미국에서

When strategies to avoid problem behaviors fail and a student engages in behaviors to get out of training, escape extinction must be used to avoid reinforcing the behavior. Do not reinforce problem behaviors by allowing them to delay or terminate the session.

The intertrial interval is the time from the end of consequence to the presentation of the SD in the next trial. Generally, it should be about 1-3 seconds. That is, it should be long enough to be distinguishable from hte last trial, but short enough so that interfering behaviors do not  begin.

In a massed practice format, numerous consecutive trials are presents. At least initially, all trials should be reinforced. However, too much reinforcement in a short period is an abolishing operation. Limiting access to the reinforcer outside of the instructional setting, moving to an intermittent schedule, and using a token system are three strategies for mitigating this effect. 

'

The best way to address uncooperative behavior is to avoid it. Establish yourself as a reinforcer, make tasks easy, make requests only when the MO is strong and the corresponding reinforcer is available. Also, do not attempt to initiate instruction when the child is already engaged in reinforcing activity. Finally, teach skills that are functional for the student-ones that yield reinforcement. 

Avid uncooperative behavior by ,

-starting instruction when the MO is strong and the reinforcer is available.

-establish yourself as a reinforcer

-waiting to begin instruction until other reinforcing activities are finished. 

-making tasks easy.

In discrete trial instruction, the intertrial interval is the time from the end of consequence to the presentation of the SD in the next trial.

Discrete trial training (DTT) involves the repeated presentation of discriminative stimuli and predeterined consequences(e.g.,"Good" for correct responses and corrective feedback for incorrect responses), or perhaps extinction. Prompting and fading procedures may also be used. 

In Discrete trial training, the presentation of the SD is controlled by the teacher. THe response may only be reinforced when the SD is presented and the opportunity to respond is available. A free operant is a response that has no such requirements. An SD is present, the behavior may occur repeatedly, and reinforcement is not specifically programmed. 

The No-No-Prompt method of error correction is used for infrequent errors after the child has demonstrated correct responding with errorless teaching. An incorrect or no response is consequated with a verbal "No," removal of material, and looking away for 2 seconds. This is repeated if there is an error or no response on the next trial. The third trial uses a prompt sufficient to produce correct responding. 

Response should only be considered correct if they begin within a few seconds. Responses that take longer to begin are typically not functional, and therefore, should not be accepted as correct. Moreover, it may indicate that control of the response by the SD is tenuous. 

The model-prompt-switch-repeat (4-step) method of error correction includes:

1) the teacher says the SD and models the response. If an error occurs,

2) the next trials uses an intrusive prompt to ensure correct responding.

3) the next trials is for a previously mastered skill(i.e., insets an interval between step 1 and 4.

4) the initial trials is repeated without a prompt. 

A "No," delivered in a neutral tone can be an efficient way of providing important feedback. This is why it is used in this program for all incorrect responses. It lets you know immediately that you made the wrong selection. When teaching students, "No," it should not be in an upbeat tone, as you would use with a praise statement-keep it neutral. 

A discrete trial training(DTT) transfer trial involves a reduction in a reduction in a controlling prompt to trasfer control to the target SD. Prompt fading should begin after the initial trial. If the correct response is emitted with the reduced prompt, a more potent reinforce is often provided. If correct responding requires the same or a greater prompt, then potent reinforcer is often provide.

Natural environment training(NET) is a free-operant arrangement for training. It uses child-directed interactions in an arranged environment that resembles free play. Often the caregiver is trained to promote language learning opportunities during free play. NET uses naturally occurring MOs to promote responding. It is often used for language training in early intervention. 

Reinforcement should be delivered within a second or two to maximize its effectiveness. During discrete trial training, it is important that session move along rapidly. Therefore, the reinforcer given should be able to be consumed within about 5 seconds or the student should be willing to set it aside after that amount of time. Examples include a single m&m, one small piece of cookie, sip of beverage, 5 seconds of music or a video. 

