Measurement

Parameters to Measure

After initial interviews to clarify referral concerns and to give you an idea of the behaviors you are to measure, set up measurement processes for the yourself or surrogate observers.  We have standardized forms for Event Recording, Duration Recording, Likert Scales for estimating the intensity or force of behavior, Sampling, etc.  The right form can streamline measurement and increase the accuracy of the results.

We may also add a blank sheet of paper.  This way, the surrogate observer can write something down whenever they want to, including questions that might want to ask us later.

Explain to the surrogate observer why measurement is necessary.  Here are three reasons you can give:

  • It clarifies the existence and the extent of the problem, justifying the expenditure of time and funds to intervene;
  • It provides an objective “baseline” level of function that enables us to measure progress.  By doing this, the evaluator may immediately gain insight into something that was previously described in subjective terms such as “a lot,” “always,” “never,” “tantrums that last an hour (when in reality a three-minute tantrum has always seemed like an hour to the Teacher), etc.;
  • It provides information that might help us understand why behavior occurs.

Parameters of Measurement

Various parameters of measurement are available, and your choice should be logical, not automatic.  Measurement is not as precise as perhaps we would like it to be.  Basically, we are looking for the most meaningful and representative baseline numbers, and ways to measure behavior that are minimally intrusive.

Frequency

Sometimes called “Rate of Behavior.”  The frequency describes how often the behavior occurs.  Frequency is typically described in terms of: number of events per period of time, or the number of target events per total events (percentage).

Examples:

Susannah tantrums average of 4x per day
The median of Philip’s aggression towards his brother is 5 times per play period.
Tran wakes up an average of 3 times per night.

Quantifying Frequency/Rate

You can report the frequency as # events/period of time, as in the above examples.

Reporting as a Percentage

# target events/# of total events can also be expressed as a percentage (e.g., Ashley responds to directions appropriately about 30% of the time; Craig has problems an estimated 60% of total transitions).

Duration

How long the behavior lasts.  It is reported in terms of clock or calendar time.

Examples:

The duration of Susannah’s tantrums averages 10 minutes.
Tran typically (mode) sleeps 5 to 6 hours per night.
Tran’s bouts with insomnia last about 3 days at a time.
The mode of Charlie’s TV watching is 4 hours per day.

Latency

This has to do with how long it takes for someone to start doing something.  Latency is just another form of duration recording – it is reported in terms of time elapsed.

Examples:

It typically takes Kristin about 10 seconds before she can respond to a question or a prompt.
It can take up to 30 minutes before Carl makes the transition between break time and beginning his work.

Topography

Topography refers to the description of what a behavior looks like.  In our measurement table, we don’t use the category “topography” per se, because there is a section for Description of the behavior.  The Description section is where topography is detailed.

Shape

The shape of a behavior has to do with what observable forms the behavior takes when it begins, how it escalates, and how it typically resolves.  This is very important when considering how the ongoing consequent conditions might be influencing the behavior in a dynamic way.

Examples of Topographical Description:

Lucia bolts out in to the street when she sees an open door.  This has occurred only 2 times because the Staff is very vigilant over this.  Nevertheless, the behavior is potentially very dangerous.

Roseanne can sometimes hit herself when she’s overwhelmed with frustration or anger.  The behavior won’t last long if she is given time and space for the emotions to subside, but if she is escalated further, she will hit herself to the point of sustaining bruises.

Intensity or Force

How serious, strong, or injurious the behavior is.  Sometimes called “force,” this describes how the behavior impacts the individual, others, or environment.  It is easiest to describe the intensity of a behavior in words, but a very useful method is to make and use a Likert Scale.

Likert Scale

You can also report intensity or force with a Likert scale.  When you use this form of reporting, you have to give your definition of the scale – usually by describing the extremes.

Example:

On a scale of 1 to 5, where 1 represents no injury at all, and 5 represents injury to the point of permanent tissue damage, Roseanne’s self-injurious behavior is rated by her Guides as ‘3’ most of the time, and occasionally at ‘4’ or ‘5.’

Setting

The “setting” is where behavior occurs.  Keep in mind that it is pointless to know where a behavior occurs unless there is some correlation between the setting and the behavior.  What you list as setting information should provide information about correlations between environmental features of the setting and behavior.  You might prefer to simply list settings in the table, and describe the correlations between individual environmental factors elsewhere in the wording of your functional analysis.

Therefore, we define setting as any information about features of the environment that have some relationship to the behavior.  Such information can include when the behavior occurs, who it occurs with, when it doesn’t occur or who sees it and who doesn’t, etc.

Examples:

(Where) Celina’s tantrums occur most often at his daycare setting.  The daycare environment has less structure than home or school, there are more changes in routines and other dynamics, and there is more noise there than at home or school.

(Who) Maya never has accidents at school.  There, she is taken to the bathroom every hour.

(When) David’s hyperactivity seems to get worse as the evening progresses into night.

Problems with Measurement

MEASUREMENT ALONE PROVIDES LITTLE UNDERSTANDING OF BEHAVIOR.  “A calf does not get fatter by measuring it over and over.”

