Students With Emotional–Behavioral Disorders as Cross-Age Tutors: A Synthesis of the Literature

Abstract

The purpose of this quantitative synthesis was to evaluate the effectiveness and related outcomes of the cross-age tutoring model when students with or at risk for emotional–behavioral disorders (EBD) serve as tutors. Research questions were posed to identify the shared and unique components (e.g., dosage, tutor training) of the cross-age tutoring model; the extent to which students with EBD can effectively serve as cross-age tutors (i.e., fidelity of implementation and tutees’ improvement); the extent to which the model was effective in promoting desired academic and/or social–emotional–behavioral outcomes for tutees and tutors with EBD; the generalization, maintenance, and social validity of the effects; and the overall methodological quality and rigor of the included studies. Findings showed common training and instructional components across interventions and that tutors with EBD can implement cross-age tutoring procedures with fidelity. The cross-age model was shown to be effective in promoting academic and social–behavioral skills for the tutees as well as the tutors. Evidence for effectiveness in improving self-concept and attitude of the tutor with EBD was inconsistent. Implications and future research considerations are discussed.

Keywords

emotional–behavioral disorders cross-age tutoring peer-mediated disabilities

Academic and behavioral needs of students with emotional–behavioral disorders (EBD) have been identified as some of the most difficult to address (Kern, 2015). However, these needs can be met through the use of effective academic planning and thoughtful selection of instructional techniques (Hughes & Fredrick, 2006). One such instructional technique, known as cross-age tutoring (i.e., an older student tutoring a younger student), shows evidence of being an effective model for teaching academic and social skills to students with disabilities, including students with EBD (Okilwa & Shelby, 2010; Spencer, 2006; Spencer, Simpson, & Oatis, 2009). In addition, this instructional technique requires minimal costs (i.e., time and materials) and can be implemented without substantial training time (Heron, Welsch, & Goddard, 2003). Given the demands placed upon special education classrooms for instructional techniques that are practical, low- to no-cost, and above all, provide effective individualized instruction, utilizing cross-age tutoring may provide a model for addressing the intensive needs of students with disabilities while also providing tutors with EBD opportunities to practice and develop social, behavioral, and academic skills in an instructional context.

Cross-Age Tutoring and Students With EBD

Research focusing on students with challenging behaviors in the role of cross-age tutor has been limited in recent years but has shown positive outcomes for the tutor, as well as for the tutee (i.e., the student receiving instruction from the tutor; Blake, Wang, Cartledge, & Gardner, 2000; Gumpel & Frank, 1999). Improvements in the areas of reading (Cochran, Feng, Cartledge, & Hamilton, 1993; Top & Osguthorpe, 1987), mathematics (Robinson, Schofield, Steers-Wentzell, 2005), spelling (Stowitschek, Hecimovic, Stowitschek, & Shores, 1982), general test scores, and grades (Maher, 1982, 1984) have been found for tutors with EBD. In addition to academic achievement, research on cross-age tutoring models also suggest positive outcomes in social, emotional, and behavioral skills, including discipline within the classroom setting and the reinforcement of peer relationships (Greenwood, Carta, & Hall, 1988; Maher, 1982, 1984), social skills (Blake et al., 2000; Gumpel & Frank, 1999), on-task behavior (Greenwood, Delquadri, & Hall, 1989; S. Hogan & Prater, 1993), self-esteem and self-worth (Lazerson, 2005; D. Miller, Topping, & Thurston, 2010), and attendance rates (Maher, 1982). Given that social–behavioral and academic skills are frequently characterized as deficit areas for individuals with EBD (Landrum, Tankersely, & Kauffman, 2003; Trout, Nordness, Pierce, & Epstein, 2003), utilizing cross-age tutoring shows promise as a possible intervention for addressing these needs.

Existing Reviews

A number of systematic reviews completed within the last few years have focused on both academic outcomes, and less frequently, social–emotional and behavioral outcomes in regard to students with disabilities and peer-mediated interventions. Most recently, Bowman-Perrott, Burke, Zhang, and Zaini (2014) conducted a meta-analysis focusing on direct and collateral effects of peer tutoring on social and behavioral outcomes for students with disabilities. Findings showed peer tutoring had a greater effect on promoting social skills and reducing disruptive behaviors than increasing academic engagement for students with disabilities. Also, cross-age tutoring was found to be more effective than same-age or reciprocal tutoring for students with EBD. Similar findings were obtained through a meta-analysis of tutoring models for literacy instruction, where cross-age tutoring was found to be more effective than adult tutoring and computer-based tutoring, especially when students with disabilities served as tutors (Jun, Ramirez, & Cumming, 2010).

Bowman-Perrott and colleagues (2013) also examined peer-tutoring effects on academic skills in a meta-analysis. Findings of this review showed the model to be highly effective and that students with EBD obtained greater benefit from the model than other disability types. Ryan, Reid, and Epstein (2004) also focused their review on the academic achievement of peer-mediated interventions for students with EBD. Overall findings of the synthesis suggest that peer-mediated interventions appear to be effective across content areas for students with EBD.

A review focused on students with EBD within cross-age and same-age peer-tutoring models found that the cross-age tutoring model to be more effective than both the same-age and reciprocal tutoring in reading but less effective than the same-age tutoring model for mathematics (Spencer, 2006). It should be noted that only 13 studies provided sufficient data to calculate effect sizes. Spencer and colleagues (2009) continued the previous review by identifying nine additional studies that included students with EBD in tutor and tutee roles within peer-tutoring models. The authors noted that although peer tutoring continues to show promise as an effective intervention for students with EBD as tutors or tutees, additional research is required for these students in secondary and generalized settings.

The number of studies containing students with EBD as cross-age tutors was limited in previous reviews (Spencer, 2006; Spencer et al., 2009). Therefore, an expanded search is required to more comprehensively evaluate the effectiveness of this tutoring model for students with challenging behaviors. Considerations for research in this area, proposed by Greenwood et al. (1988), include identifying strategies used by students with disabilities as tutors that are sufficiently developed and validated, evaluating the fidelity of tutoring interventions, comparing procedures or materials with other conditions, and identifying potential areas for standardizing tutoring models for students with disabilities. Overall, the existing reviews routinely focused on the general outcomes of peer tutoring with students with disabilities and rarely addressed the underlying, functional components of a given peer-tutoring model such as cross-age tutoring.

Therefore, the purpose of this synthesis was to examine the cross-age tutoring model components with students with EBD serving as tutors. In addition, this review identified the shared, key components of the model (i.e., tutor training and implementation of tutoring sessions), and the extent to which fidelity, maintenance, generalization, and social validity were measured across included studies. Thus, this synthesis of the literature addressed the following research questions

Research Question 1: What were the components (e.g., dosage, tutor training, instructional components) of cross-age tutoring models with students with EBD as tutors?

Research Question 2: To what extent was the cross-age tutoring model effective in promoting desired outcomes for tutees, and/or tutors with EBD?

Research Question 3: To what extent were students with EBD effective in the role of tutor (i.e., fidelity of implementation and improvement of tutees)?

Research Question 4: To what extent were generalization, maintenance, and social validity measured within cross-age tutoring studies with tutors with EBD?

Research Question 5: What was the overall methodological quality and rigor of the included studies?

Method

Search Procedures and Inclusion/Exclusion Criteria

The procedures for this synthesis were designed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; http://www.prisma-statement.org) statement. PRISMA is a process of directing and reporting systematic reviews and meta-analyses that has been mutually agreed upon by an international group of health care researchers. For the purpose of this synthesis, cross-age tutoring was defined as a school-age tutor (i.e., nonadult) who is 1 or more years older and/or in a higher grade than the student (i.e., tutee) to whom they are providing instruction.

To be included in this systematic review, the study had to utilize single-case, experimental, or quasi-experimental designs and meet the following inclusion criteria: (a) cross-age peer-tutoring model was used to deliver instruction/intervention, (b) the tutor was at least 1 year older or enrolled in a higher grade than the tutee, (c) the tutor was identified with, or at risk for developing an EBD, and (d) at least one outcome (e.g., academic, social, behavioral) was measured for the tutee(s) and/or tutor(s), including fidelity of implementation of tutoring procedures. Studies that took place outside of school or school-like settings (e.g., after-school programs), involved adult tutors (S. R. Miller, 1995), or implemented a tutoring model that was not face-to-face (e.g., online; Smet, Keer, Wever, & Valcke, 2010) were excluded. In addition, studies utilizing solely qualitative, anecdotal, or descriptive methods for identifying peer tutor outcomes were excluded, along with reviews and position papers. To account for publication bias, where studies yielding more favorable (statistically significant) results tend to be published, it was decided that there would be no restriction on the year of publication, and dissertations and unpublished manuscripts would also be considered when full-text was available (Dwan, Gamble, Williamson, & Kirkham, 2013; Pigott, Valentine, Polanin, Williams, & Canada, 2013). The search was limited to papers produced in the English language.

