Expert answer:One-Way ANOVA in Practice

Expert answer:Write a 2- to 3-page critique of the article you found. In your critique, include responses to the following:
Why did the authors use this one-way ANOVA test?
Do you think it’s the most appropriate choice? Why or why not?
Did the authors display the data?
Does the results table stand-alone (i.e., are you able to interpret the study from it?) Why or why not?
Use proper APA format, citations, and referencing.
one_way_anova.pdf

Unformatted Attachment Preview

Research in Developmental Disabilities
24 (2003) 120±138
Ef®cacy of behavioral interventions for
reducing problem behavior in persons
with autism: a quantitative synthesis
of single-subject research
Jonathan M. Campbell*
Department of Psychology, University of Memphis, Memphis, TN, USA
Received 28 January 2002; received in revised form 2 May 2002; accepted 4 November 2002
Abstract
The ef®cacy of behavioral interventions for problem behavior in persons with autism was
reviewed. One hundred and seventeen published articles representing 181 individuals with
autism were examined. Articles were selected from 15 journals. Participant, treatment, and
experimental variables were evaluated. Three effect sizes were calculated for each article.
Behavioral treatments are effective in reducing problematic behaviors in individuals with
autism. Type of target behavior and type of treatment did not moderate the average effect of
treatment. As measured by percentage of zero data (PZD), three variables were predictive of
behavioral suppression beyond that accounted for by behavioral topography and treatment
type. Reliability of observation and number of treatment data points were positively related to
PZD scores. Treatments based on experimental functional analysis (EFA) produced higher
average PZD scores than treatments that did not include an EFA. The implications of the
®ndings, study limitations, and suggestions for future research are discussed.
# 2003 Elsevier Science Ltd. All rights reserved.
Keywords: behavioral interventions; problem behavior; autism; quantitative review; functional analysis
Problematic behaviors associated with autistic disorder include self-injurious
behavior (SIB), stereotypic behavior (e.g., body rocking), aggression, and property destruction (Charlop, Schreibman, & Kurtz, 1991). Aside from causing tissue
*
Present address: Department of Educational Psychology, Research and Measurement, 325
Aderhold Hall, University of Georgia, Athens, GA 30602-7143, USA. Tel.: ‡1-706-542-5108;
fax: ‡1-706-542-4240.
E-mail address: jcampbel@coe.uga.edu (J.M. Campbell).
0891-4222/03/$ ± see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0891-4222(03)00014-3
J.M. Campbell / Research in Developmental Disabilities 24 (2003) 120±138
121
damage, SIB is considered problematic because it results in reduced attention and
interferes with educational intervention (e.g., Gorman-Smith & Matson, 1985).
Stereotypy interferes with new learning and competes with socially acceptable
behavior, such as play (e.g., Koegel, Firestone, Kramme, & Dunlap, 1974). In
published treatment studies for individuals with autistic disorder, SIB, aggression,
stereotyped behavior, and property destruction were targeted most frequently
(Lundervold & Bourland, 1988; Matson, Benavidez, Compton, Paclawskyj, &
Baglio, 1996).
Many interventions based on the principles of applied behavior analysis have
been evaluated via single-subject designs. Techniques used to reduce problematic
behaviors have included sensory extinction (e.g., Maag, Wolchik, Rutherford, &
Parks, 1986), time-out (e.g., Durand & Carr, 1987), differential reinforcement of
other behavior (DRO), differential reinforcement of alternative behavior (DRA),
overcorrection (e.g., Harris, Handleman, & Fong, 1987), and combinations of
these techniques (e.g., DRO and time-out; Rolider & Van Houten, 1985). Other
techniques have involved delivering controversial punishing agents, such as water
mist (e.g., Bailey, Pokryzwinski, & Bryant, 1983) and electric shock (e.g.,
Linscheid, Iwata, Ricketts, Williams, & Grif®n, 1990), and manipulating antecedent conditions, such as altering the physical environment (Duker & Rasing,
1989) or engaging individuals in physical exercise (e.g., Kern, Koegel, & Dunlap,
1984).
The use of aversive procedures to treat problematic behaviors in individuals
with developmental disabilities has evoked considerable controversy in the past
(e.g., Matson & Taras, 1989; Schopler, 1986). More recently, discussion has
moved from the pros and cons of aversive methods to strategies for selecting
treatment procedures that are linked to variables that maintain problem behaviors
(e.g., social attention or tangible reinforcement; Carr, Robinson, & Palumbo,
1990).
A variety of functional assessment procedures, including indirect, descriptive,
and experimental assessment methods, have been used to identify the factors
maintaining an individual’s problematic behavior (Lerman & Iwata, 1993). With
indirect assessment methods, information is gathered through interview or rating
scales (e.g., Paclawskyj, Matson, Rush, Smalls, & Vollmer, 2000). Descriptive
assessment involves direct observation of the occurrence of problem behavior in