Incidental teaching(aka in situ training, naturalistic teaching, or non-intensive teaching) occurs in the natural environment where the child would use the language. It occurs throughout the day as the teacher responds to whatever stimulus that the child selects(child wants or needs). It is considered "loose training" where the child initiates the teaching interaction and the teacher provides a brief teaching interaction which delivers the child-specified reinfercer. Incidental teaching of mands provides opportunities for language elaborations, prompts, and interaction; and it automatically promotes generalization of language use. 

Early autism intervention can include both discrete trial training(DTT) and natural environment training(NET). The NET allows the child to emit free operants, such as spontaneous mands(e.e., pointing to an item), and the teacher uses that arrangement to teach vocal language. 

Free-operant arrangements of teaching allow the child to play in an environment with naturally occurring MOs(preferred toys and actives). Access to the item may require the child performing a target behavior(speaking). Teaching may also occur in interactions about something that interests the child. For example, the child picks up a toy dog and the teacher says, "that is a dog. Say dog."

Using errorless learning methods for discrete trial training(DTT), the first SD is followed immediately by a prompt that will evoke the correct respomse(e.g. "touch the car" followed by gently grasping the child's hand and moving his hand to the card.) A "transfer trial" is a trial that uses a reduced prompt so the response is transferred from the prompt to the target SD. 

Incidental teaching- training is conducted throughout the day in the natural environment as the child plays. Training is generally single-trial.

Natural Language Paradigm-a combination of free-operant and discrete trial training(DTT). The child plays and selects a reinforcer(toy). Therapist conducts a short discrete trial training(DTT) using the toy. Continued use of toy is contingent on correct responding.

Mand Model - the child plays and selects a reinforcer(toy). Therapist conducts a 3-trial training using the toy. Child receives toy noncontingently. 

Correct responding in a mass trial format does not ensure the correct discrimination. After correct responding is obtained for a target trial, distractor trials(any other previously trained discrimination) should be interspered with the target trial. The distractor trials are first faded into the sequence of trials until the trials alternate between the target and the distractor. 

To graph the number of behaviors that occur within a specific time period, use a line graph, which includes equal interval graph, Standard Celeration Chart, and a cumulative record. 

You could graph behaviors that occur one time per day to one thousand per minute on a Standard Celeration Chart. If you used an equal interval graph, the ordinate (Y-axis) could not detect any meaningful changes in the low frequency behavior (without using a scale break). 

The vertical axis should be approximately 5/8 the length of the horizontal axis.

Scales should begin at zero. If the range of values of the dependent variable is large and some of those values are at the lower end of the scale, do not use a scale break. 

The Standard Celeration Chart best accommodates an analysis of variability at very high and very low rates. 

Horizontal Axis - time periods (e.g., minutes, years) or response opportunities(e.g., session)

Vertical Axia - dependent measure(e.g., frequency, count)

               frequency, Duration, Cumulative number

It is appropriate to use a scale break on the horizontal axis hen there are long periods with no data collected.

When necessary, use scale breaks on the vertical axis so the graph shows variability that is of social significance. For example, a difference of 1 and 7 aggressive acts per day is significant. It this behavior reached 100 on a few days, a scale break between 10 and 100 should be used so that the variability of low rates is reflected.

-Accuracy is the extent to which a measure reflects or closely approximates the true values. It can be tested by comparing a measure to the measure obtained by an established competent recorder.

A practitioner compared newly trained observer's measures of math fact performance to an expert observer's extraordinarily careful measures. 

-Accurate date reflex or closely approximate true values. 

Reliability means that repeated measurement of the same event yields the same result.

Reliability is a measure of the stability of the relationship between observed values and actual events. 

Interobserver agreement is the extent to which two independent observers agree.

Validity means that a data collection system measures what it purports to measure.

Reactivity- The presence of an observer influences behavior

Believability - The extent which others are convinced that measures are accurate.

Observer drift - over time observers modify operational definitions.

Expectancy - bias due to preconceived notions about the behavior and/or person being observed

Complexity - complex data collection system(e.g., several target behaviors or several subjects) are prone to unreliable data collection

Similar to interobserver agreement, this technique is often used to train new observers. For example, they practice scoring a video until their data matches the known true values. Also, as a way to "spot check" observer performance. A practitioner may assess the accuracy of his observers' data by scoring videos of a few sessions himself and comparing his data with theirs. 

