. And I find that frustrating too, right? Like, “I sunk a semester into learning this thing, and now you're telling me that that never gets used again?”
00:15:06 Amelia Mcnamara
So I think like the payoff of learning R in your first semester of stats is that you can use it in your next semester of stats and maybe in your ECON class or you know wherever else it's being taught. So, those were some of the attributes.
Another thing that I think is really important and came from that paper is that visualization should be like a key component, because this is almost like the kind of editor thing where if you just start running statistical tests on your data, you might make a huge mistake if you haven't made a plot of that data first.
So, statistics has all of these conditions that are required to be met for your conclusions to be valid. So, sometimes it's like the data needs to be normally distributed, it needs to look like a bell curve. And if you do a plot you might see “oh, it's not a bell curve, it's just, like, two boxes and, like, one is really tall and one is really short. Like this, maybe is not the right analysis method.”
So, some software makes it really easy to make a bunch of plots and look at your data, and some software makes it really hard. And so, then you're just gonna probably jump into doing your data analysis. So, I think that's a piece of it. And I would love to do some research about, you know, different visualization types and how they help people understand things.
00:16:24 Amelia Mcnamara
There's a very rich field in data visualization research that I haven't really dipped my toe into, but I do talk about it like in my data communication class. We talk about all this research about, you know, different graph types and colors and all this perceptual, almost like psychological research about what graphs are the most effective.
00:16:46 Mae Macfarlane
Yeah, when you were just describing that, I was kind of taken back to like my 4th grade class and learning how to do percent math, you know? 90% and using, this might be a niche maybe thing, but there are the little blocks and there's like the 10 strip and then the 100 square and they can have different colors.
And if you have the one little block compared to the big block, it's you know that's 10% of the whatever you know it's the visual -- that type of visualization and really sticks with the child because it's tactile. Listening to you say that it makes a lot of sense of the visualization is important for clarifying data and kind of getting your answers.
00:17:30 Amelia Mcnamara
I mean, this isn't my research, but there's a paper that I really love about... So, there's a probability formula called Bayes rule which has to do with conditional probability and it is like these formulas to help you figure out. Basically like complicated fractions and it often doesn't make sense in people's brains. The, like, most common example is about a particular kind of disease and how how good the testing is for that disease. So for like, a COVID test, what is the false positive rate versus the false negative rate?
If you get a positive test, what are the chances that you actually have COVID? And it turns out people are really bad at reasoning about that kind of conditional probability. But there's an amazing paper that basically says that those same -- almost like those blocks -- Like if you make a visualization…that is these rectangles, it helps basically everyone understand Bayesian reasoning. And I think the paper is like “Improving Bayesian Reasoning Without Instruction.” Like literally, you don't have to teach people anything, you just show them the picture. And they're like, “oh II get it!”
00:18:35 Mae Macfarlane
Visualization just really does make a difference. It's so fascinating.
00:18:38 Amelia Mcnamara
I mean, I think there's exceptions, right? Like some people aren't visual learners. Some people are blind.
But I think that for a lot of folks, it is a very effective way of communicating information.
00:18:49 Merrie Davidson
I think they've found that people aren't so different in the visualization and auditory learning kind of thing. It's just kind of a..
I watched Dope Sick. I don't know if you've seen it.
It's a Michael Keaton movie about the opioid epidemic and how Purdue Pharma pushed Oxycontin. And one of the things they did was to make a graph of how consistent the level of medication is in people’s systems, and they made a logarithmic Y axis. And so it looked flat when it really actually had the same dips that any other opioid.
00:19:23 Amelia Mcnamara
Yeah, I think there's like all kinds of ways that graphs can mislead people, either intentionally or unintentionally. And we talked about this in my data communication class as well. Like a lot of times, people who are, like, really into research, love log scales because it lets you show, “OK, this is about 10, this is about 100 a 1000.”