natural conditions in order to identify antecedent and consequent variables
associated with the behavior (Lerman & Iwata, 1993). Experimental functional
analysis (EFA) involves systematic manipulation of variables that may maintain
the problem behavior, such as removing task demands or providing attention after
the occurrence of behavior (Iwata, Dorsey, Slifer, Bauman, & Richman, 1982/
1994). Although EFA may be relatively complex and labor intensive, it offers
greater precision in identifying causal relationships between environmental
variables and problem behavior (Kahng, Iwata, & Lewin, 2002; Paclawskyj
et al., 2000).
Many quantitative reviews have examined the treatment of problem behavior
with individuals with developmental disabilities (DD) and mental retardation
122
J.M. Campbell / Research in Developmental Disabilities 24 (2003) 120±138
(MR) (e.g., Didden, Duker, & Korzilius, 1997; Gorman-Smith & Matson,
1985; Kahng et al., 2002; Lennox, Miltenberger, Spengler, & Erfanian, 1988;
Lundervold & Bourland, 1988; Matson & Gorman-Smith, 1986; Pelios, Morren,
Tesch, & Axelrod, 1999; Schlosser & Goetze, 1992; Scotti, Evans, Meyer,
& Walker, 1991; Sternberg, Taylor, & Babkie, 1994; Wehmeyer, 1995). Results
have indicated that studies that employed functional assessment methods,
promoted generalized behavioral change, and combined differential reinforcement with another intervention produced the best outcomes (Didden et al.,
1997; Kahng et al., 2002; Pelios et al., 1999; Scotti et al., 1991; Sternberg
et al., 1994).
To date, reviews of treatment outcomes for individuals with autistic disorder
have been either narrative (e.g., Rogers, 1998) or descriptive (e.g., Matson et al.,
1996) in nature. Although treatment outcomes of individuals with autistic
disorder have been combined with other individuals in the context of larger
reviews, individuals with autistic disorder show cognitive and behavioral
features that distinguish them from persons with DD or MR (e.g., Carpentieri
& Morgan, 1994, 1996). For example, children with autistic disorder are more
impaired in the domains of verbal reasoning, social knowledge, and communication skills than nonautistic children with MR (Carpentieri & Morgan, 1994,
1996). Compared to children with other disabilities, children with autistic
disorder also exhibit more de®cits in adaptive social behavior (e.g., Volkmar
et al., 1987).
In light of qualitative differences in social functioning, individuals with autistic
disorder may be more likely than individuals with MR to exhibit problematic
behavior maintained by variables that are not mediated by the social environment
(i.e., automatic reinforcement; see LeBlanc, Patel, & Carr, 2000 for discussion).
Results of a study by Dawson, Matson, and Cherry (1998) suggested that, in
individuals with autistic disorder, reinforcers for problem behavior are less
likely to be socially mediated when compared to reinforcers that maintain
problem behavior in individuals with MR. Problem behaviors maintained by
automatic reinforcement have been shown to be quite dif®cult to treat (LeBlanc
et al., 2000).
The present review considers four treatment-related questions. First, what is
the average ef®cacy of focal behavioral interventions in reducing problematic
behaviors in individuals with autistic disorder? Second, are some behavioral
interventions more effective than others in reducing problematic behaviors in
individuals with autism? This question is important as functional assessment
methods have been shown to increase the effectiveness of positive interventions in
individuals diagnosed with DD and MR, thereby reducing the need for treatment
based on punishment alone. Third, do participant, treatment, or experimental
variables in¯uence the average ef®cacy of behavioral interventions? For example,
do children with higher IQs receive more bene®t from behavioral interventions
than those with lower IQs? Fourth, does the presence of a pretreatment functional
assessment improve the average ef®cacy of behavioral interventions with individuals diagnosed with autistic disorder?
J.M. Campbell / Research in Developmental Disabilities 24 (2003) 120±138
123
1. Method
1.1. Study identi®cation and selection
Published treatment studies of autistic disorder from 1966 to 1998 were
identi®ed through searches of the PsycLit, ERIC, and MedLine databases. Studies
also were identi®ed by a hand search of pertinent journals, including Behavior
Modi®cation, Behavior Therapy, Behavioral Disorders, Behavioral Interventions,
Behaviour Research and Therapy, Child and Family Behavior Therapy, Education and Training in Mental Retardation, Journal of Abnormal Child Psychology,
Journal of Applied Behavior Analysis, Journal of Autism and Developmental
Disorders, Journal of Behavior Therapy and Experimental Psychiatry, Journal of
Clinical Child Psychology, Journal of Consulting and Clinical Psychology,
Journal of Experimental Child Psychology, Journal of the Association for Persons
with Severe Handicaps, Mental Retardation, and Research in Developmental
Disabilities.
Articles were selected for inclusion if the following criteria were satis®ed.
First, articles were selected if any participant was diagnosed with autistic
