The Impact of an Intensive Multisensory Reading Program on a Population of Learning-Disabled Delinquents Stephen B. Simpson Orange County Department of Education Orange CountF California

James M. Swanson University of California, Irvine, California

Ken Kunkel Tustin Unified School District Tustin, California

The high prevalence of learning disabilities in the juvenile delinquent population has been well documented, but attempts to remediate and have an impact on recidivism of this population of delinquents has produced limited results. The present study is a replication of the remediation phase of the 1976 LD/JD Project with methodological refinements to control for treatment integrity and strength of treatment. Delinquents in two detention facilities were screened for a developmental reading disorder. Subjects were selected for the study based on normal intelligence, full English proficiency, and a discrepancy of 15 points between reading achievement and IQ. Subjects in the treatment group received 90 minutes of remedial reading instruction per day using a multisensory (Orton/ Annals of Dyslexia, Vol. 42, 1992. Copyright ©1992by The Orton Dyslexia Society ISSN 0736-9387

54

MULTISENSORY READING FOR LD DELINQLrBN'rS

55

GiUingham) approach. A comparison group received 45 minutes of daily reading instruction in the regular classroom. There was no significant difference between the two groups in mean age of first arrest, mean age, and mean hours of reading instruction. Based on pre- and posttesting in reading and arrest records one year following release, the treatment group made significantly greater growth in reading (.33 year growth vs -.05 year growth per 10 hours of instruction) and had a significantly lower rate of recidivism (41 percent vs 63 percent) than the comparison group. Results were discussed in terms of hours of instruction necessary to improve reading, intervening treatment variables, and cost effectiveness of remedial program.

The prevalence of learning disabilities (LD) in the juvenile delinquent 0D) population has been estimated to be 32 percent to 36 percent (Zimmerman et al. 1981; Boder et al. 1981; and Campbell 1978), while the prevalence of LD in the general school-aged population has been estimated to be 5 percent to 16 percent (Hallahan, Keller, and Ball 1986; Zimmerman et al. 1981). While the elusive definition of LD creates problems in comparing different populations of youth with LD, it is estimated that LD is over represented in the JD population by a factor of 2 to 4 times. A recent estimate by Keilitz and Dunivant (1986) indicated that adolescent males with LD are 2.2 times more likely to be adjudicated delinquent than adolescent males without LD. Attempts to provide academic remediation for delinquents with LD has been particularly disappointing. The most extensive study was the 1976 LD/JD Project. Under the auspice of the National Institute of Juvenile Justice and Delinquency Prevention, the 1976 LD/JD Project designed and conducted a program of remedial instruction for delinquents with LD (Dunivant 1984; Keilitz and Dunivant 1986). A treatment and a control group were randomly selected. The design called for the treatment group to receive 150 hours of individualized remedial instruction (four 50-minute sessions per week for one year) in a variety of community settings. Programmatic restraints (e.g., drop-out rate and absenteeism) reduced the average number of hours of instruction to 32 (an average of two sessions per week over a six-month period). The treatment group failed to demonstrate significant academic growth in reading and arithmetic compared to the control group. Furthermore, changes in school attitude, change in self-reported delinquenc~ and rate of recidivism did not differ for the treatment and control groups. Clearly, lack of treatment integrity and strength may have limited the overall impact of remediation. The present study (Juvenile Justice Literacy [JJL] Project) was designed to provide intensive academic remediation to a population of LD delinquents, but, drawing upon the experience of the 1976 LD/JD Project, some methodological refinements were made to facilitate treat-

56

T~E OLDER DYSLEXIC

ment integrity and strength. First, we focused solely on reading remediation. A specific approach for reading remediation was not used in the 1976 LD/JD Project, so it was not possible to gauge how well the remedial reading program was implemented. In the 1989 JJL Project, we used a specific approach (Orton/Gillingham) to remedial reading. The Orton/Gillingham approach was selected for the following reasons: (1) it was clearly defined in well-documented operational steps; (2) it has been subjected to a considerable amount of research; and (3) it is well known and generally accepted by reading professionals as a treatment for reading disorders. Second, we selected a detention setting instead of a community setting for the 1989 JJL Project. Drop-out rate and absenteeism in the remediation phase of the 1976 LD/JD Project was so high that the frequency of the remedial instruction (strength of treatment) was only one-fifth of that intended. We implemented the 1989 JJL Project in a detention setting where participation and attendance in the project could be more effectively controlled than in a community setting. The 1989 JJL Project sought to answer the following research questions: (1) Are minors in a detention setting long enough to profit from an Orton/Gillingham approach to remedial reading? (2) Can an Orton/ Gillingham approach to remedial reading in a detention setting produce significant growth in reading? (3) Can an Orton/Gillingham approach to remedial reading impact recidivism? (4) Can an Orton/ Gillingham approach to remedial reading in a detention setting be a cost-effective treatment regime?