A search of the literature was conducted to identify the relevant studies utilizing six electronic databases: Academic Search Complete, Education Source, Education Resources Information Center (ERIC), Professional Development Collection, PsycINFO, and ProQuest Dissertations & Theses Global. The following terms were entered into all six databases: “cross-age* OR mixed-age* OR coach*,” “emotion* OR behavi* OR emotional-behavioral OR EBD OR social*,” “disab* OR disorder*,” and “tutor* OR support* OR mediat*.” In addition, a hand search of the past 4 years of publications (i.e., 2013–2016) was conducted for five related journals (i.e., Behavioral Disorders, Behavior Modification, Beyond Behavior, Emotional and Behavioural Difficulties, and Journal of Emotional and Behavioral Disorders). Details pertaining to the method of literature search and inclusion/exclusion of studies are shown in Figure 1 (adapted from Moher, Liberati, Tetzlaff, Altman, & The PRISMA Group, 2009).

Figure 1.

Inclusion flow diagram illustrating the results of the literature search and inclusion process.

Screening and Coding Procedures

Prior to screening, two graduate research assistants with backgrounds in special education were trained on the screening and coding manual developed by the lead author in a 90-min training session and practice articles were coded until 100% consensus was met. At the end of the session, 5 practice articles (i.e., 2 single-case design, 3 group design) were screened and coded independently, according to the manual procedures. For screening and coding procedures, interrater reliability (IRR) was calculated by summing the number of agreements and then dividing by the total number of agreements plus disagreements. IRR on training articles was found to be 100% for screening and 93% for coding categories. Two researchers then independently screened the titles and abstracts of the identified papers within the initial pool to assess if the given paper utilized a peer-mediated intervention and contained participants with or at risk for EBD (IRR = 98.3% for journal articles, 95.9% for dissertations). Disagreements were discussed until a consensus was reached. This step yielded 38 journal studies and seven dissertations for potential inclusion (n = 45 studies). The two researchers then screened the reference list titles of the included studies up to this point. They read and discussed the potential papers’ abstracts until a consensus was made to include or exclude. Finally, the full-texts of the papers were screened based on the inclusion criteria. Agreement during full-text screening was defined as both researchers approving that a paper should be included (i.e., all inclusion criteria were met) or excluded (i.e., one or more inclusion criteria were not met). If disagreements were found within a coding category, meetings were held to review the category information until a consensus was reached. The final pool included 15 papers that met the qualifications for inclusion in this synthesis (IRR = 94%; published = 11, dissertation/manuscript = 4).

Each of the included papers was then double-coded by the trained researchers according to the coding manual procedures. Descriptive information and page numbers were also noted for items that were not captured by the coding scheme (i.e., coded as “other”) and were later discussed and added to a mutually agreed upon coding category for added specificity. The following categories were used in coding the characteristics, quality, and outcomes of each study:

Design

Designs were coded as experimental (i.e., random assignment to conditions), quasi-experimental, or single-case, and descriptive information was recorded for specific design components (e.g., single group, number of control/comparison groups) as well as descriptions of how students, teachers, and/or classrooms were randomly assigned, if applicable.

Setting

Placement and treatment settings were both coded when reported. Placement setting refers to the school type and geographic location where the students attended. Geographic location was recorded as descriptive information as the authors reported it (e.g., Brooklyn, New York; central Texas). The description of the community where the school was located was also coded (urban, suburban, rural, not specified). The coding checklist of school types included public, charter, lab/university, residential, clinic, hospital, other, and not reported. Treatment settings refer to the location within the school where the tutoring sessions took place, and options within this category included general education classroom, special education classroom, hallway, office, observation room, hospital room, other, or not reported. Descriptive information was also recorded for added specificity. For example, if the setting was a special education classroom, and the authors reported it to be a resource or self-contained classroom, this description was recorded as the authors reported it.

Implementer

Implementer refers to the individual(s) providing tutor training and/or supervision of tutoring sessions. Codes included researcher, research assistant, lead teacher (specify special or general educator), paraprofessional, related school staff (e.g., school counselor, school psychologist, social worker), other, or not reported. Descriptive information was recorded if provided (e.g., number of years/experience, educational background).

Participants

Demographics for both the tutor and tutee populations were accounted for when reported. Each of the following categories was coded as reported/not reported as well as quantity/descriptive data for each population (i.e., tutors and tutees): number of participants, grade level, age, gender, race/ethnicity, socioeconomic status, language status, and disability label. If more than one grade level/age/disability label was present, all were recorded. Disability categories included emotional disturbance/EBD, specific learning disability, intellectual disability (mental retardation in older studies), autism, visual impairment, deaf-blindness, deafness, hearing impairment, multiple disabilities, orthopedic impairment, traumatic brain injury, at risk (difficulties/challenges as reported), no disability/general education/typically developing, other health impairment (OHI), other, or not specified. In addition, to add further specificity, descriptive information related to qualification criteria utilized in determining disability risk status/categorization was recorded as reported by the authors.

Tutor training

Tutor training refers to any researcher or practitioner provided instruction or practice opportunities for tutors prior to the implementation of tutoring sessions with their tutee. Frequency and duration of individual tutor training sessions was recorded when reported. In addition, the total duration (minutes) of tutor training was calculated when possible. Components of tutor training were defined as any instructional method or practice utilized to teach the tutors the procedures or strategies they would use within the tutoring sessions. The coding checklist contained the following categories: curriculum name (scripted/unscripted), teacher-developed lessons (scripted/unscripted), researcher-developed lessons (scripted/unscripted), purpose of training/introduction, greeting strategies, modeling, prompting/redirection, positive reinforcement, role-playing, lesson planning, evaluation/assessment/progress monitoring training, goal setting, problem solving/discussions, performance feedback from trainer, planning time, materials/manipulatives, review sessions, other; descriptive information for each, other (descriptive information recorded), or not reported.

Tutoring sessions

Tutoring sessions refer to the meetings where the tutor provided instruction to their tutee(s). Frequency and duration of individual tutoring sessions as well as total duration throughout the intervention was calculated when the necessary information was reported. When information was provided on how time was allocated within individual tutoring sessions, the disaggregated instructional (i.e., tutoring) time was calculated (i.e., does not include time for administration of measures). Components of tutor sessions were defined as any instructional method or practice utilized by the tutor to teach the tutee(s) the target skill or content. Descriptive information was recorded for future specification of coding categories for the following tutoring session checklist items: curriculum name (scripted/unscripted), teacher-developed lessons (scripted/unscripted), researcher-developed lessons (scripted/unscripted), purpose of tutoring/reviewing goals, introduction/greeting strategies, modeling, prompting/redirection, corrective feedback, positive reinforcement, role-playing, evaluation/assessment/progress monitoring, goal setting, problem solving/discussions, performance feedback from tutor, planning time, instructional materials, manipulatives, review sessions, reward/reinforcement system (type), tutor retraining/follow-up sessions, other (descriptive information recorded), or not reported.

Instructional focus or target skill(s)

The content area and/or social/emotional/behavioral skill targeted for instruction or skill promotion within the tutoring sessions was coded for both tutees and tutors, when applicable. Academic content area was coded as basic reading skills (e.g., decoding), reading comprehension, written expression, early numeracy skills (e.g., counting), mathematics (e.g., calculation, reasoning), history/social studies, science, other academic area, or not specified. Nonacademic skills were recorded based on authors’ operational definitions of target behaviors.

Dependent measures

Names and descriptions were coded for dependent measures used to evaluate outcomes for both tutor and tutee populations. Response categories included standardized, teacher-developed curriculum-based measures (CBMs), researcher-developed CBMs, quiz/test grades, report card grades, other academic measure, attendance, observation of operationally defined target behavior(s), researcher-developed social/behavioral rating scale, self-concept/esteem scale, other social–emotional–behavioral measure, or not reported. Measures used to evaluate the maintenance and/or generalization of effect of the intervention was assessed using the same coding checklist.

Tutors’ fidelity of implementation

Fidelity was coded if information was reported regarding the extent to which the tutors implemented the instructional procedures specified by the researcher for use within the tutoring sessions. Evidence of the type of measures used, range, and mean fidelity scores were recorded when reported.

Social validity

Social validity was defined as any measure of consumer satisfaction from a tutor, tutee, teacher, and/or parent. Social validity was coded when measured for any of the participant populations. Participant codes included tutor(s), tutee(s), supervisor(s) of tutoring program/sessions (e.g., teacher(s), paraprofessional(s), related staff), and/or parent(s). Descriptive information was recorded for all reported outcomes.

Initial IRR was found to be 93.4% for coding categories containing response options. Disagreements were found most frequently in identifying the primary, targeted skill/behavior for the tutors, as there were occasionally varying behaviors/social skills measured across tutors within the same study. Coding disagreements and descriptive data recorded under the “other” code was discussed, agreed upon, and categorized by the coders prior to analysis. Each study was also assessed to determine the extent to which it met quality indicators for the design based upon categories outlined by Gersten et al. (2005) and Horner et al. (2005).

