Social StoriesÔ for Children with Disabilities Georgina Reynhout 1,2 and Mark Carter 1 A review of the empirical research literature on Social StoriesÔ is presented, including a descriptive review and single-subject meta-analysis of appropriate studies. Examination of data suggests the effects of Social StoriesÔ are highly variable. Interpretations of extant studies are frequently confounded by inadequate participant description and the use of Social StoriesÔ in combination with other interventions. It is unclear whether particular components of Social StoriesÔ are central to their efficacy. Data on maintenance and generalization are also limited. Social StoriesÔ stand as a promising intervention, being relatively straightfor- ward and efficient to implement with application to a wide range of behaviors. Further research is needed to determine the exact nature of their contribution and the components critical to their efficacy. KEY WORDS: Social StoriesÔ; autism; Asperger’s syndrome; behavior modification; social skills; special education. INTRODUCTION In describing Social StoriesÔ, Gray (2003) has stated that: A social story is a process that results in a product for a person with autism spectrum disorder (ASD). First, as a process, a social story requires consideration of —and respect for—the person with ASD. As a product, a social story is a short story—defined by specific characteristics—that describes a situation, concept, or social skill using a format that is meaningful for people with ASD. The result is often renewed sensitivity of others to the experience of the person with ASD, and an improvement in the response of the person with ASD (p. 1). In the Social StoryÔ’s original form, three types of short, direct sentences, descriptive, directive, and perspective, were used. Descriptive sentences describe the social situation in terms of relevant social cues; directive sentences specify an appropriate behavioral response; and perspective sentences describe the feelings and responses of the student or others in the targeted situation (Gray, 2000b). Gray (2003) has suggested that perspective sentences should only be rarely used to describe the internal states of persons with autism. Originally the use of illustrations was not gen- erally recommended as ‘‘...they may be distracting, or a student may make an inaccurate interpretation of the situation based on the illustration’’ (Gray & Garand, 1993, p. 4). This recommendation was subsequently revised and the use of illustrations ‘‘...that reflect consideration of the age and personal learning characteristics of the person with ASD’’ (Gray, 2003, p. 5) is now suggested as beneficial to social understanding. The use of pictures may be viewed as consistent with other research on visual supports for children with autism (e.g., Dettmer, Simpson, Myles, & Ganz, 2000). 1 Macquarie University Special Education Centre, Macquarie University, North Ryde, NSW, Australia. 2 Correspondence should be addressed to: Georgina Reynhout, Macquarie University Special Education Centre, Macquarie University, 2109, North Ryde, NSW, Australia.; e-mail: georgina. [email protected]445 0162-3257/06/0500-0445/0 Ó 2006 Springer ScienceþBusiness Media, Inc. Journal of Autism and Developmental Disorders, Vol. 36, No. 4, May 2006 (Ó 2006) DOI 10.1007/s10803-006-0086-1 Published Online: April 26, 2006
26
Embed
Social Stories for Children with Disabilities - Autism Community
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Social Stories� for Children with Disabilities
Georgina Reynhout1,2
and Mark Carter1
A review of the empirical research literature on Social Stories� is presented, including adescriptive review and single-subject meta-analysis of appropriate studies. Examination ofdata suggests the effects of Social Stories� are highly variable. Interpretations of extant
studies are frequently confounded by inadequate participant description and the use of SocialStories� in combination with other interventions. It is unclear whether particular componentsof Social Stories� are central to their efficacy. Data on maintenance and generalization are
also limited. Social Stories� stand as a promising intervention, being relatively straightfor-ward and efficient to implement with application to a wide range of behaviors. Furtherresearch is needed to determine the exact nature of their contribution and the components
critical to their efficacy.
KEY WORDS: Social Stories�; autism; Asperger’s syndrome; behavior modification; social skills; special
education.
INTRODUCTION
In describing Social Stories�, Gray (2003) hasstated that:
A social story is a process that results in a product for
a person with autism spectrum disorder (ASD). First,
as a process, a social story requires consideration of
—and respect for—the person with ASD. As a
product, a social story is a short story—defined by
specific characteristics—that describes a situation,
concept, or social skill using a format that is
meaningful for people with ASD. The result is often
renewed sensitivity of others to the experience of the
person with ASD, and an improvement in the
response of the person with ASD (p. 1).
In the Social Story�’s original form, three typesof short, direct sentences, descriptive, directive, and
perspective, were used. Descriptive sentences describethe social situation in terms of relevant social cues;directive sentences specify an appropriate behavioralresponse; and perspective sentences describe thefeelings and responses of the student or others inthe targeted situation (Gray, 2000b). Gray (2003) hassuggested that perspective sentences should only berarely used to describe the internal states of personswith autism.
Originally the use of illustrations was not gen-erally recommended as ‘‘...they may be distracting, ora student may make an inaccurate interpretation ofthe situation based on the illustration’’ (Gray &Garand, 1993, p. 4). This recommendation wassubsequently revised and the use of illustrations‘‘...that reflect consideration of the age and personallearning characteristics of the person with ASD’’(Gray, 2003, p. 5) is now suggested as beneficial tosocial understanding. The use of pictures may beviewed as consistent with other research on visualsupports for children with autism (e.g., Dettmer,Simpson, Myles, & Ganz, 2000).
1 Macquarie University Special Education Centre, Macquarie
University, North Ryde, NSW, Australia.2 Correspondence should be addressed to: Georgina Reynhout,
Macquarie University Special Education Centre, Macquarie
University, 2109, North Ryde, NSW, Australia.; e-mail: georgina.
4450162-3257/06/0500-0445/0 � 2006 Springer ScienceþBusiness Media, Inc.
Journal of Autism and Developmental Disorders, Vol. 36, No. 4, May 2006 (� 2006)
DOI 10.1007/s10803-006-0086-1
Published Online: April 26, 2006
Over time, the suggested format for a SocialStory� has become more sophisticated. Gray (2003)now describes four basic sentence types: descriptive,directive, perspective and affirmative. ‘‘Affirmativesentences enhance the meaning of surrounding state-ments, often expressing a commonly shared value oropinion within a given culture’’ (Gray, 2003, p. 3).These four basic sentence types and a ratio thatdefines their frequency are the most importantcomponents of the Social Story�. A basic SocialStory� has a ratio of two to five descriptive,perspective and/or affirmative sentences for everydirective sentence.
Two other types of sentences can be included:control and cooperative sentences. ‘‘Control sentencesare statements written by a person with ASD to identifypersonal strategies to use to recall and apply personalinformation’’ and cooperative sentences are ‘‘...used toidentifywhat otherswill do to assist the student’’ (Gray,2003, p. 3). A ratio of zero to one directive or controlsentences to between two and five descriptive and/orperspective sentences forms the complete Social Story�ratio. Within a Social Story�, descriptive, perspective,control and directive sentences can be partial orcomplete. ‘‘Partial sentences encourage the studentwith ASD to make guesses regarding the next step in asituation, the response of another individual, or his ownresponse’’ (Gray, 2003, p. 3).
There are several considerations to be made whenwriting a Social Story� according to Gray’s format(2000a, 2000b, 2003). The author must adopt andmaintain the perspective of the child for whom thestory is written. The story should be well within thestudent’s comprehension level and clearly presentedusing vocabulary and print size that are appropriatefor the student. Behavioral responses should be statedin positive terms, for example using ‘‘I will use myquiet voice’’ instead of ‘‘I will not shout.’’ It issuggested that Social Stories� be used with students‘‘...functioning intellectually in the trainable mentallyimpaired range or higher who possess basic languageskills’’ (Gray & Garand, 1993, pp. 2–3).
Gray and Garand (1993) describe how SocialStories� can be modified to suit a wide variety ofpurposes. For example, checklist or sequence SocialStories� can teach and provide practice in thefollowing of routines, and curriculum stories can becreated by inserting pages involving academic activ-ities, such as math, into the Social Story�. Skillsrelated to specific curriculum objectives can then bepracticed in the context of real life situations relevantto the student. Generic Social Stories� can be used to
describe social situations frequently experienced bychildren with autism which can be individualized andadapted as the need arises (Gray & Garand, 1993).
A Social Story� intervention can be imple-mented in one of three ways and the approach isdependent upon the particular needs and abilities ofthe child (Gray & Garand, 1993). If the child is ableto read, the teacher can initially read the SocialStory� with the child, and the child thereafter readsthe story independently. If the child is unable to read,the story can be recorded on cassette and the childtaught to use the tape recorder, turning the pagewhen prompted by an auditory signal. In this way thenon-reader can ‘‘read’’ the story independently. Thethird method involves video modeling. The story isrecorded onto videotape and sequences depicting thesocial situation and desired behavioral responses arematched to the story. Peers can read the SocialStory� to the child and when targeted situationsarise, they are able to prompt the student toremember the Social Story�, thus reinforcing itsuse. Following the introduction of the story thestudent’s comprehension of the story is checked. Thisis a mandatory part of the process involving theteacher asking the student to respond verbally or inwriting to verbal or written questions, or to fill in achecklist (Gray & Garand, 1993). Fading a SocialStory� can be achieved for example, by extendingthe time periods between readings, or by rewriting theSocial Story�, omitting or revising selected sen-tences. Fading is highly individualized and again,dependent on the needs and abilities of the child(Gray & Garand, 1993).
