Introduction

The principal character trait setting behavior analysis apart from all other subdisciplines of psychology rests on the conceptualization of behavior. Launching contemporary behavior analysis with his masterpiece The Behavior of Organisms, Skinner wrote, “Behavior is what an organism is doing; or more accurately what it is observed by another organism to be doing” (Skinner 1938, p. 6). Skinner went on to elaborate stating behavior also could refer to the movement of an organism or of its parts.

Contemporary definitions of behavior follow Skinner’s original definition: “Behavior is anything an animal (including the human animal) does” (Malott and Shane 2014, p. 6); “Behavior is that portion of an organism’s interaction with its environment that involves movement of some part of the organism” (Johnston and Pennypacker 2009, p. 31). The key to all definitions of behavior center on movement. Indeed, movement-based definitions of behavior led to the discovery and classification of different types or classes of behavior: operant, respondent, and adjunctive (Cooper et al. 2007).

Practicing behavior analysts rely on a behavior analytic framework when formulating intervention targets. For example, an applied study defined “requests for attention” as “raising one’s hand, calling out to the teacher, or leaving one’s seat and going to the teacher to make a request for assistance or approval after the student had been instructed to work independently” (Austin and Bevan 2011, p. 453). Thrusting a hand upward, saying answers or comments, rising from a sitting position, walking to the teacher, and asking questions all constitute movement or action. Target behaviors take center stage for descriptive analyses, functional analyses, and all behavior plans and interventions. The major pillar of the science of behavior lies in formation and execution of selecting and labeling behavior for analysis and intervention.

Ogden Lindsley, the founder of Precision Teaching and one of B. F. Skinner’s students, understood the critical need to properly label or pinpoint authentic behavior. In Precision Teaching, recognizing and detecting behavior meant parents and teachers needed to learn how to discriminate action and non-action. One of the first Precision Teaching texts explained the difference between behaviors that do and do not contain movement (Kunzelmann et al. 1970). “Waiting in line” does not contain movement, but “slowly walking forward” does. “Being depressed” lacks movement while “says unhappy comments” has clear movement. The identification of movement-based versus non-movement-based behavior led to practical information.

Early Precision Teachers also acknowledged that action-based verbs required a cycle to complete the movement of behavior. As an example, “to empty” and “to suck” both represent detectable movements, but without a cycle one may ask “to empty what?” and “to suck what?” (Kunzelmann et al. 1970). A cycle must contain the action and the associated object for a person to determine when the movement (i.e., behavior) has ended (Kubina and Yurich 2012). “To empty a dishwasher” or “to suck a lollipop” forms a complete description of behavior.

Verb tenses formed another feature of the pinpointing process by adding precision to the selected verb. Many parents and teachers would choose verbs such as hitting, spitting, or yelling. All of the previous mentioned -ing verbs fall under the present progressive tense. The present progressive tense illustrates action with ongoing progress (Berry et al. 2010). Precision Teaching suggests using simple present tense requiring the addition of an “s” to the action verb. Hitting, spitting, and yelling become hits, spits, and yells. Use of the simple present tense expresses actions that repeat (Language Dynamics 2013). A verb communicated in the present progressive tense directs the person counting the behavior to look for ongoing instances of the behavior. Alternatively, Precision Teachers found parents and teachers had an easier time counting behavior that repeated itself (e.g., “hitting” refers to ongoing behavior whereas “hits” conveys a discrete instance that begins and ends).

Through the years, Precision Teachers have refined how to create movement cycles. The four steps involve: (1) selecting a precise, observable action verb (present tense); (2) choosing an object (noun) that ends the cycle of movement; (3) adding an “s” to the action verb; and (4) checking the movement cycle for observability and repeatability (Kubina and Yurich 2012). Swings hammer, kicks wall, scratches arm, licks popsicle, shakes hand, and turns page all showcase movement cycles using the previously described four steps.

Pinpointing behavior using movement cycles has helped teachers and parents identify and solve behavioral problems. Using movement cycles may also aid behavior analysts in practice and research. The advantages of movement cycles include having an explicit behavioral target easily detected. The ease of detection then leads to an enhanced capacity to count the behavior. The precision of detecting and counting movement cycles facilitates the identification of quantitative relations between an operant and features in the environment, in other words, distinguishing a contingency (Vargas 2009). The sharpened focus on movement cycles also may improve the identification of response classes. A “tantrum” may include bites hand, pinches arm, slaps face, and kicks floor. And the culmination of the behavioral measurement of movement cycles lies with the application of behavior plans and subsequent evaluation of the intervention. Decisive, reliably observed behaviors, and recorded data play a critical role in behavior analysis.

Overall, the research examining movement cycles appears limited. One study supporting movement cycles appeared in an experiment examining the differences in target behavior descriptions (Smith et al. 2013). A groups of participants watched a videotape under two different conditions. When identifying the target behavior with present tense, active voice verb, and an object that named the reception of action, target behavior detection was higher than when using an operational definition. Namely, detecting the target behavior in the videotape with a movement cycle occurred at a higher rate and the amount of false positives (i.e., detecting behavior that did not meet the target definition) were at a much lower rate.