Quality indicators

Quality indicators for single-case designs were examined for multiple components within the following categories: participant description/characteristics (e.g., gender, disability, diagnosis) and selection process/criteria described in sufficient detail to allow replication of process to obtain participants with similar characteristics; physical features/location of the setting described with enough detail for replication; dependent variable defined (operationally defined, countable index provided, evidence of validity/reliability, frequency of measurement, interobserver agreement measured/established); independent variable (operationally defined, systematically manipulated, evidence of fidelity of implementation); baseline (conditions operationally defined, stable data trend); experimental control/internal validity (a minimum of three demonstrations of effect at three different times, controlled for threats to validity, demonstrated experimental control); external validity (effects replicated across participants, settings, or materials); and social validity (provided social importance of dependent variable, magnitude in change, practicality of intervention, cost-effectiveness, and/or practitioner implementation).

Quality indicators for group designs were examined for components within the following categories: participants description/characteristics and selection process described with sufficient detail to allow replication of process to obtain a sample with similar characteristics (e.g., age, grade, disability/risk status); comparable population characteristics present across groups/conditions; differential attrition reported; setting described in enough detail for replication; dependent variable defined (operationally defined, aligned with intervention, evidence of validity/reliability, frequency of administration, interscorer agreement measured/established, multiple measures used and/or administered at different times, data collectors are blind/unfamiliar to conditions/participants); independent variable (operationally defined, comparison conditions described, fidelity of implementation measured/established); and data analysis (methods chosen are aligned to research questions, variability is accounted for, power analysis is provided). Quality indicator categories were scored as 1 (met standard without reservations; reported sufficient information for replication/outcome), 0.5 (met standard with reservations; met the minimum requirements for categories for replication/outcome), or 0 (did not meet standards; information provided was not adequate for replication/outcome; met less than half of quality indicators within a category; Institute of Education Sciences, 2014). An overall quality score was calculated for each study by dividing the score obtained by the total possible points and multiplying by 100%. A trained graduate research assistant in special education who had taken a course in quality indicators and experimental design assisted in assessing reliability. Reliability was calculated for 30% of the studies and interrater agreement was found to be above 90%.

Analysis of Outcomes

Percentage of data points exceeding the median of the baseline phase (PEM) approach was chosen to assess effectiveness in single-case design studies due to its assumed validity in assessing disruptive behaviors (Chen & Ma, 2007; H. -H. Ma, 2006), a frequently targeted skill within the single-case studies included in this synthesis. PEM’s design does not rely on the most extreme datum point and therefore is recommended for use in instances where significant outliers may be identified within the baseline data or there is some variability over time. Furthermore, when floor or ceiling data points are present, PEM is still capable of reflecting effect size and has shown utility in meta-analysis of single-case research (Y. Ma, 2009; Preston & Carter, 2009). PEM is calculated by identifying the median baseline point and drawing a median line from that point through intervention phases. The percentage of data points above or below the median line is calculated by summing all intervention data points above or below the line, depending on targeted skill or measure (e.g., increasing an academic skill, or decreasing a behavior), and dividing that sum by the total number of data points in the intervention phase. PEM results were interpreted using the following scale: 90%–100% = large or highly effective, 70%–90% = moderately effective, and < 70% = small or questionable effectiveness (H. -H. Ma, 2006).

Cohen’s d was calculated for experimental and quasi-experimental, between group designs (Cohen, 1988). When raw data were not reported or were missing, t-scores were provided in place. Cohen’s (1988) criterion was utilized for grading effect size values (i.e., .20 is small, .50–.79 is moderate, and >.80 is large). For studies containing a single group, pretest/posttest design, the mean of the preintervention assessment was subtracted from the mean of the postintervention assessment, and the result was divided by the standard deviation of the preintervention.

Results

The search procedures and inclusion criteria resulted in 15 studies being identified for this synthesis. A summary of the study characteristics is shown in Table 1. All but four of the included papers were published in peer-reviewed journals, with three of the studies being doctoral dissertations (Hamelberg, 1987; Harrigan, 1994; Holecek, 2012), and one study unpublished at the time of this review (Watts & Bryant, 2017). The years of publication across the 15 studies ranged from 1972 to 2017. Group design was the most common study design (n = 9) and included treatment-comparison (n = 6) and pretest/posttest, single group methodologies (n = 3). Single-case designs were represented less frequently (n = 6). A multiple-baseline design was employed in all studies utilizing single-case methodology.

Table 1.

Study Characteristics.

