Quantitative Synthesis and Component Analysis of Single-Participant Studies on the Picture Exchange Communication System

Abstract

The Picture Exchange Communication System (PECS) has emerged as the augmentative communication intervention of choice for individuals with autism spectrum disorder (ASD), with a supporting body of single-participant studies. This report describes a meta-analysis of 16 single-participant studies on PECS with percentage of nonoverlapping data (PND) as the metric of effect size. Results suggest that PECS was moderately effective in establishing mands (PND = 80.1) for 41 participants up to Phase IV of the system. Higher levels of manding were found when PECS was taught to individuals without ASD diagnoses versus those with ASD diagnoses and in single settings versus multiple settings; however, these differences were not statistically significant. For a smaller subset of participants for whom vocalizations were recorded, PECS appeared to facilitate speech, though considerable variability in speech acquisition was evident. While these results support PECS as an evidenced-based communication intervention, they indicate that more research is needed on speech with PECS, to establish the efficacy of PECS when implemented across settings and communicative partners, and to confirm efficacy of Phases IV, V, and VI.

Keywords

autism behavior analysis meta-analysis research methodology single-participant research methodology

Communication deficits are a core feature of autism (American Psychiatric Association, 2000). Of individuals with autism, 30% or fewer develop functional speech (Miranda-Linné & Melin, 1997), and up to 50% may remain mute as adults (Wetherby & Prizant, 2005). A variety of interventions has evolved to address communication deficits that accompany autism (Mirenda, 2003; Simpson et al., 2005). These include unaided communication systems, such as gestures and sign language, and aided communication systems, such as voice output communication aides and picture-based systems.

One picture-based system, the Picture Exchange Communication System (PECS; Frost & Bondy, 2002), is among the most popular interventions for children with autism and related disabilities. For example, Stahmer, Collings, and Palinkas (2005) found that 95% of children with autism served in California community early intervention programs received PECS, more than any other intervention methodology.

The PECS protocol comprises six phases (Frost & Bondy, 2002). The individual is taught to initiate interactions and to communicate in real-world situations in Phases I and II, whereas Phases III and IV seek to increase vocabulary and to introduce sentence structure. Phase V teaches responding to another’s request (i.e., “What do you want?”), and Phase VI establishes commenting in response to a question (e.g., “What do you see?”). Although the intent of PECS is to teach functional communication by establishing an increasingly complex repertoire of picture exchange (Bondy, Tincani, & Frost, 2004), development of speech has been reported for some users (e.g., Charlop-Christy, Carpenter, Le, LeBlanc, & Kellet, 2002; Tincani, Crozier, & Alazetta, 2006; Yoder & Stone, 2006).

Research demonstrating PECS’s efficacy, although promising, is emerging. A few recent studies have used group comparison methodologies to examine the effects of PECS on communicative skills of children with an autism spectrum disorder (ASD), finding favorable, if mixed, outcomes (Carr & Felce, 2007a, 2007b; Howlin, Gordon, Pasco, Wade, & Charman, 2007; Yoder & Stone, 2006). For example, Yoder and Stone (2006) conducted a randomized controlled trial (RCT) study comparing PECS and Responsive Education and Prelinguistic Milieu Teaching with 36 preschool children with ASD. Both interventions increased children’s generalized spoken communication at posttreatment and follow-up; however, PECS resulted in a differentially higher frequency of spoken communication for children who demonstrated high object exploration before treatment. This finding makes sense in light of an initial emphasis within PECS on using children’s preexisting interests in reinforcing items to promote communication (Frost & Bondy, 2002). Similar, Carr and Felce (2007a) reported an increase in child-to-adult initiations of 24 children who received PECS up to Phase III in comparison to 17 children who did not receive PECS. In contrast, Howlin et al. (2007) conducted a RCT study evaluating the impact of classroom-level training and consultation in PECS on communication skills of elementary-age children with ASD. Although increases in communicative initiations and picture exchanges occurred in conjunction with PECS training and consultation, these improvements diminished on cessation of classroom consultation visits. Moreover, they did not observe increases in children’s vocal communication skills with PECS, as measured by classroom observations and standardized assessments.

Although group comparison studies have yielded fruitful and promising results, significant questions remain about the efficacy of PECS as an augmentative communication system. For instance, PECS was initially developed for children with ASD and has been empirically validated for preschool- and elementary-age students. However, more recent studies suggest that PECS can be successfully used with adolescents and adults, including individuals who lack diagnoses of autism or pervasive developmental disorder, not otherwise specified (PDD-NOS; Chambers & Rehfeldt, 2003; Ziomek & Rehfeldt, 2008). It remains unclear if factors such as age and diagnosis affect acquisition of PECS and if the system is useful for individuals who are older or who lack ASD diagnoses.

Furthermore, although the PECS phases are cumulative, with advancement contingent on mastery of previous phases (see Frost & Bondy, 2002), rates of mastery may vary considerably among individuals (Bondy & Frost, 1994), and it is unclear if learners tend to advance through most or all of the system. For instance, although the beneficial effects on basic communication skills of Phases I through III are apparent (e.g., Carr & Felce, 2007a), less is known about a user’s acquisition of the more advanced communication skills taught in Phases IV through VI.

Finally, as speech has been reported as a desirable side effect for some users, more information is needed to confirm the utility of PECS in promoting speech. This is particularly important given inconsistent findings yielded by group comparison studies relative to speech (Carr & Felce, 2007a; Howlin et al., 2007).

In addition, PECS is a complex implementation system involving an array of teaching techniques derived from the field of applied behavior analysis (Bondy & Frost, 1994). The system’s relative complexity sets the occasion for procedural variations, both intended and unintended. For example, Howlin et al. (2007) speculated that “less reliable or less frequent implementation of PECS” (p. 479) may have diminished students’ communication skills following the end of classroom consultation in their study. Therefore, it is critical for investigators who are validating PECS to demonstrate fidelity with the teaching procedures outlined within Phases I through VI; however, the extent to which such fidelity measures are in place within published studies has yet to be fully quantified.