Method Subjects Subjects were selected from two detention facilities. One detention facility provides residential care for 60 wards of the court, who had an average length of stay of 92 days. The other facility provides residential care for 120 wards of the court and a day school program for another 60. The average length of stay was 37 days for the residential program and 162 days for the day school program. Teachers in the school at each facility were provided inservice training which focused on identification of adolescents with developmental reading disorders. Teachers used a symptom checklist to refer subjects for the project. Subjects referred to the project were screened for educational background and full English proficiency. The IDEA Oral Language Proficiency Test II was administered to subjects with a bilingual background to identify those who were fully English profident. Subjects who were fully English proficient were tested for cogni-

MULTISEN$ORYREADINGFOR LD DELINQUENTS

57

tive development and reading achievement. Cognitive development was measured by the Wechsler Intelligence Scale for Children Revised, administered by a licensed clinical psychologist. Reading achievement was measured by the Woodcock Test of Reading Master3~ administered by a licensed educational psychologist. A visual screening test was also used by the educational psychologist to screen for problems in visual perception. School records were used to screen out subjects with a history of hearing, vision, or other medical problems that might affect reading achievement. Based on this assessment, the criteria used to select subjects for the Project were the following: (1) full English proficiency; (2) hearing and vision within normal limits; (3) verbal or performance IQ at or above 80; (4) discrepancy between reading and verbal or performance IQ of 15 points or more; (5) history of adequate elementary school attendance. Following the completion of the remediation phase of the project, the comparison group was selected from teacher referrals. IQ testing was not done with the comparison group, since teacher estimate of intellectual ability and discrepancy between ability and achievement proved to be very accurate: all the treatment subjects referred to the project by teachers met the IQ and discrepancy criterion for inclusion in the study. The IDEA Oral Language Proficiency Test II was administered to subjects with a bilingual background to determine full English proficiency. The Woodcock Test of Reading Mastery was administered to English proficient subjects when they entered the comparison group. The initial treatment and comparison groups consisted of 55 and 61 males respectivel~ For both groups the age range was 13 to 18 years old. Procedure

One teacher and one aide were hired to deliver the specific remediation program for the treatment group at each of the two detention sites. The selected teachers had extensive experience using the Orton/Gillingham approach to remedial reading. This is a highly structured phonics approach that integrates reading, writing, and spelling through the primary modes of hearing, vision, and kinesthesia. The teachers and the JJL Project coordinator met frequently before the project was implemented and periodically during the project to standardize implementation of the treatment regime. Reading instruction for the treatment group was provided in groups of one to six students. Subjects in the treatment group received 90 minutes of reading instruction per day. Reading instruction for the comparison group was provided in a regular classroom of approxi-

58

ThE OLDER DvsrEraC

mately 12 students as part of the daily 45-minute English class. A certificated teacher taught the class using individualized and small group instruction based on the assessed needs of the student. Prior to being released from the detention facility, subjects in both groups were reassessed for growth in reading using an alternate form of the Woodcock Test of Reading Master. The County Sheriff's Central Crime Index was used to determine age of first arrest and the number of arrests for all subjects within one year of their release from the detention facility.