Study & research design	Setting, implementer (I), & quality score (Q)	Tutors (T)/tutees (t)	Tutor training^a	Tutoring sessions^b	Dependent measure: T/t target skills	Outcomes (ES)	Maintenance (Ma), generalization (G), & social validity (SV)
Blake, Wang, Cartledge, and Gardner (2000) [Experiment 2], multiple baseline	Public, suburban self-contained special education school I = researcher Q = 90.9%	T: n = 3, ages 11–13 years, Grades 6 & 7, challenging behaviors t: n = 3, ages 9–10 years, Grades 3 & 4, difficulties relating to peers	5 sessions, 30 min each; scripted social skills lessons & materials, practice, feedback	5 sessions/week, 45 min, 7 weeks; WTSSC: Tutor modeling, practice, role-playing; play informal games, feedback F = Yes (97%)	T & t: Frequency count Verbal/nonverbal abusive behaviors Verbal/nonverbal supportive behaviors	T: (a) PEM = 93.5, large; (b) PEM = 95.5; large t: (a) PEM = 93.5, large; (b) PEM = 97.9, large	T: Ma (a) PEM = 100, large (b) PEM = 100; large t: Ma (a) PEM = 100, large, (b) PEM = 100, large SV = Yes, T: +/+ (self/program) t: + (program) Te: + (tutors) P: + (tutors)
Cochran, Feng, Cartledge, and Hamilton (1993), treatment/comparison, quasi	Large, urban self-contained elementary school I = researcher Q = 95.5%	T: n^T = 4 /n^C = 4, Grade 5, 8 boys, with/at risk for EBD/struggling readers t: n^T = 4, n^C = 4, Grade 2, 8 boys, with/at risk for EBD/struggling readers	5 sessions, NR min, 5 days; instructions, modeling, role-playing, reinforcement	32 sessions, 28–30 min each, 8 weeks; tutor review, instruction, practice, reinforcement, review games, testing & charting of progress; verbal praise & tangible reinforcers provided to tutors by researchers F = NR	T & t: (a) CBM-sight words; (b) SSRS-S; (c) SSRS-T c1—social skills c2—problem behaviors c3—academic Direct observation of tutoring dyads (d) cooperative statements, (e) put-downs	T: (a) T > C, d = .28, small; (b) T = C, d = .05, NE; (c1) T > C, d = 1.1, large; (c2) T > C, d = .56, moderate; (c3) T > C; d = .54, moderate t: (a) t = C, d = .01, NE; (b) t > C, d = .26, moderate; (c1) t = C, d = .18, NE; (c2) t = C, d = −.19, NE; (c3) t < C, d = −.47, NE; (d) M change = +13.94 (e) M change = −1.4	Ma/G = NR SV = Yes, T: +/+ (self/program) t: +/+ (self/program) Te: +/+/+ (T/t/program)
Gumpel and Frank (1999), multiple baseline	Elementary school I = researcher Q = 84.1%	T: n = 2, ages 11–12 years, Grade 6, 2 boys, at risk for EBD, socially rejected & isolated t: n = 2, age 5 years, Grade K, at risk for EBD, socially rejected & isolated	6–8 sessions, 45 min each, 1.5 weeks; training on MSC procedures: Identify appropriate social stimuli, state possible behavioral options, enter into that social interaction, self-monitor, & observe environmental reaction	4 times/week; MSC procedures: Modeling, role-playing, & discussion of tutee’s self-monitoring sheet F = NR	T & t: Momentary time sampling (a) No social interaction (b) Positive social interaction	T: (a) PEM = 97.5, large; (b) PEM = 100; large t: (a) PEM = 100, large; (b) PEM = 100, large	T: Ma (a) PEM = 100, large; (b) PEM = 100, large t: Ma (a) PEM = 100, large; (b) PEM = 91.7, large SV = Yes, Te: + (program)
Hamelberg (1987)^c, multiple baseline	Public, suburban elementary school in central Ohio, resource & computer room I = researcher Q = 86.4%	T: n = 3, ages 16–18 years, males, EBD t: n = 3, ages 11–12 years, males, LD	4 sessions, 60 min each; goals/objectives, instructional steps, data collection, role-playing, review	29 possible sessions, 20 min each, 10 weeks; sight word acquisition; sight word flashcards, games, word sheets, pictorial charts, progress monitoring F = Yes (95.5%)	t: (a) Sight words read aloud T: (b) PHSC self-concept percentiles	t: (a) PEM = 88.2, medium T: (b) d = 1.6; large	t: Ma (a) PEM = 85.2, medium SV = Yes, T: +/+ (self/program) Te: mixed (tutors)
Harrigan (1994)^c, treatment/comparison, RCT	Private, nonprofit school for children with EBD & neurological impairments in New Jersey I = NR Q = 70.5%	T: n^T1 = 10/n^T2 = 10, ages 13–16 years, 16 boys, 4 girls, EBD & neurological impairments t: n^T = 20 /n^C = 10, ages 9–14 years, 26 boys, 4 girls, EBD & neurological impairments	1 session, 30 min both groups; T1: received 2 additional 30-min sessions in extrinsic motivational techniques (e.g., positive verbal reinforcement) & token reinforcement T2: received 1 additional 30-min session in spelling method based on deCharms (1972) C: BAU	4 times/week, 20–30 min, 6 weeks; Tr1 students received verbal praise & daily bonus points Tr2 included intrinsic motivation (self-selected words, goal setting) F = NR	t: CBM spelling word list	t: Tr1 < C, d = –.31, NE; Tr2 > C, d = .25, small	Ma/G = NR SV = NR
S. Hogan and Prater (1993), multiple baseline	Public, rural high school in U.S. Midwest I = trainer Q = 70.5%	T: n =1, age 15 years, boy, EBD t: n =1, age 14 years, boy, LD	3 days, min NR; individual interview, provided procedures, modeling, role-playing, & behavior self-monitoring training	4 days/week, 15 min; practice, error correction, modeling, positive reinforcement, assessment, tutors record progress monitoring data F = NR	T: (a) Frequency count disruptive behaviors t: (b) Time sampling on-task CBM: (c) spelling test scores, (d) vocabulary test scores	T: (a) PEM = 42.9, small t: (b) PEM = 100, large; (c) PEM = 100, large; (d) PEM = 100, large	t: Ma/G (b) PEM = 100, large; (c) PEM = 100, large; (d) PEM = 100, large SV = NR
Holecek (2012)^c, pre/post test, single group	Middle/high school in the Midwest I = NR Q = 25.0%	T: n = 3, Grade 1 or more above tutees,’ EBD t: n = 3, Grades 1–5, ID	NR min, 1 week; greeting, engagement, prompting, redirecting, & positive reinforcement strategies	2 days/week, 30 min, 6 weeks; content from tutees’ math class F = NR	T: (a) CBM basic math skills, (b) observation of time engaged during math class	T: (a) d = 1.0, large	T: G (b) d = 2.8, large SV = NR
Lane, Pollack, and Sher (1972), pre/post test, single group	Private, urban middle school for students with EBD, Brooklyn, NY I = research asst. Q = 52.3%	T: n = 8, Grades 8–9, EBD & struggling readers t: n = 8, Grades 3–4, 8 boys, EBD & LD	NR; IRM training; modeling, role-playing, positive reinforcement strategies	2 sessions/week, 7 months; IRM: phonic-linguistic instructional method for reading, writing, & spelling; tutors trained in social skills every other week F = NR	T & t: (a) MAT reading achievement T: (b) BBRS disruptive behaviors	T: (a) M gain (in months) = 19; (b) d = 3.4, large t: (a) M gain (in months) = 14	Ma/G = NR SV = NR
Lazerson (1980), treatment/comparison, quasi	Lab school, university campus, Buffalo, NY I = researcher Q = 47.7%	T: n^T = 20 /n^C = 20, Grades 5–8, aggressive & withdrawn t: n = 20, Grades 2–5, aggressive & withdrawn	2 sessions, NR min; T: Procedures: modeling, role-playing, positive reinforcement, & correction methods C: BAU	5 sessions/week, 20–30 min, 5 weeks; tutors had free manipulation of content & format within each session F = NR	T & t: (a) DESB behaviors, (b) LSSC self-concept	T & t: (a) T/t > C, t = 13.7 (p < .005, df = 38), SS; (b) T/t > C, t =11.29 (p < .005, df = 38), SS	Ma/G = NR SV = Yes, Te: mixed/mixed (tutors/program)
Lazerson (2005), pre/post test, single group	Public school, western New York, self-contained classroom I = NR Q = 27.3%	T: n = 3, ages: 15–16 years, 2 boys, 1 girl, EBD & LD t: n = NR, Grades 1–5, students with IEPs	2 sessions, NR min; practice/role-playing corrective reinforcement statements	2–5 sessions/week, 45–60 min, 3 months; instruction on basic reading comprehension/decoding & math skills, use of math manipulatives (e.g., blocks, sticks) F = NR	T: LSSC self-concept discrepancies	T: d = .23; moderate	Ma/G = NR SV = Yes, T: +/+ (self/program); Te: + (tutors)
Maher (1982), treatment/comparison, RCT	Public urban high school, New Jersey I = School psychologist & SPED teacher Q = 75.0%	T: n^T1 = 6, n^C1 = 6, n^C2 = 6, ages 15–18 years, 11 boys, 7 girls, EBD t: n = 6, ages 8–10 years, LD and/or ID	1 session, NR min; T: Instruction & practice on lesson planning, intervention, & setting/evaluating instructional goals ^C1Received peer tutoring ^C2Received group counseling	2 sessions/week, 30 min each, 10 weeks; 1 planning session/week (15–20 min) with SPED teacher, planned & provided instruction on reading & writing content/skills	T: (a) grades language arts, (b) grades social studies, (c)discipline referrals, (d) % of days in attendance	T: (a) T > C, F(4, 30) = 6.41 (p < .01), SS; (b) T > C, F(4, 30) = 5.06, (p < .05); SS; (c) T > C, F(4, 30) = 8.90 (p < .01), SS; (d) T > C, F(4, 30) = 20.50 (p < .01); SS	T: Ma (b) T > C, F(2, 45) = 11.12 (p < .01), SS G = NR SV = Yes, Te: + (tutors)
Maher (1984), multiple baseline	Public urban school district, New Jersey I = School psychologist & SPED teacher Q = 86.4%	T: n = 8, ages 14–16 years, EBD t: n = 8, ages 9–12 years, LD and/or ID	3 sessions; purpose/goals, tutee background information, discussions about problems/ concerns, role-playing of planning meetings with SPED teachers, role-playing of tutor sessions & performance feedback	2 sessions/week, 30 min each, 10 weeks; 1 planning session/week (15–20 min) with SPED teacher; present reading/math/language art skill(s), complete worksheets, provide feedback & reinforcement; follow-up tutor support conferences	T & t: (a) % of assignments completed, (b) test & quiz scores T: (c) disciplinary referrals	T: (a) PEM = 94.7, large; (b) PEM = 88.3, large; (c) PEM = 100, large t: (a) PEM = 93.8, large; (b) PEM = 88.3, large	T: Ma (a) PEM = 93.1, large; (b) PEM = 85.1, moderate; (c) PEM = 100, large t: Ma (a) PEM = 90.2, large; (b) PEM = 82.1, moderate SV = Yes, T: +/+ (self/ program) Te: +/+ (t/program)
Scruggs and Osguthorpe (1986) [Experiment 1], Treatment/comparison, quasi	Public, rural elementary schools in the west I = Project staff Q = 65.9%	T: n^T = 13, n^C = 20, Grades 2–6, LD & EBD t: n^T = 14, n^C = 20, Grades 1–5, LD & EBD	1–2 sessions T: Tutoring session procedures, modeling, role-playing, proximity to tutee, positive reinforcement, correction, & progress monitoring methods C: BAU	2–5 days/week, 30 min, 10 weeks; beginning reading curriculum (Harrison, 1979), directly supervised by project staff F = NR	T & t: reading skills, (a) WJ, (b) BR-I, (c) attitude toward school	T & t: (a) T & t = C, NR, NS; (b) T & t > C, t(33) = 2.46 (p < .05), SS t: (c) t > C, t(14) = 2.08 (p < .05), SS T: (c) T = C, t(13) < 1, NS	Ma/G = NR SV = NR
Top and Osguthorpe (1987), Treatment/comparison, RCT	Large, suburban elementary school district I = teaching assistant Q = 72.7%	T: n^T = 39, n^C = 39, Grades 4–6, LD & EBD t: n^T = 37, n^C = 39, Grade 1, at risk for reading LD	3 sessions, 60 min each; T: Procedures (Osguthorpe, 1984): modeling, prompting, positive reinforcement & progress monitoring methods C: BAU + additional instructional time to equal treatment group	4 days/week, 15–20 min, 14 weeks; instruction on reading skills (phonics/ sight words); paraprofessional recorded tutees’ daily performance F = NR	T & t: (a) WJ T: Self-concept, (b) PHSC, (c) SPAS, (d) ISC t: Reading skills, (e) BR-I	T: (a) T > C, d = .94, large; (b) T = C, d = .15, NE; (c) T > C, d = .45, small; (d) T = C, d = .00, NE t: (a) t = C, d = .16, NE; (e) t > C, d = .86, large	Ma/G = NR SV = NR
Watts and Bryant (2017)^c, Multiple baseline	Charter elementary school, Texas, resource room I = Researcher Q = 90.9%	T: n = 3, Grades 4–5, ages 9–11 years, with & at risk for EBD t: n = 3, Grades K–1, ages 5–7 years, with & at risk for math LD	2 sessions, 45 min each; number line board game procedures, modeling, role-playing, corrective feedback, positive reinforcement strategies, progress monitoring; performance feedback, & review	3 sessions/week, 25 min each, 12 weeks; number line board games, modeling, corrective feedback, positive reinforcement, & progress monitoring; retraining sessions provided to tutors when fidelity dropped below 80% F = Yes (88%)	T: (a) partial interval, positive interactions/ behaviors t: number sense, (b) TEMI, (c) TEMA-3	T: (a) PEM = 96.3, large t: (b) PEM = 91.7, large; (c) d = .68, moderate	T: Ma (a) PEM = 100, large t: Ma (b) PEM = 100, large SV = Yes, T: + / + (self/program); t: + / + (tutors/program) Te: +/ + (tutors/program)