Several possible explanations have been offeredfor the purported success of Social Story� interven-tion for individuals with autism. One explanationinvolves the concept of ‘‘theory of mind’’ (ToM).Garfield, Peterson, &Perry (2001, pp. 495) define ToMas describing ‘‘...whatever knowledge guides preposi-tional attitude attribution and the explanation andprediction of behaviour by means of inner states andprocesses.’’ Individuals with autism, lacking ToM, areunable to appreciate other people’s intentions, beliefs,needs and desires (Greenway, 2000). Research showsboth linguistic (Happe, 1995) and social skills (Ozonoff& Miller, 1995) support the development of ToM.Individuals with autism, demonstrating significantimpairments in both linguistic and social skillsdomains, consequently show ToM deficits (Garfieldet al., 2001). The perspective sentences in SocialStories� are seen to address this social-cognitivedeficit.
446 Reynhout and Carter
Another explanation involves the use of sharedschemata or background knowledge. Rowe (1999)suggests that a Social Story� builds a scaffold ofunderstanding for a schema (mental representation)that an individual does not yet possess. Myles andSimpson (2001) describe Social Stories� as providingaccess to a social skill’s ‘‘hidden curriculum.’’ Thiscurriculum is based on the dos and don’ts innatelyunderstood and adhered to by everyone, exceptindividuals with Asperger’s syndrome.
Social Stories� may also be seen to includeaspects of purported good practice in ASD, beingvisual, permanent, written in simple language, basedon careful assessment of the child, focused on keyareas (for example, social interaction), and factual,focusing on the perceptions of others (Smith, 2001). Ina survey of school staff, parents and caretakers trainedin the Social Story� approach, the majority reportedit to be enjoyable, practical and effective (Smith, 2001).
Kuoch and Mirenda (2003) offered a limitedreview of research of 10 Social Story� interventionsstudies prior to 2002 (including some case studies).Several limitations in the research were noted includ-ing non-conformity of many of the stories withrecommended guidelines, flawed and weak researchdesigns and confounding effects as additional inter-ventions were often employed with Social Stories�.However, Kuoch and Mirenda (2003) included somestudies that did not present learner outcome data, didnot attempt to quantify outcome strength, examinegeneralization or maintenance and did not attempt toanalyze whether intervention features were associatedwith differential outcomes. Thus, at this point, we donot have a comprehensive and detailed review ofempirical studies on Social Story� interventions.
Social Stories� have inherent attractions inbeing relatively undemanding to implement and arereported to be applicable to a wide variety ofbehavior. The use of Social Stories� has beenpopularized, widely discussed and recommended inthe literature (e.g., Backman & Pilebro, 1999; Chap-man & Trowbridge, 2000; Rowe, 1999; Simpson &Myles, 1998). Unfortunately, clinical popularity ofinterventions has not always been a good indicator ofefficacy (see Dawson & Watling, 2000; Elder, 2002;Kasari, 2002; Kerrin, Murdock, Sharpton, & Jones,1998; McWilliam, 1999). While there has been anincreasing quantity of research examining SocialStories� in recent years, no systematic and compre-hensive review of empirical literature appears to havebeen conducted to date. The present paper addressesissues related to the nature and quality of studies
conducted, the participants, characteristics of theSocial Story� interventions with relation to suggestedguidelines (see Gray, 2003; Gray & Garand, 1993),behaviors that have been targeted for intervention, theshort-term efficacy of the technique, and generaliza-tion and maintenance. The paper includes a descrip-tive review of extant research as well as applyingappropriate statistical analyses to relevant studies,including calculation of conventional effect sizes forgroup designs and single-subject meta-analysis.
IDENTIFICATION OF STUDIES
The following on-line databases were searchedfor sources appearing before December 2003: ABI/INFORM Global; Academic Research Library; Cur-rent Contents Connect; ERIC; Expanded AcademicASAP; First Search; Ingenta; Inspec; Kluwer Online;Proquest Education Complete; PsyARTICLES; psy-cINFO; Science Direct and ISI Web of Science. Amanual search of all issues of Journal of Autism andDevelopmental Disorders was made for articles after1990 and the reference sections of all located sourceswere reviewed for additional sources that did notappear in the on-line searches.
Only studies related to Social Stories� wereincluded in the review. Studies involving the use ofsocial scripts, mutual storytelling and narrative ther-apy were excluded. Descriptive cases not includingdata (Del Valle, McEachern, & Chambers, 2001;Rowe, 1999), and studies not based on learnerperformance but teacher perception were excluded(Smith, 2001). Noting that the body of researchremains relatively small, peer-reviewed journal arti-cles and unpublished dissertations were included inthe review. Eleven peer-reviewed journal articles(Bledsoe, Myles, & Simpson, 2003; Brownell, 2002;Hagiwara & Myles, 1999; Kuoch & Mirenda, 2003;Kuttler, Myles, & Carlson, 1998; Lorimer, Simpson,Myles, & Ganz, 2002; Norris & Dattilo, 1999; Rogers& Myles, 2001; Scattone, Wilczynski, Edwards, &Rabian, 2002; Swaggart et al., 1995; Thiemann &Goldstein, 2001) and five dissertations (Cullain, 2002;Feinberg, 2002; Pettigrew, 1998; Romano, 2002;Staley, 2002) were identified.
PROCEDURES
A summary of each study was prepared address-ing participants, research design, dependent variable
Social Stories� 447
class, target behaviors, use of Social Story� strategy,use of additional strategies, reliability, short-termfindings, maintenance, and generalization, for eachstudy that provided a verbatim copy of SocialStories� used in intervention (totaling 31 storiesfrom 13 of the 16 studies). Each sentence wasindependently rated for sentence type (i.e., descrip-tive, perspective, affirmative, directive, control andcooperative) using the guidelines provided by Gray(2003). Sentences were coded in multiple categorieswhere this was considered appropriate. In the case ofperspective sentences taking the format of ‘‘I feelhappy when,’’ information provided in the secondpart of the sentence was considered to be supportingthe first and was not coded elsewhere in terms ofGray’s (2000b, 2003) classification. Where perspec-tive sentences were employed, they were coded as towhether they took the perspective of the targetindividual and/or that of others. One sentence typenot described by Gray (2000b, 2003) was also coded.Sentences were also coded as to whether theyspecified a consequence of the actions of the targetindividual or others (e.g., ‘‘When I share my toys, myfriends will want to play with me again’’).
Both authors independently rated all stories toallow estimation of interrater reliability of sentencecoding. An agreement was recorded when both ratersindicated the presence of a particular coding categoryfor a sentence and a disagreement was recorded whenonly one rater coded the category. Reliability wascalculated by dividing agreements by the total ofagreements and disagreements and multiplying by100. Interrater reliability was 86.7% for descriptivesentences. 88.2% for perspective sentences, 85.7% foraffirmative sentences, 93.5% for directive sentences,85.7% for control sentences, 85.7% for cooperativesentences and 82.5% for consequence sentences.Where raters agreed on the presence of perspectivesentence, agreements on whose perspective was takenwas 96.8%.
Data on types of sentences were summarized interms of the percentage of stories containing eachsentence type and the mean percentage of eachsentence type per story. In addition, the percentageof perspective sentences taking the perspective of thetarget individual, other individuals and both werecalculated.
For group designs, presentation of effect sizesalong with inferential statistics is recommended(American Psychological Association, 2001). Consis-tent with this recommendation, effect sizes werecalculated on relevant studies by subtracting the
post-test mean of the control group from the mean ofthe experimental group and dividing by the standarddeviation of the control group (McCartney & Ro-senthal, 2000). Effect sizes were expressed positivelywhen change occurred in the predicted direction andnegatively when changes were opposite to thosepredicted. When targeting an increase in behavior,an effect size of .20 is small, .50 moderate, and .80large; the usually accepted minimum clinically accept-able effect size for educational interventions is 0.33(McCartney & Rosenthal, 2000).