To date, the majority of published studies supporting PECS have utilized single-participant designs (e.g., Charlop-Christy et al., 2002). These studies provide a potentially rich source of data to augment findings of group studies and to answer specific questions regarding the viability of PECS. Importantly, because PECS is among the most popular interventions for persons with autism and related disabilities, there is a need to synthesize single-participant studies to verify the system’s efficacy in promoting functional communication and speech and to address specific questions regarding characteristics of individuals who may benefit from the system, mastery of phases, and production of speech. Although several investigators have recently conducted summaries of PECS research including single-participant studies, these have lacked quantitative measures of effect size or have not included a component analyses to evaluate the interaction of critical variables—such as participants’ diagnosis, setting, and age—on acquisition of picture exchanges and speech (cf. Ostryn, Wolfe, & Rusch, 2008; Sulzer-Azaroff, Hoffman, Horton, Bondy, & Frost, 2009; Tien, 2008).

Meta-analysis is a procedure for quantitatively synthesizing the effects of an intervention across several studies, thus providing one measure of the efficacy of a particular treatment (Scruggs & Mastropieri, 1998). Several comparable methods exist for quantitatively synthesizing single-participant studies (Olive & Smith, 2005); however, calculating the percentage of nonoverlapping data (PND) between baseline and intervention phases is a commonly used metric. Specifically, PND is the percentage of data points in the treatment phase over the highest data point occurring in the baseline phase (Scruggs, Mastropieri, & Casto, 1987). Thus, the higher the PND, the more effective an intervention. Specifically, a PND of 90% or higher demonstrates highly effective, 90% to 70% indicates moderately effective, 70% to 50% represents mildly or questionably effective, and 50% or below demonstrates ineffective treatment (Ma, 2006; Scruggs & Mastropieri, 1998).

The purpose of the present study was to examine the efficacy of PECS in establishing functional communication and speech in individuals with autism and other disabilities. A meta-analysis of 16 peer-reviewed single-participant studies comprising 44 participants was conducted with PECS as the independent variable and picture exchange and vocalization as dependent variables. Of the 44 participants in the studies, data on picture exchange were graphically depicted for 41 participants, whereas data on vocalizations were graphically depicted for 12 participants. Nonparametric component analyses—the Mann–Whitney U and Kruskal–Wallis one-way ANOVA—were conducted to assess for interactions among participant characteristics (gender, age, disability, highest phase of PECS mastered), setting characteristics (single setting vs. multiple settings, type of setting), and picture exchange acquisition. Conclusions about the efficacy of PECS and relevance for persons with autism and other disabilities are drawn based on these results.

Method

Procedure

Peer-reviewed articles were identified by the first and second authors through searches of the PsycINFO, ERIC, and ISI Web of Knowledge databases conducted by the first author. Picture exchange communication system, PECS, and autism were the keywords used to search the databases. In addition, manual searches of the following journals were conducted to locate studies meeting the authors’ inclusion criteria: Education and Training in Developmental Disabilities, Research in Autism Spectrum Disorders, Journal of Applied Behavior Analysis, Journal of Autism and Developmental Disorders, Remedial and Special Education, and Focus on Autism and Other Developmental Disabilities.

From the articles identified through database searches, articles for analysis were selected based on the following five criteria: (a) the researchers used a single-participant research design (i.e., alternating treatment with baseline, multiple baseline, multiple probe, or ABAB reversal); (b) the PECS intervention as described in Frost and Bondy (2002) was implemented in the study; (c) the participants were reported to have autism diagnoses, PDD-NOS diagnoses, dual diagnoses, or other diagnoses of disability; (d) the effect of the intervention on picture exchange and/or vocalizations was empirically measured and graphically illustrated with clearly identifiable baseline and intervention phases; and (e) the article was published in a peer-reviewed journal.

The first and second authors identified 16 studies published between 2002 and 2009 in nine journals: Behaviour Change, Focus on Autism and Other Developmental Disabilities, Education and Training in Developmental Disabilities, Japanese Journal of Special Education, Journal of Applied Behavior Analysis, Journal of Autism and Developmental Disorders, Remedial and Special Education, Research in Autism Spectrum Disorders, and Research in Developmental Disabilities. The second author manually reviewed each of the 16 articles to confirm that they met the five inclusion criteria.

These 16 articles used single-participant designs, reporting quantitative data for picture exchange and/or vocalizations. Importantly, each of the studies employed a baseline from which the effectiveness of PECS could be evaluated and made explicit reference to Frost and Bondy’s (2002) PECS training manual in its current or previous edition within a description of the procedures. Additional refereed journal articles on the PECS found in database searches were excluded from the analysis because they were (a) descriptive or anecdotal in nature (e.g., Bondy & Frost, 1994); (b) did not meet all of the authors’ inclusion criteria, including presence of a baseline (Ganz & Simpson, 2004) and implementation of the PECS protocol (Buckley & Newchok, 2005); and/or (c) did not use a single-participant research design (Magiati & Howlin, 2003; Schwartz, Garfinkle, & Bauer, 1998). For instance, Ganz and Simpson (2004) employed a single-participant design to evaluate PECS, but their study lacked baseline measures of participants’ communication skills prior to intervention. Related, Schwartz et al.’s (1998) quasi-experimental group study on PECS employed neither baseline measures nor a comparison group. In addition, one single-participant study (Angermeier, Schlosser, Luiselli, Harrington, & Carter, 2008) was excluded from the analysis because baseline measures strongly suggested that the participant had a history of PECS training prior to the study.

Of the 44 participants included, 34 were male and 10 were female. The participants’ diagnoses included autism, PDD-NOS, nonspecific “developmental delays,” mental retardation, seizure disorder, Down syndrome, blindness, cerebral palsy, and expressive and receptive language disorder. For analysis purposes, the participants were divided into three age groups: (a) preschool age (2 to 5 years old), (b) school age (6 to 17 years old), and (c) adults (18 years and older). All 44 participants were treated with the PECS intervention; however picture exchanges (mands) were reported for only 41 participants. Vocalizations were reported for a total of 12 participants, including 3 participants for whom picture exchange data were not reported.