A stable relationship between what is observed and what actually happened refers to the reliability of the measurement system.

Data are accurate to the extent that they reflect true values. Interobserver agreement addresses the logical perspective that a report from two observers is more likely to be accurate than a report from one observer, but data are more believable when steps are taken to ensure their accuracy(ie., they are obtained using a valid and reliable measurement system designed to measure the true value of the observed event)

If someone else took the test, it does not measure what it is supposed to measure. 

To measure true values, one must have a valid and reliable measurement system. For example, a person who weighs 100 pounds is repeatedly weighed on a scale that consistently(i.e., reliably( indicates a weight of 110 pounds. That is, the scale provides a reliably inaccurate measure. Similarly, two human observers may record that a behavior occurred or didn't occur with a high degree of interobserver agreement, meaning they are recording the behavior reliably, but the accuracy of their data depends on the validity of the data collection procedures. 

Interobserver agreement can be influenced by 

-expectancy, 

-reactivity, 

-observers modifying operational definitions over time.

-complexity of the data collection system

-staff complaints about the person being observed.  

-reliability of dependent measure. (다른문제는 다시 체크)

A researcher avoids having a particular therapist who had bad experience with a student collect data to avoid the influence of expectancy. 

Data are more believable when 

-interobserver agreement is high, 

-they are obtained by reliable measures, 

-they are accurate. 

Accurate data

-can only be obtained using a reliable measurement system.

-reflect or closely approximate the true values of the observed events.

-may be recorded by one observer even if interobserver agreement is low. 

-compare them to true values.

A continuous graphical analysis of each observer's data could identify error patterns, thus enabling corrective action to be taken immediately.

Corrective and timely action can be taken by analyzing each observer's data daily.

Consider an observation of 100 intervals. Observer A records occurrences in intervals 11, 12, and 13. Observer B records occurrence, and nonoccurrence calculations yield, 98%(98/100), 67%(4/6), and 97.9%(96/98), respectively. Given the very low number of reports of the target behavior, interval-by-interval agreement is misleading by itself. It should be supplemented by a coefficient based on intervals in which one or both observers reported an occurrence. Thus, by reporting interval-by-interval agreement of 98% and occurrence agreement of 67%, the reader is presented with the most accurate summary. 

Interobserver agreement may be influenced by the number of individuals upon whom data are being collected.

Interval-by-interval agreement is likely to be high if the behavior is very high or very low. Observers may record consistent with what behavior usually occurs, rather than record based on careful observation. In other words, they "get in the habit" of scoring a certain way. 

Ideally, the only reinforcement contingencies operating on observers are related to accuracy of the data. 

Accuracy of data is the sole function of the data collector, at least ideally. Interest in the study's outcome or publication could present a bias. Learning about the experimental procedures could result in an expectancy effect. 

Interobserver agreement increases the believability of data because it measures the extent to which two people agree. It doesn't necessarily increase accuracy(which is comparison to true values) or reliability(repeated measures). 

To assure that observers are seeing and hearing the same things, they should observe from the same vantage point. .

The complexity of the procedures, number of behaviors, and number of individuals upon whom data are being collected can all influence inter observer agreement. 

A high response rate warrants reporting nonoccurrence agreement. A low rate warrants reporting occurrence agreement. 

Some researchers collect inter observer agreement data in all sessions. However, this isn't always possible or even necessary. Although opinions vary, a general rule to follow is that it should be collected minimally in 20% to 30% of the sessions and at least once per condition.

Making each interval helps observers keep a rhythm and provides a visual stimulus for where to record next. 

To ensure accurate data collection during the study, train extensively prior to the study and check occasionally during the study 

Mathematical calculations by data collectors should be closely supervised due to possible bias and errors. 

Although not always possible, observers naive to the objectives of the study will not be biased. 

Data collectors will be more accurate if they believe another observer is collection inter observer agreement. 

High response rate warrants reporting nonoccurrence agreement. A low rate warrants reporting occurrence agreement. 