Using movement cycles for behavior description could positively enhance the verbal repertoires of those conducting behavioral analyses and interventions. Future measurement (data collection) and intervention efforts could improve by describing target behaviors in ways that fall within the original account of behavior (i.e., containing movement). Therefore, at a school serving students with autism spectrum disorders, the present experiment examined whether behavior analysts and special education teachers could become fluent identifying movement cycles with an intervention called SAFMEDS. SAFMEDS stands for “Say All Fast Minute Every Day Shuffled” (Graf and Auman 2005) and has experimental evidence demonstrating its effectiveness increasing different verbal behavior pinpoints to fluency (e.g., Beverley et al. 2009; Chapman et al. 2005; Cihon et al. 2012; Hughes et al. 2007).

Because few studies have shown how SAFMEDS could impact in-service professionals, the potential participants (i.e., behavior analysts and special education teachers) received an invitation to take part in the experiment. The specific experimental questions asked whether SAFMEDS would lead to fluency for identifying movement cycles among practicing behavior analysts and special education teachers. A second question examined whether the SAFMEDS fluency intervention would foster positive outcomes with maintenance after the intervention ended.

Method

Participants

Participants consisted of eight educational or behavioral health professionals. All participants had a master’s degree in Applied Behavior Analysis, education, or psychology, and five of the eight participants were or became Board Certified Behavior Analysts during the intervention. The participants each held professional positions in which they frequently had to pinpoint behavior for various tasks (e.g., assessment, intervention, goals, individualized education programs, communication with the education team or communication with parents). Table 1 provides a summarization of participant characteristics.

Table 1 Participant characteristics
Full size table

The participants did not receive compensation for their participation in the study. Participants involved in the agreed to take part in an experiment aimed at determining whether the SAFMEDS intervention would help them learn movement cycles better and thus apply it in their respective jobs. Because participants worked at the same campus, they all received instructions not to discuss experimental procedures or results with one another until the study ended. Participants were assigned to either Group 1 or Group 2 through random selection.

Setting

The experimenters conducted the study at a private school in the eastern part of the USA. The school served children between the ages of three and 21 with a primary diagnosis of autism. All experimental sessions took place in general office locations that contained a table or desk, chairs, and other basic office equipment. Practice sessions occurred during the course of participants’ normal workdays, and therefore, sessions occurred at various times of day.

Materials

Materials consisted of two counterbalanced decks of SAFMEDS cards depicting behavior. Experimenters photographed individuals performing actions typically observed in both school and home environments. Adults and their related children as well as school students served as models. The photographs included individuals ranging in age from 1 to 20 years. The final set of 100 images was separated into Deck A and Deck B; each image illustrated a clear behavior and came from a selection of approximately 200 images.

To determine the movement cycle label, experimenters assigned the most precise and directly observable action verb and object label using both a commercially available dictionary and thesaurus. Each SAFMEDS card contained the movement cycle label on one side of the card opposite the image. An expert panel comprised mainly of senior-level behavior analysts (seven panelists in total, two Ph.D.s, four Masters, and one Bachelor’s level) independently reviewed the prospective images. The expert panel also described the movement cycles using action oriented verbs to describe the movement shown and plain English nouns to depict the object of the movement (Lindsley 1991). Panelists had to reach 100 % agreement for each SAFMEDS label. When the expert panel did not come to a 100 % consensus, they excluded the image.

The experimenters organized the selected images into two decks (e.g., Deck A and Deck B) containing fifty pictures each. Each deck was then counterbalanced across photographs of daily care routines (e.g., brushes teeth, brushes hair), academic behaviors (e.g., touches shape, colors paper), leisure activities (e.g., catches ball, places puzzle piece), and functional routines (e.g., washes dish, empties trashcan). Each counterbalanced deck contained no more than two depictions of the same action.

Images were printed on 3½ × 5 inch cardstock paper. Each full color image further included three or four consistent markings superimposed using Microsoft® drawing tools in Word. Two-dimensional still frame images have limitations for illustrating directionality or movement. Therefore, the experimenters used superimposed arrows to depict movement and circles to link the movement with the object to complete one movement cycle per card. Every image included one or two circles and one or two arrows in order to assist in communicating the correct movement cycle. Circles marked the focal point of the movement. Arrows connected movement and object or provided information of directionality of the movement.

An example of the unit of analysis appears in Fig. 1. Figure 1 shows a selected image containing a circle around the foot of a young child and a soccer ball. The circle indicated the portion of the image depicting movement. The arrow communicated movement in a horizontal direction. The image combined with the graphic notations (i.e., circle and arrow) showed a particular movement cycle, “kicks ball.”