Note. WTSSC = Working Together Social Skills Curriculum; F = fidelity; ES = effect size; PEM = percentage of data points exceeding the median; + = overall positive responses; – = overall negative responses; Te = teacher/paraprofessional/school staff/child-care worker; ^T/T = treatment/tutoring group; ^C/C = comparison group; EBD = emotional–behavioral disorder; NR = not reported; CBM = curriculum-based measure; SSRS-S = Social Skills Rating System–Student Form; SSRS-T = Social Skills Rating System-Teacher version; NE = no/negative effect; MSC = Model of Social Competence; LD = learning disability; PHSC = Piers-Harris Children’s Self-Concept Scale; RCT = randomized control trial; Tr1 = treatment group 1; Tr2 = treatment group 2; BAU = business as usual; ID = intellectual/cognitive disability; IRM = Intersensory Reading Method; MAT = Metropolitan Achievement Test; BBRS = Burk’s Behavior Rating Scale; DESB = Devereaux Elementary School Behavior Rating Scale; LSSC = Luszki & Shmuck Self-Concept Scale; SS = statistically significant; IEP = individualized education program; SPED = special education; WJ = Woodcock-Johnson Test of Achievement; BR-I = Beginning Reading Criterion Test; NS = nonstatistically significant; SPAS = Students’ Perception of Ability Scale; ISC = Inferred Self-Concept Scale; TEMI = Texas Early Mathematics Inventories; TEMA-3 = Test of Early Mathematics Ability–Third Edition.

Frequency, duration, components. ^bFrequency, duration, components, tutor supports, fidelity (F). ^cDissertation/unpublished.

Setting, Implementer, and Participants

Of the studies that reported adequate information pertaining to the setting of the intervention, the most common location was in an urban environment (n = 6), with public (n = 5), suburban (n = 3), and rural school districts (n = 2) also being represented. Private, charter, lab, and self-contained special education schools were also used as settings for cross-age tutoring interventions (n = 5). Within these school settings, the most common location for tutoring sessions was in a special education classroom (n = 6).

A researcher was frequently the primary implementer of tutor training and supervision of tutoring sessions (n = 6). It was also reported that research assistants or trained staff undertook these responsibilities as well (n = 3). Practitioners, such as special educators or school psychologists (n = 2), and paraprofessionals (n = 1) were utilized as implementers less frequently. In three studies, this information was not reported (Harrigan, 1994; Holecek, 2012; Lazerson, 2005).

Across the included studies, the number of participants was 436, with the number of students serving in the role of cross-age tutor (N = 126; comparison = 105) ranging from one to 39 per study, and the number students serving in the role of tutee (N = 132; comparison = 73) ranging from one to 37. The grade level of the tutors ranged from second to high school, and the ages ranged from 9 through 18 years, with a majority of the studies including tutors in high school (n = 7) and middle school (n = 4). The tutees’ grade levels ranged from kindergarten to middle school, and ages ranged from 5 to 14 years, with a majority of studies containing tutees in the elementary grades (n = 11).

In all 15 studies, a student with or at risk for EBD fulfilled the role of tutor. Students with EBD, or who had a comorbid label that contained EBD (e.g., learning and behavioral disorder), also commonly served in the role of the tutee in the included studies (n = 8). Equally represented were tutees with or at risk for specific learning disabilities (n = 8); occasionally, this tutee role was filled by students with cognitive or intellectual disabilities (n = 4). The tutors’ specific disability label, or area of challenge, was inconsistently defined: behavior disorder (Cochran et al., 1993; Lane, Pollack, & Sher, 1972) challenging behaviors and difficulties relating to peers (Blake et al., 2000), learning and behavior disorders (Scruggs & Osguthorpe, 1986; Top & Osguthorpe, 1987), aggressive and withdrawn (Lazerson, 1980), and socially rejected and isolated (Gumpel & Frank, 1999).

Quality of Studies

For the nine studies utilizing group designs, the quality scores varied greatly, from 27.3 to 95.5, with the mean across studies being 59.1. Single-case designs proved more rigorous and quality scores were more consistent overall, although it should be noted that the number of single-case studies included was less than the number of group design studies included. Single-case study scores ranged from 70.5 to 90.9, with a mean of 84.9 across studies.

Components of Cross-Age Tutoring

Table 2 summarizes the common and unique components of the cross-age tutoring model across the included studies.

Table 2.

Summary of Common and Unique Components of the Cross-Age Tutoring Models.

Model	Component
Model	Sessions (n)	Average session length (range)	Average training/tutoring program duration (range)	Session frequency	Common features^a	Unique features^a
Tutor training	1–8	45 min (30–60 min)	177.5 min (30–360 min)		• Procedures/objectives (60%)• Modeling (47%)• Role-playing (80%)• Corrective feedback (60%)• Positive reinforcement (60%)	• Greeting tutees (7%)• Proximity to tutee (7%)• Review of skills (7%)• Individual interviews (7%)• Scripted lessons (7%)• Lesson planning (13%)• Goal setting (13%)• Self-monitoring (13%)• Data collection (20%)• Token reinforcement (7%)• Specific behavioral or academic strategies (20%)
Tutoring session		20–30 min(15–60 min)	5 weeks–7 months	2–5/week	• Modeling (54%)• Corrective feedback (54%)• Positive reinforcement (47%)	• Planning sessions with special educators (13%)• Reteaching/retraining session (13%)• Follow-up conferences (7%)• Tangible reinforcers (7%)

Percentage of included studies.

Tutor training

The frequency of tutor training ranged from one to eight sessions, with the duration of individual training sessions ranging from 30 to 60 min. For studies that reported adequate information for determining the total number of minutes provided for tutor training, the durations ranged from 60 min (Harrigan, 1994) to 360 min (Gumpel & Frank, 1999) with the average length of training across studies being 177.5 min (n = 6). The remaining studies reported solely the number of sessions/days of training (n = 7) or no frequency/duration information at all (Holecek, 2012; Lane et al., 1972).

The components of tutor training were relatively consistent across studies. Mutual instructional features of the tutor training sessions (percentage across studies) included instructions/procedures/objectives (60%), role-playing/practice opportunities (80%), performance/corrective feedback techniques (60%), providing positive reinforcement (60%), and modeling (47%). More unique features included training on scripted lessons (Blake et al., 2000), greeting tutees (Holecek, 2012), individual interviews with tutors (S. Hogan & Prater, 1993), lesson planning (Maher, 1982, 1984), goal setting (Maher, 1982, 1984), self-monitoring procedures (Gumpel & Frank, 1999; S. Hogan & Prater, 1993; Scruggs & Osguthorpe, 1986), data collection procedures (Hamelberg, 1987; Top & Osguthorpe, 1987; Watts & Bryant, 2017), token reinforcement (Harrigan, 1994), proximity to tutee (Scruggs & Osguthorpe, 1986), review of previous skills (Hamelberg, 1987), and specific behavioral or academic instructional techniques (Gumpel & Frank, 1999; Harrigan, 1994; Lane et al., 1972).

Tutoring sessions

The frequency of tutoring sessions across all studies ranged from two (Holecek, 2012; Lane et al., 1972; Maher, 1982, 1984) to five sessions per week (Blake et al., 2000; Lazerson, 1980). The length of individual tutoring sessions varied from 15 min (S. Hogan & Prater, 1993; Top & Osguthorpe, 1987) to 60 min in length (Lazerson, 2005), with duration falling in the range of 20 to 30 min in all but five studies. The duration of the intervention was reported in all but two studies. Of those reported, the longest intervention phase was 7 months (Lane et al., 1972) and the shortest was 5 weeks (Lazerson, 1980).

All but three studies selected academic skills as the target for tutoring instruction, while the remaining studies taught social skills or gave the tutors free manipulation of the content materials. When academics were the focus of instruction, the target skills most frequently fell within the domain of reading and/or literacy (n = 10). Mathematics instruction was represented in four of the studies (Holecek, 2012; Lazerson, 2005; Maher, 1984; Watts & Bryant, 2017). The instructional skills taught within tutor training sessions (e.g., modeling, corrective feedback, positive reinforcement) were frequently utilized within tutoring sessions as the primary instructional techniques. Supports for tutors included weekly planning sessions with special educators (Maher, 1982, 1984), performance feedback or follow-up conferences (Maher, 1984), reteaching/retraining sessions (Lane et al., 1972; Watts & Bryant, 2017), and tangible reinforcers provided by researchers (Cochran et al., 1993).

Fidelity of implementation

Fidelity of implementation of tutoring procedures by students with EBD was reported by four studies (Blake et al., 2000; Hamelberg, 1987; Maher, 1984; Watts & Bryant, 2017). Across the studies reporting fidelity, the rates of implementation ranged from 88% to 97% (M = 94.2%). When fidelity was measured, the outcomes for all participants were moderate to large, with effects being maintained in each of the studies.