Accuracy-the extent to which a measure reflects the true value

Reliability-repeated measurement of the same event yields the same result.

Interobserver agreement-the extent to which two observer agreement enhances credibility of the data.

An analysis of the variables that influence agreement may indicate a need for additional training, improved definitions, clarification of recording procedures, etc. 

Minimum acceptable inter observer agreement of publication is usually 80%. However, one must consider the type of data being collected. For example, agreement on permanent product data should usually be in the 90% range; agreement on complex social interactions might be in the 70% range. 

- occurrence(scored interval) interobserver agreement considers only the intervals in which one or both observers recored an occurrence(that would be nine in this item). Divide the number of agreements by the number of intervals in which one or both observers recorded an occurrence.

- nonoccurrence(unscored interval) interobserver agreement, divide the number of intervals in which both observers recorded no occurrence by the number of intervals in which at least one observer recorded no occurrence( X 100)

- When calculating interval-by-interval inter observer agreement, divide the number of intervals in which both observers recorded an agreement by the total number of intervals(X100). 

- A frequency-within-interval calculation increases the probability that agreements are based on both recorders observing the same behaviors. The calculations is percent agreement for each interval averaged across all intervals. For example, in the first interval, observer one records two occurrences and observer two records four occurrences(50%). In the second interval, observer one records one occurrences(33%). The sum of 50%, 33%, etc., would be divided by the number of intervals. 

-Nonoccurrence(unscored interval) inter observer agreement considers only the intervals in which one or both observers recored a nonoccurrence(that would be 5 in this item). Divided the number of agreements by the number of intervals in which one or both observers recored a nonoccurrence.

-Interval-by-interval data considers agreement on ALL intervals. Divide the number of agreements by the total number of intervals. 

-If rates are high, it could be that in the quest to obtain a good agreement score, observers "drift" away from observing carefully and err on the side of recoding nonoccurrence because they predict(perhaps unconsciously) that the other observer will record nonoccurrence. 

- Divide the time recorded by the observer who recorded the shortest duration by the time recorded by the observer who recorded the longest duration(X 100)

- To calculate interobserver agreement on frequency data, divide the lowest total by the highest total(X100) to obtain a percent. 

- If behavior rates are high, it could be that in the quest to obtain from observing carefully and err on the side of recording occurrences because they predict(perhaps unconsciously( that the other observer will record an occurrence.

Measures defined

1. frequency- number of responses(within a constant time period)

2. rate- number of response per standard unit of time(e.g., seconds, minutes, hours)

3. duration-measure of time from the begging to the end of the response 

- Use when the amount of time the person engages in the behavior is important

4. latency-measure of time between an environmental event and a response. To measure the amount of time between opportunity and the beginning of a response.

When measuring latency, begin the timer at the presentation of the stimulus that is to occasion the behavior. The timer stopped at the onset of the behavior. 

If a response is occurring too late, a reasonable goal would be to decrease the latency to respond. 

5. inter-response time- measure of time between the end of one response and the beginning of the next one

-Interresponse time is the time between responses. If a response is occurring at too high a rate, increasing interresponse time is a reasonable goal. If a response is occurring at too low a rate, decreasing interresponse time is a reasonable goal. 

6. percent of occurrence- number of responses divided by number of opportunities and multiplied by 100.

-Use with discriminative operants that have a limited opportunity to occur.

7. trails to criterion- number of response opportunities to achieve a performance standard

-Trials-to-criterion is a measure of response opportunities needed to achieve a predetermined level of performance.  The frequency of opportunities to respond(learning trials) is counted instead of the frequency of the response. Target criteria are determined by the nature of the target behavior and the desired performance level.

Trials-to-criterion is a frequency measure. to yield meaningful information, frequency data must be reported relative to some other variable. For example, rate is reported in terms of the frequency of a response over a period of time. 

8. celeration-ratio of two response rates(measure of change) divided by measure of time between the two response rates.

9. Momentary time sampling - use when you can only collect data a small portion of the time and rate is sufficiently high.