Fig. 1
figure 1

Image of a SAFMEDS card. Circles indicated the portion of the image depicting movement. The arrow shows the direction of movement. The image combined with the graphic notations shows the movement cycle, “kicks ball”

Full size image

Response Measurement and Accuracy

Dependent Variable

The dependent measure consisted of correctly labeling SAFMEDS cards. The experimenters scored a verbal response as correct when the participant said the specific movement cycle assigned to the card (i.e., unique combination of action verb and object). For example, if the participant labeled a card, “turns page,” in response to seeing the image with a child turning a page in a book, the response was scored as one correct. Any stated response other than “turns page” resulted in an incorrect score. Correct and incorrect responses were tallied during the daily assessment trial and charted on an Excel version of a Standard Celeration Chart (Graf 2008). The participants did not have access to their daily frequency scores or charted data nor did they receive any feedback regarding their performance.

Accuracy

Accuracy refers to the degree of observed values correctly estimating the events that occurred in an experiment. Accuracy provides more information than interobserver agreement in regards to the accuracy and reliability of experimental data (Johnston and Pennypacker 2009). Because the experimenters videotaped all experimental sessions, every dependent variable measurement was checked for accuracy. The videotaped session became the true value. The true value represents a value or score that involves the highest degree possible for minimizing measurement error (Johnston and Pennypacker 2009). The experimenters checked all scores for measurement of the dependent variable against the videotaped sessions. Two experimenters scored each video independent of one another. As an additional measure, the two independent scores were compared against each other. Accuracy for the dependent variable was 100 %.

Independent Variable and Procedural Integrity

Independent Variable

The independent variable consisted of frequency building to a performance criterion (FBPC; Kubina and Yurich 2012). FBPC entailed daily, timed skill practice with performance feedback (Graf and Lindsley 2002). Frequency building to a performance criterion, a more precise and technical term than practice, results in systematically increasing the frequency or rate and accuracy of the target skill. Within the experiment, participants built frequency of saying movement cycles in response to individual picture cards arranged into a deck of 50. The independent variable, FBPC, increases accuracy and speed of behavior with timed repetition and performance feedback (Datchuk and Kubina 2015; Datchuk et al. 2015; Kubina and Yurich 2012). The cumulative implementations of FBPC is also purported to foster growth across time (Kubina 2005). The performance criterion (the PC of the acronym FBPC) indicates the frequency aim signaling the end or goal of frequency building. For the present study, a performance criterion of 17 correct responses in 20 s was set, or a frequency of approximately 50 corrects per minute (Graf and Auman 2005).

Each group of participants used one of two equivalent SAFMEDS decks (Deck A or Deck B). The two decks were counterbalanced and contained corresponding stimuli for each group. Both decks included no more than two pictures depicting the same movement or action and included an equal number of child or adult models. The two decks controlled for idiosyncratic pictures or some unforeseen graphic feature of a SAFMEDS card exhibiting control of responding.

Procedural Integrity

The procedural integrity measure documented the correct and consistent use of the intervention. Integrity data were collected on the following measures: (a) correct materials used, (b) proper positioning of participants, (c) distraction-free environment, (d) correct number of SAFMEDS trials implemented, (e) minimum level of daily sessions per week of the SAFMEDS intervention, and (f) videotaping the assessment trial. Procedural integrity was calculated by dividing the number of steps correctly competed by the total number of possible steps and multiplying by 100 (Gast 2010). The mean procedural integrity was 98.2 %. Errors mainly consisted of videotaping from an incorrect position.

Experimental Design

Experimenters used a multiprobe multiple baseline across participants design (Kennedy 2005) for two groups of four participants each. The first participant from each group whose data had the most stability during baseline observation started the intervention phase first. Stability was defined as low variability, flat or decelerating corrects, and/or flat or accelerating incorrects. All other participants continued within a multiple probe design consisting of one trial (a 20-s time interval), once per week, until he or she began the intervention phase. When the first participant in the intervention phase reached ten correct responses within 20 s (with a minimum of 4 days in the intervention phase), the next participant with a stable baseline began the intervention. The process continued with each participant beginning the intervention phase once the previous participant achieved 10 correct responses.

Procedure

Pre-test

During pre-test, all eight participants were tested on both Deck A and Deck B. The pre-test helped assure no participants were familiar with the information. The participants sat to the left of the experimenter. The experimenter shuffled the deck of cards and set a timer for 20 s. The participant then went through each deck of cards as fast as they could naming the movement cycle, with the timer starting upon initiation of the participant’s first response. The participant could not see the correct movement cycle written on the back of the card during the assessment trial. Participants received no feedback during assessment as to the accuracy of their responses. Videotaped assessment trials captured the cards in the participant’s hands. Later, the experimenters counted correct and incorrect responses from the recorded video clip for both decks. After participants were assigned to a group, the results of the pre-test became the first data point in baseline.

Baseline

During baseline, all participants completed one trial (a 20-s timing) each day on the deck assigned to them for four consecutive days. The scoring procedures from pre-test were replicated in baseline (i.e., the participants could not look at the movement cycles on the back of the cards, could not practice incorrect responses, did not receive any feedback, and had all trials videotaped). The first participant from each group with a stable baseline, defined as having a celeration of corrects below ×1.1 and incorrects accelerating or remaining at ×1.0, started the intervention phase. All other participants continued with a weekly baseline probe until they started intervention. As each participant met the criteria (ten correct responses in 20 s with less than one error across four consecutive data points), the participant with the most stable baseline started intervention next.