Results Of the 55 subjects who started in the treatment group, pre- and posttest results were obtained on 32 subjects or 58 percent. Of the 61 subjects who started in the comparison group, pre- and posttest results were obtained on 31 subjects or 51 percent. Posttest results were not obtained on subjects who unexpectedly left the facility. This typicaUy happened w h e n a subject escaped from the facilit~ was removed for disciplinary reasons, was transferred to another facilit~ or was released early. The percentage of subjects for w h o m complete data were available did not differ for the two groups. One-way analyses of variance were performed with Group (Treatment vs Comparison) as an independent variable. As shown in Table I, no significant difference was detected for the four background variables: mean age of first arrest, age at release from detention facility, initial reading level, and number of hours of reading instruction. However, the range of the number of hours of reading instruction was greater for the comparison group (13.5 to 129 hours) than for the control group (29 to 70 hours). The extended range of the comparison group was due to three subjects who remained in detention for exceptionally

Table I Treatment Group and Comparison Group Age of First Arrest, Age, Initial Reading Level, and Number of Hours of Instruction Mean Age Mean Age Initial Reading Mean Hours 1st Arrest at Release Level Instruction Treatment 13.7 16.2 4.3 51.9 Group (N = 32) 1.64" 1.46 1.61 13.68 Comparison 13.8 16.3 4.6 46.0 Group (N = 31) 1.05 1.23 1.79 30.98 *+ S.D. F(1,60)= .09 ns F(1,60)= .17 ns F(1,60)= .38 ns F(1,61)= .96 ns

MULTISEN$ORY READING FOR LD DELINQUENTS

59

long periods of time and, according to the design of the study, continued to accumulate instructional hours in the regular classroom. The comparison group had a higher percentage of subjects with fewer hours of instruction (71 percent of the comparison group had 50 hours or less of reading instruction vs 41 percent for the treatment group). This difference is accounted for by the design of the study. Subjects in the treatment group received a more intense remedial program (more hours of reading instruction per week) than the comparison group. The two groups did differ on reading growth as measured by differences in performanc~ in the Woodcock Test of Reading Mastery administered upon entry into the project and at the end of the project. As shown in Table II, the reading growth (in grade score units) for the treatment group (.93) was significantly greater (p < .01) than the growth score for the comparison group (.07). A correlation coefficient was calculated to relate reading improvement to hours of instruction. This correlation was significant (p < .05) for the treatment group (r = .33), but not for the comparison group (r = - .26). A correlation coefficient was calculated for the treatment group to relate reading improvement to IQ (see Table III.). These correlations were not significant for verbal IQ ( r = - .031), performance IQ (r = - .012), or full scale IQ (r = - . 0 8 6 ) .

To compare reading growth per hour of instruction for the treatment and comparison groups, regression analyses were performed to estimate the rate of growth (slope) of reading as a function of hours of instruction. The following regression equations were obtained: Treatment Group - Reading Growth = -.7863 + .0330 (hour) Comparison Group - Reading Growth = .3139 - .0055 (hour) The rate of recidivism one year following release was significantly (p < .05) lower for the treatment group (41 percent) than for the corn-

Table II Treatment Group and Comparison Group Reading Growth, Correlation Hours vs Reading Growth, and Reading Growth per 10 Hours of Instruction Mean Reading Correlation Reading Growth Growth Hours vs Reading Per 10 Hours Treatment .93 .33 .33 year Group 1.36 t(30) = 1.92 p < .05 Comparison .07 -.26 - . 054 year Group 1.05 t(29) = 1.43 ns F(1,61) = 9.78 p < .01

60

THE OLDER DYSLEXIC

Table III Treatment Group Correlation: Reading Growth vs Verbal IQ, Reading Growth vs Performance IQ and Reading Growth vs Full Scale IQ Correlation Reading Correlation Reading Correlation Reading vs Verbal IQ vs Performance IQ vs Full Scale IQ Treatment - .031 - .012 - .086 Group t(30) = .169 ns t(30) = .066 ns t(30) = .472 ns parison group (63 percent), and the average number of arrests was significantly (p < .05) less for the treatment group (.46) than for the comparison group (.99). A correlation coefficient was calculated to relate reading improvement to rate of recidivism and frequency of arrest (see Table W.). These correlations were significant (p < .05) for the treatment group (r = .33 and .38) but not for the comparison group (r = .062 and .037).

