Section C Flashcards
Measurement Systems Info
Some classes of a behavior lend themselves to a specific of recording; others to another
Data
The material that influences and evaluates behavior analytic services
- Empirically guides you
- Quantitative results of measurement
- Raw data doesn’t tell you much, this is why ABA uses graphs
Occurrence
Something that occurs, and the action or fact of happening or occurring
AKA: Repeatability; Percentage measures
Occurrence AKAs
- Repeatability
- Percentage measures
Repeatability Measures (Definition & Details)
Behavior that is counted
- Behavior must have a clear start and end
- Do not use for continuous behaviors with long durations
3 Types: Count; Rate; Celeration
AKAs: Countability; Occurrence; Event recording
3 Types of Repeatability Measures
- Count
- Rate
- Celeration
Repeatability Measures AKAs
- Countability
- Occurrence
- Event recording
Count (Definitions & Details)
- The # of occurrences
- Not enough information for programming decisions
- Use when observations time is constant across observations
AKA: Frequency
Count AKAs
Frequency
Rate (Definition & Details)
- # of occurrences / time
- Unit of time must be standard to compare data
Rate (Use)
- Use for free operant behavior
- NOT for DTT and continuous behaviors
Celeration (Definition& Details)
- Changes in rate / time
- Must be a min of 7 measures
Celeration (Use)
Use for examining rates of response change over time; fluency
Percentage Facts
- An occurrence measurement
- Proportional quantity
- Min. 30 response opportunities
- No % over 100%
Trials to Criterion (Definition)
- An occurrence measurement
- A measure of the # of response opportunities to achieve a pre-specified level of performance
Trials to Criterion (Use)
- Evaluation mastery of a class of concepts
- Comparing 2 or more procedures
- Measuring skills
- Can be measured by count, rate, duration, and latency
2 Types of Derivative Measures of Occurrence
- Percentage
- Trials to criterion
Percentage Cons
- Assumed progress
- Doesn’t reflect fluency
- Restricts limits of data
Temporal Dimensions of Behavior (Definition)
- Measure behavior using timing
- Data measures the length and/or point in time of behavior
Categories:
- Temporal extent: Duration
- Temporal locus: Latency; IRT
Temporal Dimensions of Behavior Categories
Temporal extent
- Duration
Temporal locus
- Latency
- IRT
Duration (Definition)
Length of time the behavior occurs from the onset to offset
2 Methods to Calculate:
- Total Duration
- Duration per Occurrence
AKA: Temporal Extent
Duration (Use)
Primary time is the amount of time a client engages in a behavior; High rates of behavior
2 Methods to Calculate:
- Total Duration
- Duration per Occurrence
AKA: Temporal Extent
2 Methods to Calculate Duration
Total Duration: Add up the cumulative amount of time your client engages in the response in the total session
Duration per Occurrence: Measure the duration of time that each episode of the response
Duration AKAs
Temporal Extent
Latency (Definition)
Duration of time between the onset of a stimulus and initiation of a behavior
- Reported using means and medians of response latencies per observation period
AKA: Response latency
Latency (Use)
Use for when your primary concern is in the response latencies that are too long or short
AKA: Response latency
Interresponse Time (IRT) (Definition)
Duration of time that elapse between 2 consecutive instances of behavior
- Report as: Mean, media, and range of IRT per session
Rates of responding
- Shorter IRT = Higher rate
- Longer IRT = Lower rate
Interresponse Time (IRT) (Use)
When your primary concern is the time between behaviors; identifying interval criteria for DRO schedules
- Used in: DRL and DHR
Rates of responding
- Shorter IRT = Higher rate
- Longer IRT = Lower rate
Latency AKAs
Response latency
Types of Definitional Measure
- Topography
- Magnitude
Topography (Definition)
- Measurable and changeable dimension of behavior
- Physical form/shape of behavior
- NOT function; Different topographies can have the same function
AKA: Form
Topography (Use)
Measuring behaviors that require a specific form, style, physical skill to be correct and/or effective
Topography AKAs
Form
Magnitude (Use)
Use for measuring the force of a behavior when the specific intensity is needed for successful responding, severity of dangerous behaviors