For the 41 participants for whom picture exchange was measured as a dependent variable, information was coded relative to the following: (a) diagnosis, (b) gender, (c) age, (d) highest PECS phase mastered, and (e) setting. Autism or PDD-NOS constituted 51% (n = 21) of the sample, and other diagnoses constituted 49% (n = 20). Males comprised 76% (n = 31) of the participants, females 24% (n = 10). Preschool-age participants constituted 41% (n = 17) of the sample, 32% (n = 13) were school age, and 27% (n = 11) were adults. Only 7% (n = 3) of the sample mastered only up to Phase I, 15% (n = 6) mastered up to Phase II, 49% (n = 20) mastered up to Phase III, 20% (n = 8) mastered up to Phase IV, and 9% (n = 4) did not report mastered phases. Intervention was performed in single settings (e.g., home, school, training center) for 83% (n = 34) of the sample and in multiple settings for 17% (n = 7) of the sample.

For the 12 participants for whom vocalizations were recorded as a dependent variable, the following characteristics were recorded. Individuals with autism or PDD-NOS constituted 100% (n = 12) of the sample. Males comprised 92% (n = 11) and females 8% (n = 1). Preschool-age participants constituted 50% (n = 6) of the sample, and 50% (n = 6) were school age. None mastered up to Phase I, none up to Phase II, 33% (n = 4) up to Phase III, 42% (n = 5) up to Phase IV, and 25% (n = 3) up to Phase VI. Intervention was performed at school for 50% (n = 3), and for 50% (n = 3) it was performed in multiple settings.

In addition to the aforementioned variables, it was coded whether or not each study included quantified procedural fidelity data. A study was coded Y if the authors reported that an observer monitored the PECS intervention and reported the percentage of steps correctly implemented. A study was coded N if the authors did not report that an observer monitored the PECS intervention and did not report the percentage of steps correctly implemented.

Table 1 presents a brief description of the 16 peer-reviewed studies, including participant characteristics, research design, phases of PECS intervention received, results, whether procedural fidelity measures were taken, and average PND for picture exchange and/or vocalizations across participants.

Table 1.

Peer-Reviewed Single-Participant Studies on the Picture Exchange Communication System (PECS)

Authors	Research Design	Participant Characteristics	PECS: Phases Received and Results of Training	Fidelity? (Y/N)	Picture Exchange PND	Vocalization PND
Anderson, Moore, and Bourne (2007)	A-B-C-D design	1 school-age male w/ autism	I–IV. Zero levels of picture exchange and very low levels of vocalization observed during baseline. Increases in picture exchange, vocal initiations, and new words spoken following PECS training. Increases in play also observed.	N	70.6	Initiations: 82.4
Bock, Stoner, Beck, Hanley, and Prochnow (2005)	Alternating treatment design w/ baseline	6 preschool-age males w/ developmental delays^a	I–III; all children demonstrated stable, low baselines before intervention was initiated; all children advanced to Phase II; 5 children met criterion and moved to Phase II; 2 children met criterion for completion of Phase III.	Y	93.0	Not reported
Carré, Le Grice, Blampied, and Walker (2009)	Multiple baseline design across participants	3 children (2 males, 1 female) between 5 and 6 years old; 2 w/ autism, 1 w/ Down syndrome and “autistic features”	I–III; participants demonstrated zero levels of picture exchange at baseline; all participants acquired at least Phase II during training. Participants demonstrated moderate and varying levels of generalization of picture exchange in nontraining settings; higher rates of picture exchange were observed in classroom generalization settings than home generalization settings.	Y	50.4	Not reported
Chambers and Rehfeldt (2003)	Alternating treatment design w/ baseline	4 adults (2 males, 2 females): 1 w/ MR and expressive and receptive language disorder; 1 w/ severe MR, cerebral palsy, and seizure disorder; 1 w/ Down syndrome, MR, and seizure disorder; 1 w/ MR	I–III; all adults were not able to mand (picture exchange) for any items during baseline; 3 participants performed with 100% accuracy using PECS; 1 participant showed improvement from baseline but did not continue training because of acquired infectious disease.	N	100.0	Not reported
Charlop-Christy, Carpenter, Le, LeBlanc, and Kellet (2002)	Multiple baseline design across settings and participants	2 preschool-age and 1 school-age males w/ autism	I–VI; phases were acquired but results for picture exchange were not reported; all participants displayed an increase of spontaneous speech or imitation (vocalizations) from pretraining to posttraining; during 1-year follow-up sessions, spontaneous speech gains were maintained.	N	Not reported	Spontaneous: 62.8, imitative: 75.2
Ganz, Simpson, and Corbin-Newsome (2008)	Multiple baseline design across participants	3 preschool-age: 1 male w/ autism, aphasia, global developmental delay, and ADHD (P1); 1 male w/ autism (P2); 1 female w/ global developmental delay and seizure disorder (P3)	I–IV; participants demonstrated no picture exchanges during baseline; 2 participants acquired Phase IV; 1 participant acquired no PECS phases and demonstrated low levels of responding throughout intervention; participants did not significantly increase word approximations or intelligible words with PECS training.	N	64.1	P1: 9.6 (vocal approximations) 7.7 (intelligible words); P2: 4.6 (vocal approximations), 2.8 (intelligible words); P3: Not reported (0)
Jurgens, Anderson, and Moore (2009)	A-B-C-D-E design	1 preschool-age male w/ autism	I–IV; picture exchanges were measured in generalization settings only. Participant demonstrated zero levels of picture exchange during baseline; picture exchanges remained at low levels during PECS intervention; however, vocal initiations and vocal mands increased in generalization settings.	Y	20.0	Vocal initiations: 60.0, vocal mands: 35.0
Kravits, Kamps, Kemmerer, and Potucek (2002)	Multiple baseline design across settings	1 school-age female w/ autism	I–III; picture exchanges rarely occurred during baseline; picture exchange increased when PECS implemented; effects were consistent across settings; vocalizations reportedly increased, however, results for vocalizations were not separately graphed.	N	89.8	Not reported
Lund and Troha (2008)	Multiple baseline design across participants	3 school-age (2 males, 1 female) w/ autism, blindness, and cognitive impairment	I–III with tactile modifications; no participants demonstrated picture exchanges during baseline; 1 participant demonstrated mastery of Phase III and 2 participants demonstrated mastery of Phase I with tactile modifications of PECS.	N	55.0	Not reported
Marckel, Neef, and Ferreri (2006)	Multiple baseline design across descriptors	2 preschool-age males w/ autism	Not specified; during baseline, picture exchanges rarely occurred, if at all; mands steadily increased and were maintained at high levels as PECS was implemented.	Y	98.0	Not reported
Sigafoos, Ganz, O’Reilly, Lancioni, and Schlosser (2007)	Multiple probe design across participants	2 school-age males: 1 w/ PDD-NOS and MR, 1 w/ autism	Not specified; participants maintained zero level baselines; a rapid increase in mands picture exchanges were demonstrated when training was implemented.	Y	100.0	Not reported
Stoner et al. (2006)	ABAB, reversal design	5 adults (4 males, 1 female): 2 w/ Down syndrome, 1 w/ developmental delays,^a 1 w/ cerebral palsy, and seizure disorder, 1 w/ seizure disorder	I–IV; 3 of 5 participants successfully completed PECS through Phase IV, other 2 participants did not progress past Phase III; successful participants: 2 demonstrated stable baselines, other participant engaged in low level of picture exchange; 1 participant maintained ability to use PECS even after second baseline, other participants showed a decrease in performance; unsuccessful participants: 1 engaged in low level of picture exchange during initial baseline, other participant demonstrated stable baseline; both demonstrated difficulties progressing through phases, 1 discontinued PECS training, other participant showed reduction in accuracy after seizure, was reinstated at Phase 1, and then discontinued training shortly after.	N	78.1	Not reported
Tincani (2004)	Alternating treatment design w/ baseline and final best-treatment phase	1 preschool-age male w/ autism and MR; 1 school-age female w/ PDD-NOS	I–III; participants’ picture exchanges increased overall; 1 participant demonstrated more success with other treatment; therefore, PECS was not chosen for final best-treatment phase; the other participant demonstrated 3 times more mands in PECS than other treatment; therefore, PECS was chosen for final best-treatment phase; both participants vocalizations increased overall.	Y	86.4	P1: Not reported (0), P2: 100 (word vocalizations)
Tincani, Crozier, and Alazetta (2006)	Delayed multiple-baseline design across participants	2 school-age males w/ autism	I–IV; during baseline, picture exchanges rarely occurred, if at all; for both participants, exchanges increased during Phases I–IV and were maintained during generalization; vocalizations reported for only 1 participant, who displayed a substantial increase in speech while PECS was implemented.	Y	97.5	P1: 62.5 (word approximations), P2: not reported (0), P3: 100 (word approximations)
Yokoyama, Naoi, and Yamamoto (2006)	Multiple-baseline with changing criterion	2 preschool-age and 1 school-age w/ autism	I–III; 2 participants demonstrated low levels of independent picture exchanges during baseline, 1 participant demonstrated a high level of independent picture exchanges, thus lowering overall PND. All participants acquired PECS skills up to Phase III. Collateral increases in speech were also reported.	N	64.2	Increases reported for all participants
Ziomek and Rehfeldt (2008)	Alternating treatments design w/ baseline	3 adults (1 male, 2 females) w/ MR and multiple disabilities	I–III; participants demonstrated zero levels of picture exchange during baseline. 2 participants acquired Phase III; the other participant acquired only Phase I. For the participants who acquired Phase III, picture exchanges generalized across people, settings, and functions.	N	97.6	Not reported