Effectiveness of Cross-Age Tutoring

Table 3 provides a summary of participant characteristics and common effects for targeted skill categories/content areas. Across studies, the measures used, in order of prevalence, were curriculum based (n = 9), standardized (n= 7), direct observation (n = 6), researcher-developed (n = 2), and school records (n = 2). Of the standardized measures, social–emotional scales, including attitude and self-concept assessments, were the most commonly used (n = 5), followed by behavior and social skill scales (n = 3) and academic tests (n = 3). Studies utilizing group designs contained effect sizes (Cohen’s d) for tutors and tutees ranging from null, nonsignificant effects to large, statistically significant outcomes. Single-case design studies reported effect sizes (PEM) for tutors ranging from 42.9 (S. Hogan & Prater, 1993) to 100 (Blake et al., 2000; Gumpel & Frank, 1999; Maher, 1984; Watts & Bryant, 2017), and for tutees, from 88.2 (Hamelberg, 1987) to > 90 (Blake et al., 2000; Gumpel & Frank, 1999; Maher, 1984; Watts & Bryant, 2017).

Table 3.

Summary of Participant Characteristics, Outcomes, and Common Effect Size Grades.

Group	Participants (n)		Academic overall^a		Reading/literacy^a		Mathematics^a		Social–behavioral^a		Self-concept/attitude^a
Group	Grade range	Disability/at risk category	SCD	Group	SCD	Group	SCD	Group	SCD	Group	SCD	Group
Tutees	• Elementary = 163• Middle school = 41• High school = 1	• LD = 90• Comorbid = 80• EBD = 25• ID = 10	Moderate/large(88.2–100)	Mixed(.01–.86)	Moderate/large(88.2–100)	Mixed(–.3 to .86)	Large(91.7)	Moderate(.68)	Large(93.5–100)	Mixed(–.19 to .26)	NA	Moderate/large(.26–SS)
Tutors	• Elementary = 133• Middle school = 45• High school = 53	• Comorbid (EBD+) = 150• EBD = 81	Large(88.3–94.7)	Large(.28–1.0)	NA	Large(.28–.94)	NA	Large(1.0)	Large(42.9–100)	Large(.56–3.4)	NA	Mixed(.00–1.6)

Note. SCD = single-case design; Group = group design; LD = learning disability; EBD = emotional–behavioral disorder; ID = intellectual disability; SS = statistically significant effect.

Most frequent/common effect size interpretation across included studies (range of effect sizes; SCD = PEM; Group = Cohen’s d).

Academic outcomes

Tutees were most frequently assessed on academic skills and showed gains in three studies (S. Hogan & Prater, 1993; Lane et al., 1972; Maher, 1984), moderate gains in two studies (Hamelberg, 1987; Watts & Bryant, 2017), small gains in one study (Harrigan, 1994), null effects in one study (Cochran et al., 1993), and mixed outcomes in three studies (Harrigan, 1994; Scruggs & Osguthorpe, 1986; Top & Osguthorpe, 1987). When tutors were assessed for academic outcomes, across studies, they showed large gains in five studies (Holecek, 2012; Lane et al., 1972; Maher, 1984, 1982; Top & Osguthorpe, 1987), small gains in one study (Cochran et al., 1993), and mixed outcomes in one study (Scruggs & Osguthorpe, 1986).

Breaking down the effects by instructional content, the most frequently addressed skills fell in the domains of reading, spelling, and language arts. Effects on reading skills ranged greatly from study to study with null effects to significant increases being found for both tutees and tutors (Cochran et al., 1993; Hamelberg, 1987; S. Hogan & Prater, 1993; Lane et al., 1972; Maher, 1982; Scruggs & Osguthorpe, 1986; Top & Osguthorpe, 1987). Spelling outcomes ranged from no effects to small effects in the group design study (d = −.31 to .25; Harrigan, 1994) and large effects were found in the single-case study (PEM = 100; S. Hogan & Prater, 1993). Two studies that focused on basic mathematics and number sense during tutoring sessions showed moderate to large effects for tutees (d = .68; PEM = 91.7; Watts & Bryant, 2017), and also large effects for tutors (d = 1.0; Holecek, 2012).

Social/emotional/behavioral outcomes

Tutors were more frequently assessed in the areas of social skills and behavioral outcomes and showed significant gains in seven studies (Blake et al., 2000; Gumpel & Frank, 1999; Holecek, 2012; Lazerson, 1980; Maher, 1984, 1982; Watts & Bryant, 2017), moderate gains in one study (Cochran et al., 1993), and small gains in one study (S. Hogan & Prater, 1993). Self-concept and attitude measures were regularly administered to this population as well and findings show significant, positive changes in two studies (Hamelberg, 1987; Lazerson, 1980), moderate gains in one study (Lazerson, 2005), null effects in two studies (Cochran et al., 1993; Scruggs & Osguthorpe, 1986), and mixed outcomes in one study (Top & Osguthorpe, 1987). Tutees showed significant differences on an attitude measure compared with a control population (Scruggs & Osguthorpe, 1986) and moderate gains on a self-assessment of their own behavior (Cochran et al., 1993). Tutees were also assessed for behavioral outcomes and showed significant gains in four studies (Blake et al., 2000; Gumpel & Frank, 1999; S. Hogan & Prater, 1993; Lazerson, 1980), and mixed effects in one study (Cochran et al., 1993).

Maintenance, generalization, and social validity

Maintenance was measured inconsistently across the included studies (n = 7), and generalization of targeted skills was measured even less frequently (n = 2). Studies that administered follow-up and distal measures found that effects were readily maintained (Maher, 1982) and/or generalized to other settings/skills (Blake et al., 2000; Gumpel & Frank, 1999; Hamelberg, 1987; S. Hogan & Prater, 1993; Holecek, 2012; Maher, 1982, 1984; Watts & Bryant, 2017).

Nine of the 15 studies measured social validity, whether qualitatively or quantitatively. Measures included participant perceptions of tutee outcomes (n = 2), tutor outcomes (n = 8), tutor self-assessments (n = 6), and overall tutoring program ratings (n = 8). Across studies, almost all social validity measures reported overall positive responses from students, teachers, parents, and trainers regarding perceived outcomes for the participants and the tutoring program in general (n = 8), with the exceptions being Hamelberg (1987) and Lazerson (1980).

Discussion

The purpose of this synthesis was to evaluate the effectiveness and related outcomes of the cross-age tutoring model when students with or at risk for EBD serve as tutors. The findings related to the first research question showed the prevalent use of modeling, role-playing, feedback, and positive reinforcement as common instructional components in both the tutor training sessions and the tutors’ instructional procedures. Considering that these components are also common instructional features within other peer-mediated instructional models (e.g., Peer Assisted Learning Strategies; Fuchs & Fuchs, 2005; classwide peer tutoring; Greenwood et al., 1989), these findings could provide an opportunity for creating more standardized, systematic procedures for the cross-age model with tutors with EBD. Furthermore, although it was noted that the techniques of self-monitoring and self-assessment were used infrequently, the studies that included these components showed consistently large effects in promoting the academic and behavior skills targeted for tutoring instruction (Gumpel & Frank, 1999; S. Hogan & Prater, 1993). These self-management techniques have been shown to be effective strategies for students with EBD (Mooney, Ryan, Uhing, Reid, & Epstein, 2005) and may prove to be beneficial supports within the training and tutoring sessions of this model moving forward.

Results and findings from 11 studies reported intervention phases of 10 weeks or less. Findings from one study (Lane et al., 1972), which had the longest duration of the intervention phase (i.e., 7 months), yielded consistent, positive outcomes across both academic and behavioral measures. This duration finding aligns with previous research findings of positive effects for intensifying intervention dosage and duration to meet the intensive needs of students with disabilities (Bryant et al., 2011; Vaughn et al., 2012). It is suggested that future research is undertaken to determine effective intervention phase durations in regard to tutor training and tutor instructional time, and additionally, utilize more rigorous reporting procedures to determine accurate relations to participant outcomes (Conn & Chan, 2016; Conn & Groves, 2011).

In addressing Research Question 2, results from this review are similar to previous findings, which suggest the cross-age tutoring model can be effective in promoting academic and behavioral outcomes for both the tutee and the tutor (Robinson et al., 2005). Reading and spelling skills, the most frequently assessed academic outcomes, showed varying levels of effectiveness across studies, and also across targeted skills (e.g., fluency, comprehension), while instruction in mathematics showed the most consistent, positive outcomes for participants across the limited number of studies focusing on this content. The findings from this synthesis related to content area differences are comparable with the results from a previous review of peer-mediated instruction for students with EBD, which found larger effects for mathematics skills than reading skills (Ryan et al., 2004). Perhaps the procedural steps for some mathematics skills are conducive to the cross-age tutoring model; however, instruction in conceptual understanding remains with the teacher. Furthermore, additional empirical research on cross-age tutoring with students with EBD in the areas of mathematics, science, writing, and social studies instruction is apparent.

