APPLIED BEHAVIOUR ANALYSIS

Operational Definitions

Lesson 2
LESSON 2

Operational Definitions and Data Collection

1
SECTION 1

Operational Definitions

2
 Steps to Creating Behaviour Change g.! Evaluate the effect of the treatment.
(Lesson 5)
Plans
Did you notice that choosing an
There is a process that one follows when
intervention does not happen until the
creating any behaviour change plan.
4th step?
Here are the basic steps:
(In further courses you will elaborate on this process.)

a.! Operationally define the target Even before you decide how you will
behaviour to be changed (taught, change someone’s behaviour, there are
increased, or decreased). (Lesson 2) several things that must be done.
b.! Take baseline data. Graph it. (Lesson 3) In this lesson we will cover the first 2
c.! Determine a behavioural objective. steps.
(Lesson 3)

d.! Decide on your intervention. (Lessons 3, 4, 6

- 8, 10-12; Also other courses in this program)

e.! Implement the intervention.

f.! Continue taking data. Graph it. (Lesson 3)

3
 Before seeking to change behaviour, be it
your own or someone else’s, you must
first be very clear about what you are
trying to change. You must write an
operational definition of the target
behaviour. Remember from lesson 1 that
Operational Definitions
behaviour is something that can be
observed and measured.

An operational definition is simply a

description of what the
target behaviour looks or sounds like.

4
Operational definitions do not include behaviour is. You all want to be
any suggestions about how we will collecting data on the same behaviour.
accomplish that behaviour – no program
3.You may be delivering reinforcement or
ideas are included yet. punishment contingent (based on if the
behaviour occurs) on the behaviour.
You want to make sure that everyone on
the team is reinforcing or punishing the
Why are Operational Definitions same behaviour!
Important:

Some questions you may ask while

1. You must be clear about the target developing your operational definition:
behaviour to measure it correctly and
Do the teacher and the classroom aide
consistently through the behaviour
both understand and agree that the
change procedure.
operational definition has captured what
2.Because it is a team approach, there they mean when they use the same
must be agreement on what the target words?

5
If we are asking the parents to become •typically include only one behaviour
involved in a version of the same per definition
program for the home environment, are
•begin by giving a name to the
they also clear and in agreement with behaviour
what we mean?
•describe the behaviour only in terms
Is the behaviour defined well enough that of what we can see or hear
all staff are able to agree when it occurs
and when it does not? •include all the necessary aspects of the
behaviour
•be reasonable
•explain whether to include certain
Operational likely exceptions or to stop recording
definitions should: when these exceptions occur

•be written in sentences

•be as short as possible, as long as
necessary

6
 Review 1.1 Operational Definitions
Operational definitions
should not include: Question 1 of 6
Operational Definitions should be written in point form:

•any discussion as to why the behaviour

might be happening
•any details of the program to change
the behaviour A. True

•any statements about eventual B. False

outcomes of the program

•any descriptors of the behaviour that
cannot be seen or heard

Check Answer

7
 Creating an Operational Definition Example of Operational Definition:
Let’s begin operational definitions by exploring what we might
mean by being seated. (scroll through these pictures)
The target behaviour of sitting in a chair
occurs whenever John meets all of the
Gallery 1.1 Being Seated Example
following criteria:
a)! he has his buttocks on the horizontal
part of the chair
b)! his legs and feet are under his desk or
chair
c)! his head is above the top of the desk.
If he does not meet the above criteria for
What might an operational definition of “being seated” need to in- 3 seconds or longer, then he is considered
clude. (Some ideas follow but…no looking first!)
“out of seat”.

Being Seated

8
Important Points about Operational more than 3 seconds) to go outside of the
Definitions: criteria without the observer scoring “out
of seat”. This would allow him to bend
•! In the above definition the target
over and pick up a pencil from the floor if
behaviour is given a name –“sitting in a
needed but not enough time to walk over
chair”.
to someone else in the classroom and talk
•! In order for observers to say John is to him or her.
sitting in his chair, John must meet the
•! Not programmatic. In the operational
three criteria listed above, all of which are
definition, no explanation is offered as to
observable criteria.
how to help John remain in his seat. This
•! The criteria are reasonable – John is will be stated later in our treatment
not required to sit with his back against program. No expectation is stated about
the back of the chair. He is not required how long or how often we want John to
to keep his feet flat on the floor. (Are remain seated as this will also be stated
yours right now? If they are, will they be later in our program.
continually on the floor while you are
•State Exceptions. For this behaviour
reading and working on this lesson?) He
there is no exception needed. However,
is allowed a very short period of time (no