disorder; participants described as “autistic-like” or engaging in “autistic-like
behavior” were excluded. If the article included multiple participants, only
those individuals diagnosed with autistic disorder were included in the review.
Second, only articles that reported the results of single-case studies were selected;
articles using group designs were excluded. Third, single-case studies were
selected if baseline and treatment phases were present in the study and if repeated
data points, not mean scores, were reported. Studies containing less than two
baseline data points were excluded for data analysis purposes (e.g., Didden et al.,
1997). Fourth, studies were included if treatment targeted reduction of SIB,
stereotypy, aggression, or property destruction. A total of 117 articles met these
criteria.
1.2. Estimating effects of behavioral treatments
Disagreement exists regarding how to quantify results of single-case data for
meta-analytic purposes (see Allison & Gorman, 1994; Scruggs & Mastropieri,
1998 for discussion). Frequently reported summary methods have involved the
calculation of: (a) a percentage of reduction of behavior from baseline (e.g.,
Kahng et al., 2002), identi®ed as the mean baseline reduction (MBLR) procedure
for the purposes of this review; (b) percentage of nonoverlapping data (PND;
Scruggs, Mastropieri, & Casto, 1987); and/or (c) percentage of zero data (PZD;
Scotti et al., 1991). The MBLR is calculated by subtracting the average observation in the treatment phase from the average observation in the baseline phase,
dividing by the average observation in the baseline phase, and multiplying by 100.
PND summarizes single-subject treatment ef®cacy by calculating the percentage
of treatment data points that do not overlap with the highest or lowest baseline
data point. PZD is calculated by locating the ®rst data point in the treatment
124
J.M. Campbell / Research in Developmental Disabilities 24 (2003) 120±138
phase that reaches zero and calculating the percentage of data points recorded in
the treatment phase, including the ®rst zero, that remain at zero (Scotti et al., 1991,
p. 238).
Compared to MBLR and PND, PZD has been considered a more stringent
indicator of treatment ef®cacy, representing the degree of behavior suppression as
opposed to behavior reduction (Scotti et al., 1991). Also, PND and PZD have been
found to be uncorrelated with each other, thereby justifying their use as independent measures of treatment outcome (Scotti et al., 1991). Thus, MBLR, PND,
and PZD were calculated for each participant that met selection criteria within
each article. For the calculation of MBLR, graphically represented data were
transformed to raw data via a drafting divider. The distance between each data
point and the abscissa was calculated in millimeters and rounded to the nearest
0.5 mm. The data conversion procedure has been used with a high degree of
interrater reliability (e.g., Allison, Faith, & Franklin, 1995). For the PND statistic,
if a baseline phase reported one or more data points of zero, the same number of
zero data points was excluded in the treatment phase prior to calculation of the
PND (Didden et al., 1997). For convenience, the MBLR, PND, and PZD statistics
are referred to as effect sizes throughout the remainder of the article.
When more than one problem behavior was targeted for a participant, the
average effect sizes for the participant were calculated by weighting each
behavior according to the number of data points reporting on the behavior. When
data points represented averages of multiple target behaviors (e.g., eye gouging,
kicking wall, and hitting sibling), behaviors were weighted according to the
number of total observations contributing to the data for that participant; the
averaged target behaviors were coded according to procedures described below.
Within each article, effect sizes were weighted according to the number of data
points per participant, then averaged for all participants to yield three effect sizes
per article. All effect sizes were calculated by comparing the ®rst baseline phase
to the ®nal treatment phase.
1.3. Variables coded and reliability
The following participant information was coded: age, gender, IQ score, level
of intellectual functioning, level of verbal ability, and diagnostic criteria used. The
following intervention information was coded: target behavior, type of intervention technique used, presence of follow-up data, length of follow-up interval,
attempt to generalize treatment effects, parental involvement in treatment, length
of treatment per session, and presence of functional assessment. Coding systems
from previous reviews (Didden et al., 1997; LaGrow & Repp, 1984; Matson et al.,
1996) were adopted so that results could be directly compared to data reported
previously. Target behaviors were coded as either internal maladaptive (e.g., SIB;
stereotypy), external destructive (e.g., aggression; property destruction), or
combinations of internal maladaptive and external destructive. Behavioral
techniques were coded as utilizing either (a) punishment (e.g., overcorrection;
time-out), (b) positive procedures (e.g., DRO; noncontingent reinforcement),
J.M. Campbell / Research in Developmental Disabilities 24 (2003) 120±138