Single-subject studies are typically interpreted byvisual inspection of graphed data. Such visual inspec-tion of single subject studies can be subjective and theobjective aggregation of results is problematic. Cal-culation of the percentage of non-overlapping data(PND) has been suggested as an alternative (Scruggs,Mastropieri, & Casto, 1987). Using a graphical dataplot of results from baseline to treatment,the percentage of non-overlapping data (PND) isthe number of treatment data points that exceed thehighest (or lowest, if appropriate) baseline data point,divided by the total number of treatment data pointsand multiplied by 100 (Scruggs et al., 1987). Whilethe procedure has been subject to criticism on severalgrounds (e.g., Salzberg, Strain, & Baer, 1987; Strain,Kohler, & Gresham, 1998; White, 1987), it has gainedcurrency and has been applied across a range of areas(e.g., Didden, Duker, & Korzilius, 1997; Mathur,Kavale, Quinn, Forness, & Rutherford Jr., 1998;Scruggs, Mastropieri, Forness, & Kavale, 1988; Xin& Jitendra, 1999).
In the present analysis, a PND statistic wascalculated for each study using the pooled number ofnon-overlapping data points across all subjects and allconditions. In addition, a PND statistic was calcu-lated for each relevant story. A PND between 91 and100 indicates a highly effective intervention, between71 and 90 moderately effective, between 51 and 70mildly effective, and between 0 and 50, non-effective(Mastropieri, Scruggs, Bakken, & Whedon, 1996).
RESULTS
A summary of the participants, research designand duration of study, target behaviors and settings,use of Social Story� strategy and use of additionalstrategies was constructed and is presented in Table I.Table II provides descriptive information on reliabil-ity, short-term findings, study effect size or PND,maintenance, and generalization results.
A summary of the results of the analysis of thesentence types is presented in Table III.
Table IV provides a summary of aspects of storyconstruction and implementation in relation to PNDfor individual stories for which that metric could becalculated. All stories involved one participant exceptthose of Staley (2002), who used five participants fortwo Social Stories�, and Thiemann and Goldstein(2001), who used five participants for one story.
Social Story� use comprises Social Story�construction and implementation. Gray (2003)defines two Social Story� types, those adhering tothe basic Social Story� ratio, and those adhering tothe complete Social Story� ratio. Gray (2003) hasnoted that directive sentences may not be necessary insome instances so stories with a lower than recom-mended ratio of directive to other sentences wereconsidered to be appropriately modified. Based onanalysis of the Social Stories�, four categories ofSocial Story� were identified: (a) stories that adheredto the basic Social Story� ratio; (b) stories thatadhered to the complete Social Story� ratio; (c)appropriately modified Social Stories�, where thenumber of descriptive, perspective, affirmative or co-operative sentences is greater than that recommendedin the basic or complete ratios, but still acceptableaccording to Gray’s recommendations (see Gray,2003), and (d) inappropriately modified SocialStories�, where the number of directive or controlsentences is greater than that recommended in thebasic or complete ratios. The results of the analysisare presented in Table V.
Participants
Fifteen of the 16 studies involved a child orchildren with autism or Asperger’s syndrome; ofthese children, 60 were boys and 17 were girls, aged3–15 years. Pettigrew (1998) provided the only study
Table
II.Continued
Study
Interobserver
reliability
Proceduralreliability
Reported
short-term
re-
sults
Effectsize
(ES)orPND
Maintenance
and
generalization
Thiemann
and
Goldstein
(2001)
Calculated
for
30
%of
sessions.
Mean
agree-
ments
for
participants
92%,89%,92
%,93%
and
90%.
Procedural
reliability
measuredfor20
%ofses-
sionsat89%.
Increase
intargeted
social
skills
from
baselinewhen
treatm
entim
plemented.
Mean
PND
forDan,70,
Greg28,John
22,Casey
77,andIvan17.
StudyPND
45.
Maintenance:
Lim
ited
maintenance
insocial
skills
dem
onstrated
by
threeparticipants.
Generalization:one
par-
ticipant
dem
onstrated
generalization
of
new
lyacquired
socialskills
toonenew
setting.
Abbreviation:SCBE,SocialCompetence
andBehaviorEvaluation-PreschoolEdition.
*Dissertation.
Table III. Social Story� Sentence Analysis
Sentence type
Percentage of
stories containing
sentence type
Mean percentage of
sentence type per
story (range)
Descriptive 97 42 (0–91)
Perspective 87 23 (0–60)
Affirmative 22 3 (0–21)
Directive 100 28 (9–89)
Control 10 1 (0–14)
Cooperative 13 2 (0–29)
Consequence 90 24 (0–50)
Note: All percentages are rounded.
458 Reynhout and Carter
Table
IV.Story
ImplementationandConstructionVariablesandPND
Socialstory
�Behavior
Percentageofsentence
type:
SSratio
COMP
AdditionalStrategies
PND
DES
PER
AFF
DIR
CON
COOP
CONS
Bledsoeet
al.(2003)
Foodspill,napkin
use
25
23
025
00
50
Basic
No
None
16
Brownell(2002;Brian)
Follow
directions
50
17
033
00
33
Inappropriately
modified
No
None
88
Brownell(2002;Justin)
Usingquietvoice
50
33
017
00
33
Basic
No
None
90
Brownell(2002;Nathan)
Usingquietvoice
57
14
029
00
29
Inappropriately
modified
No
None
80
Brownell(2002;Peter)
Echolalia
29
43
029
00
29
Inappropriately
modified
No
None
90
Cullain
(2000;Sander)
Playinggames
38
44
019
00
25
Basic
No
None
80
Cullain
(2002;Beth)
Talkingwithfriends
13
60
027
00
13
Basic
No
None
0
Cullain
(2002;Derek)
Computeruse
14
43
021
029
29
Complete
No
None
26
Cullain
(2002;Douglas)
Differentfriends
60
00
40
00
0Basic
No
None
37
Cullain
(2002;Kevin)
Playinggames
29
47
012
14
14
24
Complete
No
None
53
Hagiwara
andMyles
(1999)
Handwashing
011
089
00
0Inappropriately
modified
No
Verbalprompts,
physicalprompts
38
Kuoch
andMirenda
(2003;Neil)
Playinggames
45
15
21
18
00
27
Basic
(appropriately
modified)
No
None
75
Kuoch
andMirenda
(2003,Andrew)
Sharing
60
20
10
10
00
30
Basic
(appropriately
modified)
No
None
66
Kuoch
andMirenda
(2003;Henry)
Eating
45
32
514
50
40
Complete
No
None
100
Kuttleret
al.(1998)
Behavioratlunch
38
13
050
00
25
Inappropriately
modified
No
Reinforcers
(stickers/prize
bag)
90
Kuttleret
al.(1998)
Workingforsticker
67
17
017
00
50
Basic
No
Reinforcers
(stickers/prize
bag)
100
Lorimer
etal.(2002)
Talking
53
13
020
13
07
Complete
No
None
43
Lorimer
etal.(2002)
Waiting
91
90
90
09
Basic
(appropriately
modified)
No
None
43
Scattoneet
al.
(2002;Howard)
Shouting
43
14
029
014
29
Complete
Yes
Verbalprompts
57
Scattoneet
al .
(2002;John)
Lookingatgirls
50
10
10
30
00
10
Basic
Yes
Verbalprompts
92
Scattoneet
al.
(2002;Kenny)
Sittingonchair
50
30
020
00
50
Basic
Yes
Verbalprompts
100
Staley(2002)
Chew
ing
65
419
12
04
27
Complete
Yes
Verbalprompts,
modeling,reinforcer
(edibles)
32
Staley(2002)
Napkin
use
78
011
11
00
11
Basic
(appropriately
modified)
Yes
Verbalprompts,
modeling,reinforcer
(edibles)
39
Swaggart
etal.
(1995)
Greeting
40
00
60
00
40
Inappropriately
modified
No
Verbalprompts,
physicalprompts
88
Swaggart
etal.
(1995)
Appropriate
behavior
50
17
033
00
33
Basic
No
Verbalprompts,
physicalprompts,
reinforcer
(edibles)
0
Note:DES,Descriptive;
PER,Perspective;
AFF,Affirm
ative;
DIR
,Directive;
CON,Control;COOP,Cooperative;
CONS,Consequence;SSratio,SocialStory
�ratio;COMP,Comprehension
assessed.Allpercentages
are
rounded.
