Subscribe
Share
Share
Embed
A podcast on statistical science and clinical trials.
Explore the intricacies of Bayesian statistics and adaptive clinical trials. Uncover methods that push beyond conventional paradigms, ushering in data-driven insights that enhance trial outcomes while ensuring safety and efficacy. Join us as we dive into complex medical challenges and regulatory landscapes, offering innovative solutions tailored for pharma pioneers. Featuring expertise from industry leaders, each episode is crafted to provide clarity, foster debate, and challenge mainstream perspectives, ensuring you remain at the forefront of clinical trial excellence.
Judith: Welcome to Berry's In the
Interim podcast, where we explore the
cutting edge of innovative clinical
trial design for the pharmaceutical and
medical industries, and so much more.
Let's dive in.
Scott Berry: Uh, welcome
everybody back to, in the interim,
I'm your host, Scott Berry.
I have a really, uh, uh,
interesting guest today.
Uh, I will introduce him, but I
wanna mention this podcast is jointly
sponsored by the A SA and the Journal of
Statistics and Data Science Education.
Today we're gonna talk with Jim Albert.
Uh, Jim Albert is emeritus.
Professor from Bowling Green State
University, he's done a, a incredible
amount of really cool work in
Bayesian, uh, statistical computing
and sports and statistical education.
And so we're gonna talk today to Jim
about teaching statistics using sports.
Welcome Jim.
Jim Albert: Thanks for having me.
Scott Berry: So it.
It.
My interest in statistics largely grew
outta sports, uh, watching sports,
understanding and understanding
the game, forecasting its strategy.
it, to this day, much of what I do in, in
clinical trials, I, I tie to, to sports.
What's your introduction to sports?
Your introduction to statistics and the,
and the, the, the tying the two together.
Jim Albert: Well, I grew up in
Philadelphia suburbs and uh, you
know, I liked to always like sports.
I mean, I was always was a baseball fan
and I played, um, a number of sports.
I liked basketball, I tried baseball.
Wasn't real good at it.
Uh, my tennis, I, I was born in a
tennis family, so played a lot of tennis
and so it was sorta, and I always was
good at math, although I didn't know.
What that would eventually lead to.
And so, you know, I played, uh, baseball
simulation games like Strat Amatic and
All Star Baseball, and I kept stats
in our basement and I typed up, you
know, I kept records of games and I
had standings and things like that.
So, uh, yeah, I, I think long ago
I was trying to combine my interest
in math with my interest in sports.
Scott Berry: uh, so, and, and you
still play tennis to this day.
Jim Albert: Yeah, I'm very lucky.
Uh, my body is held up and, uh, the US I
I have a lot of local people my age who
like to play and they, I'm still active in
USTA, um, which is the tennis Association.
We have leagues and, uh, it's, it's fun.
Yeah, it's a nice, I really, I think,
I don't, that's how he exercise.
I play tennis.
Yeah.
Scott Berry: Yeah.
Yeah.
Uh, very good.
So, so Jim is.
a distinguished university professor from
bowling at Bowling Green State University.
Uh, he's won the Founders Award from
the American Statistical Association,
fellow of the American Statistical
Association, uh, won the contributor
for Statistics and Sports Award.
So that's, that's sort of
the backdrop of all of this.
When, when did you start and,
and go back to when you started
doing statistics in sports?
It was incredibly different era than now.
Now sports analytics is huge,
but you started doing this long
before this was an industry.
Jim Albert: Well, I was sort of
lucky because I was in a, I mean,
obviously I got hired at Bowling Green.
That was my first job.
In fact, my only job in academia,
and really honestly, although I
liked sports and I could think of
good ways of applying statistics.
That was not the way.
Writing sports papers was
not the way to get tenure.
So obviously I started
doing Bayesian research.
Uh, I would throw in some baseball
examples in my as, as illustrations, but
I never really focused on sports problems.
But once I got tenure, then I
started to think a little more.
I wasn't really in a consulting
environment, so when I was thinking
about applying Bayesian methodology,
I, I just naturally thought
about sports as the application.
So then I started writing, and of
course I gotta mention your, your
father, Don Berry, uh, was someone I,
I looked, I really, I really, um, um,
really thought he did a remarkable work
in Bayesian statistics, but also he
loved sports and he often used sports
to illustrate statistical thinking.
Um, we had him speak, in fact at a
meeting in 1987 at Bowling Green,
where he talked about, you know.
