Latest / Why We Do What We Do / Mini: Weird Blue Chart

Transcript

Mini: Weird Blue Chart · Why We Do What We Do

0:03
Speaker 1: You're listening to Why we Do what we Do. Welcome
0:15
Speaker 1: to Why we Do what we Do Mini. I'm your
0:18
Speaker 1: host Abraham and I'm Shane. This is a short form
0:21
Speaker 1: series that we do that comes out on Mondays where
0:24
Speaker 1: we take a topic, do a quick dive on that
0:26
Speaker 1: topic for like fifteen maybe ten ish minutes, and then
0:30
Speaker 1: get out of there.
0:31
Speaker 2: Yeah, and today is one that definitely doesn't need more
0:35
Speaker 2: than ten or fifteen minutes. Ever.
0:37
Speaker 1: Disagree to an easy, easy topic.
0:41
Speaker 2: Everybody understands it. Nobody has anxiety about this thing. Yeah,
0:45
Speaker 2: we're talking about the Standard Celeration Chart. Super simple and easy, right.
0:49
Speaker 1: Yeah. Yeah, this is going to be both, I think
0:51
Speaker 1: for the hardcore nerds and scientists, but also I'm going
0:55
Speaker 1: to try and get everyone interested in this thing. I mean,
0:58
Speaker 1: not really. What I'm gonna do is I'm just going
1:00
Speaker 1: to explain, explain what it is and why we use it. Sure,
1:03
Speaker 1: and most of you are going to hear that and
1:04
Speaker 1: then forget that I ever talked about it. And that's okay,
1:06
Speaker 1: you know, thanks for being here.
1:08
Speaker 2: Yeah, I mean, maybe there's a plug somewhere in here
1:10
Speaker 2: for the Standard Celeration Society for people to check out
1:12
Speaker 2: at some point in time.
1:13
Speaker 1: But it's a great a great plug. Yep, going in
1:15
Speaker 1: the links perfect.
1:16
Speaker 2: I love it. So, I mean, you know, everybody, I
1:19
Speaker 2: think in our field, if you're a behavior analyst and
1:21
Speaker 2: you're listening, then you know about the blue chart, and
1:24
Speaker 2: it maybe stresses you out, but it's actually quite simple
1:27
Speaker 2: when you understand it.
1:28
Speaker 1: Yeah, so let's go ahead and give give the set
1:30
Speaker 1: up here, which is the understanding. In science, a huge
1:33
Speaker 1: amount of scientific work and scientific discovery and research is
1:38
Speaker 1: communicated using visual depictions that show relationships among data graphs.
1:46
Speaker 1: We use a lot of graphs. Is just the sort
1:48
Speaker 1: of the way to say that how we understand what
1:50
Speaker 1: are going on in data. There's a lot of ways
1:52
Speaker 1: to do it, but one way we commonly do it
1:55
Speaker 1: is by using various graphs.
1:57
Speaker 2: Yeah, and there's you know, there's charts and graphs like
2:00
Speaker 2: we use pie charts, the J curve, climate change graphs,
2:03
Speaker 2: scatterplot graphs, and we use tables which take numerical data
2:07
Speaker 2: and the arrange them along a matrix of one or
2:09
Speaker 2: more dimensions. So we do try to communicate data in
2:11
Speaker 2: a lot of different ways, but typically pictorial images tend
2:15
Speaker 2: to be the most effective. People tend to like pictures,
2:18
Speaker 2: which is why people are on Instagram more than like
2:22
Speaker 2: X or whatever you know.
2:25
Speaker 1: Indeed, so the selection of a graphical representation in data
2:32
Speaker 1: is important because it's it's an efficient way to understand
2:35
Speaker 1: a lot of data and the relationship to each other
2:38
Speaker 1: by showing where the data are on some given measure
2:41
Speaker 1: like that is, like the scale of the graph might
2:44
Speaker 1: be something like percent or frequency, and then where that
2:48
Speaker 1: those data fall would be the level. So if you
2:51
Speaker 1: had like a zero to one hundred percent, if it
2:53
Speaker 1: was at fifty percent, the data were at fifty percent,
2:56
Speaker 1: that shows you what the level is of those data
2:58
Speaker 1: on that graph, right, And it shows also that as
3:02
Speaker 1: data are collected, if they're moving in a particular direction
3:05
Speaker 1: along that scale, that'd be trend. So if the data
3:07
Speaker 1: go from fifty to fifty five to sixty, et cetera,
3:10
Speaker 1: you see a general increasing trend, versus if it were
3:12
Speaker 1: going down forty thirty five, thirty, et cetera, you'd see
3:15
Speaker 1: a decreasing trend. And then the last aspect of those
3:19
Speaker 1: data is the relative degree of change across data points,
3:23
Speaker 1: which we usually talk about as being variability, assuming that
3:26
Speaker 1: the change is not directional like a trend but moves around.