10. Partial-interval - use when responses are too frequent to count each instance or they occur at various durations. 

-Intend is not observable. If collecting duration or frequency data, behaviors without a clear onset and offset will not be reliably recorded. 

-Duration is measured from the time the behavior begins until it ends. 

-Total duration is the sum of the durations.

-Average duration per occurrence is calculated by dividing the total duration by the number of occurrences of the behavior.

-Frequency measures are most appropriate for behaviors that have discrete beginnings and endings, do not vary significantly in duration, and should not be too frequent to count. The rate makes it possible to count each response. 

-Rate is count divided by time. Frequency is used to indicate simply the number of times a behavior  occurs, which is an appropriate measure as long as observations are always the same length. Rate is necessary when sessions are not a constant length. 

-Proficient responding is both accurate and rapid. Rate is the only measure that reflects both of these dimensions. 

-If the percent of occurrences is based on too few opportunities, results will be over-or underestimated. For example, say the true percent of correct answers is 70 out of 100, or 70%. If a sample is made of only five observations, results are likely to yield 60% or 80%, but never 70%. Error is decreased proportionally to the number of opportunities. '

Generally the lowest accepted number of opportunities upon which to base a percent is 30. 

-If the percent of occurrences is based on too few opportunities, results will be over-or underestimated. For example, say the true percent of correct answers is 70 out of 100, or 70%. If a sample is made of only five observations, results are likely to yield 60% or 80%, but never 70%. 

- Latency is a measure of the time that passed between the opportunity to respond and the onset of the response. If a response is occurring too soon, a reasonable goal would be to increase the latency to respond. 

how much time passes between the opportunity to emit a behavior and when the behavior occurs. 

-trals-to-criterion data can be used to assess a learner's increasing competence in acquiring a related class of concepts. For example, if a learner achieves mastery of each newly introduced color in fewer trial blocks than it took to learn previous color discriminations, this can be regarded as evidence of the student's increasing ability to generalize the concept to novel stimuli. 

-trals-to-criterion data can be used to compare the relative efficiency of two or more treatments or instructional methods. Sometimes trials-to-criterion data are supplemented with other measures, such as the amount of instruction time needed to achieve predetermined performance criteria. 

-interresponse : how much time passes between consecutive occurrences of responses in the same response class

-Latency :  how much time passes the opportunity to emit a response and when the response occurs 

- 

- 

-Select a time sampling method that will produce a conservative measure. 

When the goal is to increase the behavior, select whole-interval recording because it underestimates the duration of the behavior. When your goal is to decrease a behavior, select partial-interval recording because it overestimates the duration of the behavior.

-for behaviors that are largely continuous or do not have clear onset or offset, if you goal is to reduce behavior, the behavior should be measured using partial-interval recording. 

-Generally " discontinuous measurement" and "time sampling" are synonymous terms, although there are some exceptions. Momentary time sampling and partial and whole-interval recording are discontinuous methods. 

- Consider an hour-long class. On-task behavior occurs for 20 minutes, stops for 6 minutes, occurs for 30 minutes, and then stops for 4. The total actual duration would be 50 minutes. However, a 15-minute whole-interval procedure would yield 2 out of 4 intervals(it only occurred entirely during the first and third intervals)-50% of the intervals yield a duration estimate of 30 min, which is 20 min short of the actual duration.

-despite some inherent inaccuracy, time sampling procedures such as momentary time sampling and partial and whole-interval data collection procedures are popular. Generally, the data resulting are reported as percent of observations or percent of intervals. However, they should NEVER be reported as percent of time," which is reserved for duration measures. 

-Partial-interval recording tends to underestimate high frequency behavior and overestimate duration. Consider an observation period of 5 minutes divided into 5 one minute intervals. One student exhibits a target behavior one time in four of the five intervals-a total of 4 responses. A second student exhibits a target behavior three times in four of the five intervals- a total of 12 responses. Yet the data sheets of both individuals would indicate responding in 80% of the intervals. Thus, the summary data for the second student is not sensitive to the high frequency of responding. 