Note: PND = percentage of nonoverlapping data; MR = mental retardation; ADHD = attention-deficit/hyperactivity disorder; P1 = Participant 1; P2 = Participant 2; P3 = Participant 3; PDD-NOS = pervasive developmental disorder, not otherwise specified.

Diagnostic criteria not available.

Analyses

The PND between baseline and treatment phases was calculated for the first to last phases of PECS received per participant. PND was calculated by dividing the number of picture exchange or vocalization data points exceeding the highest baseline data point by the total number of data points in the treatment phase and multiplying the sum by 100 (Scruggs et al., 1987).

For each article reviewed, PND scores for picture exchange (mands) were calculated between baseline and intervention phases for each participant. The PNDs for each participant were then averaged to provide an aggregate PND score for each study (see Table 1). In the case of the ABAB reversal design, PND was calculated for each AB of the ABAB design, considering Kazdin’s (1982) observation that “it is quite possible that behavior will not revert toward baseline levels once the intervention is withdrawn or altered” (p. 121). The individual PNDs for each AB of the reversal design were averaged to provide one estimate of the acquired phases for each participant. The first author and a graduate student simultaneously coded and calculated PNDs for 5 of the 16 studies (31%). Interobserver agreement was 100%.

Researchers employed varying operational definitions and measurement procedures for the 12 participants for whom vocalization data were reported. For instance, Charlop-Christy et al. (2002) measured spontaneous and imitative speech in play and academic demand settings, whereas Tincani (2004) and Tincani et al. (2006) measured words and word approximations during PECS training. Furthermore, some participants demonstrated no improvements in speech on implementation of PECS; hence, speech data were not graphically depicted for these participants. Vocalization data thus could not be aggregated in the same manner as for picture exchange. Instead, for Charlop-Christy et al. (2002), PND speech scores were averaged for spontaneous and imitative speech across their three participants. For Anderson, Moore, and Bourne (2007), Ganz, Simpson, and Corbin-Newsome (2008), Jurgens, Anderson, and Moore (2009), Tincani (2004), and Tincani et al. (2006), PND speech scores were reported individually by participant, and the type of vocal response was indicated. For additional participants in these studies for whom vocalization data were not reported, a PND of zero denoted failure to acquire any speech. In the study by Yokoyama, Naoi, and Yamamoto (2006), vocalization data were graphically depicted; however, the method of graphing used by these authors did not permit calculation of PND. Therefore, it was simply indicated that increases in vocalization were reported for the participants in this study (see Table 1).