3:31
Speaker 1: Sometimes it's higher, sometimes it's lower.
3:33
Speaker 2: That we call variability, right, And so if we're looking
3:36
Speaker 2: at a simple xy coordinate plane graph with ploted data
3:39
Speaker 2: points and trend lines, or even simply connecting the dots,
3:42
Speaker 2: what this does allows users to immediately ascertain these values
3:45
Speaker 2: and glean something about the data that are indicating in
3:48
Speaker 2: a way. And I think we can see very clearly
3:51
Speaker 2: on that picture, Like we could see what's going on
3:53
Speaker 2: in a much faster and less efforable way than trying
3:56
Speaker 2: to explain data in a narrative form or laying out
3:59
Speaker 2: data in a sequel and trying to determine levels, trends,
4:01
Speaker 2: and variabilities. That visual depiction is so important. It makes
4:05
Speaker 2: it so much easier to communicate data in such a
4:08
Speaker 2: rapid way. So that's often why we use graphs. They
4:10
Speaker 2: are super helpful and very allows for a very rapid
4:13
Speaker 2: interpretation of the data.
4:15
Speaker 1: Yeah, I mean, if you can imagine the amount of
4:17
Speaker 1: time it would take for me to describe what data
4:20
Speaker 1: are doing and what it should mean, versus how long
4:24
Speaker 1: it takes you to look at a few data points
4:26
Speaker 1: trending up or down very different times. One of them
4:29
Speaker 1: is podcast length the other one is like one second
4:33
Speaker 1: or less where you can look at it. So it's
4:36
Speaker 1: just it's very efficient and across many professionals and many
4:40
Speaker 1: fields of study in science and research or even in
4:43
Speaker 1: clinical practice, they have standardized their data displays because doing so,
4:50
Speaker 1: having a standard way of displaying their data comes with
4:52
Speaker 1: some advantages that a lot of people like. One of
4:55
Speaker 1: them is that it has a very quick interpretation. Now,
4:59
Speaker 1: most people who maybe aren't trained to look at a
5:01
Speaker 1: particular graph, let's just say an EKG, they might not
5:04
Speaker 1: necessarily know what they're looking at, and it would take
5:07
Speaker 1: a while to get used to it. But people who
5:09
Speaker 1: work in that field who use those data regularly, they
5:13
Speaker 1: have they're used to looking at ekg's. It's a standardized display.
5:16
Speaker 1: They can look at it and immediately interpret what those
5:19
Speaker 1: data are saying. They know what the scales and the
5:21
Speaker 1: features of the graph are.
5:23
Speaker 2: And it's also easy to analyze patterns in the data
5:26
Speaker 2: by just simply because I understand what is being displayed
5:29
Speaker 2: on that data. If I'm familiar with that graphical display,
5:32
Speaker 2: then I know what to look for. I know what's
5:34
Speaker 2: off or different or what to kind of interpret that
5:37
Speaker 2: in that display.
5:39
Speaker 1: Another thing that people like about standardized displays is it
5:42
Speaker 1: prevents obfuscating or misrepresenting data because they're standardized, Like, the
5:48
Speaker 1: graphs cannot be manipulated to emphasize or de emphasize aspects
5:52
Speaker 1: of the data. They just are what they are. They're
5:54
Speaker 1: on there, like, you can't change the axes or the
5:56
Speaker 1: values of the axes and make them say something else,
6:00
Speaker 1: going to show whatever they show, and then you sort
6:02
Speaker 1: of have to take the data as they are and
6:05
Speaker 1: not make it look like it's saying something and then
6:07
Speaker 1: other than what it might be saying.
