E1 Calculate and Summarize Behavior Data | RBT® Task List Explained

Okay, we're going to talk about

calculating and summarizing behavioral

data. So, we talked about how do you

take the data last module. So, now you

have a lot of data. You did it. You

collected it correctly, accurately. Now,

what do we do with that

data? So, why do we calculate behavioral

data? One, it helps us track progress.

Behavioral data tells us what's working

and what's not. It allows us to make

decisions based on real numbers. So

simple calculations, provides really

powerful information. It shows patterns

and trends over time. It identifies

patterns and helps us guide intervention

and it will help us adjust goals and

interventions. So it makes it so we're

doing that datadriven decision making to

better outcomes. It's really the

interventions are great that we work

with and the science is very powerful

but science the backbone and the body of

the science is the data. Collecting the

data is the most important part of

anything. Without it nothing's accurate.

You don't know what's going

on. We talked about rate prior but we

will talk a little bit more about it.

One, it's confusing in that moment

because you just learn

frequency. And two, it belongs sort of

in both places. So, I like going over

again. And three, it's math. So,

everybody has a lot of people have more

difficulty with math concepts than other

concepts. If you're that person, this is

specifically probably for you. Rate is a

measure of how behavior occurs in a

specific time period. It's calculated by

dividing the numbers of responses by the

time observed to give us a metric to

track it. So when behavior happens

multiple times, you might want to

calculate rate. So when it's useful when

analyzing behaviors that occur

repeatedly, such as asking for help,

displaying aggression, or attempting

tasks. And then you're going to compare

sessions of different lengths. So rates

allows you to standardize the

measurement of behaviors across all

sessions with varying durations. It

enables meaningful comparisons and then

common examples where we might use rate

is like frequency of requesting

assistance, episodes of aggression, the

number of tasks and attempted and of

course this is once you have a frequency

count you have different length

observations and you need you have this

number like the behavior occur this

much. this is how much I observe, but

you don't know what to do with it next.

I'm going to give you a bunch of

examples. So, for

example, asking for help. We're taking

the time. We rarely take it to one

minute, but if you have very frequent

behaviors, we often take it down to the

hour. Sometimes we might take it down to

the day, week, or month in some

situations. Example one, we're asking

for help. She observed over 20 minutes

and she saw the behavior occur for 10

times through the 20 minutes. So that's

a very frequent frequent behavior. So we

take it down to the minute. The rate is

10 / 20 which will equal 0.5. So she

requests but she makes 0.5 requests per

minute. So now we know that when this

girl is asking for help a little bit too

much, it's hard for the teacher to teach

because every 20 minutes she's asking

about 10 times. We want to increase her

autonomy and make it so she asks for

help less. We can give her more

directions at the beginning. We can

provide her with different kinds of

supports like supports on her desk to

help answer her questions. If she was

really asking for help, if it was poor

attention from the teacher, maybe the

teacher will increase the amount of

attention she's giving her to reduce

this asking for help. We do an

intervention. You'll learn the

interventions later on. What you're

going to do is every 20 minutes you'll

see how many occurrences and you'll keep

calculating that. For

example, let's say we got five. So you

start your intervention, maybe you let

it run a few days, and now we're going

to take some more data. In 20 minutes,

we saw five times she asked for help. So

that's good. We were trying to reduce

it. So 5 / 20 is

0.25 requests per minute. So we know we

had a reduction. Maybe you observed 30

minutes and you got 10. Did she improve?

If you observed 30 minutes and you got

10, 10 divided by 30 is.33 requests per

minute. So our second example is hitting

others. We did an observation of a full

hour, 60 minutes and we had six hits.

How often is it happening per minute? 6

/ 60 is 0.1. So 0.1 hits per minute. And

we want to reduce it. So we introduce an

intervention and we let's say we observe

30 minutes and in 30 minutes she hits

twice is our reduction. 2 / 30

is

06 hits per minute. So that is a

reduction from 0.1. So there was a

reduction. What if you observed for 20

minutes and you got six hits? So 6 / 20

is.3. So.3 would not be a reduction from

that 0.1. So it actually

increased. These are all minute

examples. I meant to give you an hour

example, but we'll just do minutes.

Okay. So hand raising. So you observed

15 minutes and they rose their hands 12

times. And we'll say, you know, the

teacher is struggling with that. So

let's try we want to reduce that. All

these examples, I put minutes cuz that's

what I was thinking of. They're

uncommon. I haven't seen a lot of kids

who ask for help 10 times in 20 minutes

or raise their hand 12 times in 15

minutes. These would be very frequent

behaviors. You might want to use minute

for very frequent behaviors, but a lot

of times we might be using hour. Okay,

we have 15 minutes. We have 12 instances

divided by 15 is 0.8 raises hand per

minute. Almost one, but not quite.

Our next way, we're doing three ways to

pull data together. That was rate. Now,

our second is mean duration. So, you

have a lot of duration data. If you have

frequency, you're going to find a rate.

And that's how we're going to compare

our data. If you have a duration or a

latency or an IRT, you're going to find

a mean

duration. Mean duration is the average

length of time that the behavior lasts.