Social Stories� 459
not involving children with autism or Asperger’ssyndrome. The criteria for inclusion in this study wasa language delay of at least 6 months and an age of3–4 years (Pettigrew, 1998). In most instances, only adiagnostic label of autism or autism spectrum disor-der was provided. Only three of the studies (Feinberg,2002; Hagiwara & Myles, 1999; Thiemann & Gold-stein, 2001) provided standardized data on theposition of the participants on the autism spectrum.The communication skills of participants in thestudies ranged from those who were non-verbal(using augmentative device, communication board,sign language or gestures), to those able to commu-nicate verbally (Table I). In seven studies (Bledsoeet al., 2003; Feinberg, 2002; Lorimer et al., 2002;Norris & Dattilo, 1999; Pettigrew, 1998; Scattoneet al., 2002; Thiemann & Goldstein, 2001), referencewas made to cognitive abilities of the participants;only four of these provided information based on theresults of standardized tests (Bledsoe et al., 2003;Feinberg, 2002; Scattone et al., 2002; Thiemann &Goldstein, 2001). Only five studies (Bledsoe et al.,2003; Brownell, 2002; Kuoch & Mirenda, 2003;Staley, 2002; Thiemann & Goldstein, 2001) provideddescription of the reading ability of the participants;four of these presented results of standardized tests(Bledsoe et al., 2003; Kuoch &Mirenda, 2003; Staley,2002; Thiemann & Goldstein, 2001). As the partic-ipant description provided in many studies consistedof little more than a diagnostic label, it was effectivelyimpossible to determine if any specific participantcharacteristics were associated with interventioneffectiveness.
Research Design
The experimental designs of the single-subjectstudies ranged in complexity from simple AB (Cul-lain, 2002; Norris & Dattilo, 1999; Swaggart et al.,1995) to withdrawal (Bledsoe et al., 2003; Brownell,2002; Kuoch & Mirenda, 2003; Kuttler et al., 1998;Lorimer et al., 2002), to multiple baseline acrossparticipants (Scattone et al., 2002; Staley, 2002),skills (Thiemann & Goldstein, 2001) or settings
(Hagiwara & Myles, 1999). Simple pre-test–post-testdata was reported in some instances (Rogers &Myles, 2001; Swaggart et al., 1995 for some partic-ipants). In a small number of studies group designswere employed including a pre-test–post-test control-group design (Feinberg, 2002; Romano, 2002) and apre-test–post-test control and comparison-groupdesign (Pettigrew, 1998). There was wide variationin study duration within and among the researchdesigns employed in the studies.
Interobserver and Procedural Reliability
Twelve of the 16 studies (Table II) providedmeasures of interobserver reliability. In general,reliability measures in the studies ranged from 85–100%, calculated for between 15 and 100% of sessions(20% being the conventional minimum standard). Onone occasion, in the study conducted by Norris andDattilo (1999), the interobserver reliability droppedto 64%, resulting in a revision by the authors of theirdefinitions of appropriate, inappropriate and absenceof social interactions, to reflect a greater level ofspecificity. Procedural reliability measures werereported for only four studies (Kuoch & Mirenda,2003; Norris & Dattilo, 1999; Scattone et al., 2002;Thiemann & Goldstein, 2001).
Target Behaviors and Settings
The studies addressed various behaviors. Sixstudies targeted disruptive or challenging behaviors(Cullain, 2002; Kuttler et al., 1998; Lorimer et al.,2002; Romano, 2002; Scattone et al., 2002; Swaggartet al., 1995), nine studies targeted social skills (Bled-soe et al., 2003; Feinberg, 2002; Kuoch & Mirenda,2003; Norris & Dattilo, 1999; Pettigrew, 1998;Romano, 2002; Staley, 2002; Swaggart et al., 1995;Thiemann & Goldstein, 2001), four studies targetedcommunicative behaviors (Cullain, 2002; Kuttleret al., 1998; Lorimer et al., 2002; Romano, 2002)and four studies targeted on-task behaviors (Brow-nell, 2002; Hagiwara & Myles, 1999; Kuoch &Mirenda, 2003; Rogers & Myles, 2001).
Twelve studies took place in a school setting(Bledsoe et al., 2003; Brownell, 2002; Hagiwara &Myles, 1999; Kuttler et al., 1998; Norris & Dattilo,1999; Pettigrew, 1998; Rogers & Myles, 2001;Romano, 2002; Scattone et al., 2002; Staley, 2002;Swaggart et al., 1995; Thiemann & Goldstein, 2001),one study took place in a home setting (Lorimeret al., 2002), two studies took place in a school andhome setting (Cullain, 2002; Kuoch & Mirenda,
Table V. Summary of Story Analysis
Category of Social Story� Number of stories
Basic 12
Complete 6
Appropriately modified 4
Inappropriately modified 9
460 Reynhout and Carter
2003), and one study took place in a ‘‘game room’’ atan unidentified location (Feinberg, 2002).
Effectiveness
The authors of nine studies (Bledsoe et al., 2003;Brownell, 2002; Cullain, 2002; Kuoch & Mirenda,2003; Kuttler et al., 1998; Lorimer et al., 2002; Norris& Dattilo, 1999; Rogers & Myles, 2001; Romano,2002; Scattone et al., 2002) reported an appropriatereduction in target behaviors and eight studies(Bledsoe et al., 2003; Feinberg, 2002; Hagiwara &Myles, 1999; Lorimer et al., 2002; Pettigrew, 1998;Rogers & Myles, 2001; Swaggart et al., 1995; Thie-mann & Goldstein, 2001) claimed an appropriateincrease in targeted behaviors from baseline tointervention, or from pre-test–post-test. Two studiesshowed no change in some behaviors when SocialStories� were used in isolation (Norris & Dattilo,1999; Staley, 2002) and two studies showed an actualincrease in disruptive behaviors post-intervention(Kuttler et al., 1998; Lorimer et al., 2002).
Effect sizes were calculated for three studies(Feinberg, 2002; Pettigrew, 1998; Romano, 2002).Based on total social skills scores for six trials(Feinberg, 2002) a mean effect size of .99 wascalculated, suggesting Social Story� interventionwas highly effective in increasing social skills in thisstudy. Romano (2002) failed to provide standarddeviations for the experimental and control groups atpost-test. The mean effect size at maintenance was1.59 (range .94 to 2.31) indicating that Social Story�intervention was highly effective in decreasing thetargeted behaviors for this group. These data shouldbe treated with extreme caution as the post-testresults indicated a massive experimental effect in thecontrol group, which was reversed at maintenance,raising serious questions about experimental controlin the study. Romano (2002) suggested that thismight have been a product of contamination of theintervention as control group students may haveoverheard the Social Stories�. Analysis of thePettigrew (1998) study (see Table II) revealed lowmean effect sizes for the experimental-control con-trasts of .34 and for the experimental-comparisoncontrasts of 0.26. In addition, these data also need tobe treated with caution, as there were clear discrep-ancies between tabular and graphic data. Staley(2002) has also pointed out other serious discrepan-cies in the data and analysis.
A total PND value was obtained for 12 studies(i.e., Bledsoe et al., 2003; Brownell, 2002; Cullain,
2002; Hagiwara & Myles, 1999; Kuoch & Mirenda,2003; Kuttler et al., 1998; Lorimer et al., 2002; Norris& Dattilo, 1999; Scattone et al., 2002; Staley, 2002;Swaggart et al., 1995; Thiemann & Goldstein, 2001).The calculated mean PND was 43 (range 16–95). Insome cases the data overlap may not provide anaccurate measure of treatment effectiveness, forexample, when baseline data shows an inappropriatetrend, or when ‘‘floor or ceiling’’ effects occur(Scruggs et al., 1987). Possible floor and ceilingeffects were certainly present in the data examinedin the present review (i.e., Bledsoe et al., 2003;Cullain, 2002; Kuoch & Mirenda, 2003; Lorimeret al., 2002; Norris & Dattilo, 1999; Staley, 2002;Swaggart et al., 1995).
To evaluate the possible effects of such effects afurther analysis was conducted. Data were eliminatedfor behaviors, subjects or phases that were expectedto decrease that contained a zero data point and,similarly, data were eliminated for behaviors thatwere expected to increase and a baseline ceiling wasevident in one data point (i.e., 100% of opportuni-ties). When these data were excluded and the PNDrecalculated, a total PND of 51(range 20–95) wasobtained. The mean PND values place the SocialStory� in the non-effective range or at very best, inthe low end of the mildly effective range (seeMastropieri et al., 1996). The most striking featureof the data, however, was the degree of inconsistency,including variation in responses to interventionacross participants or behaviors (see Table II).
Story Construction
Examples of Social Stories� used for interven-tion were not provided in three studies (i.e., Norris &Dattilo, 1999; Pettigrew, 1998; Romano, 2002).Examples of Social Stories� were provided for 13of the 16 studies (i.e., Bledsoe et al., 2003; Brownell,2002; Cullain, 2002; Feinberg, 2002; Hagiwara &Myles, 1999; Kuoch & Mirenda, 2003; Kuttler et al.,1998; Lorimer et al., 2002; Rogers & Myles, 2001;Scattone et al., 2002; Staley, 2002; Swaggart et al.,1995; Thiemann & Goldstein, 2001). Detailed exam-ination of these stories (see Table III) revealed thataffirmative, control, and cooperative sentences wereused infrequently while descriptive, perspective anddirective sentences were the most commonlyemployed. In contrast, with Gray’s (2003) recom-mendation that perspective sentences should onlyoccasionally be written from the viewpoint of theperson with autism, 47% of perspective sentences
Social Stories� 461
were written from this viewpoint, 47% were writtenfrom the viewpoint of others and 6% were writtenfrom both perspectives. Consequence sentences werenot described by Gray (2003) but were identified in90% of stories and contributed a mean of 24% (range0–50%) of sentences per story.