Discussion

The primary purpose for conducting the 1989 JJL Project in detention settings was to control for drop-out and absenteeism. Subjects in the Project who were in residence in the detention facilities (N = 20) had 90.1 percent attendance, while subjects who were in the dayschool program in the one facility (N = 12) had 67 percent attendance. These percentages of attendance closely mirror those of the two general populations in the facilities as well as those of the subjects in the comparison group. As might be expected, subjects with poor attendance generally demonstrated little reading growth. The drop-out of subjects was 8 percent for residential subjects and 17 percent for dayschool subjects. The level of attendance and program completion for the 1989 JJL Project were significantly better than the figures reported in the LD/JD Project. In the 1976 LD/JD Project, reading improvement was negligible for subjects with less than 30 hours of remedial instruction. A similar phenomenon was found in the present study. The negative intercept in the regression analysis suggests that for less than 23 hours of remedial instruction zero or negative growth would be expected. Most subjects who made large gains in reading had extensive remediation: 80 percent of the subjects w h o made a year or more growth in reading had more than 50 hours of remedial reading instruction. But, 63 percent of the subjects in the initial treatment group (N = 55) were not in the detention facility long enough to receive 50 hours of instruction (see Figure 1.). However, as the project was implemented, we observed fre-

.63% .49

Comparison Group

Z = 1.77 p < .038 F(1,61) = 4.25 p < .05

.41% .49

Treatment Group

Rate of Recidivism

.99 .99

.46 .62

Frequency of Arrest

.062 t(29) = .335 ns

.33 t(30) = 2.02 p < .015

Correlation Reading vs Recidivism

Correlation

.037 t(29) = .20 ns

.38 t(30) = 2.43 p < .05

Reading vs A r r e s t

Table W Treatment G r o u p a n d Comparison G r o u p Rate of Recidivism, F r e q u e n c y of Arrest, Correlation Reading vs Recidivism, a n d Reading vs F r e q u e n c y of A r r e s t

62

THE OLDERDYSLEXIC

greater than 66 57 to 66

e-0 o

47 to 56

t= t-

37 to 46

O

0

-1-

27 to 36 less than 27

N = 55

0.0%

10.0%

20.0%

30.0%

40.0%

Percentage of Students

F~u~I.

Distribution of hours of reading instruction for treatment group

quent movement of subjects between detention facilities for a variety of reasons (e.g., transfer, escape attempt, additional charges). If similar remedial programs were available in other detention facilities, these subjects could have continued in their remedial program. Thus the first research question of whether minors are in a detention facility long enough to profit from an Orton/Gillingham approach to remedial reading could not be dearly answered by the study. The second research question of whether an Orton/Gillingham approach to remedial reading in a detention facility could produce significant growth in reading was clearly demonstrated. Reading growth was significantly greater (p > .01) for the treatment group (.93 year growth vs .07 year growth). Since the average time between tests was about three months (one-fourth of a year), the expectation of improvement would be .25 year. The comparison group, which had a reading growth of .07 year did not meet this expectation. The treatment group, despite an initial setback at the beginning of the instruction (as reflected by the negative intercept), showed a higher than expected rate of improvement in reading. Over the three months, this group showed a grade score improvement of .68 year greater than expected due to normal growth (.93 - .25 = .68). The regression analyses indicated that for every ten hours of r e -