SAFMEDS Intervention

Intervention included daily (i.e., each day at work) frequency building or practice with the assigned SAFMEDS deck. During the intervention phase, each session consisted of several steps. First, the participants conducted 20 s timings similar to baseline. During the trial, however, they were able to turn the card over to view the back and see the correct label. While going through the cards and naming the movement cycle, the participants made their own piles for cards they labeled correctly and incorrectly.

When the 20 s trial ended, the experimenter gave the incorrect pile to the participants and instructed them to examine their incorrects and practice the correct movement cycle names for 1 min. After the 1 min of self-correction/self-feedback time came to an end, the experimenter shuffled all the cards back into the deck and repeated the 20 s trial. Shuffling occurred so each trial offered an opportunity to respond on any card in the deck. A shuffled deck appears to produce greater behavioral stability due to decreasing order effects (Graf and Auman 2005). The procedure of self-correction/self-feedback repeated itself for the second and third trial.

After the third SAFMEDS trial, the experimenters implemented the dependent variable in which the participants did not flip the cards over or make correct and incorrect piles; they instead made one pile without seeing the answer. The experimenters scored the videotaped dependent variable sessions later to obtain the score for correct and incorrect responses for that day. Participants continued in the intervention phase reaching fluency criteria of 17 or more correct responses and 0–1 incorrect responses per 20 s for two out of three consecutive days. The results (Figs. 2, 3) showed the number of daily sessions completed for each participant. While the experimenters encouraged daily application of the dependent and independent variable, occasional absences occurred (e.g., sick days, weekends, holidays).

Fig. 2
figure 2

Four Standard Celeration Chart segments displaying baseline, FBPC for SAFMEDS, and maintenance data for Group 1 participants. The data appear in calendar time to reflect the effects of absences and time off

Full size image
Fig. 3
figure 3

Four Standard Celeration Chart segments displaying baseline, FBPC for SAFMEDS, and maintenance data for Group 2 participants

Full size image

Post-test Extension Measure

The participants completed an extension measure on the same day they reached the intervention criteria. Participants completed the extension measure on the non-intervention deck (e.g., a participant who practiced Deck A during intervention completed the extension measure with Deck B). The extension measure mirrored pre-test in that participants did not flip over the cards and did not receive any feedback on the accuracy of their responses. The experimenters used the videotaped sessions to score the extension measure and report pre- and post-test accuracy results.

Maintenance

During maintenance, the participants completed one 20 s timing on their assigned deck once a week for the 4 weeks following their last intervention data point. Participants completed one trial (a 20 s timing) without the self-correction/self-feedback time as they did in the intervention phase. The experimenters scored the videotaped sessions later. Participants also completed a follow-up trial with a single 20 s timing conducted 6–7 months after post-test without self-correction/self-feedback.

Data Display

The experimenters displayed data (Figs. 2, 3) on sections taken from Standard Celeration Charts (SCC; Graf and Lindsley 2002; Lindsley 2005; Pennypacker et al. 2003). The figures have an exact portion of the SCC that places behavior within calendar time, shows proportional changes of behavior, and permits the quantification of important change measures (Pennypacker et al. 2003). Displaying participant data within the context of time demonstrates how the intervention affected each participant as it occurred temporally with breaks shown for weekends and sick days. Additionally, the SCCs quantify the following important features of visual analysis: trend, trend stability, variability, immediacy of an effect, and the consistency of data patterns across phases (Barlow et al. 2009; Fisher et al. 2003; Kazdin 2011; Kennedy 2005; Kratochwill et al. 2013; Parsonson and Baer 1978). In the present study, the following measures enhance visual analysis: for trend, celeration; for immediacy of an effect, frequency multipliers (or frequency jumps); for quantification of change in trend from baseline to intervention, celeration multipliers (or celeration turns); and accuracy improvement measure (or A.I.M.) to quantify the accuracy of behavior within a phase.

Celeration is a standard unit of measurement and quantifies change in performance frequency over time (Johnston and Pennypacker 2009). For example, a learner who read 20 words per minute on Monday and reads 40 words per minute on the following Monday has doubled her performance, or has a celeration of ×2.0 [7 days]. Celeration values include the number of calendar days (e.g., the length of the phase) shown in brackets to express the length of time data were measured. Celeration values also communicate the magnitude of behavior change and offer enhanced quantitative and visual analysis power (Datchuk and Kubina 2011). Data also suggest celeration lines enhance the accuracy of visual analysis (Bailey 1984; Stocks and Williams 1995).