  • Measurement describes the forms of behavior but not the functions of behavior.  Correlations between behavior and environmental events (antecedents and consequences) are not necessarily causal when it comes to behavior.
  • Measurement in the real world is fraught with error.  While we want to provide information as accurately as possible, we do not have the luxury of laboratory conditions.
  • First of all, people may not be careful recorders.  They are busy with other things, and they often do not witness or record all of the events in question.
  • Secondly, non-professionals (and professionals) can lack objectivity.  This is why it is so important to define the behavior operationally (see Operational Definition of Behavior) for them.  Even when you do this, behavior is experienced in different ways emotionally.  It is a good idea to check for agreement between Guides regarding the operational definition of a given behavior.

The Uncertainty Principle

Besides the obsession with data that some may have (and their subsequent choices to teach skills that are easier to measure), there can be an inverse relationship between measurement and accurate or valuable knowledge.

The Uncertainty Principle comes to us from Physics or more specifically, Quantum Mechanics, which I know almost nothing about.  But it was brought to my attention in a book entitled Investigations in Behavioral Epistemology, which featured highly regarded Behaviorists and contributors to the Journal of Experimental the Analysis of Behavior (JEAB) debating the problems Behaviorism has with analyzing complex behavior.

Basically, the principle explains that the more attention you pay to one parameter, the less you may be able to pay attention to another parameter, perhaps a more important one.  Here’s the example most often cited:

If you want to measure parameters of a ball’s flight through the air, you can:

  1. Measure its velocity (the speed in which it travels) or;
  2. Measure its position (where it is at any moment)

If you concern yourself with the speed of the ball, you are not going to measure its position easily…
If you concern yourself with the position of the ball, you are not going to measure its speed easily.

How does this apply to behavior?  Well, it applies most when you measure rapidly changing stimuli, such as those found in chains of reciprocal interacting, changes in affect, and other complex behaviors.  If you stop to record data, you interrupt the flow of ongoing social communication.  If you pay attention to the frequency in which an individual skill is noted in a long chain, spontaneous and frequently changing social interaction, you are in danger of losing sight of the context variables at play.

We have forms that are based on how developmentalists record data that are useful enough in determining progress and effectiveness of treatment, but that may not yield the kind of precision so easily afforded in stop/start trials or cued responding.  These are certainly more useful when using techniques to promote long chains of reciprocal and dynamic social interaction, and they are easier for parents and Teachers to fill out.

Measurement Forms

We supply a spreadsheet that includes many different forms for measurement.  You can use those, or you can make up your own (you stand the chance of my rejecting the form as inaccurate or otherwise inappropriate if you do not check with me first).   All forms should include the operational definition of the behavior observed.

Event Recording

Technically, all recordings of behavior are event recordings.  Sometimes called a “tally” or “frequency count,” an event record shows how many times something occurred.

An Event Recording can give you a total of the number of times an event has occurred.  Since events always occur within some time period, it is almost impossible to have a pure event record – you almost always come up with a rate of occurrences /(over) time.

A rate of behavior can be calculated by taking the number of behavioral events and dividing it by a period of time.

Examples of Results of Event/Rate Recording:

The average of Susannah’s tantrums is 4 per day.

A rate of behavior can also be calculated by taking the number of behavioral events and dividing it by the total number of opportunities:

Kenneth complied with 6 of 10 requests during the observation period, or 60% of the time.  Guides report that this is consistent with their estimates of his rate of compliance most of the time.

Event Records are best used when:

  • Behaviors are fairly discrete, that is, they have a clear beginning and ending to them, and they are easily discriminated from other events.
  • Behaviors occur at a fairly low-frequency, and are easily counted.
  • When intervention targets the reduction or increase of the number of target behavior events.

All observational recording methods and forms should show:

  • An operational definition of the behavior
  • The beginning and ending times or dates of the measurement period.

Event Records are probably not best when:

  • Behaviors occur at very high frequencies (e.g. looping behaviors such as rocking or hand-flapping)
  • Behaviors occur once, but last for extended periods of time (e.g. watching television)
  • Behaviors that tend to occur every time (e.g. picky eating that happens at every meal).
Interval Recording

This is typically used for higher-frequency behaviors.  As the name implies, behavior is observed in order to determine whether it occurs or not within a given interval.  There are 3 main types of interval-type records commonly used: Interval Records; Time Samples, and; Probes.

Interval Recording can estimate how many times a behavior occurred within the observation period only, and does not produce the most reliable rate of behavior.  It should not be used for an extrapolation of rate across all settings.

  • IRs can provide a sense of the continuity of a behavior – rather than the number of times it occurs per a [representative] period of time.
  • Interval Recording and Time Sampling differ in several ways, which makes them useful for different purposes.
  • Interval Records use small intervals, usually no longer than a minute.  The intervals are equal in duration.

Time Samples use larger intervals, usually minutes rather than seconds, and the behavior is recorded only if it occurs at the end of the interval.  This makes them useful for seeing the distribution of behavior over a period of time.

The forms for Interval Recording and Time Sampling look the same.  They both consist of rows or columns of boxes for making notations of the presence or absence of behavior.