9
consider the definition of on task In our program we may be trying to
behaviour, in which children are change more than one behaviour but it
expected to be seated at their desks and is best if each one is described
doing their assigned work. Here the separately.
target behaviour includes two •If we had said in the above definition
components – the students will only be that John also needed to be on task
scored as being “on task” if they are while seated, we would have introduced
both seated and are doing their work.) If a second behaviour into this one
a student asked for and got permission definition. The problem this might
to get up and sharpen his pencil, would create is that after observing John for an
he still be on task? Would this be an hour we may find that he never met the
allowable exception or would it be criteria. The question then arises – why
counted as off task? Should we simply did John not engage in the behaviour of
stop recording until he is back in his being seated? If we have a videotape of
seat? the observation period to play back, we
may see that John actually was properly
Describe one behaviour at a time. The in his chair for the whole time but,
definition only describes one behaviour.

10
 because we had also said he had to be Wording
on task to meet our criteria for There are a number of ways to word an
recording, he scored zero.
operational definition.
•If the behaviours had been separately “Being seated is defined as …..”
defined and recorded we could have
clearly seen that the sitting was not the “The behaviour of being seated is defined
problem and would have seen the as …”
opportunity to reward John for sitting in “Johnny is considered to be seated when
his chair. Important data would have …”
been lost if our definition had included
“Being seated is considered to occur
more than one behaviour. It would be
when…”
better to write a separate definition for
each behaviour and record each
behaviour separately for the same time
period.

11
Review 1.2 Operational Definitions Watch this video and choose one behaviour that you are
going to operationally define. Write your operational
Question 1 of 3 definition and we will review it later.
In the classroom, the teacher asks you to measure Tony’s
learning in math, spelling, and writing behaviour. You go to
write an operational definition. How many operational defini-
tions would you have to write?
Movie 1.1 Clip 44 - Severe Autism

A. 1

B. more than one

Check Answer

12
SECTION 2

Data Collection - Part I

13
 3.! Data taken during treatment show you whether or not your
Data collection – Why do it? treatment is affecting the behaviour. They allow you to decide if
Having operationally defined the behaviours of concern, the next any changes are needed.
step is to collect data about the occurrence of the behavior(s). You 4.! If your treatment is effective, data will help show others –
must collect data before you start a program to change the staff, parents, and teachers, funding sources - that the treatment or
behaviour. Data collected before the behaviour change program is teaching approach is successful and that you should continue your
begun are called baseline data. You will also need to continue to efforts.
collect data during your program.
5.! If progress is slow, the progress may not be detected by those
Why collect all this data? who are working day to day. Data can show if long term progress
1.! Baseline data allow you to assess whether a problem really is being made, and thus be a source of encouragement. There is
does exist: something very reinforcing about seeing a graph that shows
progress!
Is there a significant difference between the behaviour of this When do you collect data?
person and that of his or her peers? For example, the behaviour of
a child with autism may be seen as different or deficient, when Data need to be collected before a program is started, during the
data collection shows the child to be more similar to peers on some program and ideally some months after the program has finished.
measures or in some situations.
• Before: to know where to start

What is the current level/ frequency of this person’s behaviour? Is • During: to know if the program is working
it below that expected of peers – a behavioural deficit -- or is it
• After: to know if the changes were permanent
more frequent than that of peers – a behavioural excess?

2.! Baseline data help you determine what reasonable first

objectives should be. You will not set your first objective at 60 Note that it is also possible to take data on your own behaviour.
minutes of independent reading, if the child is only managing 1 This is called self monitoring.
minute at baseline. It may be more reasonable to set 1 minute as
your first objective.

14
 Which dimension you choose will depend on the target behaviour,
Dimensions of Behaviour and the dimension you are most interested in. Sometimes you are
What aspect of a behaviour do you measure? There are number of interested in many dimensions at once. You may be interested in
dimensions that you may be interested in. how often your client aggresses (frequency), how long he
aggresses for (duration), and how intense the aggression is
(intensity). If you are measuring a child’s play initiations, you may
interested in how long it takes before she asks another child to play
Frequency (how often it occurs)
from the time she arrives in the playground (latency), and how
often she initiates play in a 30 minute play period (rate).