To observe and quantify immediacy of an effect across phases (e.g., from baseline to first treatment phase), chart readers relate the value of the last data point in baseline to the first data point in an adjacent treatment phase. The resulting value expresses the magnitude of change due to the presentation of the independent variable (i.e., intervention) on the dependent variable. A frequency multiplier (or frequency jump) quantifies the change in value from one measured performance to the next (across phase lines). For instance, a ×3.5 jump up means the intervention frequency immediately grew or multiplied 3.5 times (e.g., a baseline data point of 4 grew to 14, a substantial 3.5 “jump up” in frequency).

Another measure used to quantify change across phases comes in the form of a celeration multiplier. Similar to a frequency multiplier that measures immediacy of behavior change after a condition line, a celeration multiplier quantifies the direction and magnitude of changes following the introduction of a variable. For example, a decelerating pattern of correct responding during baseline may change to an accelerating pattern during intervention (described as the celeration “turning up” or a turn up). The change from deceleration to acceleration across a condition line shows the celeration change. A multiplication or division symbol written before the numeral indicates the direction of celeration change. A ×3 celeration multiplier means the celeration change between two conditions increased by a factor of three (i.e., the celeration improved or “turned up” by a factor of three). Conversely, a ÷3 celeration multiplier describes a significant reduction in speed between conditions (i.e., the celeration decelerated or “turned down” by a factor of three).

The last measure used to evaluate change illustrates the relationship between correct and incorrect responding, called the accuracy improvement measure or A.I.M. An improvement over percentage-based measures, A.I.M. communicates how well learning has improved over a condition by calculating the change in accuracy (e.g., the relationship between the celeration of correct and incorrect responses) across successive performance frequencies. The higher the A.I.M. value, the more significant the learning change (Kubina and Yurich 2012). A performance with ×2.0 [10 days] A.I.M. shows significant improvement from the beginning of the measurement period to the end (e.g., the accuracy of performance doubled in 10 days).

Results

Tables 2 and 3 display all the baseline, intervention, maintenance, and extension data for participants in Group 1 (Mack, Thomas, Natalie, and Rachael) and Group 2 (Darla, Penny, George, and Hanna) on Standard Celeration Chart segments. The solid black dots on the charts represent correct performance frequencies (e.g., the number of correct pictures labeled during a 20 s timing). The Xs represent incorrect performance frequencies per timing (e.g., the number of pictures labeled incorrectly during a 20 s timing). The dashed and solid vertical lines signify condition changes (e.g., from baseline to intervention, from intervention to maintenance). The diagonal lines drawn through performance frequencies within each condition (celeration lines) provide a visual representation of the growth or decay of performance frequencies across each condition. Tables 2 and 3 provide change measures that quantify participant performances within baseline (celeration and A.I.M.) and illustrate performance changes following the SAFMEDS intervention (celeration, A.I.M., frequency multiplier, and celeration multiplier).

Table 2 Quantitative values for celeration and accuracy improvement measures (A.I.M.) used with Group 1 and Group 2
Full size table
Table 3 Quantitative values for the intervention measures frequency and celeration multiplier used with Group 1 and Group 2
Full size table

Group 1: Baseline, Intervention, and Maintenance Results

Figure 2, and Tables 2 and 3 display the data for the participants in Group 1. Baseline celerations measure the successive frequencies for correct and incorrect responses by participants in calendar time. Participant celerations for correct responses ranged from ×1.0 [7 days] (i.e., flat, no change in the frequency of responding across baseline) to a deceleration of ÷3.4 [15 days]. Mack and Rachel’s celerations for correct responses during baseline were ×1.0 [7 days] and ×1.05 [44 days]. Thomas and Natalie’s correct responses decelerated during baseline, ÷3.4 [15 days] and ÷1.05 [30 days], respectively. Mack, Thomas, Natalie, and Rachael’s incorrect responses during baseline revealed flat or growing celerations of ×1.3 [7 days], ×1.3 [15 days], ×1.2 [30 days], and ×1.0 [44 days], respectively.

The accuracy of participant responding was calculated using the accuracy improvement measure (A.I.M.). Overall, the A.I.M.s produced in Group 1 show stable or worsening performances. Mack’s, Thomas’s, Natalie’s, and Rachael’s A.I.M.s all show a deterioration of accuracy for baseline. The minimal threshold for significant A.I.M. starts at ×1.3 (Kubina and Yurich 2012). Each separate performance shows a worsening of accuracy across baseline.

The SAFMEDS intervention began after 7 (Mack), 15 (Thomas), 30 (Natalie), and 44 (Rachael) days of baseline. The visual pattern produced by the celerations for correct and incorrect responses during intervention showed the same form for all participants. Correct responses accelerated sharply during intervention and incorrects decelerated rapidly. Specifically, celerations for correct responses during intervention came to ×1.45 [18 days], ×1.4 [17 days], ×3.4 [7 days], and ×1.8 [9 days]. The decelerations for the incorrect responses of Mack, Thomas, Natalie, and Rachael’s were ÷2.45 [18 days], ÷1.6 [17 days], ÷7.4 [7 days], and ÷3.8 [9 days], respectively.