MULTISENSORY READING FOR LD DELINQUENTS

63

medial instruction, the treatment group achieved about one-third of a year growth in reading, while the comparison group made no growth (or regressed) in reading. The negative intercept of the regression equation for the treatment group (-.7863) and the value of the slope (.0331) suggest that 23.75 hours of instruction were required to maintain entry reading level (0 = - .7863 + .0331 [23.75 hours]). Using the regression equation, we estimated that to achieve a reading growth of one year, 53.97 hours would be required (1 year = -.7863 + .0331 [53.97 hours]). The variance in the responses to treatment (as reflected by reading improvement scores) was large. Not all subjects in the treatment group who received extensive instruction significantly improved in reading. A cut-off of one year growth in reading was selected to reflect significant improvement. Of the 32 subjects in the treatment group over half (18) received extensive reading instruction (over 50 hours), but only eight subjects exceeded the cut-off value of one year growth in reading. Thus, given over 50 hours of remedial reading, the conditional probability of significant reading growth was .45 (P = 8/18). On the other hand, very few subjects improved who did not receive over 50 hours of remedial instruction. Only two of the 14 subjects in the treatment group who did not receive 50 hours or more of reading remediation showed a significant improvement in reading growth of a year or more. Thus, given reading growth of one year or more, the conditional probability of over 50 hours of instruction was .80 (P = 8/10). These conditional probability analyses of significant improvement suggest that exposure to intensive reading instruction is not sufficient to produce a significant growth in reading. It also suggests that exposure to intensive reading instruction may be an essential (but not sufficient) ingredient for significant improvement in reading. Twenty-five percent of the subjects in the treatment group made no growth in reading or regressed. There are many factors that may determine reading growth in addition to hours of instruction. One obvious factor is IQ. However, the correlation of reading growth with IQ was not significant (r = - .086). We did not collect data in this study to consider any other variables. However, the attitude and perception toward change the subject brings into the remedial setting may have a significant impact on reading growth. In a separate project, Lerner (1990) evaluated the personal attitudes and perceptions of juvenile delinquents toward change, using the transtheoretical change theory of Prochaska and DiClemente (1982). This theory hypothesizes five stages-of-change: precontemplation (lacking awareness or motivation to change), contemplation (thinking about change), action (determining action plan for change), maintenance (implementing action plan), and relapse (returning to earlier stage-of-change). It is possible that

64

ThE OLOER DrsLExxc

subjects in the treatment group who failed to profit from remediation were resistant to change because they had entered the contemplation or action stage-of-change. The 1989 JJL Project remedial program may provide the impetus for juvenile delinquents to move from the contemplation or action stage-of-change. The negative intercept in the regression analysis may be interpreted as the time required for subjects to progress from the contemplation and action stages-of-change. The slope of the regression equation may represent the effect of the maintenance stage-of-change. A successful remediation program (i.e., one that significantly improves reading) may need to provide skills which help keep juvenile delinquents in the maintenance stage-of-change and prevent relapse. The processing skills the learner brings into the remedial setting may also have a significant impact on reading growth. Woods et al. (1991) found that children with developmental reading problems are characterized by deficits in linguistic processes (i.e., the ability to organize, store, and retrieve words and to analyze and synthesize sounds and syllables in words). Deficits in linguistic processes vary along a skill continuum. It is possible that subjects who failed to profit from the remedial reading program did not have the entry level linguistic processing skills necessary to profit from the program. Under these assumptions, the negative intercept of the regression analysis may be interpreted as the time required for subjects to acquire the necessary linguistic processing skills to profit from the remedial program. For the remediation program to be successful, it may be necessary to identify the level of linguistic processing skills required for each subject to profit from the remedial instruction. Subjects' attitudes toward change and linguistic processing skills are probably interacting variables. Subjects with severe deficits in linguistic processing skills are more likely to experience frustration and failure in reading and to have a lower expectation for change. This lower expectation for change may be seen as resistance to entry into the contemplation and action stages-of-change. The strength of the remedial program might be significantly enhanced by identifying subjects whose deficits in linguistic processing and/or attitudes toward change are barriers to reading improvement. Providing these subjects with supplemental skill training and/or counselling to deal with these barriers may be an effective enhancement of the reading remediation program. The third research question of whether an Orton/Gillingham approach to remedial reading could impact recidivism was dearly demonstrated. The rate of recidivism for the comparison group (63 percent) closely mirrored that of the general population (66 percent) in the detention facilities, while the rate of recidivism for the treatment