It was of interest to examine whether specificaspects of story construction were associated withPND. Before proceeding with presentation of thesedata, it is appropriate to note that the analysis shouldbe treated with caution as is based on a relativelysmall number of stories (26) on which PND can becalculated. Gray (2003) offers clear guidelines on howSocial Stories� should be constructed. Inspection ofTable V indicates that analysis of the Social Stories�reveals 10 (39%) of the stories deviate from recom-mended basic or complete Social Story� ratios. Oneobvious question was whether deviation from Gray’sguidelines (Gray, 2003) would differentially affectoutcomes. From the data in Table IV, which islimited to stories where PND was available, it can bedetermined that the 14 stories with basic or appro-priately modified basic ratios yielded a mean PND of55.6. The six stories with complete story ratiosyielded a mean PND of 51.8 and the six inappropri-ately modified stories, with a combination of lessdescriptive and more directive sentences, yielded amean PND of 79.0.
In order to determine whether particular sen-tence types corresponded with differential outcomes,initially data in Table IV were ranked according tothe frequency of each type of main sentence. Storieswere then divided into approximate thirds and aPND statistic was calculated for the nine stories withthe highest frequency of each sentence type, the eightstories that were middle ranked and the remainingnine sentences with the lowest rank. There was noevidence of an effect for perspective sentences with aPND of 64 for the highest ranked stories, 60 formiddle ranked stories and 56 for the lowest rankedstories. The pattern was not consistent for directivesentences with a PND of 67 for the highest rankedstories, 45 for middle ranked stories and 66 for thelowest ranked stories. There was also inconsistency inthe pattern for descriptive sentences with a PND of59 for the highest ranked stories, 75 for middleranked stories and 48 for the lowest ranked stories.There was some evidence of a pattern in PND forconsequence sentences with a PND of 70 for thehighest ranked stories, 65 for middle ranked storiesand 47 for the lowest ranked stories. It was alsointeresting to note that the actual percentage of
consequences sentences was very low for the bottomthird of stories (11.0% vs. 26.2% for middle rankedstories and 41.0% of highest ranked stories).
Story Delivery
In 15 of the studies, implementation of the SocialStory� was traditional, the Social Story� being readto or by the participant(s) (Table I). In one study(Hagiwara & Myles, 1999) the Social Story� waspresented on a computer; in another study (Brownell,2002) the Social Story� was presented in bothmusical and traditional formats. In 11 studies theauthors used visual symbols (Hagiwara & Myles,1999; Kuoch & Mirenda, 2003; Kuttler et al., 1998;Lorimer et al., 2002; Norris & Dattilo, 1999; Petti-grew, 1998; Rogers & Myles, 2001; Romano, 2002;Swaggart et al., 1995; Thiemann & Goldstein, 2001)or photos (Bledsoe et al., 2003) to enhance theinformation provided in the Social Story� text.
There was variation in the number of SocialStories� used in some studies. Norris and Dattilo(1999), for example, randomly used three differentstories with one participant, to provide interest,targeting social interaction only. Romano (2002)used three different stories, each one targeting differ-ent behaviors (those associated with communication,aggression and socialization). Other variations inimplementation procedures were also evident. Thefirst variation involved the inclusion of some form ofcomprehension exercise, undertaken by the studentfollowing the reading of the Social Story�, asrecommended by Gray and Garand (1993). In fivestories from two studies (Scattone et al., 2002; Staley,2002) for which PND data were available, compre-hension sessions were included as a component ofSocial Story� intervention to enhance student under-standing. From Table IV, it can be seen that thestories where authors reported a comprehensioncomponent yielded a mean PND of 64 and the 20that did not yielded a PND of 60. The second relatesto the student’s access to the Social Story� followingits reading. In seven studies (Bledsoe et al., 2003;Feinberg, 2002; Norris & Dattilo, 1999; Scattoneet al., 2002; Staley, 2002; Thiemann & Goldstein,2001) the Social Story� remained accessible to thestudent following its initial reading. During twostudies the Social Story� was taken home for theparents to read to the child (Cullain, 2002; Feinberg,2002). Examination of Table IV reveals that the PNDfor the seven accessible stories was 64 and for the 19remaining stories the PND was 63.
462 Reynhout and Carter
Use of Additional Strategies
Evaluation of the efficacy of Social Stories� wasconfounded in some instances by the use of addi-tional intervention strategies. Researchers used verbaland/or physical prompting in seven studies (Feinberg,2002; Hagiwara & Myles, 1999; Rogers & Myles,2001; Scattone et al., 2002; Staley, 2002; Swaggartet al., 1995; Thiemann & Goldstein, 2001) andtangible reinforcers such as stickers, or edibles wereused in three studies (Kuttler et al., 1998; Staley,2002; Swaggart et al., 1995). The study of Staley(2002) provided specific comparison of additionalstrategies and Social Stories�. Other strategies,including teacher modeling and rehearsal (Pettigrew,1998; Thiemann & Goldstein, 2001) and self-evalua-tion using video-feedback (Thiemann & Goldstein,2001), were also employed. From Table IV it can bedetermined that the 13 stories where additionalstrategies were employed yielded a PND of 59 andthose that did not a PND of 62.
Maintenance and Generalization
Maintenance results were only reported for threeof the 16 studies (Kuoch & Mirenda, 2003; Romano,2002; Thiemann & Goldstein, 2001). Thiemann andGoldstein (2001) describes maintenance data acrossparticipants as ‘‘not compelling’’ and suggests anumber of factors including length of training, skilldifficulty, and adult prompt and visual cue fading toaccount for this finding. Romano (2002) collectedpost-intervention data on both the treatment andcontrol group 6 weeks after discontinuation of theintervention for five consecutive days. Positive resultswere obtained for the treatment group, and negativeresults were obtained for the control group, that isthere was an increase in inappropriate behaviors frompre-test and post-test to maintenance. As previouslynoted, however, the data from the Romano (2002)study needs to be interpreted circumspectly. Limitedgeneralization in terms of targeted behaviors or newlyacquired skills being demonstrated in more than onesetting were reported in six studies (Cullain, 2002;Hagiwara & Myles, 1999; Kuoch & Mirenda, 2003;Kuttler et al., 1998; Swaggart et al., 1995; Thiemann& Goldstein, 2001). Cullain (2002) investigated SocialStory� intervention with three participants at schooland two participants at home; the study by Kuttleret al. (1998) was conducted in two school settings,morning work time and lunchtime, and (Swaggartet al., 1995) reported increased social interaction ofthree participants across a variety of settings.
Social Validity
Only three of the studies reviewed examined anaspect of social validity. Scattone et al. (2002) usedthe Intervention Rating Profile (Martens, Witt,Elliott, & Darveaux, 1985) with teachers to evaluatethe acceptability of the treatment with the interven-tion falling well into the acceptable range. Thiemannand Goldstein (2001) asked 13 teachers and graduatestudents to view videotapes of social interactionrecorded before and after intervention and to ratespecific social behaviors for the target children andtheir peers; improvements in reciprocal social behav-iors were reported by all raters. Hagiwara and Myles(1999) asked five educators and professors to verifythe appropriateness of story construction.
DISCUSSION
Twelve of the 16 studies examined in this reviewused single subject designs. A major criticism of singlesubject designs is that they have low external validity.This weakness can be addressed by replication(Tawney & Gast, 1984). It is difficult to conductreplications and draw sensible conclusions aboutexternal validity if one is unsure of the characteristicsof participants in existing research. A general criti-cism of the studies is that adequate descriptions ofparticipant’s communicative and cognitive skills werenot always provided. Gray and Garand (1993, p. 2)state ‘‘Social stories are most likely to benefit studentsfunctioning intellectually in the trainable mentallyimpaired range or higher who possess basic languageskills.’’ Documentation of the level of cognitive andcommunicative functioning is therefore relevant andimportant, particularly considering the extremevariation possible in ASD. It is possible that theintervention is suited to participants with specificcharacteristics that will only be identified if the natureof the participant(s) is known. It is worthy of notethat in all but one study (Pettigrew, 1998), partici-pants were diagnosed with autism or Asperger’ssyndrome. The efficacy of Social Stories� in pro-moting behavior change in children with disabilitiesother than autism or Asperger’s is yet to be substan-tively researched. While almost all students in thestudies reviewed were identified as being on theautism spectrum, quantification of the degree ofautism was not provided in the vast majority ofinstances. Again, this may be an important variableimpacting on the efficacy of Social Stories�.