The concurrent celerations showed a visual pattern of improvement and, due to the ratio scale of the SCC, also led to quantification. A.I.M.s calculated during intervention show that all participants improved the accuracy of their responses (i.e., correct responses grew and incorrect responses decayed) throughout the SAFMEDS intervention. A.I.M.s for Mack, Thomas, Natalie, and Rachael came to ×3.6 [18 days], ×2.25 [17 days], ×25.2 [7 days], and ×6.85 [9 days], respectively.

The frequency multiplier shows and quantifies the immediate impact of the SAFMEDS intervention by comparing the last performance frequency in baseline with the first performance frequency in intervention. An increase in performance frequencies across phases is denoted using a “×” or times symbol and is referred to as a “jump up,” while a decrease in frequency across phases is written with a “÷” or division symbol and is referred to as a “jump down.” All Group 1 participants showed a jump up in correct frequencies between baseline and intervention (Table 3). Frequency multipliers for corrects were ×13, ×18, ×6.1, and ×2.8 for Mack, Thomas, Natalie, and Rachael. Frequency multipliers for incorrect responses all jumped down, quantified as ÷1.4, ÷3.9, ÷1.9, and ÷2.0, respectively.

Celeration changes between baseline and intervention are expressed using the celeration multiplier. All participants experienced celeration changes with a “turn up” in which correct responses accelerated faster when compared to baseline (Table 3). Celeration multipliers for correct responses are reported as Mack with a ×1.45, Thomas with a ×4.75, Natalie with a ×3.6, and Rachael showing a ×1.7. Incorrect celerations turned down for all participants. Celeration multipliers for incorrect responses were ÷3.2 (Mack), ÷2.1 (Thomas), ÷2.85 (Natalie), and ÷3.8 (Rachael).

Maintenance data in Fig. 1 revealed three out of four participants continued with consistently high performance frequencies for correct responses and low performance frequencies for incorrect responses. Thomas, Natalie, and Rachael all showed correct frequencies that were near the intervention performance criteria (i.e., 17 in 20 s). Incorrect frequencies were low initially for Thomas and Natalie and increased slightly during the last 2 weeks of maintenance. Rachael and Mack showed more variability with incorrects during maintenance. Mack’s frequencies for correct responses were initially low with high incorrects. However, the data points reversed at the last maintenance probe when Mack’s performance frequencies met the performance criteria for corrects (i.e., 17 in 20 s).

Group 2: Baseline, Intervention, and Maintenance Results

Figure 3, and Tables 2, and 3 display the data for the participants in Group 2. Baseline celerations for corrects included ×1.0 [7 days], ×1.0 [16 days], ×1.05 [30 days], and ×1.1 [50 days], respectively, for Darla, Penny, George, and Hanna. Celerations for incorrect responses for Darla, Penny, George, and Hanna included ×1.1 [7 days], ÷1.0 [16 days], ×1.05 [30 days], and ×1.05 [50 days]. All participant data showed stable, relatively flat patterns of responding, represented in Fig. 3 as level celeration lines for both corrects and incorrects (dots and X’s). The baseline A.I.M.s indicated very low to no change in accuracy for each participant across baseline (Darla, Penny, George, and Hanna), range from ×1.0 [16 days] to ÷1.1 [7 days].

All participants in Group 2 showed similar celeration change patterns in their response to the SAFMEDS intervention. Specifically, Darla, Penny, George, and Hanna accelerated their corrects and decelerated their incorrects during intervention. Corrects accelerated at ×1.2 [38 days], ×1.5 [28 days], ×2.05 [20 days], and ×1.8 [8 days] for Darla, Penny, George, and Hanna. Additionally, incorrects decelerated for each participant, respectively, with a ÷1.4 [38 days], ÷1.5 [28 days], ÷1.7 [20 days], and ÷7.6 [8 days].

Frequency multipliers calculated between baseline and intervention showed jump ups for correct responses in three out of four participants (Table 3). Frequency multipliers for Group 2 were ×18, ×6.2, ÷1.7, and ×2.9. George’s first three data points in intervention revealed lower performance frequencies (for correct responses) than during baseline, though George’s performance data quickly accelerated across the fifth to final data point during intervention. Three out of four participants exhibited a jump down for incorrects as soon as the SAFMEDS intervention began. The frequency multipliers for incorrects for Darla, Penny, and Hanna came to ÷2.8, ÷1.7, and ÷1.8. George’s incorrects did not jump up or down between baseline and intervention showing a frequency multiplier of ×1.0.

Celeration multipliers for Group 2 revealed the celeration changes between baseline and intervention for Darla and Penny as modest changes for correct responses, ×1.2 and ×1.5, and incorrect responses, ÷1.4 and ÷1.5. Hanna showed modest changes for correct responses, ×1.65, and a large change in the celerations of incorrects from baseline to intervention, at a ÷8.0. George’s performance illustrated a more pronounced effect with a celeration multiplier for corrects at ×1.95 and a ÷1.8 celeration multiplier for incorrects.