6:08
Speaker 2: Right, which then also gives you guidelines on how to
6:11
Speaker 2: collect and present the data because the graphs can only
6:13
Speaker 2: display certain kinds or certain amounts of data. So you
6:16
Speaker 2: can't go and make up a whole bunch of stuff
6:18
Speaker 2: and just throw it onto a graph, because the graph
6:20
Speaker 2: itself has designed an explicit way to make it so
6:23
Speaker 2: that it's interpretable a very specific way with those data
6:27
Speaker 2: right now.
6:28
Speaker 1: Another feature that people sometimes can see as beneficial from
6:31
Speaker 1: using standardized graphs is that it can create both normative
6:36
Speaker 1: and criteria driven interpretation of the data. And what I
6:40
Speaker 1: mean by that is that, like it's how things typically appear,
6:43
Speaker 1: so sort of what is normative what you normally expect
6:46
Speaker 1: to see on those data, as well as a criteria
6:49
Speaker 1: against which those can be measured. So you can also say,
6:52
Speaker 1: not only am I used to seeing it this way,
6:54
Speaker 1: but it has to be at least at this threshold
6:56
Speaker 1: for to mean one thing or another. And so it
6:59
Speaker 1: have that standardized display gives you sort of multiple options
7:03
Speaker 1: in terms of how to look at those data and
7:05
Speaker 1: meaningfully draw normative and criteria based conclusions about what those
7:09
Speaker 1: data are telling you, right, And it.
7:12
Speaker 2: Also allows for a rapid transfer of data or the
7:15
Speaker 2: data display to somebody who's a trained user, because they
7:18
Speaker 2: can just pick that up and they can move with it.
7:20
Speaker 2: That standardized format allows us to communicate information in a
7:24
Speaker 2: really rapid way, especially to people who are trained, and
7:27
Speaker 2: we can transfer those data over, We can carry it
7:29
Speaker 2: over into different displays and whatnot without too much effort
7:32
Speaker 2: or too much training. Yeah, you know, too much additional training.
7:35
Speaker 1: I should say, once you know how to use it,
7:36
Speaker 1: you can just put it on there. You don't have
7:38
Speaker 1: to like create a new graph every time. It's just
7:40
Speaker 1: put the data into the existing sort of template, if
7:43
Speaker 1: you will. So let's take a quick break for some
7:46
Speaker 1: advertisements and then continue talking about graphs. Yay, that was
7:58
Speaker 1: maybe one of those instances where people we're actually looking
8:00
Speaker 1: forward to the advertisement break because they're like rain Meldy,
8:05
Speaker 1: They're like, uh.
8:06
Speaker 2: Graphs, why are we doing with data?
8:08
Speaker 1: Yeah? Anyway, I think they're cool.
8:11
Speaker 2: They are cool.
8:12
Speaker 1: In the field of behavior analysis, we use a lot
8:15
Speaker 1: of graphs, and there's really only one tool designed for
8:19
Speaker 1: standardized display, and this is called the Standard celeration chart
8:23
Speaker 1: or also sometimes called the standard behavior chart, and the
8:28
Speaker 1: features of this are that it can encompass a range
8:31
Speaker 1: of human activity from as infrequently as once per day
8:35
Speaker 1: all the way up to as often as a thousand
8:37
Speaker 1: times per minute. And it's that sounds like a crazy
8:41
Speaker 1: range that would be impossible to interpret. We'll explain how
8:44
Speaker 1: it works because it's really like that range sounds way
8:47
Speaker 1: more difficult than what it actually looks like. We'll explain
8:50
Speaker 1: that in a moment. But then it also contains up
8:52
Speaker 1: to one hundred and forty days worth of data, which
8:55
Speaker 1: means that your graph can cover You can use the
8:58
Speaker 1: same graph for about two and a half times per
9:01
Speaker 1: year because it goes for one hundred and forty days,
9:04
Speaker 1: or if you're using an online platform. It goes for
9:06
Speaker 1: infinity days.
9:07
Speaker 2: Yeah, which is a lot of fun. So this is
9:09
Speaker 2: called a semi logarithmic display, and what that means is
9:13
Speaker 2: the y axis is logarythmic, it's based on tens, and
9:15
Speaker 2: the x axis is linear. So each data point is
9:19
Speaker 2: a frequency per minute on a typical chart. So like
9:22
Speaker 2: we're looking at frequency per minute, which calculates three.