AKAs: Strength; Force; Intensity; Severity
Magnitude AKAs
- Strength
- Force
- Intensity
- Severity
Direct Measure of Behavior (Definition & Use)
- Measuring behavior as it is occurring
- Use for: Free operant; Discrete; Minor displacement in space/time
Indirect Measures of Behavior (Definition & Use)
- When behavior can’t/ isn’t being measured as it is happening; Used when there are no means of getting direct access to a behavior
- Inferences are made about what occurred vs what actually occurred
- Violates the APPLIED dimension of ABA
Product Measures (Definition & Ways)
- Measure behavior after it has occurred by looking at changes in the environment
- Can be measured: Naturally (bylooking); Event recording; Timing; Time sampling
Product Measures Pros
- Freedom to do other things
- Allows data collection for inconveniently time behaviors (sleeping)
- Increasingly, correct, complete and continuous data (rewatching a video)
- Allows multiple measurers; IOA
- Can examine behaviors with multiple response classes and intricacies
- No reactivity
Should you use product measures? (4Qs)
- Is real time better? If data needs to be measured immediately > No
- Can this behavior actually be measured by product measure? If each response does not produce the same permanent produce AND other response can produce the product > No
- Will a contrived product negatively impact the behavior? (Camera reactivity)
- Is it costly? If it is too costly and not worth it > No
Time Sampling Procedures (Definition)
Direct methods to recording behaviors during intervals or at specific moments in time
3 Forms:
- Whole
- Partial
- MTS
AKAs: Interval recording; Discontinuous measurement system
Time Sampling Procedures (Use/Nonuse)
- Use for continuous behavior or high rate behavior
- Behavior with no clear start / end (Not discrete)
- Don’t use for: Specific critical (but infrequent) behaviors
Whole Interval Recording (Details)
- A form of time sampling
- Record behavior during the ENTIRE interval
- Produces an estimate of total duration
- Report the % of total intervals
Whole Interval Recording (Use/Nonuse & Outcomes)
- Use for: When you want to increase behavior because
- Don’t use for: Behaviors you want to decrease
- Underestimates the % of a behavior’s occurrence
- Longer intervals are more likely to produce underestimations
Time Sampling AKAs
- Interval recording
- Discontinuous measurement system
Partial Interval Recording (Details)
- A form of time sampling
- Overestimates the % of a behavior’s occurrence AND total duration
- Underestimates the rate of high-rate behaviors
Partial Interval Recording (Use/Nonuse & Outcomes)
- Overestimates the % of a behavior’s occurrence AND total duration
- Underestimates the rate of high-rate behaviors
- More conservative; use when you want decrease a behavior
- Don’t use: When the target behavior is one you want to increase
Momentary Time Sampling (Details)
- A form of time sampling
- Record behavior at the END of the interval ONLY
- Report the % of total intervals in which the behavior occurred
- Misses a lot of behavior
Momentary Time Sampling (Use/Nonuse)
Use
- When you are not able to continuously measure throughout the entire interval
- Easily identifiable behaviors
- When measuring multiple behaviors at a time
Don’t use for
- You care about the behavior being accounted for
- Low count and short duration
Planned Activity Check
A variation of MTS for groups
AKA: PLACHECK
3 Indicators of Trustworthy Measurement
- Validity: Data is valid when it represents the relevant dimension of the behavior
- Accuracy: Data is accurate when the observed value is the true value
- Reliability: Data is reliable when repeat measurements produce the same result
3 Indicators of Trustworthy Measurement (Description)
- Relative concepts that range from low to high
- Must be valid and accurate to be trustworthy
Ranging low to high
- Validity
- Accuracy
- Reliability
Validity (Defintion)
- An indicator of trustworthiness
- Represents the relevant dimension of the behavior
- Most important; without it accuracy and reliability are moot
- Inaccuracy makes Valid data Invalid
Threats
- Indirect measurements
- Measurement artifacts
- Measuring the wrong things
Validity (3 Fundamentals)
- Relevant dimension
- Directly measuring socially significant target behavior