Nonparametric tests were used to analyze the relationship, if any, between participant characteristics and participants’ acquisition of picture exchange (Scruggs et al., 1987). To investigate whether the dichotomous variables of diagnosis (autism or PDD-NOS vs. other diagnosis), gender, and setting (single vs. multiple) affected PECS acquisition, separate Mann–Whitney U tests were conducted using individual participants’ picture exchange PND scores grouped accordingly as dependent variables. The effects of age (preschool, school age, or adult) and highest PECS phase (I, II, III, IV, V, or VI) mastered were examined with separate Kruskal–Wallis one-way ANOVAs, with individual participants’ picture exchange PND scores also serving as the dependent variable. Nonparametric tests were not conducted for vocalizations because of the small number of participants, 12, for whom vocalization data were reported and variations in operational definitions and measurement of speech evident in these studies.

Results

The meta-analysis resulted in 41 unique PND scores for picture exchange. The overall mean PND for picture exchange across participants was 80.1 (n = 41, SD = 26.7). According to Scruggs and Mastropieri’s (1998) criteria, a PND value of 80.1 represents a moderately effective treatment.

PND scores for picture exchange by participant characteristics are presented in Table 2. Acquisition of PECS was substantially similar across both male (PND = 80.8) and female (PND = 77.6) participants; however, variability in acquisition of PECS was apparent for individuals with differing diagnoses and ages, those with differing phases of PECS mastered, and those taught PECS in single versus multiple settings. Specifically, higher levels of manding were found when PECS was taught to individuals without autism or PDD-NOS diagnoses (PND = 85.0) versus those with autism or PDD-NOS diagnoses (PND = 75.4). PECS also produced differing levels of acquisition for preschool (PND = 80.5), school-age (PND = 71.3), and adult (PND = 89.7) participants. In addition, PECS produced substantially lower levels of acquisition for the three participants who mastered only Phase I (PND = 45.5) compared to those who mastered Phase II (PND = 81.1), Phase III (PND = 82.5), and Phase IV (PND = 76.7). Finally, individuals who were taught PECS in single settings demonstrated higher levels of acquisition (PND = 81.8) compared to those taught PECS in multiple settings (PND = 71.5).

Table 2.

PND Score Means and Standard Deviations for Picture Exchange Communication System (PECS) by Participant Characteristics

	PND
Participant Characteristic	n	M	SD
Diagnosis
Autism or PDD-NOS	21	75.4	28.8
Other	20	85.0	24.1
Gender
Male	31	80.8	25.8
Female	10	77.6	30.8
Age
Preschool	17	80.5	23.8
School age	13	71.3	34.8
Adult	11	89.7	17.3
Highest PECS phase mastered
I	3	45.5	32.2
II	6	81.1	19.9
III	20	82.5	26.8
IV	8	76.7	28.6
Not reported	4	99.0	2.1
Setting
Single	34	81.8	26.4
Multiple	7	71.5	28.7

Note: PND = percentage nonoverlapping data; PDD-NOS = pervasive developmental disorder, not otherwise specified.

Mann–Whitney U tests were conducted to assess for significant differences among groups, with individual participants’ picture exchange PND scores grouped according to diagnosis, gender, and setting. Kruskal–Wallis one-way ANOVA tests were also conducted, with participants’ picture exchange PND scores grouped according to age and highest phase of PECS mastered. No statistically significant differences were found when PND scores were grouped according to diagnosis (Mann–Whitney U = 165.50, p = .15), gender (Mann–Whitney U = 153.00, p = .96), age (χ² = 2.74, p = .25), highest phase of PECS mastered (χ² = 8.18, p = .08), and setting (Mann–Whitney U = 84.50, p = .222), indicating that PND scores did not significantly differ on the basis of diagnosis, gender, age, highest phase of PECS mastered, and setting.

Vocalization PND scores were not aggregated across studies. As shown in Table 1, there was considerable variability between participants in the degree of speech acquired with PECS, with some studies indicating moderate to substantial improvement and others indicating little or no improvement. For instance, the participant in the Anderson et al. (2007) study had a PND score of 82.4 for vocal initiations, and participants in the Charlop et al. (2002) study had, on average, PND scores of 62.8 for spontaneous speech and 75.2 for imitative speech, suggesting that PECS was mildly to moderately effective for increasing speech in these investigations. In contrast, in the Tincani (2004) study, Participant 1 demonstrated no speech with implementation of PECS, whereas Participant 2 demonstrated a substantial increase in word vocalizations (PND = 100). In Tincani et al. (2006), Participant 1 demonstrated a mild increase in word approximations with PECS (PND = 62.5), Participant 2 demonstrated no speech, and Participant 3 demonstrated a substantial increase in word approximations (PND = 100) with PECS.

Finally, procedural fidelity data were reported for 7 of the 16 studies analyzed. Researchers used differing procedures to calculate procedural fidelity, reflecting varying levels of specificity, and most did not report quantitative fidelity data (e.g., as a percentage of implementation steps accurately completed). For example, Bock, Stoner, Beck, Hanley, and Prochnow (2005) reported that procedural fidelity was documented on “a form that was created for each phase of . . . PECS” (p. 269) without additional detail, whereas Marckel, Neef, and Ferreri (2006) reported that procedural fidelity was scored with a 13-step checklist. In contrast, Tincani (2004) explicitly stated that the procedural fidelity checklist used in this study followed the procedures outlined in the phases of PECS, providing an example of the Phase I checklist within the report and reporting procedural fidelity as an average percentage of steps accurately completed (97.1).

Discussion

The findings of this meta-analysis support the PECS as an effective intervention to promote functional communication for individuals with ASD and other disabilities. Regardless of participants’ diagnosis, gender, age, highest phase of PECS mastered, and setting, PECS training, on average, produced moderate improvements in communication, as measured by increased picture exchange during intervention in comparison to baseline (PND = 80.1). This result is consistent with previous quantitative reviews of augmentative communication systems, which demonstrated enhanced communication for individuals with autism, mental retardation, and other disabilities (Millar, Light, & Schlosser, 2006; Schlosser & Wendt, 2008).