About problems in, uh, in
sports, you know, in using stats.
Scott Berry: Yeah, and I, I, I think
that was a, a, a banner year in
1987 where the Minnesota Twins won
their first World series, uh, which
was, which was a, a great year.
So I, so numerous articles in baseball.
Bayesian.
Let's talk a little bit about teaching.
It, uh, did you, uh, you started teaching,
was it something that you did a lot of
sports examples was something I did.
So initially in, in teaching you,
you brought in sports examples.
Is that the introduction to the tying
together the education with sports?
Jim Albert: I think sports
is a natural way of talking.
The nice thing about sports is that
people understand, a lot of the
students understand the context and so
I think especially intro stats where
they don't really understand why.
Why statistics is useful, but they
do understand things about sports.
And so I thought, I said,
why don't we make the course?
since the students don't quite follow
the applications we thought might
find interesting, why don't we just
devote the whole course to sports?
And so I, developed a, intro stats, a
baseball version of Intro Stats course.
And, that was a fun course.
Unfortunately, I didn't get to teach
it a lot, but I got to write a book.
I got a NSF grant for it.
And.
Really, it was fun because you
would walk in and you would just
talk, you would just talk sports.
You would just talk baseball and you would
introduce the, the students to historical
people like Babe Ruth and BA Barry
Bonds and, Whitey Ford and Sandy Koufax.
And and then you would, of course, you
always would bring in statistical ideas,
but they were, that never was though.
You would always talk,
in the context of sports.
And I think it's funny
because students afterwards.
Thought the course was useful, which
of course is a little silly to say
because talking about baseball, I think
what they were saying was they really,
they understood a little more what
we were talking about because we were
doing it in the context of baseball.
Scott Berry: Yeah, and I, I, part
of it is the, the real application.
You know, we, statistics can suffer
from talking about earns and spinners
and dice and flipping a coin.
But to have a real application where
you're talking about the difference
between a parameter, the intrinsic
ability of somebody as opposed to
their data and, and all of those parts.
I always found it was, uh, and it was, as
you described, most people understand it.
There's always the issue of culture
and the different sports and, you know,
if you're talking about is bunting a
good strategy, do people understand
the sport and, and all of that, but.
But partly the passion that, that,
that comes with sports and partly
the, the clarity to which you can
explain so many sports concept, uh, so
many statistics, concepts in sports.
Jim Albert: Yeah, I always think,
I mean, there's certain issues in
sports, like, uh, talking about
clutch ability, for example.
People love to talk about that, or they
get, they're so excited about streaks,
you know, streakiness and performance.
And so I guess that raises the issue.
What, what does that really mean?
I mean, we.
We, for example, we, I understand
there are clutch performances, right?
Somebody hit a home run with a bases
loaded in a ninth beating or a game.
But does it mean that that person
has some extra ability to, uh,
perform better in the clutch?
You know, um, I once talked to Tom
Tipt who, um, you know, who created
a, um, a baseball game called Diamond
Mind Baseball, and I asked Tom,
does he believe in clutch ability?
He says, no, but.
My, my, my people who by my game
want that to be part of the game.
So he has to put in the clutch ability
button or something, you know, which,
uh, so, 'cause people want that, people
believe in those things even though
there isn't much evidence to suggest
there's clutch ability in sports.
Scott Berry: Yeah, so even Strat
Amatic added that at one point, uh,
to it where, uh, two out spaces loaded
clutch situation that it, it either
added or subtracted hits from it.
So I, I think the audience would
be really interested in this.
And one of the, uh, one of the
initial papers that I think
you wrote was, uh, was on.
And I think you, one of your
heroes, Mike Schmidt, uh, you were
very interested in whether or not
he was a streaky home run hitter.
So explain the difference and,
and do you think Mike Schmidt
was a streaky home run hitter?
Jim Albert: Well, that
I don't, I don't know.
I think what's interesting is that if you
look at Mike Schmidt, his pattern, the way
he's, he played, the way he swung, I think
it was a bit different through his career.
I think he swung, had bigger swings at the
beginning of his career, and then later
on he realized he didn't need a swing.
As long to get his home runs.
And so I think you, you saw a little
bit of that in the, so what you looked
at was the, um, number of, uh, played
appearances between, between home runs.
And that seemed a little more
irregular, I think in the early
part of his, uh, the season.