- A guarantee that the data represents the behavior occurring in the conditions of interest
Threats
- Indirect measurements
- Measurement artifacts
- Measuring the wrong things
Threats to Validity
- Indirect measurements; Measure by proxy
- Measurement artifacts: Does not provide a meaningful representation of the behavior (Misleading)
- Measuring the wrong things / dimension (Most threatening to Validity)
Measurement Artifacts (Definition & Causes)
A threat to Validity
Does not provide a meaningful representation of the behavior (Misleading)
3 Causes
1. Discontinuous measurements - Produces over and underestimations
2. Poorly scheduled measurement periods
3. Limiting measurement scales - An artificial floor / ceiling
Accuracy (Not on Task List)
- An indicator of trustworthiness
- The observed value is the true value
- Helps us to identify and fix issues with IOA
- If measurement is not Valid, makes Accuracy moot
- If true values can’t be established, rely on reliability
Threats
- Human error (#1 Threat to Accuracy)
- Weak measurement system
- Poor observer training
- Measurement bias
- Reactivity
Threats to Accuracy & Reliability
- Human error (#1 Threat to Accuracy)
- -Poorly designed measurement systems
- -Bad observer training
- -Unintended influences on observer
- Weak measurement system
- Poor observer training
- Measurement bias
- Reactivity
Reliability
- An indicator of trustworthiness
- Dependability of measurement
- Repeat measurements produce the same result
- Poor reliability = Problems with accuracy
- Reliable does not mean valid or accurate
Threats
- Human error
- Weak measurement system
- Poor observer training
- Measurement bias
- Reactivity
Threats to Reliability
Human error (#1 threat to reliability)
Weak measurement system
Poor observer training
Measurement bias
Reactivity
Weak Measurements Systems
Threats to Accuracy & Reliability
- The more behaviors being recorded, the worse your data
- Use simple measurement systems to reduce errors
Poor Observer Training
Threats to Accuracy & Reliability
- Train to a standard of established competency
- Provide ongoing training to minimize observer drift
Observer Drift
Threats to Accuracy & Reliability: Poor Observer Training
- When observers have a shift / drift in how they interpret the operational definition of the target
- To minimize, collect IOA
- If needed, retrain observers on the correct operational definitions
Measurement Bias
Threats to Accuracy & Reliability
- Nonrandom measurement error
- Measurement that is influenced by an expectation / belief that behavior will / won’t occur under certain conditions
- Data that overestimates or underestimates the true value of an event
Observer Reactivity
When an observer’s data collection is influenced by the awareness that they are being monitored and evaluated
- Can be minimized by observing unobtrusively
Interobserver Agreement (IOA)
- Most common mean to ensure quality of measurement
- The degree to which 2 observers report the same values
- High IOA = High believability
IOA (Benefits & Uses)
- Assess a new staff member’s competence
- Identify observer drift
- Increases / decreases confidence that the definition and measurement code wasn’t too difficult
- Supports that any data variability isn’t due to who was recording, but a change in behavior
IOA (Information)
- Reported in a %
- Minimum of 20% of session across time
- Goal: 100% (Minimally 80%)
Methods for Calculating IOA
Repeatability: Total count; Mean count per interval; Exact count per interval; Trial by trial
Temporal/Duration: Total duration: Mean Duration (or IRT) per occurrence
Time Sampling: Interval by interval; Scored interval; Unscored interval
Total Count IOA
Method of Repeatability
How: (Total smaller count / Total larger count) * 100
- Simplest repeatability measure
- Overestimates agreement
Mean Count Per Interval IOA
Method of Repeatability
How: ((Int 1 Total IOA + Int 2 Total IOA + Int 3 Total IOA) /3 (# of Total Int)) * 100
- Calculates the agreement between the count within each interval
Exact Count Per Interval IOA
Method of Repeatability IOA
How: ((# of Int with 100% IOA) / Total # of Int) * 100
- Strictest IOA method
- % of total intervals in which both observers record the same thing
Trial by Trial IOA
Method of Repeatability IOA
How: ((# of trials in agreement) / Total # of trails) * 100)
- % of agreement between observers who measure occurrence or nonoccurrence of behavior