The findings of this meta-analysis support group comparison studies for which the results demonstrated that Phases I through III of the PECS enhanced children’s communicative initiations with others (Carr & Felce, 2007a, 2007b; Yoder & Stone, 2006). For the most part, the reviewed studies did not assess or report acquisition data on PECS Phases IV through VI. More data are needed to evaluate (a) whether users readily acquire the skills taught in these latter phases and (b) what, if any, enhancements to individuals’ communicative skills the phases provide. Therefore, although these single-participant data are supportive of PECS as a basic augmentative communication system to establish mands, the utility of PECS in promoting more advanced communication skills, including tacts (i.e., expressive labels) and intraverbals (i.e., conversation), has not been established.

Although PECS was initially designed to promote communication in young children with autism, it is noteworthy that 20 of the 41 participants for whom picture exchange was a dependent variable were not reported to have a diagnosis of autism or PDD-NOS. This finding tentatively suggests that PECS is effective for individuals with diagnoses beyond the autism spectrum; however, it should be noted that specific diagnostic information was not reported for seven individuals, who were described as having nonspecific “developmental delays” (Bock et al., 2005; Stoner et al., 2006). Thus, it is possible that individuals described as having “developmental delays” met the criteria for an ASD diagnosis though this was not specifically reported.

Participants demonstrated higher levels of picture exchange when PECS was taught in single setting (PND = 81.8) versus multiple (PND = 71.5) settings. Although these differences were not statistically significant, this finding is not surprising given difficulties with stimulus overselectivity and stimulus generalization intrinsic to persons with ASD (Chiang & Carter, 2008). As PECS is purported to be a functional communication system with utility across environments, this finding underscores Frost and Bondy’s (2002) emphasis within the PECS protocol on the importance of teaching across multiple trainers, environments, and reinforcers to promote generalized communication. As few studies in this review examined users’ generalized acquisition of picture exchange and vocalizations across settings and communicative partners, more research is needed to establish the efficacy of PECS in promoting generalized communication skills.

There was considerable variability among participants in the PECS phases mastered. Only one study reported that participants acquired all six phases (Charlop-Christy et al., 2002). In contrast, the majority of participants for whom picture exchange was graphically measured, 29 or 41, mastered only up to Phases I, II, or III. Three factors could account for participants’ limited acquisition of PECS. First, variability in participants’ acquisition rates could preclude experimenters from progressing through all the system’s phases by the conclusion of the experiment. For example, participants in the Charlop-Christy et al. (2002) study were described as mastering all six PECS phases within an average of just 246 total trials. By comparison, one of the participants in the study by Tincani et al. (2006) required an average of 358 trials to master only Phases I and II. Second, Stoner et al. (2006), who taught PECS to Phase IV with four of their participants, reported that Phases V and VI could not be completed because of limits in implementers’ training schedules. Thus, experimental time limitations could have prevented participants in some studies from progressing through the latter phases of PECS. Third, although participants’ diagnoses of autism or PDD-NOS versus another disability did not appear to influence acquisition of picture exchange, participant’s cognitive abilities could have influenced acquisition rates. Unfortunately, because most studies lacked quantitative and complete descriptions of participants’ cognitive functioning levels, it was not possible to assess for the influence of IQ score, for example, as an independent variable on acquisition of PECS. Nonetheless, empirical support for Phases V and VI of the system is lacking.

PECS training produced varying levels of speech among the limited number of participants for whom speech was documented. Six participants evidenced little or no improvements in speech, whereas another ten evidenced mild, moderate, or substantial improvements in speech. This finding is consistent with Schlosser and Wendt (2008), who found varying, yet generally modest, improvements in speech with augmentative communication systems, including PECS. Group studies have yielded similar varying findings with respect to speech (Carr & Felce, 2007a, 2007b; Howlin et al., 2007). One possible source of variability in speech acquisition could be differing exposure to PECS among participants in terms of duration and intensity of intervention. Importantly, because the primary focus of PECS is to teach functional communication and not speech per se, the absence of speech development for some users should not be viewed as a deficit of the system. Moreover, there is no evidence within the reviewed studies to suggest that PECS inhibited speech; to the contrary, if any effect was observed, it was facilitative rather than inhibitory.

Limitations

There are several potential limitations to the investigation that should be considered in relation to the results. The first limitation is the relatively small number of participants, 44, for whom picture exchange and/or speech data were reported. Given the sample size, caution should be exercised in generalizing results to the general population of individuals with ASD and other disabilities. Particular caution should be used in generalizing the speech data given the small number of participants, 12, for whom speech data were documented.

Second, the PECS protocol is a complex system requiring myriad teaching procedures, including most-to-least prompting, least-to-most prompting, shaping, chaining, and error correction (see Frost & Bondy, 2002). Indeed, Howlin et al. (2007) found that teachers had difficulty maintaining communicative gains achieved with PECS when classroom consultation visits ended, perhaps in part because expert consultation did not fully establish complex teaching repertoires necessary to implement PECS without external support. Although 7 of 16 studies in this review reported procedural fidelity suggesting conformity with the PECS procedures described by Frost and Bondy (2002), it is not unreasonable to assume some degree of procedural variation from the PECS protocol in these studies given the system’s complexity. Several studies, including those by Lund and Troha (2008), Marckel et al. (2006), Sigafoos, Ganz, O’Reilly, Lancioni, and Schlosser (2007), and Tincani (2004), involved procedural variations not explicitly delineated within the PECS protocol. Therefore, caution should be exercised in generalizing results from these studies to field-based implementations of PECS. For practitioners and parents, this finding indicates that, in some cases, the PECS system may need to be modified from Frost and Bondy’s (2002) protocol to accommodate the needs of individual learners, for example, those with visual impairments, fine and gross motor difficulties, and problems with discriminating between picture symbols.

A third limitation involves the vocalization data collected for 12 participants. Although some researchers reported increases in vocalizations with implementation of PECS, there was considerable variation in how vocalizations were measured across participants. For example, Charlop-Christy et al. (2002) measured spontaneous speech and imitative speech in play and academic demand settings, whereas Tincani (2004) and Tincani et al. (2006) measured words and word approximations emitted during PECS training. Furthermore, in one study (Kravits, Kamps, Kemmerer, & Potucek, 2002), vocalizations were measured and were reported to have increased, but vocalization data were not graphed separately from picture exchange data. Differing measurement and data displays across studies limits conclusions that can be drawn about the efficacy of PECS in promoting speech.