9:25
Speaker 1: Right, So as I said, like, because that can be
9:27
Speaker 1: the permitt it can be all the way down to
9:29
Speaker 1: point zero zero, you know, whatever times per minut it
9:32
Speaker 1: is once per day to you know, a thousand times
9:35
Speaker 1: permitted if you are somehow able to do something that quickly.
9:39
Speaker 1: For those of you who have never seen it before,
9:41
Speaker 1: on this chart, all the lines are blue. The reason
9:44
Speaker 1: that they are blue is that that that color was
9:46
Speaker 1: empirically derived when they tested the colors of different charts
9:51
Speaker 1: that people could use the most quickly and efficiently. Blue
9:54
Speaker 1: was the color. So they just tested all these different colors.
9:57
Speaker 1: Blue was the fastest and easiest, so they went with blue.
10:00
Speaker 1: And so if you ever see it, it's all blue
10:03
Speaker 1: lines on a white background. You know, the contrast is decent.
10:06
Speaker 1: It probably would it'd be better if it was black,
10:08
Speaker 1: I cou'd be more contrasted, but the blue allows for
10:11
Speaker 1: the black dots that are the data points to sort
10:14
Speaker 1: of pop out as well. So that's the reason that
10:17
Speaker 1: I think blue worked. Well.
10:18
Speaker 2: Yeah, you got to see the dots when you drop
10:20
Speaker 2: them exactly, drop it like it's dot. Yeah, drop it
10:23
Speaker 2: like it's dot. I love that. See, I know, I
10:25
Speaker 2: know just enough about it to be dangerous. You know,
10:28
Speaker 2: it's really fascinating that. Like, scientists argue use and non use, right, So,
10:33
Speaker 2: supporters of the standard celeration chart argue that the nature
10:36
Speaker 2: of the graph forces relatively simple symbolism. The idea is
10:40
Speaker 2: that we don't need a lot of interpretation. It's very
10:42
Speaker 2: simple to interpret once you use it, once you understand it,
10:45
Speaker 2: you can kind of look at it and go, ah,
10:47
Speaker 2: I know what's going on here.
10:48
Speaker 1: Yeah, Like you're going to be basically have two or
10:51
Speaker 1: maybe three different symbols most of the time, and rather
10:54
Speaker 1: than having to orient to a lot of different symbols
10:56
Speaker 1: that overlap and cross each other, the standard pract this
11:00
Speaker 1: is that with a standard display, you go with a
11:02
Speaker 1: simple legend for your graph. Another argument that they make,
11:06
Speaker 1: or they argue in support of the use of the
11:08
Speaker 1: standard celeration chart is that the logarithmic nature of the
11:12
Speaker 1: y axis means that the data points are proportional as
11:15
Speaker 1: they change, and what that means is that the emphasis
11:18
Speaker 1: on changes in data, they're most easily observed when they're
11:22
Speaker 1: sort of moving proportional to one another. And sorry, that's
11:26
Speaker 1: probably very unclear. As an example of this, if you
11:29
Speaker 1: go from five to ten, right, so we had five
11:32
Speaker 1: actions or five activities or five behaviors to ten, that
11:36
Speaker 1: would be a one hundred percent change. It doubled, right,
11:38
Speaker 1: And when it went up, right, went up by an
11:40
Speaker 1: absolute number of five, it increased by five. But proportionally,
11:44
Speaker 1: that is a doubling or a one hundred percent change,
11:47
Speaker 1: and that would be very noticeable in someone's behavior. When
11:50
Speaker 1: parents notice that their kids go from speaking five words
11:53
Speaker 1: to ten words, that is a noticeable increase, right, right.
11:57
Speaker 1: But let's say you take another absolute change from five
12:00
Speaker 1: five hundred to five hundred and five. That is still
12:03
Speaker 1: a change of five, but that's like a one percent change,
12:07
Speaker 1: and you probably wouldn't really notice that that change even
12:11
Speaker 1: took place. If something increased from five hundred, like five
12:13
Speaker 1: hundred words to five hundred and five words, right, And
12:16
Speaker 1: So the argument here is that proportional change is good
12:20
Speaker 1: because it emphasizes the points at which that change is
12:23
Speaker 1: going to be the most meaningful.