- Use for DTT
Total Duration IOA
Method of Temporal Measure IOA
How: (Shorter Duration / Longer Duration) * 100
- % of agreement regarding total duration
- Overestimates actual agreement
Mean Duration Per Occerrence
Method of Temporal Measure IOA
How: ((Dur IOA B1 + Dur IOA B2 + Dur IOA B3) /3 (# of Behaviors)) * 100
- Most precise temporal / duration IOA
Interval by Interval
Method of Time Sampling IOA
How: ((# of Int in agreement / Total Int) * 100 )
- # of scored and unscored Int with 100% IOA
- Likely to overestimate agreement with behaviors that occur at a very high / low rate
AKA: Point by Point IOA; Point by Point Agreement Ratio IOA
Interval by Interval IOA AKAs
- Point by Point IOA
- Point by Point Agreement Ratio IOA
Scored Interval IOA
Method of Time Sampling IOA
How: ((# of Int with agreed occurrence) / # of Int with at least 1 recorded occurrence) * 100
- Use for behaviors that typically occur approximately 30% of intervals or fewer
- Only uses intervals in which both observers scored an occurrence
- Minimizes overestimation by ignoring intervals in which measure by chance is highly likely
Unscored Interval IOA
Method of Time Sampling IOA
How: ((# of Int with agreed nonoccurrence) / # of Int with at least 1 recorded nonoccurrence) * 100
- Stricter for high rates of behavior
- Use for behaviors that typically occur approximately 70% of intervals or more
- Minimizes overestimation by ignoring intervals in which measure by chance is highly likely
- Only uses intervals in which both observers scored a nonoccurrence
Graphs
- Reveals relations between a series of measurements and relevant variables
- Visually allows you to examine quantitative info
- For organizing, storing, interpreting, and communicating results
5 Types of ABA graphs
Equal Interval: Line graph; Bar graph; Cumulative record; Scatter plot
Non-Equal Interval: Standard Celeration Chart
Equal Interval Graph (Definition)
The distance between any 2 consecutive points on both the x and y axis are always the same
- All intervals are the same size
Line Graphs
- Most common graph in ABA
- Based on the Cartesian plane
- Show the level of some quantifiable feature of the DV in relation the IV
- Y-axis should be shorter than the x-axis
- Maximum of 5 different data paths on 1 set of axes
Bar Graphs
- Use for: Group summative performance; Separate data sets that aren’t related
- Don’t use for: Trend; Variability
- Based on Cartesian plan
- No successive time is displayed on x-axis
Cumulative Record (Definition & Details)
- Skinner in EAB 1957
- Total # of behaviors since the start of data collection
- Behavior is only recorded once per observation period
- Steeper slope = Faster rate of responding
- Gradual slope = Slower rate of responding
2 Types of Cumulative Record Response Rates
- Overall response rate = An average rate of response over a given time period
- Local response rate = An average rate of response during periods of time smaller than that for which an overall response rate has been given
Scatter Plot
- Illustrates the relative distribution of each measurement
- Unconnected data points; Different symbols representing specific times
- Use for examining patterns in the temporal distribution of the behavior
AKA: Pattern Analysis
Scatter Plot (Pros & Cons)
Pros: Identifies time periods when the challenging behaviors occur
Cons:
- Correlational data only (No ABC)
- Only a hypothesis of function
- Subjective
- No replacement behaviors offered
Standard Celeration Chart
Ogden Lindsley (Precision Teaching)
4 Fundamental Properties of Behavior Change
- Data points
- Level
- Variability
- Trend
Data Point
A Fundamental Properties of Behavior Change
- More data and time = More confident measurement
- High variability needs more data points; fewer needed with less variability
Level
A Fundamental Properties of Behavior Change
- A change in Level = A change in the data average
- Line is drawn at the mean or median level
- May camouflage variability
Variability
A Fundamental Properties of Behavior Change
- A bounce in the data
- Use to determine how steady the data is
- High variability = Low control over the elements that effect your client
Trend
A Fundamental Properties of Behavior Change
- Overall direction of a data path
- Direction: Increasing; Decreasing; Zero Trend
- Degree: Gradual; Steep