Um, and later on it was a little more,
the, the gaps weren't quite as extreme.
Now what's interesting is that Henry Aaron
great home run hitter, he didn't have.
Streaks or have long streaks,
but he was considered more of
a, um, a line drive hitter.
So he didn't really ha hit up with
this big upper cut that people like,
um, like Kyle Schwarber now hits with.
But what was interesting, so when you
look at, um, Hering, it doesn't seem
to be a big pattern in his streaks.
So I think there are, there is something
going on there, but I think it's a
very subtle, um, effect and I think.
I think we're, um, it's hard
to, to view these things because
there's so much transferability
or coin tossing variability that's
making it much harder to see.
Scott Berry: Yeah, and it's,
it's one of those topics in
teaching a course that I find.
Once you bring it up, of course the
rest of your, your hour or hour and a
half, you're talking about this topic.
But it's such a neat, a, a neat
demonstration of randomness as
opposed to signal, and it's one of the
examples that I, I do sometimes is I
put up a sequence of zeros and ones.
And then you put up random sequences
of, of random zeros and ones, and you
ask people, which one is the human?
up zeros in one and which
one are the random one?
And invariably, when you ask somebody
to write down a sequence of random
zeros and ones, it goes 0 1, 0 1, 1 0.
Yeah, it it, there are too few
streaks, the natural variability
of a coin flip, if you show that to
people, they say your coin is streaky.
Uh, and it has momentum
at different times.
And so thinking about the game,
how often do we attach that?
I have a, a, a buddy of
mine who loves to follow.
The, the win probability that's
being posted now in games in the
NBA, uh, uh, basketball games.
And he sends me all kinds of
examples of where a team has a 98%
chance of winning and then losing
and the streaks aspect of it.
Uh, so it's understanding that it, it's
such a, a, a great example in class.
Jim Albert: Right.
I think, uh, one of my, I had
a student who was collecting,
uh, we were collecting, uh.
S every, like in BA college basketball
games, we were collecting the actual,
you know, streaks of performance.
And it's remarkable that of all the
momentum shifts you see in sports.
And, uh, again, the question
is, I mean, what does that mean?
I think a lot of it is a sort
of natural kind of randomness.
It doesn't mean that these players are
suddenly getting, but of course you
watch, you play a video game and you know
a person is hot and they put a big, um.
Some fire on his head or something
indicate that he's, he's in a
different mode or something.
But, uh, and we love people,
love to believe that.
I mean, it's a great story.
You know, it's a great story about.
Yeah.
Scott Berry: Yeah.
So, uh, another topic you've written
quite a bit about, uh, and I guess
I'll introduce it by saying we're
at, we're it's May 20th Aaron Judge
is hitting 4 0 1 May 20th, uh, 2025.
What do you think Aaron Judge
will hit by the end of this?
Assuming he plays a full season,
what do you think his batting average
will be at the end of the year?
Jim Albert: I am guessing
it'll be about three 20.
I mean, the point is that, and this
is just a empirical observation,
that people's, uh, batting averages
at the first part of the season
tend to be very variable and, um.
I think I just read that, um, on
the nationals, Josh Bell is hitting
like, like one 50 or something.
He's like the worst regular player
Now in baseball, certainly he's been
struggling, but he's gonna get better.
I mean, it's gonna catch up.
Likewise, Aaron Aaron's not gonna get
as he maybe as lucky with his hitting.
I mean, uh, and so baseball batting
averages are, tend to, I mean,
we'll never see a 400 hitter again
in baseball for a full season.
That's not gonna happen.
You know, I think, uh, there are too
many variables that are in terms of
strikeouts, for example, that are just,
that will hurt someone's batting average.
I mean, but, right.
But him to hit 400, even for this part
of the season is still pretty remarkable.
Scott Berry: So
Jim Albert: Yeah.
Scott Berry: also is a intersection
between sports teaching statistics,
regression to the mean and Bayesian.
It's such a natural
Bayesian hierarchical model.
and Carl Morris was, was the first
to show this, it's one of the
examples when I, and I've taught
actually sessions that are what drug
developers could learn from sports.
And this regression to the
mean is such an incredible.
Concept in sports that drug
developers, it's such a a, important
thing about forecasting what happens
from a phase two trial to a phase
three trial within its subsets.
And it's such a natural thing in sports
that people generally understand.
Jim Albert: Yeah.
Another thing that's interesting is that
it really depends on what you measure.