Fourth, use of PND as a metric to evaluate effect size across studies has potential limitations that should be considered. Specifically, this metric evaluates the degree of nonoverlap between baseline and intervention phases but does not consider other relevant indicators of effect size, including latency and magnitude of behavior change across phases. Although in most cases the reviewed studies yielded PNDs of 70 to 100 for picture exchange, participants’ performance levels were typically at low or zero levels prior to intervention. Therefore, marginal and clinically insignificant performance improvement could have produced high PND scores. However, it is important to consider that PECS requires a criterion of at least 80% independence for progression to the next phase. Therefore, participants who exhibited marginal levels of improvement with PECS would not progress beyond Phase I of the system. Because a majority of participants in the review acquired at least Phases I, II, and III, this suggests a higher and clinically significant magnitude of behavior change, though PND does not necessarily reflect this magnitude of change.

Finally, although the current review considered effect size for PECS intervention in conjunction with phases acquired, it did not systematically evaluate for quality indicators of single-participant research (Horner et al., 2005) beyond inclusion of procedural fidelity measures. Quality indicators include complete descriptions of participants and settings, at least three demonstrations of experimental control, and social validation of dependent variables. Given the importance of quality indicators to establishing the efficacy of a particular intervention, future reviews should carefully and systematically consider these factors.

Future Research

These results yield several directions for future research. First, given that the majority of participants taught PECS mastered only up to Phase III, additional research is needed to establish efficacy of the latter phases, particularly Phases V and VI, for which no data were graphically reported. Second, additional research is needed to confirm the effectiveness of PECS in promoting speech. Given the observed variability among studies in the manner in which speech was defined and measured, future researchers should consider replicating the operational definitions and settings used in the current studies (Charlop-Christy et al., 2002; Tincani, 2004; Tincani et al., 2006) in systematic replications. Third, additional research on PECS as taught across settings is needed. Research addressing this particular issue is especially critical given that PECS is purported to be a functional communication system with utility across settings. Furthermore, in several studies it was apparent that researchers, rather than peers, parents, or teachers, were both the trainers and recipients of communication. Therefore, future studies should investigate the effects of PECS as taught in the context of natural behavior change agents to determine the feasibility of the system for secondary consumers. In addition, researchers should carefully document critical outcome variables related to PECS that were not described in most of the reviewed studies, specifically, the number of trials or sessions required to acquire each of the six phases and the total number of pictures or words acquired. Documentation of these important variables would contribute to an understanding of the utility of PECS as a functional augmentative communication system. Finally, future studies should employ procedural fidelity measures that demonstrate direct correspondence with the procedures outlined by Frost and Bondy (2002) to validate the PECS protocol.

Footnotes

The authors declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

The authors received no financial support for the research and/or authorship of this article.

References

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text revision). Washington, DC: Author.

*Anderson

Moore

D. W.

Bourne

(2007). Functional communication and other concomitant behavior change following PECS training: A case study. Behaviour Change, 24, 173–181.

Angermeier

Schlosser

Luiselli

Harrington

Carter

(2008). Effects of iconicity on requesting with the Picture Exchange Communication System in children with autism spectrum disorder. Research in Autism Spectrum Disorders, 2, 430–446.

*Bock

Stoner

Beck

Hanley

Prochnow

(2005). Increasing functional communication in non-speaking preschool children: Comparison of PECS and VOCA. Education and Training in Developmental Disabilities, 40, 264–278.

Bondy

Frost

(1994). The Picture Exchange Communication System. Focus on Autistic Behavior, 9, 1–19.

Bondy

Tincani

Frost

(2004). Multiply controlled verbal operants: An analysis and extension to the Picture Exchange Communication System. Behavior Analyst, 27, 247–261.

Buckley

Newchok

(2005). Differential impact of response effort within a response chain on use of mands in a student with autism. Research in Developmental Disabilities, 26, 77–85.

Carr

Felce

(2007a). Brief report: Increase in production of spoken words in some children with autism after PECS teaching to Phase III. Journal of Autism and Developmental Disorders, 37, 780–787.

Carr

Felce

(2007b). The effects of PECS teaching to Phase III on the communicative interactions between children with autism and their teachers. Journal of Autism and Developmental Disorders, 37, 724–737.

10.

*Carré

A. J.

Le Grice

Blampied

N. M.

Walker

(2009). Picture Exchange Communication (PECS) training for young children: Does training transfer at school and to home? Behaviour Change, 26, 54–65.

11.

*Chambers

Rehfeldt

(2003). Assessing the acquisition and generalization of two mand forms with adults with severe developmental disabilities. Research in Developmental Disabilities, 24, 265–280.

12.

*Charlop-Christy

Carpenter

LeBlanc

Kellet

(2002). Using the Picture Exchange Communication System (PECS) with children with autism: Assessment of PECS acquisition speech, social-communicative behavior, and problem behavior. Journal of Applied Behavior Analysis, 35, 213–231.

13.

Chiang

Carter

(2008). Spontaneity of communication in individuals with autism. Journal of Autism and Developmental Disorders, 38, 693–705.

14.

Frost

Bondy

(2002). The Picture Exchange Communication System. Newark, DE: Pyramid Educational Products.

15.

Ganz

J. B.

Simpson

(2004). Effects on communicative requesting and speech development of the Picture Exchange Communication System in children with characteristics of autism. Journal of Autism and Developmental Disorders, 34, 395–409.

16.

*Ganz

J. B.

Simpson

R. L.

Corbin-Newsome

(2008). The impact of the Picture Exchange Communication System on requesting and speech development in preschoolers with autism spectrum disorders and similar characteristics. Research in Autism Spectrum Disorders, 2, 157–169.

17.

Horner

R. H.

Carr

E. G.

Halle

McGee

Odom

Wolery

(2005). The use of single-subject research to identify evidence-based practice in special education. Exceptional Children, 71, 165–179.

18.