Social Stories� 463
Effect sizes in the small number of studies thatused group designs were highly variable. The major-ity of studies used single subject designs and the PNDstatistic was calculated. The PND method sets arelatively high standard for evaluation of graphicdata with the most extreme data point in baselineproviding the point of comparison for intervention.Nevertheless, small n designs are not generally suitedto detecting small treatment effects (unless data isexceedingly stable) and the practitioners are mostinterested in treatments with powerful clinical effects.The overall PND of 43 indicated Social Stories�were an ineffective intervention according to thecriteria of Mastropieri et al. (1996) and was only 51(marginally effective) if zero or ceiling comparativedata in baseline were removed. While some interven-tions evidence both high PNDs and reasonably cleardemonstration of experimental control (e.g., Brow-nell, 2002; Kuttler et al., 1998), effects were relativelymodest (and/or inconsistent) in relation to baselinevariation in other studies (e.g., Hagiwara & Myles,1999; Norris & Dattilo, 1999), accounting for therelatively low PND figures. High levels of baselineinstability in relation to the magnitude of interven-tion effects were certainly evident in a number ofstudies. In addition, Kuoch and Mirenda (2003) havenoted that behavior change following Social Story�intervention may not be fully reversible. Thus, it ispossible that some studies using withdrawal compo-nents may not have been ideally suited to examiningthe efficacy of the intervention. There was alsoenormous variation in responses of specific partici-pants to intervention. Considering the relatively smalldata set variation in PND across and within studies,the mean PND value should be treated with a highdegree of caution. Some studies or interventions withspecific students appeared to be highly effective withclear demonstrations of control and other studieswere relatively ineffective, indicating the possibilitythat specific participant or intervention featuresinfluencing effectiveness.
Social Stories� are claimed to be capable ofaddressing a variety of applications (Gray, 2003). TheSocial Stories� used in the studies examined in thepresent review certainly addressed a variety of behav-iors, supporting this claim. None of the SocialStories� employed in the studies were of the curric-ulum type (see Gray & Garand, 1993). Thus, theefficacy of teaching academic skills through SocialStory� intervention remains unaddressed.
Considerable variation was seen in the use ofSocial Stories� in the studies in terms of construc-tion, implementation and use of additional strategiesand these issues will be addressed in turn. Criticism ofprevious research has been offered by Kuoch andMirenda (2003) on the basis that many stories failedto conform to the recommended construction. In asummary of the composition of Social Storied� usedin a selection of 10 case studies or experimentaldesigns published in journal articles prior to 2003,Kuoch and Mirenda (2003) found five studies tocontain Social Stories� not conforming to applicableSocial Story� guidelines, although the impact of thison Social Story� efficacy as determined by, forexample, is not determined by the authors. Analysisof the Social Stories� from the 16 experimentalstudies in the present review that included studiesconducted prior to 2004, confirmed that a number ofthese also deviated considerably from the construc-tion prescribed by Gray (2003). There was, however,no evidence that this impacted negatively on PND. Infact, the small number of stories that varied fromGray’s (2003) guidelines (higher ratio of directive todescriptive sentences) seemed to be associated withconsiderably higher mean PND. The relationshipbetween frequency of particular sentence types andPND was also of interest. There was no clear patternof evidence that a higher absolute percentage ofdescriptive sentences impacted positively on PND.Stories in the middle third for descriptive sentencefrequency were associated with considerably higherPNDs than those in the upper and lower third. Therewas preliminary evidence that the level of conse-quence sentences might affect efficacy, with storiesdemonstrating very low levels of consequence sen-tences being associated with lower PNDs. In hind-sight, this is not surprising as strategies involving thesystematic use of reinforcement has been identified asone of the most powerful interventions in specialeducation (see Fuchs & Fuchs, 1986; Kavale &Forness, 1999; Skiba & Casey, 1985; Walberg &Wang, 1987) and have been consistently and demon-stratively effective in facilitating behavior changewith individuals with autism (see Heflin & Alberto,2001; McConnell, 2002). The potential value ofreinforcement is recognized by Gray (2003) whostates:
Social stories have another purpose that is equally
important: acknowledging achievement. In fact a
child’s first social story should describe a skill or
464 Reynhout and Carter
situation that is typically successful and problem-free.
Written praise may be far more meaningful for
children with ASD than its verbal counterpart. At
least half the social stories developed for the child
with ASD should bring attention to positive
achievements (pp. 1–2).
In fact, consequences of actions were describedin 90% of the Social Stories� examined in the presentstudy, suggesting the role of consequences may wellbe more important than suggested by Gray (2003).The exact role of consequences sentences in SocialStories� awaits further empirical investigation but itis at least plausible that their role in identifyingnatural or artificial consequences of actions may beimportant. Analysis of the effects of Social Story�composition should be interpreted carefully due tothe small number of stories available, the fact thatthese variables were not experimentally manipulatedand the percentages of various sentence types variedconsiderably. Nevertheless, there is some suggestionthat some sentence types may impact on story efficacyalthough not necessarily in the way that might bepredicted by Gray (2003).
In relation to reinforcement, it should be notedthat three studies employed tangible reinforcers suchas stickers or edibles (Kuttler et al., 1998; Staley,2002; Swaggart et al., 1995) and a further study usedself-evaluation with video-feedback (Thiemann &Goldstein, 2001). One study (Staley, 2002) is ofparticular interest in this regard. Five participantswere read two Social Stories� intended to increasetheir chewing with their mouth closed and increasetheir napkin use. The researchers also gave partici-pants edible reinforcers for chewing with their mouthclosed and for appropriate napkin use during lunch.Using a reversal and multiple-baseline designs tocompare the effects of Social Stories� and reinforc-ers, results clearly showed no effect of Social Sto-ries�, but an almost immediate effect of reinforcers(Staley, 2002).
A range of other additional interventions, mostfrequently physical and verbal prompting, wereemployed in addition to Social Stories�. In studieswhere prompting and/or reinforcement are used asadditional strategies, it becomes difficult to ascertainwhich of the treatments, prompting, reinforcement orSocial Story� is the critical component of the inter-vention, or whether a combination of these has thegreatest effect. While there was no clear evidence of adifference between stories using additional strategiesand those that did not in the present analysis, manyconfounding variables were evident. There is a very
clear research need to examine the extent to whichSocial Stories� contributed additional impact to thesewell-validated intervention strategies.
The use of more than one Social Story� totarget one social situation, or the use of several storiesto target several behaviors, the frequency with whicha specific Social Story� is reviewed, the inclusion ofcomprehension activities and the access of the par-ticipant to the Social Story� outside the interventionphase are all variables that could affect interventionefficacy. The significance of these issues in terms ofthe effectiveness of the treatment requires furtherinvestigation.
Researchers used visual symbols in 11 of thestudies examined. Originally, Gray and Garland(1993) advised against the use of illustrations,describing them as distracting and liable to makethe student misinterpret the situation. This recom-mendation has subsequently been revised and the useof illustrations ‘‘...that reflect consideration of the ageand personal learning characteristics of the personwith ASD’’ (Gray, 2003, p. 5) is now suggested asbeneficial to social understanding. The use of visualsymbols is in keeping with current Social Story�guidelines, and is consistent with other research onvisual learning and children with autism (e.g., Dett-mer et al., 2000).
From the preceding discussion, it is obvious thatSocial Stories� are multifaceted interventions. It isunclear from the present review that the prescribed(and complex) story construction is necessary to theefficacy of the intervention, which components arecritical to effectiveness and whether Social Stories�necessarily add to the effectiveness of other interven-tions. The confounding of Social Story� interven-tions with other strategies is a problem in manyexisting studies (Kuoch & Mirenda, 2003).
Maintenance and generalization were also inad-equately addressed in the studies. These issues are ofpivotal importance for children with ASD. On thebasis of the limited information provided in thestudies, it is not possible to draw conclusions aboutthe efficacy of Social Story� intervention in terms ofmaintenance and generalization. Programming formaintenance and generalization are typically essentialcomponents of an effective intervention. The inherentflexibility and portability of Social Stories� maypredispose them to be of use over time and in anumber of settings.