The targeted skills for tutors with EBD were found to be more frequently socially or behaviorally oriented, and typically addressing either targeted negative behaviors or self-concept. In the social or behavioral domain, the cross-age tutoring model proved to be the most effective, as a majority of the studies reported significant gains in tutors’ social and/or behavior skills. For tutees, cross-age tutoring appears to be equally effective in developing desired behavioral and social skills, with consistent decreases in negative behaviors and increases in prosocial skills/behaviors. These findings support this model’s utility in providing opportunities to practice social and behavioral skills in natural, one-on-one interactions with other students in need of behavioral supports. Henceforth, findings related to fidelity of implementation show that students with EBD are able to implement tutoring procedures with a high level of fidelity, and in conjunction with the positive findings of tutee outcomes, the cross-age tutoring model shows promising effectiveness when students with EBD serve as cross-age tutors. This finding aligns with previous research results demonstrating that students with EBD can function effectively as tutors when provided with the appropriate training and supervision (Heron et al., 2003), and with further research, this evidence may support practitioner use of the cross-age tutoring model for providing individualized instruction to younger students as well as practice opportunities for social–behavioral skills for students with EBD.

Findings from this synthesis for generalization and maintenance of target skills demonstrated consistent and lasting impact of tutoring instruction on target skills. The infrequent measurement of these distal outcomes across studies, especially for tutors, indicates the need for future research methodologies designed to directly assess the impact of tutor training and implementation on tutors’ academic, social, and behavioral skills in generalized settings (i.e., outside of the tutoring environment), as these are commonly identified areas of deficit for students with EBD. Furthermore, for tutors, the theory that providing the responsibility of tutoring another student to elicit gains in self-concept and/or self-esteem has mixed support in this review’s findings (Allen, 1976; D. M. Hogan & Tudge, 1999). These results could be related to the sensitivity of the emotional self-assessment measures used in the earlier studies, which contained overly general characteristics (e.g., anger, friendliness, studiousness), making it difficult to determine emotional change directly related to performing as a cross-age tutor (Hamelberg, 1987; Lazerson, 1980; Scruggs & Osguthorpe, 1986; Top & Osguthorpe, 1987). Interestingly, compared with tutors, tutees showed consistent gains in self-concept and positive attitude, which may be related to the uniqueness of the model, where attention, reinforcement, and consequently, motivation, are facilitated by an older student with a disability rather than the typical adult/teacher model (Jun et al., 2010; Topping & Ehly, 1998).

Study Limitations

Findings pertaining to the fifth research question (i.e., What is the overall methodological quality and rigor of the included studies?) must be considered in the interpretation of the results of this synthesis, as the overall quality of the studies included in a systematic review will affect the comprehensiveness of the findings. The majority of the studies reviewed for this synthesis were published before 2005 when recommendations for the indicators for quality in research studies were initially developed (Gersten et al., 2005; Horner et al., 2005). Included studies with low quality scores (i.e., less than 75%) frequently omitted descriptions of any specific training (e.g., amount of training, training to a criterion) or qualifications (e.g., professional credential) required to implement the intervention, reports of reliability (e.g., internal, interobserver, test–retest, parallel-form), and procedures for measuring implementation fidelity. Taking into account the variability in methodological rigor, this review shows promise for cross-age tutoring models utilizing students with EBD as tutors as an evidence-based practice. Henceforth, it is suggested that future research is undertaken in alignment with rigorous quality standards for design and reporting (i.e., Council for Exceptional Children, 2014). Moreover, considering the limited number of studies identified, and that half of these studies were published prior to 1990, it is possible that the findings of this review may not provide an accurate representation of the current population of students with disabilities, and thus may not attain external validity. Similar findings were found in a related review showing a decline in the number of peer-mediated intervention studies for students with EBD in recent years, including the use of the cross-age model (Ryan et al., 2004). Although the number of studies devoted to the cross-age tutoring model containing tutors with EBD is limited, a majority of these studies showed moderate to large academic and/or behavioral effects for participants.

Another limitation is that researchers rather than school personnel were the most common implementers of the tutor training and supervision during the tutoring sessions. Overall social validity outcomes showed practitioners’ positive perceptions of the effectiveness and benefits of the model, and in the two studies where teachers reported mixed perceptions of student outcomes (Hamelberg, 1987; Lazerson, 1980), they still described the tutoring intervention as beneficial and stated that they would continue to utilize the model. These findings are promising, as when teachers and students perceive a tutoring model positively and see it as effective, they are more likely to continue using the practice (Zimmerman & Risemberg, 1997). To determine whether this model, with this unique population of students serving as tutors, is feasible and effective for practitioner implementation, further evaluations need to be undertaken.

By analyzing the findings of 15 cross-age tutoring interventions containing students with EBD as tutors, this synthesis makes a necessary contribution to the field of special education research and the literature base of peer-mediated interventions. Although findings show students with EBD to be able and effective cross-age tutors, and that the instructional model can facilitate positive outcomes for both tutee(s) and the tutor(s), there is still much research to be undertaken. The small number and varying rigor of the included studies provides useful insight into the need for more comprehensive research in determining whether the cross-age instructional model provides unique advantages for tutors with EBD, such as promoting generalized improvements in deficit social, emotional, and/or behavioral areas, in relation to more formalized, same-age peer support systems. Finally, to identify the extent of model’s utility, further empirical research is required to assess effectiveness under practitioner implementation and supervision, and also in providing instruction in content areas other than reading and spelling domains. These future areas of research will deepen our understanding of the potential of the cross-age tutoring model for tutors with EBD and their tutees.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Gavin W. Watts

References

Allen

V. L.

(1976). Children as teachers. New York, NY: Academic Press.

*Blake

Wang

W. Q.

Cartledge

Gardner

(2000). Middle school students with serious emotional disturbance serve as social skills trainers and reinforcers for peers with SED. Behavioral Disorders, 25, 280–298. Retrieved from ed http://www.jstor.org.ezproxy.lib.utexas.edu/stable/43153755

Bowman-Perrott

Burke

M. D.

Zhang

Zaini

(2014). Direct and collateral effects of peer tutoring on social and behavioral outcomes: A meta-analysis of single-case research. School Psychology Review, 43, 260–285.

Bowman-Perrott

Davis

Vannest

Williams

Greenwood

Parker

(2013). Academic benefits of peer tutoring: A meta-analytic review of single-case research. School Psychology Review, 42, 39–55.

Bryant

D. P.

Bryant

B. R.

Roberts

Vaughn

Pfannensteil

K. H.

Porterfield

Gersten

(2011). Early numeracy intervention program for first-grade students with mathematics difficulties. Exceptional Children, 78, 7–23. doi:10.1177/001440291107800101

Chen

C.-W.

H.-H.

(2007). Effects of treatment on disruptive behaviors: A quantitative synthesis of single-subject researches using the PEM approach. The Behavior Analyst Today, 8, 380–397. doi:10.1037/h0100629

*Cochran

Feng

Cartledge

Hamilton

(1993). The effects of cross-age tutoring on the academic achievement, social behaviors, and self-perceptions of low-achieving African-American males with behavioral disorders. Behavioral Disorders, 18, 292–302.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum. doi:10.4324/9780203771587

Conn

V. S.

Chan

K. C.

(2016). How much, how often, how long? Addressing dosage in intervention studies. Western Journal of Nursing Research, 38, 3–4. doi:10.1177/0193945915605067

10.

Conn

V. S.

Groves

P. S.

(2011). Protecting the power of interventions through proper reporting. Nursing Outlook, 59, 318–325. doi:10.1016/j.outlook.2011.06.003

11.

Council for Exceptional Children. (2014). Council for exceptional children standards for evidence-based practices in special education. Retrieved from http://www.cec.sped.org/~/media/Files/Standards/Evidence%20based%20Practices%20and%20Practice/EBP%20FINAL.pdf

12.

deCharms

(1972). Personal causation training in the schools. Journal of Applied Social Psychology, 2(2), 95–113. doi:10.1111/j.1559-1816.1972.tb01266.x

13.

Dwan

Gamble

Williamson

P. R.

Kirkham

J. J.

(2013). Systematic review of the empirical evidence of study publication bias and outcome reporting bias—An updated review. PLoS ONE, 8(7). doi:10.1371/journal.pone.0066844

14.

Fuchs

L. S.

(2005). Peer-assisted learning strategies: Promoting word recognition, fluency, and reading comprehension in young children. The Journal of Special Education, 39, 34–44. doi:10.1177/00224669050390010401

15.

Gersten

Fuchs

L. S.

Compton

Coyne

Greenwood

Innocenti

M. S.

(2005). Quality indicators for group experimental and quasi-experimental research in special education. Exceptional Children, 71, 149–164. doi:10.1177/001440290507100202

16.

Greenwood

C. R.

Carta

J. J.

Hall

R. V.

(1988). The use of peer tutoring strategies in classroom management and educational instruction. School Psychology Review, 17, 258–275.

17.

Greenwood

C. R.

Delquadri

J. C.

Hall

R. V.

(1989). Longitudinal effects of classwide peer tutoring. Journal of Educational Psychology, 81, 371–383. doi:10.1037/0022-0663.81.3.371

18.

*Gumpel

Frank

(1999). An expansion of the peer-tutoring model: Cross-age peer-tutoring of social skills among socially rejected boys. Journal of Applied Behavioral Analysis, 32, 115–118. doi:10.1901/jaba.1999.32-115

19.