125
(c) combinations of these techniques, or (d) extinction/sensory extinction. The
following experimental variables were coded: number of initial baseline observations, experimental design (e.g., A±B, reversal, multiple baseline), interrater
reliability for behavioral observations, number of ®nal intervention observations,
units of measurement reported for data (e.g., days, sessions) and year of
publication (the coding form used in the review is available from the author
upon request).
Eighteen of 117 articles (15.4%) representing 27 participants (14.9% of the
sample) were selected randomly for independent coding by an advanced graduate
student in Speech and Language Pathology, who had experience working with
individuals with autism, and interrater agreement was calculated. Pearson product±moment correlations were calculated for equivalence of continuous variables, such as age of participant. For categorical variables, reliability was
calculated across each coded variable (e.g., behavioral technique used, treatment
setting) by the percent agreement method (number of agreements/
number of agreements ‡ number of disagreements† 100). As shown in
Table 1, interrater agreement for continuous variables was generally quite good
with the exception of agreement for length of treatment per session. Percentage of
agreement exceeded 80% for all categorical variables, a widely reported standard
for acceptable agreement (e.g., Lundervold & Bourland, 1988).
1.4. Statistical analyses
Statistical analyses were conducted at both the article (N ˆ 117) and participant (N ˆ 181) levels. First, the overall ef®cacy of behavior techniques was
examined by contrasting the mean of each effect size with zero (i.e., no treatment
effect) using six, one-sample t-tests (Rosenthal, 1995). Due to the presence of
multiple contrasts, Bonferroni correction was used to determine if effect sizes
differed from zero; therefore, p-values below .0083 (.05/6) were interpreted
as signi®cant. Next, six, two-way ANOVAs were used to test for the main effects
of target behavior and treatment type as well as their interaction for each effect
size. Due to the large number of contrasts involved in the ANOVAs, Bonferroni
correction was used to determine if signi®cant main effects and interactions
were present in these analyses. Therefore, p-values below .0028 (.05/18) were
interpreted as signi®cant. This was followed by a series of hierarchical multiple
regression analyses with target behavior and treatment type entered on the
®rst step and other variables entered on the second step. Variables were grouped
into three sets: (a) participant characteristics (e.g., age, IQ, gender); (b) characteristics of behavioral treatment (e.g., presence of follow-up data, attempt to
generalize, treatment length); and (c) experimental factors (e.g., type of experimental design, number of baseline data points, reliability of observational data).
Each set of variables was used in a separate analysis for each effect size at both
the article and participant level for a total of 18 hierarchical multiple regression
analyses; articles and participants were excluded from analysis via listwise
deletion.
126
J.M. Campbell / Research in Developmental Disabilities 24 (2003) 120±138
Table 1
Inter-rater reliability for continuous and categorical variables
Variable
Pearson’s r
p
Continuous variablesb
Age of subject (n ˆ 27)
Year of publication (n ˆ 18)
No. of baseline points (n ˆ 27)
Extraction of raw datac (n ˆ 18)
Inter-rater reliability (n ˆ 24)
No. of treatment points (n ˆ 27)
Length of follow-up interval (n ˆ 9)
Length of treatment/session (n ˆ 24)d
1.00
1.00
1.00
.998
.973
.950
.909
.703
<.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 Categorical variablese Participant gender Type of experimental design Presence of follow-up data Presence of functional assessment Diagnostic criteria used Category of target behavior Level of mental retardation Parental involvement in treatment Type of intervention Units of measurement reported for data Attempt to generalize Level of verbal ability Percent agreementa 100.0 100.0 100.0 100.0 96.3 96.3 92.6 92.6 92.6 92.6 88.9 85.2 a Percent agreement based upon number of agreements divided by number of agreements plus number of disagreements multiplied by 100. b Number of comparisons reported in parentheses beside variable name; Pearson's correlations are two-tailed. c Correlation represents an average of 18 studies weighted by the number of observations per study (15.4% of total number of articles). d Agreement was lowered primarily due to two disagreements regardin ... Purchase answer to see full attachment

How it works

1. Paste your instructions in the instructions box. You can also attach an instructions file
2. Select the writer category, deadline, education level and review the instructions 
3. Make a payment for the order to be assignment to a writer
4.  Download the paper after the writer uploads it 

Will the writer plagiarize my essay?

You will get a plagiarism-free paper and you can get an originality report upon request.

Is this service safe?

All the personal information is confidential and we have 100% safe payment methods. We also guarantee good grades