12:25
Speaker 2: Well, and I was just thinking about the standard Teleration
12:27
Speaker 2: chart and the difference between five and ten versus like
12:31
Speaker 2: five hundred and five, Like, looking at that on the graph,
12:34
Speaker 2: like you would not even see. Now you would basically
12:36
Speaker 2: put the dot on the same line for five hundred
12:39
Speaker 2: and five hundred and five because they are so minuscule
12:42
Speaker 2: a change.
12:43
Speaker 1: Right, Yeah, they'd be right next to each other. It'd
12:45
Speaker 1: be undistinguishable. Versus a five to a ten would be
12:49
Speaker 1: a very big difference on the chart.
12:50
Speaker 2: Huge jump, huge jump on the chart. Now, another thing
12:53
Speaker 2: that supporters argue is that interpretation is faster and more
12:56
Speaker 2: conservative because the display is standardized and the display cannot
12:59
Speaker 2: be altered. And I think that's really important when you
13:02
Speaker 2: are looking at equal whatever line graphs. I mean, one
13:04
Speaker 2: of the things they always talk about, at least what
13:06
Speaker 2: I teach it. When I say that, I mean me.
13:08
Speaker 2: One of the things that I always talk about is
13:10
Speaker 2: the fact that you can squish down your wy axis,
13:12
Speaker 2: you can expand it. You can do that and it
13:14
Speaker 2: will massage your data in a way that will make
13:17
Speaker 2: it look a certain like you can make it look
13:19
Speaker 2: how you want it to look on any other display.
13:21
Speaker 2: You can't do that on a standardceleration chart because the
13:24
Speaker 2: axes never change.
13:25
Speaker 1: They're locked in place. If you had data that went
13:28
Speaker 1: from eighty to one hundred percent, and your axis started
13:31
Speaker 1: at eighty percent and ended at one hundred percent, then
13:34
Speaker 1: the variability and the data points could look gigantic. But
13:37
Speaker 1: if you then when if you went from zero to
13:39
Speaker 1: one hundred percent, then that shrinks the variability because now
13:41
Speaker 1: it only takes up the top one fifth of the
13:44
Speaker 1: graph as opposed to taking up the entire graph. And
13:47
Speaker 1: so yeah, that's it. That's stretching and compressing can really
13:51
Speaker 1: affect how you interpret the data. And then one of
13:54
Speaker 1: the other arguments, and there's actually more than what we
13:56
Speaker 1: have listed here, but I think one of the other
13:58
Speaker 1: main arguments that people may in support of the use
14:02
Speaker 1: of the standard celeration chart is that those trends in
14:05
Speaker 1: the data as it's going up or down, they can
14:07
Speaker 1: be quantified as a rate of change because of the
14:10
Speaker 1: logarithmic scale, and that's actually what they call celeration. The
14:13
Speaker 1: rate of change is celeration which is where the name
14:15
Speaker 1: standard celeration comes from. Then it could go up or down,
14:19
Speaker 1: and that's why it's not acceleration or deceleration, it's just celeration.
14:23
Speaker 1: It's change, right, because you can then apply a specific
14:26
Speaker 1: number to that. It is a you know, a one
14:29
Speaker 1: point five or one hundred and fifty percent change versus
14:32
Speaker 1: a two hundred percent change versus a you know, a
14:36
Speaker 1: point five would be you know, it went down percent change,
14:39
Speaker 1: you know, times change or whatever. Those are the different
14:42
Speaker 1: ways you can specifically quantify the rate of that change.
14:45
Speaker 1: And so, in addition to all the other analysis you'll
14:48
Speaker 1: get just from looking at the data points, having a
14:50
Speaker 1: quantitative value for how how quickly things are changing can
14:53
Speaker 1: be very useful too.
14:55
Speaker 2: Yeah, most definitely. On the other side of that, detractors
14:58
Speaker 2: will argue that it's difficult to learn by the way
15:00
Speaker 2: not and also it should be empirically validated for use,
15:04
Speaker 2: even though none of the other GRAPHL displays that we
15:07
Speaker 2: use are empirically validated for use the same way. So
15:10
Speaker 2: they're expecting a higher standard to use this thing for
15:14
Speaker 2: arguably a more accurate data display without holding our other
15:19
Speaker 2: data displays to the same standards.