I think a batting average, for
example, is remarkably chancey.
I think there are so many
things that influence that.
The balls that are put in play,
it's a lot of luck involved.
So I think some measures like batting
averages are very chance driven, while
other measures like strikeout rate.
Those tend to be a little more
stable, So I think a lot of
baseball is understanding, which
measures are predictive of future
performance and which ones are not.
And even though we love talking about
batting average, it's a pretty crummy
measure for predicting the future.
Scott Berry: Yeah, and, and
interestingly, over the time in
which you've been studying sports.
Batting average when you started doing
this in the nineties was, was kind of
the stat, and now it's largely discounted
in part because of analytics, uh, and
the value of batting average as opposed
to OnBase slugging percentage, OPS,
uh, all, all of these other measures.
as analytics has become huge in
baseball, but all sports within that, um.
And, and there have, there are some that
think that this has been a negative.
What, what is your take on
the explosion of analytics?
And I guess I say this to the extent
that, well, you wrote for some 40
years about statistics in sports.
You did education.
You are now employed by a sports team, uh,
the Houston Astros, and so you are part
of this analytics for better, for worse.
Jim Albert: Yeah, I think, I mean, for
example, fielding, I think in the old,
old days all we had pretty much was.
Number of opportunities and
errors, number of errors.
And we now know that those are
relatively, especially number errors,
are really very subjectively scored.
But now we have a lot more understanding
about fielding because of analytics.
We know, where you were in the field.
We know, how your, how the
fielder is moving toward the ball.
We can talk about things
like catch probability.
So I think now we can measure
fielding And to me that's, exciting.
I think, to me, we're just
learning more about the game.
it's not taking things away.
It's actually adding a new
dimension to the sport.
Scott Berry: yeah.
so uh, maybe a little bit about.
Working for the Astros and, implementing
many of these Bayesian ideas.
how is that going?
Jim Albert: it's going well.
I think the reason why teams are more
interested in Bayesians is because
they understand the point is the, these
teams are getting much more analytical.
they're building up their groups.
To me, what's exciting is that these
groups ha are all basically young
people all from different majors, but
they all have a passion for baseball.
So you have a lot of very smart
young people working with these
teams and they're just trying to,
get better In terms of predicting
future performance, a big thing.
The, like the aaa, which is the, highest
level of the minor leagues is still a very
different environment than major leagues.
So they, I think it even, I mean there's
so many things that affect baseball stats.
You're talking, you can talk
about the, uh, uh, the opposition.
You can talk about the pitching,
you can talk about the stadium.
You know, um, whenever you
think about, like, Philly's last
night played in Chorus Field.
Well, chorus Field defines
ballpark effects, right?
Because they can always
talk about the effect there.
Well, the minor league parks have
their own vibe, and so to me.
I think teams like the Astros
have to properly adjust, you
know, the minor league measures.
'cause obviously they wanna predict, you
know, how these teams will, these players
will do when they go to the majors.
And I think that's a big
part of what they're doing is
understanding that process better.
Scott Berry: Yeah.
And, and now, uh, understanding a, a, um.
prospect that's in college, a
prospect that's in minor leagues,
a prospect that's in high school.
I imagine all of that is trying to,
to put these prospects on a similar
scale who's better, uh, is fascinating.
Jim Albert: Yeah, I think college
stats and high school stats, those
are very hard to make sense of
because there's so many variables.
Other variables that are
influencing those stats.
And, uh, again, and it's also they're
playing against other players, high
school players or other college players.
And so you have to somehow
adjust for that too.
Scott Berry: Yeah.
The, I, I've also loved the parallel
of some, uh, continuous sports like
ice hockey or even soccer where they
now know where a player is on the ice.
E every.
10th of a second, and they've got
this amazing date of where the
puck is, where every player is
continuously throughout the A game.
largely, I don't know if we really
know what to do with that yet.
And in clinical trials, we now have
wearables, for example, where we have an
enormous amount of data on individuals.
Uh, their, their biomarkers, their
behavior, their gait, uh, their, their,
their, you, you can judge cognitive
ability by how they use their phone.
But we have no idea what to do
with this in clinical trials yet.
It's almost this continuous amount of
data that sports has to deal with, and
it's a very different thing, but clinical
trials and health, we've gotta, well,
what does it all mean at the same time?