Social validity usually refers to validation ofgoals, procedures and outcomes (Wolf, 1978) by
Social Stories� 465
consumers, who can include participants, implement-ers, as well as members of the broader community.Hawkins (1991) and Schwartz and Baer (1991) haveargued for more objective functional validation ofintervention (habilitative validity), which may includecomparison of performance with community stan-dards. For example, Hughes, Harmer, Killian andNiarhos (1995) collected social comparison data onregular peers when teaching social initiation skills tostudents with intellectual disabilities. Thus, as well asdemonstrating that the skills of interest could betaught, they also established that performance waswithin the broad functional range of regular peers.Smith (2001) has presented data suggesting thatteachers viewed Social Stories� as effective butobjective data on learner performance was notprovided, hence the study was not included in thepresent review. Only three of the studies reviewedaddressed social validity, one examining acceptabilityof the intervention to teacher (Scattone et al., 2002),the second the appropriateness of the Social Story�construction and the third provided subjective vali-dation of skill improvement in social skills by blindraters (Thiemann & Goldstein, 2001). Interestingly,none of the studies examined habilitative validity,although social comparative data would seem rele-vant to many of the skills addressed.
Some aspects of reliability measures in thestudies were problematic. While interobserver reli-ability was reported in 12 studies but proceduralreliability reported in only four studies. The relativesimplicity of Social Stories� intervention may lead toan assumption that procedural reliability does notneed to be measured. Examination of Social Stories�revealed deviation from recommended constructionin a substantial proportion of instances and it wouldseem at least possible that there may have beendeviations in the intended implementation. Reliabil-ity measures, including measures of procedural reli-ability, are a quintessential component of goodresearch design; the lack of these undermines confi-dence in the research.
An important limitation of the existing reviewshould be acknowledged. Results should be treatedwith caution given the modest data set available. Inparticular, analysis of specific study features shouldbe treated circumspectly. This analysis was based ona subset of studies that yielded PND data and thusshould be considered preliminary in nature. Interpre-tation is further limited by the degree of variation instory construction and implementation. Within theresearch reviewed there were several clear examples of
well-controlled studies suggesting the interventionwas effective. Nevertheless, the intervention was notrobustly effective across studies, participants andbehaviors as reflected in mean PND. Consequently,the existing research does not demonstrate unequiv-ocally that Social Story� intervention is consistentlyeffective in facilitating behavior change in childrenwith autism. The majority of studies used singlesubject designs and the mean PNDs were marginal atbest although there was substantial variation inresults across studies, with some interventions prov-ing very effective. This raises the possibility thatundetermined factors (such as participant character-istics or story construction and implementation) maybe important to the success of the intervention. Thisfinding alone provides justification for further inves-tigation of the strategy.
FUTURE RESEARCH
While there is a number of issues that need to beaddressed in future research into Social Story�intervention, several stand out as being of particularimportance. Firstly a number of other proceduralconsiderations should be addressed in future researchon Social Stories�. Inadequate participation descrip-tion in extant research has made it difficult todetermine whether participant related variables mod-erate the effect of Social Story� interventions.Adequate participant description, in terms of resultsof standardized tests, providing information pertain-ing to cognitive ability, position of the participant(s)on the autism spectrum, and language should be anintegral component of future research investigatingSocial Story� efficacy. The procedural integrity andsocial validity of Social Story� interventions has notbeen examined in much of the extant research. Boththese features are hallmarks of good research practiceand should be routinely incorporated into futurestudies. In particular, habilitative social validity(Hawkins, 1991) should be examined to provide ameaningful framework for interpreting behaviorchange. In addition, there is some question whetherthe intervention is typically fully reversible and theuse of reversal designs should be questioned.Repeated demonstrations of experimental controlthat are characteristics of strong small n researchdesigns can be achieved using the multiple baselinedesign, without requiring interventions to be effec-tively reversible. Finally, very high levels of baselineinstability in relation to intervention effects that was
466 Reynhout and Carter
evident in much of the research examined may beinterpreted as indicating that future researchers mightdo well to concentrate more on establishing baselinestability.
Research on individuals with ASD and signifi-cant intellectual disabilities is extremely limited. Grayand Garand (1993) have suggested that the interven-tion may be suitable for individuals with a moderatelevel of intellectual disability (‘‘trainable’’ in theirterminology) who possess ‘‘basic language skills’’ butvery few participants in the existing studies clearly fitthis profile. Thus, the applicability of the interventionof students with ASD and significant intellectualdisabilities needs to be explored further. Similarly,there is little research examining the potential forSocial Story� interventions to be used with studentwho do not present with ASD but do have majorsocial skills deficits. Many interventions that havebeen successful with individuals who present withASD (e.g., reinforcement, response prompting, visualsupports) have also been effective with other groups.The extent to which Social Stories� are effective withother populations or deferentially effective with ASDremains to be explored.
There remains a question as to whether theconstruction of Social Stories� contributes to thehigh variability in efficacy reported in the literature.Further research should be conducted to determinewhether recommended story construction affectsoutcome or whether particular sentence types play acritical role. Noting that the recommended construc-tion of Social Stories� is not based on theoretical orempirical rationale (Kuoch & Mirenda, 2003), thisissue should be considered a priority for futureresearch. In addition, the comprehension componentof Social Story� intervention may be important toefficacy. Further research is necessary to explore thispossibility.
The findings of many of the existing studies havebeen confounded by the use of Social Stores� incombination with other empirically verified proce-dures, such as prompting and operant reinforcement.Thus, there is a pressing need for further research intothe effectiveness of Social Stories� in isolation.
Great variability was evident in the researchreviewed in terms of story construction and imple-mentation, including significant deviations from rec-ommended practice. If such variation was evident in(presumably) relatively controlled research environ-ments, questions are raised as to how the interventionmight be being applied in the field. Noting that itcould be argued that practice often outstrips the
research base, it would be of some interest toinvestigate the extent of Social Story� use in thefield as well as how the strategy is being applied.
Finally, research addressing issues of mainte-nance and generalization is extremely limited. This issurprising given that Social Story� interventions arerelatively simple and would seem suited to use overextended time periods as well as across persons andplaces. Future studies should be extended to incor-porate investigation of maintenance and generaliza-tion a matter of priority.
CONCLUSION
A small corpus of research literature pertainingto the use of Social Stories� raises many questionsregarding their effectiveness in facilitating behaviorchange in children with autism. The present review ofthe literature highlights the need to adopt a system-atic, rigorous scientific approach to future research ifthese questions are to be answered, and the efficacy ofSocial Stories� determined.
REFERENCES
American Psychological Association.. (2001). Publication Manualof the American Psychological Association (5th ed.). Wash-ington, DC: American Psychological Association.
Backman, B., & Pilebro, C. (1999). Augmentative communicationin dental treatment of a nine-year-old boy with Aspergersyndrome. ASDC Journal Of Dentistry For Children, 66, 419–420.
Bledsoe, R., Myles, B. S., & Simpson, R. L. (2003). Use of a SocialStory� intervention to improve mealtime skills of an adoles-cent with Asperger syndrome. Autism, 7, 289–295.
Brownell, M. D. (2002). Musically adapted social stories to modifybehaviors in students with autism: four case studies. Journal ofMusic Therapy, 39, 117–144.
Chapman, L., & Trowbridge, M. (2000). Social stories for reducingfear in the outdoors. Horizons, 11(3), 39–40.
Cullain, R. E. (2002). The effects of social stories on anxiety levelsand excessive behavioral expressions of elementary school-agedchildren with autism. Unpublished doctoral dissertation, TheUnion Institute Graduate College, Cincinnati, Ohio.
Dawson, G., & Watling, R. (2000). Interventions to facilitateauditory, visual, and motor integration in autism: A review ofthe evidence. Journal of Autism & Developmental Disorders, 30,415–421.
Del Valle, P. R., McEachern, A. G., & Chambers, H. D. (2001).Using social stories with autistic children. Journal of PoetryTherapy, 14, 187–197.
Dettmer, S., Simpson, R. L., Myles, B. S., & Ganz, J. B. (2000).The use of visual supports to facilitate transitions of studentswith autism. Focus on Autism and Other Developmental Dis-abilities, 15, 163–169.
Didden, R., Duker, P. C., & Korzilius, H. (1997). Meta-analyticstudy on treatment effectiveness for problem behaviors withindividuals who have mental retardation. American Journal onMental Retardation, 101, 387–399.
Social Stories� 467
Elder, J. (2002). Current treatments in autism: Examining the evi-dence and clinical implications. Journal of NeuroscienceNursing, 34(2), 67–73.
Feinberg, M. J. (2002). Using social stories to teach specific socialskills to individuals diagnosed with autism. Unpublished doc-toral dissertation, California School Of Professional Psychol-ogy, San Diego, California.
Fuchs, L. S., & Fuchs, D. (1986). Effects of systematic formativeevaluation: A meta-analysis. Exceptional Children, 53, 199–208.
Garfield, J., Peterson, C., & Perry, T. (2001). Social cognition,language acquisition and the development of the theory ofmind. Mind & Language, 16(5), 494–541.