15:21
Speaker 1: Yeah, that's basically it. Those are the arguments that I've
15:24
Speaker 1: found is people don't want to use it because it's
15:27
Speaker 1: there's a lot of lines and it's hard to learn,
15:30
Speaker 1: and I'm like, I tak kids how to use it,
15:33
Speaker 1: but okay, anyway, But being it's not as bad as
15:36
Speaker 1: it appears. If you are interested in learning more about
15:38
Speaker 1: the standard celeration chart, you can go to celeration dot Org.
15:42
Speaker 1: There's a whole bunch of resources there and you can
15:44
Speaker 1: see what it is I'm talking about. You can also
15:47
Speaker 1: google that and you'll see a picture of it. I
15:49
Speaker 1: find it useful and how I do and conduct my
15:52
Speaker 1: work inside of the field of behavior analysis and even
15:55
Speaker 1: doing you know, personal projects where I'm sort of working
15:58
Speaker 1: on something for myself, I use that chart to graph
16:01
Speaker 1: my data. So that's I think what I have to
16:04
Speaker 1: say about it. Is there anything you'd like to add
16:06
Speaker 1: or anything I forgot before we wrap this one up?
16:09
Speaker 2: Yeah, I mean, I think this is one of those
16:11
Speaker 2: things where I think the reason it's scary is because
16:13
Speaker 2: it's not commonly used. Yeah, And it's that thing that
16:16
Speaker 2: is like we are, it's unfamiliar. But like the truth
16:18
Speaker 2: is is that when you understand the nuts and bolts
16:20
Speaker 2: of it is one of the easiest charts to use.
16:22
Speaker 2: And I think, at the end of the day, if
16:25
Speaker 2: we are going to do data and I remember, I'll
16:27
Speaker 2: never forget.
16:28
Speaker 1: I believe.
16:29
Speaker 2: I heard a lecture from doctor Kim Baron's say something like,
16:34
Speaker 2: unless you're doing a standard toleration chart, you're only doing
16:36
Speaker 2: data display. The only data analysis that you do is
16:40
Speaker 2: standard celeration charts, like everything else is a data display. Obviously,
16:44
Speaker 2: that was like kind of like a quick bullet point.
16:46
Speaker 2: That's a sound bite, like she gets more into it
16:48
Speaker 2: and explains a little bit. It's not that aggressive. That
16:50
Speaker 2: stuck with me as somebody who spends a lot of
16:52
Speaker 2: time looking at data. It's like, no, I want to
16:54
Speaker 2: do data analysis, so like I want to spend time
16:56
Speaker 2: actually doing the analysis of part of this. I don't
16:58
Speaker 2: want to just look at a graph and go it's
16:59
Speaker 2: going on up or something like that. I want to
17:01
Speaker 2: do more. And that was really insightful. That was kind
17:04
Speaker 2: of a good motivator for me to learn more about
17:06
Speaker 2: the standards celerection chart and spend some time on that.
17:09
Speaker 1: And as I said, there's some online platforms now that
17:11
Speaker 1: make it pretty easy where you really don't have to
17:13
Speaker 1: know a lot about how to get the graphs or
17:16
Speaker 1: get the data depicted on the graph, and it'll quantify
17:19
Speaker 1: all the information for you. So I think the barrier
17:22
Speaker 1: to entry has never been lower. So if you're interested,
17:25
Speaker 1: you know, go out and learn more. It's a pretty
17:27
Speaker 1: useful tool for me at least. And I think we're
17:30
Speaker 1: running along on this mini, so we should probably call
17:32
Speaker 1: it there. Yeah, so thank you all for joining us.
17:35
Speaker 1: Go check out more about this. But I think that's it.
17:37
Speaker 1: So this is Abraham and it's Shane.
17:39
Speaker 2: Why we Do what we Do mini is out why.
17:43
Speaker 1: You've been listening to Why We Do what We Do.
17:45
Speaker 2: You can learn more about this and other episodes by
17:48
Speaker 2: going to WWDWWD podcast dot com. Thanks for listening, and
17:53
Speaker 2: we hope you have an awesome day.