Review 1.3 Dimensions of Behaviour

Rate (frequency divided by time)

Question 1 of 5
You want to measure the number of times that a child puts
up his hand. What dimension would you measure?

Duration (how long it lasts)

A. Rate

B. Latency
Intensity (how much force is used)
C. Frequency

D. Duration
Latency (how long it takes for it to start)
E. Intensity

Check Answer

15
 Direct and indirect sources of data
There are a number of recording systems to choose
from depending on the nature of the behaviour or the
resources available.

Indirect sources of data include second hand

information reported by someone else. Indirect data
come from reading case files or interviewing people
asking them to remember the past accurately. You must
be aware that the data you get this way may or may not
be accurate. The person making the statements may
have been skilled at observing or very poor at
observing. Many factors can influence the accuracy of
the data.

Direct observation is a more accurate way of collecting

data. This involves directly observing and recording
the behaviour with your own eyes and ears, and is
more accurate. An appropriate recording system can be
chosen and used. In ideal conditions, a second
independent observer is also used to verify the results.

16
SECTION 3

Data Collection Part II

Direct Observational Systems 3. product recording

There are four different types of recording meth- 4. interval recording or time sampling.
ods:

1. continuous recording

2. percentage of opportunities

17
1. Continuous Recording
Continuous recording involves watching the client throughout an Count this behaviour in the movie using CONTINUOUS RECORD-
observation period, and recording each time a behaviour occurs. ING!

You may record the frequency, duration, latency, or intensity of the

Movie 1.2 Frequency Behaviour
behaviour using this recording method.

Here is an example from our data recording that we did at Camps

through the City of London:

Count this behaviour in the movie using CONTINUOUS

RECORDING!
Notice the column on the right that measures the frequency of inter-
actions in an entire 15 minute period.
Now right down your number that you counted on a piece of pa-
per for the activities!

18
2. Percentage of Opportunities Write down the percentage of opportunity count on a piece of pa-
per to bring to the activities. This should be recorded in a percent-
Some behaviour may only occur when the opportunity presents it-
age.
self.
Movie 1.3 Collect Data Using Percentage of Opportunity
For example, catching a baseball can only occur when the baseball
comes towards you. Answering a question can only occur when a
question is asked. Compliance to instruction can only be done
when an instruction is given. For these types of behaviours it does
not make sense to record the frequency or rate of the behaviour. It
makes most sense to record the percentage of opportunities.

Percentage of opportunities is calculated by dividing the number

of times the person did the behaviour, by the number of times they
had a chance to do it, and multiply by 100%. For example, if the
ball was thrown to the child 5 times, and she caught it 4 times, this
works out to be 4/5 X100% = 80%.

For example if you were measuring compliance, you would record

how many instructions an individual was given, and how many
the individual complied to. Then you would calculate the percent-
age of compliance. If Billy was given 10 instructions and he com-
plied with 6 of them, percentage of compliance is
Collect Data on the following behaviour using PERCENTAGE OF OP-
6/10 X 100% = 60%.
PORTUNITY.
Other behaviours that would be recorded in this way include cor-
Note - They use a little different terminology here.
rectly answering questions, and responding to peers’ request to
play.

19
3. Product recording Using Permanent Product Recording, record the number of ques-
tions the child completed on her math sheet.
When the behavior produces a permanent change in the environ-
ment, then the results of the behaviour can be recorded after the be-
haviour itself has occurred.

If the task is to wash the dishes, then the end result is observable
anytime after the cleaning has occurred; the dishes in the drying
rack may simply be counted. The same is true for making a bed,
doing math questions, or writing a paper for school. (Product re-
cording is sometimes referred to as “permanent product record-
ing”.)

Product recording is convenient in that the observer need not neces-

sarily observe the person as the behaviour is occurring. The ob-
server need not even be present when the behaviour occurs. The
one drawback of product recording is that, as the observer, you can-
not always be sure of whose behaviour resulted in the product you
are measuring. If you are measuring a child’s writing behaviour
by counting the number of sentences she has written, are you sure
that the child wrote the sentences herself, or did she get her friend
to write some? If you are measuring a client’s self injury by count-
ing the number of bruises on his arms, are you reasonably sure that
the bruises are all self inflicted?