Gray, C. (2000a). The new Social Story� book. (Illustrated edi-tion). Arlington, TX: Future Horizons, Inc.
Gray, C. (2000b). Writing social stories with Carol Gray. [Video-tape and workbook]. Arlington, TX: Future Horizons, Inc.
Gray, C. (2003). Social Stories. Retrieved April 13, 2003, fromhttp://www.thegraycenter.org.
Gray, C., & Garand, J. D. (1993). Social stories: Improvingresponses of students with autism with accurate social infor-mation. Focus on Autistic Behavior, 8(1), 1–10.
Greenway, C. (2000). Autism and Asperger syndrome: Strategies topromote prosocial behaviours. Educational Psychology inPractice, 16, 469–486.
Hagiwara, T., & Myles, B. S. (1999). A multimedia Social Story�intervention: Teaching skills to children with autism. Focus onAutism and Other Developmental Disabilities, 14, 82 .
Happe, F. G. E. (1995). The role of age and verbal ability in thetheory of mind task performance of subjects with autism.Child Development, 66, 843–855.
Hawkins, R. P. (1991). Is social validity what we are interested in?Argument for a functional approach. Journal of AppliedBehavior Analysis, 24, 205–213.
Heflin, L. J., & Alberto, P. A. (2001). Establishing a behavioralcontext for learning for students with autism. Focus on Autismand Other Developmental Disabilities, 16, 93–101.
Hughes, C., Harmer, M. L., Killian, D. J., & Niarhos, F. (1995).The effects of multiple-exemplar self-instructional training onhigh school students generalized conversational interactions.Journal of Applied Behavior Analysis, 28, 201–218.
Kasari, C. (2002). Assessing change in early intervention programsfor children with autism. Journal of Autism and DevelopmentalDisorders, 32, 447–461.
Kavale, K. A., & Forness, S. R. (1999). Efficacy of special educationand related services. Washington, DC: AAMR.
Kerrin, R., Murdock, J., Sharpton, W., & Jones, N. (1998). Who’sdoing the pointing? Investigating facilitated communication ina classroom setting with students with autism. Focus on Autismand Other Developmental Disabilities, 13, 73–79.
Kuoch, H., & Mirenda, P. (2003). Social Story� interventions foryoung children with autism spectrum disorders. Focus onAutism and Other Developmental Disabilities, 18, 219–227.
Kuttler, S., Myles, B. S., & Carlson, J. K. (1998). The use of socialstories to reduce precursors to tantrum behavior in a studentwith autism. Focus on Autism and Other Developmental Dis-abilities, 13, 176–182.
Lorimer, P. A., Simpson, R. L., Myles, B. S., & Ganz, J. B. (2002).The use of social stories as a preventative behavioral inter-vention in a home setting with a child with autism. Journal ofPositive Behavior Interventions, 4(1), 53–60.
Martens, B. K., Witt, J. C., Elliott, S. N., & Darveaux, D. (1985).Teacher judgements concerning the acceptability of schoolbased interventions. Professional Psychology: Research andPractice, 16, 191–198.
Mastropieri, M. A., Scruggs, T. E., Bakken, J. P., & Whedon, C.(1996). Reading comprehension: A synthesis of research in
learning disabilities. Advances in Learning and BehavioralDisabilities, 10B, 201–227.
Mathur, S. R., Kavale, K. A., Quinn, M. M., Forness, S. R., &Rutherford, Jr., R. B. (1998). Social skills interventions withstudents with emotional and behavioral problems: A quanti-tative synthesis of single-subject research. Behavioral Disor-ders, 23, 193–201.
McCartney, K., & Rosenthal, R. (2000). Effect size, practicalimportance, and social policy for children. Child Development,71, 173–180.
McConnell, S. R. (2002). Interventions to facilitate social interac-tion for young children with autism: Review of availableresearch and recommendations for educational interventionand future research. Journal of Autism and DevelopmentalDisorders, 32, 351–372.
McWilliam, R. (1999). Controversial practices: The need for a re-acculturation of early intervention fields. Topics In EarlyChildhood Special Education, 19, 177–188.
Myles, B. S., & Simpson, R. L. (2001). Understanding the hiddencurriculum: An essential social skill for children and youthwith Asperger Syndrome. Intervention in School and Clinic, 36,279–286.
Norris, C., & Dattilo, J. (1999). Evaluating effects of a SocialStory� intervention on a young girl with autism. Focus onAutism and Other Developmental Disabilities, 14, 180–186.
Ozonoff, S., & Miller, J. (1995). Teaching theory of mind: A newapproach to social skills training for individuals with autism.Journal of Autism & Developmental Disorders, 25, 415–435.
Pettigrew, J. D. C. (1998). Effects of the modeling of verbal and non-verbal procedures for interactions with peers through socialstories and scaffolded activities on the social competence of 3-and 4-year-old children with specific language impairments.Unpublished doctoral dissertation, Texas Woman’s Univer-sity, Denton, Texas.
Rogers, M. F., & Myles, B. S. (2001). Using social stories andcomic strip conversations to interpret social situations for anadolescent with Asperger Syndrome. Intervention in School andClinic, 36, 310–313.
Romano, J. (2002). Are social stories effective in modifying behaviorin children with autism? Unpublished doctoral dissertation,Fairleigh Dickinson University, Teaneck, New Jersey.
Rowe, C. (1999). Do social stories benefit children with autism inmainstream primary schools?. Special education forwardtrends, 26(1), 12–14.
Salzberg, C. L., Strain, P. S., & Baer, D. M. (1987). Meta-analysisfor single-subject research: When does it clarify, when does itobscure?. Remedial and Special Education, 8, 43–48.
Scattone, D., Wilczynski, S. M., Edwards, R. P., & Rabian, B.(2002). Decreasing disruptive behaviors of children with aut-ism using social stories. Journal of Autism and DevelopmentalDisorders, 32, 535–543.
Schwartz, I. S., & Baer, D. M. (1991). Social validity assessments:Is current practice the state of the art?. Journal of AppliedBehavior Analysis, 24, 189–204.
Scruggs, T. E., Mastropieri, M. A., & Casto, G. (1987). Thequantitative synthesis of single-subject research: Methodologyand validation. Remedial and Special Education, 8, 24–32.
Scruggs, T. E., Mastropieri, M. A., Forness, S. R., & Kavale, K. A.(1988). Early language intervention: A quantitative synthesis ofsingle subject research. Journal of Special Education, 22, 259–283.
Simpson, R. L., & Myles, B. S. (1998). Aggression among childrenand youth who have Asperger’s syndrome: A different popu-lation requiring different strategies. Preventing School Failure,42, 149–153.
Skiba, H. M., & Casey, A. (1985). Intervention for behaviorallydisordered students: A quantitative review and methodologicalcritique. Behavioral Disorders, 10, 239–252.
468 Reynhout and Carter
Smith, C. (2001). Using social stories to enhance behavior in chil-dren with autistic spectrum difficulties. Educational Psychologyin Practice, 17, 337–345.
Staley, M. J. (2002). An investigation of social-story effectivenessusing reversal and multiple-baseline designs. Unpublished doc-toral dissertation, University of Kansas, Lawrence, Kansas.
Strain, P. S., Kohler, F. W., & Gresham, F. (1998). Problems inlogic and interpretation with quantitative syntheses of single-case research: Mathur and colleagues (1998) as a case in point.Behavioral Disorders, 24, 74–85.
Swaggart, B., Gagnon, E., Bock, S. J., Earles, T. L., Quinn, C.,Myles, B. S. et al. (1995). Using social stories to teach socialand behavioral skills to children with autism. Focus on AutisticBehavior, 10, 1–16.
Tawney, J. W., & Gast, D. L. (1984). Single subject research inspecial education. New York: Merrill.
Thiemann, K. S., & Goldstein, H. (2001). Social stories, writtentext cues, and video feedback: Effects on social communication
of children with autism. Journal of Applied Behavior Analysis,34, 425–446.
Walberg, H. J., & Wang, M. C. (1987). Effective educationalpractices and provisions for individual differences. In M. C.Wang, M. C. Reynolds, & H. J. Walberg (Eds.), Handbook ofspecial education: Research and practice – Learner character-istics and adaptive education (Vol. 1, pp. 113–128). Oxford:Pergamon.
White, O. R. (1987). Some comments concerning ‘‘The quantitativesynthesis of single-subject research’’. Remedial and SpecialEducation, 8, 34–39.
Wolf, M. W. (1978). Social validity: The case for subjective mea-surement or how applied behavior analysis is finding its heart.Journal of Applied Behavior Analysis, 11, 203–214.
Xin, Y. P., & Jitendra, A. K. (1999). The effects of instruction insolving mathematical word problems for students with learn-ing problems: A meta-analysis. Journal of Special Education,32, 207–225.