Howlin

Gordon

Pasco

Wade

Charman

(2007). The effectiveness of Picture Exchange Communication System (PECS) training for teachers of children with autism: A pragmatic, group randomized controlled trial. Journal of Child Psychology and Psychiatry, 48, 473–481.

19.

Magiati

Howlin

(2003). A pilot evaluation study of the Picture Exchange Communication System (PECS) for children with autistic spectrum disorders. Autism, 7, 297–320.

20.

*Jurgens

Anderson

Moore

D. W.

(2009). The effect of teaching PECS to a child with autism on verbal behaviour, play, and social functioning. Behaviour Change, 26, 66–81.

21.

Kazdin

A. E.

(1982). Single-case research designs: Methods for clinical and applied settings. New York: Oxford University Press.

22.

*Kravits

Kamps

Kemmerer

Potucek

(2002). Brief report: Increasing communication skills for an elementary-aged student with autism using the Picture Exchange Communication System. Journal of Autism and Development Disorders, 32, 225–230.

23.

*Lund

Troha

(2008). Teaching young people who are blind and have autism to make requests using a variation on the Picture Exchange Communication System with tactile symbols: A preliminary investigation. Journal of Autism and Developmental Disorders, 38, 719–730.

24.

(2006). An alternative method for quantitative synthesis of single-subject researches: Percentage of data points exceeding the median. Behavior Modification, 30, 598–617.

25.

*Marckel

Neef

Ferreri

(2006). A preliminary analysis of teaching improvisation with the Picture Exchange Communication System to children with autism. Journal of Applied Behavior Analysis, 39, 109–115.

26.

Millar

D. C.

Light

J. C.

Schlosser

R. W.

(2006). The impact of augmentative and alternative communication intervention on the speech production of individuals with developmental disabilities: A research review. Journal of Speech, Language, and Hearing Research, 49, 248–264.

27.

Miranda-Linné

F. M.

Melin

(1997). A comparison of speaking and mute individuals with autism and autistic-like conditions on the Autism Behavior Checklist. Journal of Autism and Developmental Disorders, 27, 245–264.

28.

Mirenda

(2003). Toward functional augmentative and alternative communication for students with autism: Manual signs, graphic symbols, and voice output communication aids. Language, Speech, and Hearing Services in Schools, 34, 202–215.

29.

Olive

Smith

(2005). Effect size calculations and single subject designs. Educational Psychology, 25, 313–324.

30.

Ostryn

Wolfe

Rusch

(2008). A review and analysis of the Picture Exchange Communication System (PECS) for individuals with autism spectrum disorders using a paradigm of communication competence. Research and Practice for Persons With Severe Disabilities, 33, 13–24.

31.

Schlosser

Wendt

(2008). Effects of augmentative and alternative communication intervention on speech production in children with autism: A systematic review. American Journal of Speech-Language Pathology, 17, 212–230.

32.

Schwartz

Garfinkle

Bauer

(1998). The Picture Exchange Communication System: Communicative outcomes for young children with disabilities. Topics in Early Childhood Special Education, 18, 144–159.

33.

Scruggs

T. E.

Mastropieri

M. A.

(1998). Summarizing single-subject research: Issues and applications. Behavior Modification, 22, 221–242.

34.

Scruggs

T. E.

Mastropieri

M. A.

Casto

(1987). The quantitative synthesis of single-subject research: Methodology and validation. Remedial and Special Education, 8, 24–33.

35.

*Sigafoos

Ganz

O’Reilly

Lancioni

Schlosser

(2007). Assessing correspondence following acquisition of an exchange-based communication system. Research in Developmental Disabilities, 28(1), 71–83.

36.

Simpson

de Boer-Ott

Griswold

Myles

Byrd

Ganz

. (2005). Autism spectrum disorders: Interventions and treatments for children and youth. Thousand Oaks, CA: Corwin Press.

37.

Stahmer

Collings

Palinkas

(2005). Early intervention practices for children with autism: Descriptions from community providers. Focus on Autism and Other Developmental Disabilities, 20, 66–79.

38.

*Stoner

Beck

Bock

Hickey

Kosuwan

Thompson

(2006). The effectiveness of the Picture Exchange Communication System with nonspeaking adults. Remedial and Special Education, 27, 154–165.

39.

Sulzer-Azaroff

Hoffman

Horton

Bondy

Frost

(2009). The Picture Exchange Communication System (PECS): What do the data say? Focus on Autism and Other Developmental Disabilities, 24, 89–103.

40.

Tien

(2008). Effectiveness of the Picture Exchange Communication System as a functional communication intervention for individuals with autism spectrum disorders: A practice-based research synthesis. Education and Training in Developmental Disabilities, 43, 61–76.

41.

*Tincani

(2004). Comparing the Picture Exchange Communication System and sign language training for children with autism. Focus on Autism and Other Developmental Disabilities, 19, 152–163.

42.

*Tincani

Crozier

Alazetta

(2006). The Picture Exchange Communication System: Effects on manding and speech development for school-aged children with autism. Education and Training in Developmental Disabilities, 41,177–184.

43.

Wetherby

A. M.

Prizant

B. M.

(2005). Enhancing language and communication development in autism spectrum disorders: Assessment and intervention guidelines. In Zager

(Ed.), Autism spectrum disorders: Identification, education, and treatment (3rd ed., pp. 327–365). Mahwah, NJ: Lawrence Erlbaum.

44.

Yoder

Stone

(2006). A randomized comparison of the effect of two prelinguistic communication interventions on the acquisition of spoken communication in preschoolers with ASD. Journal of Speech, Language, and Hearing Research, 49, 698–711.

45.

*Yokoyama

Naoi

Yamamoto

(2006). Teaching verbal behavior using the Picture Exchange Communication System (PECS) with children with autistic spectrum disorders. Japanese Journal of Special Education, 43, 485–503.

46.

*Ziomek

M. M.

Rehfeldt

R. A.

(2008). Investigating the acquisition, generalization, and emergence of untrained verbal operants for mands acquired using the Picture Exchange Communication System in adults with severe developmental disabilities. Analysis of Verbal Behavior, 24, 15–30.