Write down your permanent product recording on a piece of paper

for your activity centres.

20
3. Interval recording There are two ways of recording interval data: whole interval
and partial interval recording.
Another way to measure behaviour does not involve recording
every instance of the behaviour. Instead, it is possible to divide up A. Whole interval recording
the observation period into small intervals, and simply record if
For the behaviour to be recorded it needs to occur throughout the
the behaviour happened in the interval or not.
entire or whole interval.
To do this, the observer will need a timer or other device set to sig-
If the behaviour does not occur for even a small part of the interval
nal the end of the interval. When the timer goes off, the observer
then it is not recorded in that interval as having occurred at all.
checks off on an interval recording sheet whether or not the behav-
Whole interval recording gives an estimate of the duration of the
iour occurred.
behaviour. The shorter the interval, the more accurate the estimate
Interval recording results in a less accurate measure of behaviour, is.
but it has its advantages. With interval recording is not necessary
to be able to detect exactly when the behaviour starts and stops, as
it is with frequency or duration or latency recording. This can be B. Partial interval recording
useful when the beginning and ending of a behaviour is not clear.
For example, if you are observing a behaviour (e.g., finger flicking) If the behaviour occurs at any point within the interval then it is re-
that occurs so rapidly that you cannot count the number of times it corded.
has happened, you may opt to use interval recording. Traditionally the behaviour is only recorded the first time it occurs,
Interval recording is also useful for busy observers who have other so this system is not a complete record of the frequency of the be-
tasks to attend to, such as teaching students or caring for children. haviour. Partial interval recording can be useful if you are record-
ing behaviour that is quick (such as hits or yells), but that occurs at
Another advantage of interval recording is that, unlike continuous too high a rate to allow for accurate frequency counts. Partial fre-
recording, it can show the distribution of the behaviour over time. quency data is an estimate the occurrence of the behaviour, but it is
This may offer some clues as to why the behaviour is occurring useful when frequency or duration recording is not practical.
and therefore how to help. Again, the shorter the interval, the more accurate the estimate is.
(You will notice when you start to look at research articles that par-
tial interval recording is often used in research.)

21
If more than one behaviour is being recorded, this may be done by 4. Time sampling
adjusting your data sheet to allow for coded interval recording.
In time sampling the time observation period is again divided into
Each behaviour to be recorded is given a code and these codes are
intervals. However the behaviour is only observed at the end of
put into the interval boxes on the data sheet. Each time the behav-
the interval. The observer sets a timer for the time interval. When
iour occurs, its code letter can be crossed off.
the timer goes off, the observer records the occurrence or non-
occurrence of the behaviour. All the other times the behaviour is
not recorded. Time sampling is not a very accurate representation
of the behaviour. However it is a very quick and non-time consum-
ing method of data collection. It is far easier to do time sampling
at the same time as cooking dinner, than it is to do continuous re-
cording!

Review 1.4 Data Collection

Question 1 of 6
The number of times the person did the behaviour, by the
number of times they had a chance to do it