MULTISEN$ORY READING FOR LD DELINQUENTS

65

group was significantly lower (41 percent). Subjects in the treatment group were significantly less likely to be arrested one year following their release than subjects in the comparison group, and they were significantly less likely to have multi-arrests one year following their release. There was also a significant relationship between reading growth and rate and frequency of recidivism for the treatment group, but not for the comparison group. The final research question of whether an Orton/Gillingham approach to remedial reading was cost-effective was addressed by measuring the saving in cosE of detention due to a reduction in recidivism. The treatment group had a rate of recidivism 22 percent lower than the comparison group (41 percent vs 63 percent). In Orange County, California the estimated cost for arresting, convicting, and providing residential custody for a juvenile delinquent for 90 days is approximately $8,400. The one-year cost for Orton/Gillingham remedial reading program for 60 students would be $63,720 or $1,062 per student. Based on these assumptions, if the remedial program resulted in 22 percent fewer of these 60 students being rearrested the year following their release from a detention facility, this would result in a reduction in institutional expenditures of $110,800 (60 x 22 percent x $8,400). Thus an institutional savings of $47,160 ($110,800 - $63,720) would result from a program like the 1989 JJL Project being provided 60 LD delinquents in detention facilities. In terms of dollars, for every $1.00 invested in the reading remediation program, a $1.75 return would be expected as a result of reduced institutional expenditures. This represents a 75 percent return on investment. The 1989 JJL Project demonstrates that effective remedial programs in detention facilities can significantly affect academic growth for LD delinquents and be part of a cost-effective treatment regime to reduce delinquency. The results for the total treatment and comparison groups in the 1989 JJL Project are in contrast to the findings of the 1976 LD/JD Project which failed to demonstrate a significant growth in reading or a reduction in recidivism between their treatment and control groups. The failure of the 1976 LD/JD Project to demonstrate significant results may have been related to variables that affected treatment strength and integrity. The 1989 JJL Project was designed around a high-strength intervention model (Kazdin 1988) that enhanced treatment strength by selection of treatment setting (detention facility vs. community setting), treatment intensity (90 minutes vs. 50 minutes of instruction per session), and treatment frequency (five remedial sessions per week vs. four sessions per week). The 1989 JJL Project enhanced treatment integrity through the use of a highly-structured Orton/Gillingham approach to reading in contrast to the eclectic approach used in the 1976 LD/JD Project. The degree these treatment

66

THE OLDER DYSLEXIC

variables contributed to significant reading growth and are related to reduced recidivism will require investigation in another study controlling for these variables. While the 1989 JJL Project addressed the research questions it sought to answer, it uncovered additional questions: (1) What roles do attitudes toward change and linguistic processing skills play in successful reading remediation? (2) Can subjects' success or failure in the remediation program be predicted? (3) What treatment regime would be appropriate for subjects who are unsuccessful in the remediation program? (4) How can savir{gs in institutional expenditures be translated into implementation of remedial programs for LD delinquents? The first three questions are research questions that could be incorporated in a larger scale replication of the present study. The last is a social/political question for politicians, program directors, and business managers. Research may provide documentation for these individuals to take appropriate action.

References Boder, P. K., Dunivant, N., Smith, E. C., and Sutton, L. P. 1981. Further observations on the link between learning disabilities and juvenile delinquency. Journal of Educational Psychology 73:838-850. Campbell, E B. 1978. The definition and prevalence of learning disabilities. Paper read at the 15th International Conference of the Association for Children with Learning Disabilities, March 1978, Kansas City, MO. Dunivant, N. 1984. Improving academic skills and preventing delinquency of learningdisabled juvenile delinquents: Evaluation of the ACLD remediation program. National Center for the State Courts, Williamsburg, VA. Hallahan, D. P., Keller, C. E., and Ball, D. W. 1986. A comparison of prevalence rate variability from state to state for each of the categories of special education. RASE 7:814. Kazdin, A. 1988. Child Psycholotherapy: Developing and identifying effective treatments. New York: Pergamon Press. Keilitz, I. and Dunivant, N. 1986. The relationship between learning disability and juvenile delinquency: Current state of knowledge. RASE 7:18-26. Lerner, C. E 1990. The transtheoretical model of change: self-change in adolescent delinquent behavior. Ph.D. diss., University of Rhode Island, RI. Prochaska, J. O. and DiClemente, C. C. 1982. Transtheoretical therapy: toward a more integrative model of change. Psychotherapy: Theory, Research and Practice 19:276288. Woods, F., Felton, R., Flowers, L., and Naylor, C. 1991. Neurobehavioral definition of dyslexia. In D. Drake and D. Gray (eds.). The Reading Brain. Parkton, Maryland: York Press. Zimmerman, J., Rich, W., Keilitz, I., and Broder, E 1981. Some observations on the link between learning disabilities and juvenile delinquency. Journal of Criminal Justice 9:1-7.

The impact of an intensive multisensory reading program on a population of learning-disabled delinquents.

The high prevalence of learning disabilities in the juvenile delinquent population has been well documented, but attempts to remediate and have an imp...
693KB Sizes 0 Downloads 0 Views