Example: Interval Recording of 3 total minutes for hand-flapping behavior, where each interval represents 30 seconds:

1

2

3

4

5

6

X

O

O

O

X

X

Interval Recordings are best used when:

  • Behaviors occur at frequencies that are too high for simple Event Recording.
  • You can count on behavior occurring during the observation period AND the observation period is representative of conditions throughout a “typical” day or activity.

Interval Recordings are probably not best when:

  • Behaviors occur at relatively low frequencies.

If so, there is a chance that behavior will not occur at all during the observation period.  Or, if behavior is sparsely episodic, an extrapolated estimate of the total number of events in a “typical” day or activity would be misleading.

Time Sampling

As mentioned, the notation for an interval in a Time Sample is typically done at the end of the period.  The observation of behavior is done only at the end of the period. This is good when the recorder has to do other things.  Another convenience is that intervals in a TS do not have to be equal.  The lengths of the intervals could be averaged into an average interval. This might be helpful when asking a Teacher to record behavior.

Probes

A probe is another interval method, where behavior is observed in short intervals that can be spread across an entire day or longer.  For instance, the Teacher can record behavior “here and there” throughout the day, observing in short intervals (consisting of anywhere from a few seconds to several minutes each).  Probes are observations made at different times, rather than a single block of time, compared to IR and TS, which use contiguous intervals.

When using this method, Guides must be advised that they should not record only when behavior occurs.  That is, they might see behavior occurring when they were not recording, quickly get the record, and then enter the data.  This would result in a misleadingly high estimation of the occurrence of the behavior.  It is better to select probe times beforehand, preferably in a randomized way.

Limits to Interval, Time Sample, and Probe Recording
  • Because high frequency behaviors occur so quickly, the behavior can occur while the observer is making the notation.  Therefore, it is easy to miss occurrences of behavior when using IR.
  • IR methods can be somewhat formal and unnatural for family life, and if the method is not made easy, there is a likelihood the surrogate observers will fail to follow through.
  • With IR or TS, you can record a frequency of behavior, where the data is applicable only to the interval observed.  This is not a reliable means of extrapolating a rate.
  • As with Event Recordings, IR methods yield a quantity of behavior, but do not necessarily offer insight into why behavior occurs.

Recording of Controlled Presentations

This is the typical method for scoring “trials” of behavior, such as in direct teaching (ABA/Special Education techniques).  The observer records whether the trial was correct or incorrect, or any other parameter specified.  A key must be provided on the form that explains the symbols that will be used to fill it out.

Typically, the following criteria are examined when recording controlled presentations (teaching trials):

Correctness

The accuracy of the response is recorded: e.g., Correct (+); Incorrect (-); Partially Correct or Approximation (±), etc.

Support Needed

The directness or intensity of the guidance required from the teacher is recorded: e.g., Full physical prompt or motor guidance (FP); Partial physical prompt (PP); Mild physical prompt – such as a tap on the shoulder or hand (MP); Gestural prompt or pointing (GP); Modeling or demonstration (M).

Verbal Prompts also vary in their directness and intensity: e.g., Direct verbal prompt: “Put it in the box…”  “Do this.” (DVP); Indirect verbal prompt: “Where does it go?”  “What do you say?” “Try again.” (IVP)

Presence or Absence of Interfering Behavior

Since direct teaching methods often require compliance from the Student, uncooperative, resistant, or failure to try (no attempt) behaviors can be recorded.

Mastery Criteria

Direct teaching methods often have separate records indicating the dates in which objective criteria have been met.

  • There should be forms for recording mastery of all objectives throughout the intervention, if not assessment periods.

Anecdotal Recording

Anecdotal records can be subjective, but they are the typically the most natural form of recording for Guides.

It is perfectly OK to ask Guides to indicate why they think the behavior occurred.  If anything, this gives insight into the attributions (see Parent Attributions of the Behavior, below) the Teacher has regarding the behavior.

It is appropriate to ask the Parent, Teacher or Caregiverwhythey think the behavior occurs.  Right or wrong, what they say is clinical information to be considered.  What is wrong is to consider the opinion as fact.  This information is potentially relevant anywhere in the report where a discussion of Stakeholder responses to behavior is concerned.  Again, this is due to the belief that people behave according to their perceptions and conceptions of reality, or in other words – their attributions of the behavior (of course, data could do a lot to change these perceptions).  A concise summary of Stakeholder beliefs and attributions of the behavior goes into the section in the report entitled “Stakeholder Attributions

Parent, Teacher or Caregivers may claim that they don’t know why a behavior occurs, or they may remind you that it is your job to figure it out.  It is not your job alone.  Figuring out why the behavior occurs is a collaborative effort, where only a hypothesis can be formed.  The strength and potential accuracy of the hypothesis comes from the data collected and the consensus formed between the stakeholders and the Analyst.

It is a good idea to say something to the Teacher such as, “Write down any information you feel would help us understand why this behavior occurs.  Look for patterns of behavior and associations with other events if you can. Try to tell us why you think it happened – even if you don’t really know – tell us what your gut feeling was…”