Jim Albert: Yeah, I think, I think
it's, I think it's exciting because I,
especially, I think soccer, of course,
that's the number one sport in the world.
And I think, and really honestly, what
happens in soccer is not the scoring,
it's the movement of players, right?
I mean, I think that's really
where teams get advantages.
Like, um, I'm a tennis player, so I know.
The importance, I mean, when you get
older, movement is not as obvious,
not as easy as when you're younger.
So I, I always talk to my tennis
friends and it's, and it's
not about hitting the ball.
The problem is getting to the position
where you want to get a comfortable
swing, you know, get a good swing.
And it's, so, I think
that's, this is exciting.
And I think for people that want
to go into the, these professional
sports who have statistical trainings,
especially in spatial modeling.
I mean, this is a wonderful
opportunity for them because of
that, you know that that modeling is
very useful in those, those sports.
Scott Berry: Yeah.
Yeah.
So the, over the course of your career.
was about teaching statistics, teaching
data science, teaching the concepts
through the applications, through sports,
uh, individual's interest in, that may
get passionate about, about statistics.
That was my interest in math and
statistics was I was a hockey
goalie and I could sit back and
think about the analytics of
ice hockey at different times.
'cause I had.
Good breaks in a hockey game.
It was very much my interest in it.
But now this is an industry.
there are people that, their
career are sports analytics.
They do this professionally.
There are training programs in college.
So it's, it's, you've seen this
go from in application to teach
statistics and data science to this
is a field that people are going into.
Jim Albert: Yeah, I believe, I mean,
a lot of, I'm getting more inquiries
from students who want to work for a
baseball team or work for Anno Sport, and
I tell them, I, I basically go for it.
I mean, I think the.
I can speak in terms of baseball,
there are more opportunities
to present your work.
For example, one illustration
example is Sabre Seminar.
This is a conference
that's held every summer.
It's currently being be
held this summer in Chicago.
And the point is to allow college
students and high school students
opportunities to present their work.
And a lot of the baseball
teams are represented at these
meeting, at this meeting.
So it gives a great way of connecting.
You know, professional people
with, with these young people.
And there are internships available.
Um, you know, teams are always expanding.
Uh, you know, it's just, I also went
to a, Sabre has a meeting in the Saudi
for American baseball research, has a
analytics meeting in spring in Phoenix.
I was there this year and I
saw a lot of young people.
They got special
competitions for students.
So I think, um.
If you, if the students have the
passion and they have the ability
to communicate and put together like
little projects, they really have good
opportunities to get jobs for these teams.
Scott Berry: Yeah.
So what suppose, suppose there's
somebody out there that's, uh.
Uh, sophomore in college, and
they're interested in this.
They, they're interested in sports
statistics, sports analytics.
skillset should they focus on, uh,
as an undergraduate and then maybe
even a potential graduate program?
What are their important skillsets?
Jim Albert: Well, they need some
background in statistics and I I, that
includes some experience in modeling.
Um, and of course modeling can be a
regression, of course is a good step.
But then obviously there are
extensions to that, like generalized
additive models, um, spatial models,
but some experience in statistics.
Um, they need some, you know,
background in, in the sport.
Like either they could be working
for a team, a college team, or they
could be, um, you know, have an
internship or something with that kind.
I think the ability to.
Communicate what you do is so important
nowadays because I think it's one
thing to do your own analysis, but
to me the value it is' invaluable.
Unless you can explain to somebody who
may be more, it's more of a decision
maker may not know the statistics
that well, but they wanna see what,
what are you doing that's useful to
what, what we're doing, you know?
So I think the ability to
do that is pretty important.
Scott Berry: Yeah, so,
uh, statistical inference.
Statistical modeling, presumably
there's programming, the ability to
program, ability to handle data because
now the explosion of data, then the
ability to communicate all of those
things, uh, together is the focus.
Jim Albert: Yeah, I think, um, even
what's interesting is that, um,
we taught a course in data science
at Bowling Green, which a lot
of it was coding, and I think I.
It's one thing to code, but you
want to code with a purpose.
And I think was a, David Donahoe had a
book about data science, about the growth
of data science, and he said a great
thing to do, but if you're interested
in baseball, is put together a book
like, you know, um, we have a, I have a
book co-authored with a Max Marchy and
Ben Ballmer on analyzing baseball data.
R Well, that tells you about the data
and teaches you r but then get exposed
to some of the baseball problems.
Um.