It sort of lets you show those orders of magnitude. But I think the average person, when they look at a graph, they expect that the scale is linear. And they maybe even if they see the labels, they might not process that it the data has been transformed in that way.
00:19:58 Amelia Mcnamara
And then I think you were mentioning some like news programs that show graphs, a lot of the cable news get caught out. And again, is it intentional or is it just because they're working on a really tight deadline?
But you'll see things on social media of you know, it's a bar chart, but the axis doesn't start at 0, so it's trying to make the difference between like 56 and 57 look enormous when it's just like one number off.
00:20:23 Mae Macfarlane
No, that's a very important point. And this kind of goes into the like misunderstanding of graphs.
And I think that it's important for labeling, you know, even if someone is blind, like you mentioned earlier. So visualization is not necessarily part of what they're understanding, but having correct descriptions of things is really important and for the consumer of statistics and data for people to understand what they're looking at, if they don't come from the background of, say, working at the CDC or working in agriculture or whatever, or reading the statistics that they're being presented..
Is there any accessibility research being done for people that say, they have English isn’t their first language, and but they're living here in the United States and trying to intake all this information regarding COVID especially.
Or is there any part of the statistics community I guess at large that is working for accessibility within communicating this information?
00:21:19 Amelia Mcnamara
Yes, there definitely is. In terms of other like human languages and people who don't speak English, one of the first things that comes to mind for me is actually not related to visualization, but related to like software and programming.
So basically every programming language, whether it's R or C or HTML or JavaScript or whatever programming language you can think of, It's based on English. So if you are writing a function name in R, it's like “mean,” “median,” “mode,” right.?
Those are the English words for those statistics. And so there is like this higher barrier, cognitive load, for people for whom English is their second language.
And there have been some movements to have programming languages that are in someone’s native human language. Felina Armans is a researcher in that field who I think is doing some awesome work.
And then in terms of disability and accessibility, I actually collaborate with some researchers -- Andrea Stefik is kind of a principal investigator on this grant that I'm collaborating on and he has written a programming language called Quorum which is super cool because it is all evidence-based. Every decision that they've made in the programming language has been based on doing a study versus like are people just made it up?
They're like, I think this is how it should be and wasn't based on anything.
So quorum is all evidence-based and Quorum has been designed from day one to be super accessible to people who are blind and low vision. So it works really well with a screen reader. It has less punctuation than a lot of programming languages do.
And then the reason why I am sort of brought on as a consultant on this grant is they're adding data science functionality to Quorum and they're like, “OK, graphs are super Important how are we going to put those in for people who are blind or low vision?” And I'm not an expert in that, but we have folks who are and, you know, we're talking about what should the screen reader say.
How should you navigate this graphic? Should bar chart and a pie chart are sort of displaying the same data like does the screen reader need to say something particular that's different?
00:23:22 Amelia Mcnamara
And then there's also, and I know less about this, but there's a team associated with that grant that's also doing haptic feedback.
So you could kind of tactically feel the graph through vibrations on a tablet or a smartphone.
00:23:34 Mae Macfarlane
That's genius. Yeah, no, that's very interesting. And making this information and to create this data, the accessibility to help create it and understand it, that's really cool.
00:23:47 Mae Macfarlane
OK, that was kind of the last clarifying thing I needed for my brain.
Thank you so much for joining us today, Dr. McNamara.
00:23:54 Amelia Mcnamara
Of course, thanks for having me.
00:23:56 Mae Macfarlane
To learn more about Doctor McNamara's work. You can find her on the Saint Paul campus at the University of St. Thomas. The Professor Podcast is brought to you by the St. Thomas libraries and made possible with funding from the College of Arts and Sciences. I'm your host, Mae Macfarlane, a 2022 graduate.
00:24:12 Mae Macfarlane
The producers and library staff are Merrie Davidson, Andrea Keoppe and Trent Brager.
Thank you so much to our guests. And you, our listeners.
0 Comments.