All Group 2 participants had intervention A.I.M.s superior to baseline A.I.M.s (Table 2). Darla’s A.I.M. changed from a baseline ÷1.1 [7 days] to ×1.7 [38 days] during intervention. Penny improved from a baseline A.I.M. of ×1.0 [16 days] to ×2.25 [28 days] in intervention. George changed from ×1.0 [30 days] to ×3.5 [20 days], and Hanna, who showed the most improvement in accuracy, increased from ×1.05 [50 days] in baseline to ×13.7 [8 days] in intervention.

Figure 2 shows Darla to have had the most stable maintenance within Group 2. Darla’s corrects and incorrects remained at or above the intervention performance criteria (i.e., 17 correct and one or less incorrect in 20 s) across 4 weeks post-intervention. The remaining Group 2 participants showed moderate-to-strong maintenance. Penny and George maintained intervention level performance frequencies across both corrects and incorrects in three out of four data points. However, Hanna showed a moderate decline in performance across 4 weeks post-intervention.

Extension Measure

Table 4 displays the pre- and post-test accuracy measures. Group 1 participants who received the SAFMEDS intervention with Deck A responded to Deck B (called an extension deck). Similarly, Group 2 participants who built frequency with Deck B responded to Deck A. All of the participants in Group 1 showed gains in accuracy from pre- to post-test (range of change = +17 to +27 %). Three out of four participants in Group 2 also had gains in accuracy on the extensions deck (range of change = −24 to +40 %). Overall, seven out of eight participants had accuracy gains with an extension deck of SAFMEDS.

Table 4 Quantitative values for the intervention measures frequency and celeration multiplier used with Group 1 and Group 2
Full size table

Discussion

The clinical efficacy of intervention relies on the accurate and consistent detection of behavior (Smith et al. 2013). All topographies and functions of behavior involve movement. Therefore, the ability to reliably identify targets that fulfill the definition of behavior forms an important goal for behavior analysts and special education teachers. The present experiment directly examined whether practicing behavior analysts and special education teachers could become fluent through the use of SAFMEDS in identifying movement cycles, a Precision Teaching term for a behavior (i.e., an active verb and object such as “shakes hand”). The experiment also investigated whether experimental effects would persist through time after the intervention ended.

A multiprobe multiple baseline across participants design answered the experimental questions for two groups of four participants each. A clear, evident experimental effect occurs when performance changes only during the implementation of the intervention (Kazdin 2011). Both visual analysis and the quantitative analysis of the Standard Celeration Chart segments provided the basis evaluation of the present experiment.

The charted data reveal that for all participants, the incorrects were at a higher level than the corrects during baseline. Additionally, five out of eight participants had worsening conditions demonstrated by A.I.M. Two other participants had A.I.M.s of ×1.0 which meant their performance did not change. Only one participant exhibited a gain of ×1.05 for 50 days of baseline. A 5 % improvement in accuracy over an almost 3-month period of baseline, however, does not indicate a meaningful improvement. Furthermore, no one made a significant improvement using movement cycle labels during baseline and the majority worsened considerably across time.

Upon implementation of the intervention, all four participants in Group 1 had substantial jump ups in frequency (range ×2.8 to ×18). Furthermore, the incorrects indicated very abrupt jump downs in the amount of incorrects upon initiation of the intervention (range ÷1.4 to ÷3.9). In Group 2, three out of the four participants also showed very large, immediate performance jump ups for corrects (range ×2.9 to ×18) and jump downs for incorrects (range ÷1.7 to ÷2.8). Only George had a jump down in corrects during the first day of the intervention with incorrects staying the same. The evidence suggested SAFMEDS had an immediate, robust effect. Participants said more correct and fewer incorrect movement cycles when seeing picture cards representing active behavior.

While immediacy of effects speaks to the initial impact of the SAFMEDS intervention, the subsequent celerations indicate whether the effect was sustained. All eight participants had accelerating corrects (range from ×1.2 to ×3.4 or 20–240 % gain) and decelerating incorrects (from ÷1.4 to ÷7.6 or 29–87 % reduction). The celerations, along with the notable A.I.M.s, provide visual and quantitative evidence that very strong gains occurred in Group 1 and Group 2. Each multiple baseline contains three replications of the SAFMEDS intervention. All participants sustained their learning of identifying movement cycles. The data suggest the frequency building combined with self-feedback as part of the SAFMEDS intervention lead to a replicable, positive outcome.

Each group used a different but equivalent group of pictures of behavior and said the movement cycle depicted on the SAFMEDS card. Because each group had different decks with different pictures, the experiment showed SAFMEDS producing the same effect twice in a counterbalanced fashion. Taken in whole, the visual analysis and subsequent quantitative values support the use of SAFMEDS for fostering behavioral fluency with the behavior analysts and special education teachers in the experiment.

When people achieve behavioral fluency, an effect called retention occurs (Kubina and Yurich 2012). Retention refers to the relation between two behavior frequencies taken at two different points in time after an individual has not had opportunity to emit the behavior (Binder 1996, 2005; Johnson and Layng 1996). A number of experimental studies have demonstrated that fluent behavior persists at high levels even after an intervention has been terminated (e.g., Bucklin et al. 2000; Dermer et al. 2009; Hughes et al. 2007; Kubina et al. 2008). In the present experiment, seven out of the eight participants had corrects which maintained a high level after the intervention ended, supporting the finding that behavioral fluency leads to a maintenance of stimulus control.