Tom Tango co-authored a book called The
Book, which is basically describes a
lot of interesting baseball problems.
And so if you put the data science
knowledge together with understanding
some of the important problems in
the sport, that's a good combination.
Scott Berry: Yeah.
Nice, nice.
Is there a, what, was there ever a
downside to teaching, uh, using a lot
of sports examples in, in teaching stat?
Jim Albert: Well.
Scott Berry: I, of
course, I, yeah, go ahead.
Jim Albert: Well, I think, I
don't think people understand
baseball, for example, that well.
So even if you're talking to a bunch
of football players, they, I, I think
baseball, you know, it's, it's sort
of a, it's a hard sport to watch if
you're not familiar with the, the rules,
you know, because a lot of it is the
encounter between the, the pitcher and
the batter, and if you are not aware of
the different pitch types, you know, and
all those small things that are happening.
It was much like I went, I spent a year
in England and I watched a game of cricket
and I had no idea what was happening.
I mean, it was very, you know, it was
even harder to me in baseball because the
action was in the middle of the field.
It wasn't near the you the stands,
and so likewise, I think baseball,
you, you need to obviously explain
the sport well enough so that
people appreciate what you're doing.
Scott Berry: Yeah.
I, I, uh, I was in the UK and
I was watching, uh, at a pub.
Cricket.
I said, this is amazing.
I, everybody says it's kind of dull.
And somebody said to me, you're
watching a highlight show.
Uh, I had no idea it was a highlight show
as opposed to the, the, the action of it.
with within that.
So what, what, um, uh, now you're,
you're working for the Astros.
You're continuing, you're, you also have
an interest in po potentially a journal.
sports applications, uh, and partly
concerned about the accessibility
of the work that's being done.
Jim Albert: Well, right now if you
publish a paper in sports, the main
outlets are the Journal, journal,
quantitative Analysis Sports, and
the Journal of Sports Analytics.
Those are the two main, uh,
journals that's focused on
statistical applications in sports.
The problem is that they're both owned by.
Companies, and they really have
a payroll wall behind this.
So anybody who wants to read these
papers is not able to read them.
You know, there's this cost
involved and that, I think.
So now there's a movement among the
people, my, my group, which is the
people who are doing sports research
to maybe create a new journal
where you have more open access.
Um, and now it'd be sort of some
journals like the Journal of Sports,
uh, education and Data Science.
Those are more open.
Um, and I think we want to create
a journal, a sports statistics
journal that's the same thing.
So, so for example, you, you
wanna learn about baseball
research, you go to FanGraphs or
you go to Baseball Prospectus.
But maybe you go to a new journal where
you find these articles and you know,
access to be free no matter where you go.
Scott Berry: Yeah.
Yeah.
Nice.
A, a, a a great effort, uh, in that.
And how about teaching?
Do you get any more outlet, uh,
after, uh, teaching for 45 years?
Do you get that outlet, do you get
that outlet through the Astros, or is
that, uh, uh, any outlets to teaching?
Jim Albert: Well, I think, I do think
that the Astros might benefit if I try to.
I mean, whenever you pres I, what I
do for the Astros is I make writeups
of problems I've been working on.
And I guess one concern when you do
that is you wonder if the level of
what your presentation is meeting their
needs or if they're at the right level.
So I think there might be an opportunity
to do little more tutorials for them.
Um, there's also a program called Yugo.
It's a, it's a program to connect.
Professors and high school students who
wanna learn about different applications.
And so I've had the opportunity to work
with these students one-on-one, and
they wanna, they're just interested
in just high school students who
wanna do some sort of project related
to baseball or, or basketball.
So what I do is I meet with
them periodically and they
put together a project.
So that's a good opportunity too.
So I, I enjoy special things.
I think the thing I miss.
About being retired is I do miss the
interaction with the students and,
uh, especially the more, the more
the face-to-face, uh, interaction.
I, I retired in 2020 and that was
probably a good time to retire because,
uh, teaching was more, um, online.
Our one green gravitated to a, uh, online
format and it took a lot of effort.
The tee shut way, I didn't
see the benefits as much.
Because I missed that
kind student interaction.
Scott Berry: Yeah.
Uh, I imagine now a huge part of
statistical education, data science
education is how to do it online.
I, I mean, invariably we're sort of
here in the presentation of material.
The classroom interaction is one thing,
but the ability to do it well online is
a whole different skillset, I imagine.