A. Time Sampling

B. Permanent Product

C. Partial Interval

D. Percentage of Opportunities

Check Answer

22
Interobserver Agreement (IOA) other recording the behaviour) must agree 80% of the time on the
occurrence or non-occurrence of the behaviour in baseline before a
(Also know as IOR – Inter-observer Reliability or
treatment or teaching program should begin.
IRR- Inter-rater Reliability)
If they are not able to achieve that level of agreement, then they
When most people first begin to record behaviour, they are likely
need to either review how the recording system works or they
to make mistakes. In fact, even veteran data-collectors make mis-
need to review the operational definition to ensure that they both
takes. Data collection mistakes can come about for a number of rea-
understand it in the same way. A revision to the operational defini-
sons. Since data are used to make important decisions in behav-
tion may be necessary to clarify certain aspects of the behaviour so
ioural programming, problems can occur if data collection is inac-
that the observers are more able to record the exact same behav-
curate.
iours.
If mistakes in data collection are being made, this can be improved
If research is being conducted on new applications of behavioural
upon. However, first you have to know if mistakes are being made.
techniques or new techniques are being explored, then the stan-
To do this, whenever possible, good programs should have a sec-
dard IOA is even higher – 90%.
ond independent observer recording the behaviour at least part of
the time. This person could either make their recordings at the All published research articles will report the inter-observer agree-
same time as the first recorder or they could do their observations ment calculations and ideally they should have IOA data for each
later by watching a video of the individual. If permanent product phase of an intervention – baseline, treatment and follow-up.
recording was being used, then the second person would just have
to count the products produced. Formulas for IOA
For most types of measurement, calculating IOA is very simple.
The data collected by the two independent observers can then be
Take the smaller measurement and divide it by the larger one, and
compared. Did they see the same amount of behaviour? If they
multiply by 100%. This works for frequency, duration, latency,
did, there is a good chance that they were collecting data accu-
permanent product and percent opportunities measures.
rately. If they didn't, one or both observers' data are inaccurate.
(No, you can't tell who was right and who was wrong!) The formula is:!smaller total counted X 100%
! ! ! larger total counted
How similar should the results be between the two recorders?

Over the years, standards have evolved to evaluate the accuracy of

the recordings. For most programs, the criterion is 80%. That
means that the two independent observers (who can’t see each

23
Example 1 (permanent product recording): Calculating IOA for Interval and Time Sampling Recordings

Observer 1 counted 33 toys not put away after the child was asked There are several ways for calculating IOA for interval and time
to put them away. sampling data. This is the simplest method:

Observer 2 counted 25 toys not put away. ! # of intervals both agreed ! ! ! X 100%
! total # of intervals observed
! ! 25 X 100% = 82.5%
! ! 33 To determine the number of intervals agreed upon, a count is made
The IOA for this recording is 82.5% agreement. This would be ac-
of the number of intervals both observers agreed EITHER the be-
ceptable.
haviour did occur OR the behaviour did not occur. In either case,
Example 2 (latency): the two observers have agreed. This total of agreed upon intervals
is then divided by the number of intervals in the recording session
Observer 1 found the average time taken to start to follow a re-
quest from John’s parents was 2 minutes and 55 seconds (or 175
seconds).

Observer 2 found the average time taken to start to follow a re-

quest from John’s parents was 2 minutes and 12 seconds (or 132
seconds).

The formula is: 132 s X 100% = 75.4%

! ! ! 175s
This result does not meet the acceptable standard and therefore fur-
ther skill training in using latency recording needs to occur for the
observers or further clarification of the operational definition
needs to be done.

24
 Interactive 1.6 Whole Interval

Application Time!
With a small group, go through the following stations to complete
your agenda for the day:

1. Direct Observation Systems

From the activity in the lesson, take the three pieces of data that
you counted and compare it with others in the group. If you got
the same number - congratulations. If you did not... go back and Behaviour: The child is circling one or both hands or opening and
watch the movie and see where you went wrong. closing.

Now, watch this video below as a group. Either do one large group
or split into two groups. Have someone in your group indicate eve- Interactive 1.5 Partial Interval
rytime there is a 15 second interval.

Watch the first video and do whole interval data collection:

In the boxes below each interval, put a check or an X to indicate if
the behaviour occurred throughout the WHOLE interval.

MAKE SURE YOU ARE CONNECTED TO THE INTERNET TO

WATCH THE CLIP!
Watch the second video and do partial interval data collection:
In the boxes below each interval, put a check or an X to indicate if
the behaviour occurred throughout the interval at least once
Behaviour: The child flaps one or both hands in an up and down move-
ment. The flapping occurs when there is 3 consecutive up-down move-
ments.

xxv
2. Inter observer Agreement

a. Watch the first video with a partner. Count the FREQUENCY of

the number of times the child correctly puts the shape on the corre-
sponding shape on the book. No peeking at eachothers.

Calculate your IOA.

Interactive 1.7 IOA

b. Watch the second video IOA for frequency data

with a partner. Both of you
Interactive 1.8 IOA
collect data. Time every 15 sec-
onds when the video actually
begins. Indicate if the behav-
iour occurs at all during that in-
terval. No peeking at eacho-
ther’s

Calculate your IOA.

IOA for partial interval data

xxvi