It's calculated by dividing the total

duration of the behavior by the number

of times it occurred. This metric is

useful in tracking continuous behavior

such as tantrums or time spent working

on a task to understand how long they

last. When to use mean duration behavior

is continuous once it starts. So they

have a long behavior. You want to

measure how long how long the behavior

lasts. So mean duration gives you the

average length of the behavior which can

help you identify patterns or track

progress. Common examples, tantrums,

time spent working, crying or screaming

are all possible mean durations. And the

only other thing you might do with

duration, I'll show you the examples in

a second, is if you had you might add

them up to total duration. So you have

like one behavior occurred 3 minutes,

another behavior occurred 2 minutes, and

then the last one was 1 minute in your

10-minute observation. And someone might

be, what's the total length of behavior?

And you'll add those up and give them

the number. That's the only and that

doesn't happen as often. Most the time

you're doing mean, duration. So we

have three tantrums. You took data and

they're perfect minutes. Okay, so the

first one was 6 minutes. I just made

them whole minutes to make it easy. 6

minutes. Then you had an 8 minute

tantrum and then you have a 10-minute

tantrum. What's the average duration of

this child's tantrums? You're going to

add up 6 + 8 + 10 and then you're going

to divide that by

3. 6 + 8 + 10 is 24. So you get 24

/ 3 and that's 8. So on average, they

spend 8 minutes tantruming. So now we

want the tantrum duration to get less.

So the next observation you take the

durations of the tantrums and you see if

now the average duration is less or more

by doing the exact same

calculation time spent off task. We have

3 minutes they were off task then 4.5

minutes and then 2.5 minutes. You can

round durations up to make them easier.

I often let RBTs do that or behavior

texts. The total is 10 minutes if you

add those all together and there's

three. We're getting that three from

there's three times they engaged in the

off task behavior. So 10 / 3 is 3.33. On

average when they go off task they spend

about 3.33 minutes off task. So that's

the length. So we want to reduce that

now if it's off task. If it was on task

we might want to increase it. So

independent play. So you're seeing how

long someone engages in independent

play. First they spend 12 minutes in

independent play, then 10 minutes, then

they spend 14 minutes in independent

play, and then they spend 8 minutes. So

now you have four, you saw them

independently play with toys four times

in your observation. Those are the

lengths. So it was a total they spent a

total of 44 minutes

in four different examples. So 44

minutes divided by 4 is they're spending

about 11 minutes engaging in in

independent play right now. So now we're

going to do an intervention to increase

that. You're going to see if you can get

that 11 minutes

longer. Our last way is percent correct.

This is how we take interval data. This

is also sometimes you take data where

they have an opportunity to do something

and it's just a yes or no. So they did

it or they didn't. So, for example, a

lot of times is answering questions

correctly. I didn't talk about this type

of data. So, say someone's like, I wish

they would answer questions correctly

more. They always just say blah when I

ask them a question. A teacher complains

about that. So, what you would do is you

when you were observing, you would go,

how many opportunities do they have to

answer the question? Every time the

teacher asks a question, you might put

one. And then did they do it correctly

or not? Yes or no? Plus or minus.

Another question. Yes or no? Plus or

minus. You would also use this for that

type of data as well. I always call it

opportunities correct. And also

criterion data, which is when you have a

we'll talk about that in task list. When

you have a task list, how many of the

steps did they do correctly? When to use

percent correct? There's a clear right

or wrong answer. Just like interval

data, it's either a yes or no. Any data

where it's a yes or a no, you're using

percent correct. You want to measure

accuracy. Someone performs a test task.

And then common examples are matching

pictures, answering questions,

completing steps are all situations

where percent correct is a helpful

metric. Discrete trial training, which

we'll talk about, they have an

opportunity to correctly answer

something and you put yes or no. So, for

example, you'll put two cards down,

point to the bird, they either do it

right or they do it wrong. There's no

other data you need to collect. So, it's

yes or no. It's un

opportunities. So, they're identifying

color. So, you're putting crayons out

and saying, "What color is this?" They

might say red, which is correct, or they

say, "I don't know," or "Green, which is

incorrect." So, then you'll have your

trials. So they had eight correct

responses within the 10 trials. And

you'll see more how this looks for

discrete trial training. 8 / 10 is8 *

100 is they got 80% correct. So this

one's like a task analysis. There's a

fivestep hygiene routine and they follow

multiple directions. You have a box next

to each step and you write yes they did

it or no. For example, brushing teeth.

Did they take the toothbrush, put the

tape paste on it, brush all four

quadrants, spit the toothpaste out, you

know, whatever your multi-step direction

is. So, they did four steps correct in

this. So, but they had five steps. So,

it would be four / 5, which would be8 *

100 would be they got it 80% correct.

So, we'd want to increase that. Same

with discrete trial training. We'd be

trying to increase that. And then

spelling test is the academic one we

should be familiar with just from being

in school. You had 15 vocab words. They

got 12 correct. 12 divided by 15 * 100

is 80%. You have like 10 intervals. They

engaged in behavior in six of them. So

it would be 6 / your total intervals

gives you that

percentage. So why are we doing this?

We're doing this. Those are the three

ones that as a behavior tech you'll use

most if not all the time. So you do need

to be familiar with them. Behavior data

helps you monitor change. These

calculations give us a way to compare

and guide interventions. It shows us the

patterns like are things increasing? Are

they decreasing? If I just say oh she

hit I observed for 3 hours and she hit

twice. That rate will make it easier for

everyone and it helps us refine those

goals.