*Hamelberg

L. L.

(1987). Effects of a cross-age tutoring program on the sight word acquisition of elementary learning disabled tutees and on the self-concept of severe behavior handicapped adolescent tutors (Unpublished doctoral dissertation). The Ohio State University, Columbus.

20.

*Harrigan

K. V.

(1994). The effects of cross-age tutoring and tutor training on the spelling performance of mildly handicapped students (Unpublished doctoral dissertation). Rutgers The State University of New Jersey, New Brunswick.

21.

Harrison

G. V.

(1976). Structured tutoring: Antidote for low achievement. In Allen

V. L.

(Ed.), Children as teachers (pp. 169–175). New York, NY: Academic Press.

22.

Heron

T. E.

Welsch

R. G.

Goddard

Y. L.

(2003). Applications of tutoring systems in specialized subject areas: An analysis of skills, methodologies, and results. Remedial and Special Education, 24, 288–300. doi:10.1177/07419325030240050401

23.

*Hogan

Prater

M. A.

(1993). The effects of peer tutoring and self-management training on on-task, academic, and disruptive behaviors. Behavioral Disorders, 18, 118–128. Retrieved from http://www.jstor.org.ezproxy.lib.utexas.edu/stable/23887087

24.

Hogan

D. M.

Tudge

J. R. H.

(1999). Implications of Vygotsky’s theory for peer learning. In O’Donnell

A. M.

King

(Eds.), Cognitive perspectives on peer learning (pp. 39–65). Mahwah, NJ: Lawrence Erlbaum. doi:10.4324/9781410603715.

25.

*Holecek

D. L.

(2012). Cross age/cross disability peer tutoring: A strategy for math instruction (Unpublished doctoral dissertation). University of Wisconsin-Superior, Superior.

26.

Horner

R. H.

Carr

E. G.

Halle

McGee

Odom

Wolery

(2005). The use of single-subject research to identify evidence-based practice in special education. Exceptional Children, 71, 165–179. doi:10.1177/001440290507100203

27.

Hughes

T. A.

Fredrick

L. D.

(2006). Teaching vocabulary with students with learning disabilities using classwide peer tutoring and constant time delay. Journal of Behavioral Education, 15, 1–23. doi:10.1007/s10864-005-9003-5

28.

Institute of Education Sciences. (2014). What works clearinghouse: Procedures and standards handbook (Version 3.0). Retrieved from http://ies.ed.gov/ncee/wwc/handbooks

29.

Jun

S. W.

Ramirez

Cumming

(2010). Tutoring adolescents in literacy: A meta-analysis. Journal of Education, 45, 219–238. doi:10.7202/045605ar

30.

Kern

(2015). Addressing the needs of students with social, emotional, and behavioral problems: Reflections and visions. Remedial and Special Education, 36, 24–27. doi:10.1177/07419232514554104

31.

Landrum

T. J.

Tankersley

Kauffman

J. M.

(2003). What’s special about special education for students with emotional or behavioral disorders? The Journal of Special Education, 37, 148–156. doi:10.1177/00224669030370030401

32.

*Lane

Pollack

Sher

(1972). Remotivation of disruptive adolescents. Journal of Reading, 15, 351–364. Retrieved from http://jstor.org.ezproxy.lib.utexas.edu/stable/23890485

33.

*Lazerson

D. B.

(1980). I must be good if I can teach! Peer tutoring with aggressive and withdrawn children. Journal of Learning Disabilities, 13, 152–157. doi:10.1177/002221948001300309

34.

*Lazerson

D. B.

(2005). Detention home teens as tutors: A cooperative cross-age tutoring pilot project. Emotional & Behavioural Difficulties, 10, 7–15. doi:10.1177/1363275205050881

35.

H. -H.

(2006). An alternative method for quantitative synthesis of single-subject researches percentage of data points exceeding the median. Behavior Modification, 30, 598–617. doi:10.1177/0145445504272974

36.

(2009). Family socioeconomic status, parental involvement, and college major choices: Gender, race/ethnic, and nativity patterns. Sociological Perspectives, 52, 211–234. doi:10.1525/sop.2009.52.2.211

37.

*Maher

C. A.

(1982). Behavioral effects of using conduct problem adolescents as cross-age tutors. Psychology in the Schools, 19, 360–364. doi:10.1002/1520-6807(198207)

38.

*Maher

C. A.

(1984). Handicapped adolescents as cross-age tutors: Program description and evaluation. Exceptional Children, 51, 56–63. doi:10.1177/001440298405100108

39.

Miller

Topping

Thurston

(2010). Peer tutoring in reading: The effects of role and organization on two dimensions of self-esteem. British Journal of Educational Psychology, 80, 417–433. doi:10.1348/000709909X481652

40.

Miller

S. R.

(1995). Cross-age peer tutoring: A strategy for promoting self-determination in students with severe emotional disabilities/behavior disorders. Preventing School Failure, 39, 32–37. doi:10.1080/1045988x.1995.9944640

41.

Moher

Liberati

Tetzlaff

Altman

D. G.

, & The PRISMA Group. (2009). Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement, PLoS medicine, 6(7),e1000097. doi:10.1371/journal.pmed1000097

42.

Mooney

Ryan

J. B.

Uhing

B. M.

Reid

Epstein

M. H.

(2005). A review of self-management interventions targeting academic outcomes for students with emotional and behavioral disorders. Journal of Behavioral Education, 14, 203–221. doi:10.1007/s10864-005-6298-1

43.

Okilwa

N. S. A.

Shelby

(2010). The effects of peer tutoring on academic performance of students with disabilities in grades 6 through 12: A synthesis of the literature. Remedial and Special Education, 31, 450–463. doi:10.1177/0741932509355991

44.

Pigott

T. D.

Valentine

J. C.

Polanin

J. R.

Williams

R. T.

Canada

D. D.

(2013). Outcome-reporting bias in education research. Educational Researcher, 42, 424–432. doi:10.3102/0013189X13507104

45.

Preston

Carter

(2009). A review of the efficacy of the picture exchange communication system intervention. Journal of Autism and Developmental Disorders, 39, 1147–1486.

46.

Robinson

D. R.

Schofield

J. W.

Steers-Wentzell

K. L.

(2005). Peer and cross-age tutoring in math: Outcomes and their design implications. Educational Psychology Review, 17, 327–362. doi:10.1007/s10648-005-8137-2

47.

Ryan

J. B.

Reid

Epstein

M. H.

(2004). Peer-mediated intervention studies on academic achievement for students with EBD: A review. Remedial and Special Education, 25, 330–341. doi:10.1177/07419325040250060101

48.

*Scruggs

T. E.

Osguthorpe

R. T.

(1986). Tutoring interventions within special education settings: A comparison of cross-age and peer tutoring. Psychology in the Schools, 23, 187–193. doi:10.1002/1520-6807(198604)23:2<187::AID-PITS2310230212>3.0.CO;2-7

49.

Smet

M. D.

Keer

H. V.

Wever

B. D.

Valcke

(2010). Cross-age peer tutors in asynchronous discussion groups: Exploring the impact of three types of tutor training on patterns in tutor support and on tutor characteristics. Computers & Education, 54, 1167–1181. doi:10.1016/j.compedu.2009.11.002

50.

Spencer

V. G.

(2006). Peer tutoring and students with emotional or behavioral disorders: A review of the literature. Behavioral Disorders, 31, 204–222.

51.

Spencer

V. G.

Simpson

C. G.

Oatis

T. L.

(2009). An update on the use of peer tutoring and students with emotional and behavioural disorders. Exceptionality Education International, 19, 2–13. Retrieved from http://ir.lib.uwo.ca/eei/vol19/iss1/2

52.

Stowitschek

C. E.

Hecimovic

Stowitschek

J. J.

Shores

R. E.

(1982). Behaviorally disordered adolescents as peer tutors: Immediate and generative effects on instructional performance and spelling achievement. Behavioral Disorders, 7, 136–148.

53.

*Top

B. L.

Osguthorpe

R. T.

(1987). Reverse-role tutoring: The effects of handicapped students tutoring regular class students. The Elementary School Journal, 87, 413–423. doi:10.1086/461505

54.

Topping

Ehly

(1998). Peer-assisted learning. Mahwah, NJ: Lawrence Erlbaum. doi:10.4324/9781410603678

55.

Trout

A. L.

Nordness

P. D.

Pierce

C. D.

Epstein

M. H.

(2003). Research on the academic status of children with emotional and behavioral disorders: A review of the literature from 1961 to 2000. Journal of Emotional and Behavioral Disorders, 11, 198–210. doi:10.1177/10634266030110040201

56.

Vaughn

Wexler

Leroux

Roberts

Denton

Barth

Fletcher

(2012). Effects of intensive reading intervention for eighth-grade students with persistently inadequate response to intervention. Journal of Learning Disabilities, 45, 515–525. doi:10.1177/0022219411402692

57.

*Watts

G. W.

Bryant

D. P.

(2017). Promoting number sense through number line board games delivered by older students with EBD. Unpublished manuscript, The University of Texas at Austin.

58.

Zimmerman

B. J.

Risemberg

(1997). Becoming a self-regulated writer: A social cognitive perspective. Contemporary Educational Psychology, 22, 73–101. doi:10.1006/ceps.1997.0919