-Counting each occurrence(frequency) is the most accurate. Short partial-interval recording is a more sensitive measure than long partial-interval recording. Momentary time sampling measure is even less sensitive for short, discrete behaviors because it captures only a momentary snapshot of behavior.

event recording < short partial-interval recording < long partial-interval recording < momentary time sampling.

-A crying spell, for example, may begin with a whimper and gradually escalate. The individual may stop momentarily or for a few or several seconds. with such behavior it is difficult to determine the exact onset and offset, therefore making frequency a accurately determining the start or end of the target behavior, time sampling provides the most reliable data. 

-frequency and duration require observation to occur continuously during an entire observation period. 

-Time sampling measurement procedures involve recording the occurrence or non-occurrence of a behavior during part or all of an interval, or recording of it occurs at a specific moment at the end of an internal. Continuous measurement involves recording all instances of a behavior.

-Time sampling measurement is a means of recording the behavior when it occurs any time during an interval(partial-interval recording), during an entire interval(whole-interval recoding), or at a particular moment at the end of an interval(momentary time sampling). Note that what is recorded is the occurrence/nonoccurrence of the behavior-not each instance of the behaviors is the case with continuous measurement.

is used where the data collector cannot observe all of the time.

-Continuous measurement is more accurate. Discontinuous measurement provides an estimate of frequency or duration. with discontinuous measurement, the shorter the interval, the more accurate the estimate. Both continuous and discontinuous measurement yield publishable data, although momentary time sampling measurement may be easier because it doesn't require the observer's attention at all times. Interval recording has an advantage of allowing the observer to record data of several individuals. All methods of time sampling can produce more reliable data for responses that do not have a clear beginning or end. 

Data could include occurrence (count), duration, interresponse time, magnitude, and.or latency.  

date is taken on every occurrence of a taken on every occurrence of a target behavior for a specified time period. 

every occurrence of a target behavior during a particular observation period.

-Frequency and duration are most accurate measures. However, they require complete attentiveness 100% of the time. Partial-interval and whole interval recording also require attentiveness 100% of the time. Momentary time sampling does not require the data collector to observe continuously. Time sampling is not appropriate for infrequent behaviors(i.e., less than once in 15 minutes). 

-Discontinuous measurement refers to any measurement that does not include all occurrences of a target behavior during an observation period. It typically includes recording that the behavior occurred at least one time during an interval(partial-interval recording), during an entire interval(whole-interval recording), or at a particular moment at the end of an interval(momentary time sampling). Note that what is recorded is the occurrence/nonoccurrence of the behavior-not each instance of the behavior, as is the case with continuous measurement. 

-The task of "Measure frequency" refers to count only. 

-The task of "measure rate" refers to count per unit of time. Although the Task List defines frequency as the number of events, frequency measures should always be reported with a relevant unit of time. 

-Frequency is the number of occurrences of a behavior. Remember that frequency can also be defined as simply the count of the events. 

-Frequency- number of responses per unit of time

-Duration - time between the onset and cessation of a response.

-Latency- the time from the presentation of a stimulus to the onset of the response, Time between the onset of a stimulus and the initiation of a response. 

-Inter-response Time : the time between the cessation of one response and the onset of another response .

-Duration measure : the number of minutes an individual spends on a task.

-Measurement by permanent products is actually a measurement of a change in the environment produced by behavior. Permanent products can be natural(such as a spelling test) or contrived(such as a photo of natural permanent products)

1. Can you give an example of why an ecological assessment may be used over another form?

2. For this weeks discussion post, it said a minimum number of words needed. Is there a plenty for being under the required minimum?

3. Can you please give an example, other than the one in the reading of empirism,

4. Is the exam on both chapters and the article?

5. Any specific areas of the Baer article we should focus on?

6. When going over behavioral cusps and pivotal behavior, I was wondering if they both can be considered generalized behavior change?

7. Will this weeks quiz be all written responses like last weeks quiz?

1. What behavior prompted/brought you in to request help for your child or for assistance?

2. WHen do you feel like the behavior occurs most often?

3. What area or environment does the behavior affect your child most?

tantrum : any instance of child yelling, crying with or without tears, screaming, rolling around on flood that persists for second or more.