Jim Albert: Yeah.
You wanna somehow replicate
like interactive computing?
I.
The idea, one thing that's so nice about,
uh, packages or software like our, is
that you can do things on the fly and
you make changes and you know, sort of go
through a series of data exploration steps
and you wanna mimic that somehow online.
And that's a little harder.
I mean, 'cause really, uh, data analysis
is not a polished thing where you
start from the beginning to the end.
Rather, you know, you go forward,
you've got some step backs.
You think again about what you're doing.
You, you revise and
you try something else.
And I mean, so, uh, a paper often
doesn't, res doesn't really represent
the process of, of learning from data.
Scott Berry: Yeah, that's
the, that's so true.
That, um, it's that continually iterative,
what is the question I'm trying to answer?
How does this data help me?
How could a model help me?
And the process of that is so different
than the final output, uh, in many ways.
Jim Albert: Right.
Scott Berry: but what, what you're
trying to get through there.
So, I, I, I, The, the value of sports
in many ways is that ev people's
interest, and now of course sports
is huge business, uh, these teams
are worth, uh, billions of dollars.
Contracts are, you know.
Tens to hundreds of millions of dollars.
Do you sign a particular player?
How does he work on our team?
Is he worthwhile?
What's the value there?
There's huge value in this,
but in in other areas.
So my son is a division three
baseball player and now their team
is incredibly interested in ranking
teams because the NCAA now has gone
to using what they call the NPI.
Uh, to decide who makes the postseason.
And it's entirely an algorithm
that they've created that
decides which 64 teams make the
postseason, uh, uh, in this.
And so they're incredibly interested in
what are the components of it, what are
the value of a win, strength of schedule.
And this is so important to them
now that a number of them are really
interested in diving in and comparing
the algorithm and actually figuring out.
What, what this does and, and, and who
to schedule and, and is this good or not?
So the world is drastically different
than, you know, 20, 30 years ago.
Jim Albert: Yeah.
In fact, it's interesting 'cause I'm
a tennis player and I'm re currently
rated as a three five tennis player.
Well, the question is, where
is that three five come from?
Right.
It's a rating of your ability.
Well, nobody actually, long time
ago somebody actually watched me
play and at that time I was a four.
Oh.
But you know, the system, to me it's
not, it's sort of a, a mysterious
about how they rate players and uh,
you know, it's, um, I had a player
who was a four oh player and he
wanted to play in a three five league.
So he just asked the USDA as a
simple process to revise the rating
and they were quickly agreeable.
To lower your rating.
Um, also when you're an older player,
they're happy to lower your rating too.
And I don't, I don't quite understand.
It'd be nice if I, as a statistician,
I would like to learn a little
more about the mechanics of that.
They don't release, they don't explain,
you know, if you, if you basically, if
you play a league and you win every single
match, your rating might rise, but I
don't know exactly what the algorithm is.
So,
Scott Berry: Oh, interesting.
So it seems like chess does it much
better than, than tennis, but even
golf, uh, in terms of turning in
scores in your handicap and there's
all science to the golf handicap.
We don't get into that.
I, I'm surprised tennis doesn't
quote unquote, do it better.
Jim Albert: but of course there
are, I mean, it's hard to change
the system because people.
It's fun because you get to
compete at a local level and then
if you win your championship,
you can compete at a state level.
I, my team in the fall, last fall
went to Indianapolis for sectionals
and if we'd won that level, we
would've gone to a national.
So it's a, it's fun to have those
opportunities and I don't think you
want to change the system too much.
'cause right now it's very popular
to have those opportunities.
Scott Berry: All right.
Well, Jim, thank you very much,
uh, for joining in the interim.
Uh, it, it must be so neat for you to
spent a career within statistics in
sports, seeing the explosion actually of
Bayesian methods computation of Bayesian.
Sports from being kind of,
uh, a side interest to many to
now it is a full-fledged field
and, uh, uh, academic pursuit.
So, uh, thank you for all your
efforts and appreciate you joining,
uh, US here in the interim.
Jim Albert: Well, great.
Well, thanks for having me.
It's, I think, uh, I, I agree.
It's, I remember the days when I used
to actually copy stats from a, you
know, from a book to a spreadsheet.
So things have come a
long ways, just exciting.
Scott Berry: Yep.
Yep.
All right, well thank you all
and uh, thanks everybody for
joining us on the interim.