During maintenance, Mack showed the most variability with corrects and incorrects. Because Mack had an illness precluding measurement of his first maintenance point, he did not receive another measure until 2 weeks after the intervention ended. His first and second maintenance points show the incorrects rising with corrects falling. During the last measure of maintenance, Mack claimed he remembered how to form movement cycles. It appeared that the main source of variability was Mack not adding an “s” to the end of the action verb.

Examining the incorrects for maintenance, four participants had accelerating trends (i.e., Thomas, Natalie, Darla, and Hanna). Mack, Rachel, Penny, and George had a low level, or deceleration, of incorrects. After closely inspecting the incorrect performance, the experimenters discovered a pattern among all the participants; most incorrects the participants exhibited focused on the object part of the movement cycle. For example, the picture depicting “places slinky” was identified as “places toy.” Even though the participants used the proper action verb with an “s” and did say an object, the experimenters recorded an incorrect. The specificity with the object label in a number of pictures was not retained even though the participants did say an object in the generic form. Occasionally, the participants did say the verb incorrectly. Nevertheless, all participants demonstrated the correct movement cycle form in maintenance, defined as saying an observable action verb in the present tense, choosing an object that ends the cycle of movement, and adding an “s” to the action verb.

The extension measure for all participants reflected the previously discussed finding. At the end of the experiment, participants from each group responded to a novel deck of movement cycles. Seven out of eight participants demonstrated gains in accurately responding to a new deck. Upon closer inspection, the incorrect responses occurred on the side of the object. All participants had responses where they used an active verb, present tense, and an object receiving the action. Meindl et al. (2013) conducted an experiment and also found responses to novel decks not positively affected by the SAFMEDS frequency-building intervention. Meindl et al. incorporated multiple exemplar training and noted an improvement in regards to accuracy. While Meindl et al. used definitions in their intervention different from the pictures used in the present study, the discovery of multiple exemplar training suggests another reason why participants did not do even better on the extension decks; Deck A and Deck B did not have sufficient multiple exemplars which may have led to better transfer.

Frequency building to a performance criterion (FBPC) has two parts, the process of frequency building (i.e., timed repetition of performance and performance feedback) and a frequency aim associated with the critical learning outcomes of retention, endurance, and application (Kubina and Yurich 2012). In the present experiment, all participants worked until they reached the performance criterion suggested by Graf and Auman (2005) for SAFMEDS, 17 correct per 20 s. It took one participant, Natalie, as little as 7 days to reach aim while Darla took the longest, 38 days. With all the participants attaining the performance criterion and in an efficient manner and demonstrating strong maintenance, it appears the 17 correct per 20 s represents a justifiable frequency aim.

The experimental literature shows SAFMEDS research has a very wide range of applicability. As an example, in a sixth-grade class, SAFMEDS lessened the time spent learning unit taxonomy and increased time working with specimens (e.g., kingdom, phyla, class; Clorfene et al. 1998). A SAFMEDS intervention also helped students with mild disabilities pass minimal competency requirements for earning a high school diploma (Byrnes et al. 1990). Using SAFMEDS also resulted in university students becoming fluent with complex statistical information (Beverley et al. 2009). The present study systematically extends the SAFMEDS database by promoting fluency identifying movement cycles for behavior analysts and special education teachers. The present study also replicated many of the critical learning outcomes obtained in the previous studies: attainment of fluent responding with the targeted subject matter, long-term retention, and extension to novel content.

Limitations

During the self-correction interval for participants, it became obvious that each person started to improve, they required less time to go over incorrects. Participants did not enjoy having to sit for 1 min when they only had one or no errors. The imposed 1 min of feedback for all applications of frequency building may have affected the social validity in terms of participant satisfaction with the methods. The present study also assessed the dependent variable after the participants engaged in the FBPC intervention. It is possible a priming effect occurred which led higher performance frequencies compared to assessing the dependent variable before the intervention.

Future Research

The present experiment demonstrated practicing behavior analysts and special education teachers could become fluent identifying movement cycles when seeing behavior portrayed on SAFMEDS cards. Future research could focus on the how well behavior analysts use movement cycles to conduct functional assessments or craft behavior plans. Additionally, continuing the search for the best active verbs and using generic objects instead of specific targets (e.g., toy vs. transformer) may produce more efficient outcomes. Determining how multiple exemplars further support application to novel content could further advance the use of FBPC and SAFMEDS to behaviors important to the science of behavior. And further modifications to the frequency-building procedure may result in more substantial maintenance. Finally, examining whether movement cycles represent superior alternatives to operations definitions could impact detection and measurement of behavior in both applied and experimental settings. Changing the static SAFMEDS cards to dynamic videos depicting active movement may also lead to greater outcomes.