The Effects of Computer-Based Attribution Retraining on the Attributions, Persistence, and Mathematics Computation of Students with Learning Disabilities Cynthia M. Okolo
The purpose of the present study was to examine the impact of attribution retraining, embedded within a mathematics computer-assisted instructional (CAI) program, on students' attributions, persistence, and mathematics computation. Twenty-nine school-identified students with learning disabilities from five urban schools participated in the study. The sample's mean age was 13.3 years. After blocking on initial attributional patterns, students were randomly assigned to a mathematics CAI program that provided either attribution retraining or neutral feedback. Students used their assigned program for eight 30-minute sessions. Results did not support the contention that attribution retraining would have a significant impact on students' attributions. However, students who participated in the attribution retraining condition completed significantly more levels of the program than their counterparts who received neutral feedback. Attribution retraining students also obtained significantly higher scores on a test of problems practiced during the CAI program. These results suggest that attribution retraining may be a desirable addition to the type of feedback typically provided by CAI programs. However, they also highlight the need for further research that examines the conditions under which specific attributions are most advantageous.
S
tudents with learning disabilities usually are identified and referred for special education services because of a specific academic deficit. However, research amply documents that these students also experience motivational deficits that further interfere with their academic performance or that may partially contribute to their learning handicaps (Deci & Chandler, 1986). One particularly important aspect of students' motivation for school tasks is their belief about the reasons they succeed and fail, or their causal attributions. Research has consistently demonstrated that students who attribute successes to their ability and effort and failures to their lack of effort tend to fare well on measures of academic success, including persistence (Andrews & Debus, 1978; Dweck
& Reppucci, 1973), expectancies for future success (Weiner, 1974, 1985), strategic behavior (Diener & Dweck, 1980), teacher ratings (Kistner, Osborne, & LeVerrier, 1988), and scores on achievement and criterion-referenced tests (Kistner et al., 1988; Newman & Stevenson, 1990). Research regarding the types of attributions made by students with learning disabilities for their academic successes and failures defies a concise synthesis. Researchers have studied different samples, have used varying instruments and procedures, and often have obtained equivocal results. In general, however, studies have shown that students with learning disabilities are more likely than their nondisabled peers to attribute their successes not to effort and ability, but to luck or the
JOURNAL OF LEARNING DISABILITIES VOLUME 25, NUMBER 5, MAY 1992 Downloaded from ldx.sagepub.comPAGES at UNIVERSITE DE MONTREAL on June 9, 2015 327-334
ease of a task (Jacobsen, Lowery, & DuCette, 1986; Pearl, 1982; Pearl, Bryan, & Donahue, 1980; Tollefson et al., 1982). Similarly, these students are more likely to attribute their failures to lack of ability rather than lack of effort (Diener & Dweck, 1978; Jacobsen et al., 1986; Palmer, Drummond, Tollison, & Zinkgraff, 1982; Thomas & Pashley, 1982). These attributional patterns have important implications for the academic progress of students with learning disabilities. Unless students believe that their successes are due to their abilities and efforts, they are unlikely to experience enhanced feelings of pride and selfesteem or feel confident about their chances for future success (Weiner et al., 1971). Furthermore, students who attribute their failures to lack of ability are more likely to have pessimistic expectations for future success, less likely to persist at academic tasks, and more likely to be debilitated by failure (Dweck & Reppucci, 1973; Graham, 1991; Peterson & Barrett, 1987; Weiner et al., 1971). These less-thanoptimal attributional patterns are of particular concern when one considers that students with learning disabilities must persevere in the face of failure to master many basic academic skills. Dweck (1975) demonstrated that maladaptive attributional patterns are amenable to change. She taught 12
328
JOURNAL OF LEARNING DISABILITIES
students, who were characterized as ' 'learned helpless" and whose performance deteriorated in the face of failure, to attribute their failures to lack of effort. Subsequently, these students were able to persist and even increase their efforts after failure experiences. Since Dweck's classic study, many researchers have replicated the general finding that students' attributions can be modified. Although most of these attribution retraining studies have focused on academic tasks, they also have addressed behaviors as diverse as littering (Miller, Brickman, & Bolen, 1975), interpersonal persuasion (Anderson, 1983), assembly work (Zoeller, Mahoney, & Weiner, 1983), and dropping out of school (Wilson & Linville, 1982,1985). Student with learning disabilities and nondisabled students who have participated in attribution retraining interventions have shown increased expectancies for future success, increased persistence, and enhanced performance on target tasks. In most studies, students also experience changes in attributions for tasks similar to those used in the study (Forsterling, 1985; Pearl, 1985). Changes in more general attributional patterns or dispositions have been more difficult to achieve (e.g., Borkowski, Weyhing, & Carr, 1988), leading researchers to conclude that attributions are domainspecific (Newman & Stevenson, 1990). Most attribution retraining studies have attempted to engender more frequent effort attributions among participants (Forsterling, 1985; Graham, 1991), but the studies have varied widely in their methods. Often, researchers have modeled adaptive attributions during success and failure at a target task (Dweck, 1975; Schunk, 1982, 1984; Thomas & Pashley, 1982). In other cases, peers have modeled adaptive attributions, often on videotape (Shelton, Anastopoulos, & Linden, 1985; Zoeller et al., 1983). Many researchers have required students to state the desired attributions and sometimes have reinforced them for doing so (Andrews & Debus, 1978; Thomas & Pashley, 1982). Other researchers have found that indirect
methods, in which students listen to or read attributions, are as successful as more direct methods (Fowler & Peterson, 1981). Attribution retraining research has been criticized for its overreliance on laboratory studies (Graham, 1991), and researchers have questioned whether it is feasible to implement attribution retraining in school settings (Pearl, 1985; Stipek & Weisz, 1981). Computer-based activities offer a promising but little-researched medium for the delivery of attribution retraining within classrooms. Computers are becoming more widely available in elementary and secondary schools, and students with learning handicaps often use computer-assisted instruction (CAI) to practice basic skills (Becker, 1990; Becker & Sterling, 1987; Cosden & Abernathy, 1990). Because computer software can analyze students' performance on an ongoing basis and provide highly individualized feedback, it may be an extremely effective vehicle for providing students with adaptive attributional feedback. Moreover, some studies have found that students perceive the computer's assessment of their performance to be fairer and more credible than teachers' (Okolo, Rieth, & Bahr, 1989). However, only one study could be found that examined the effect of computer-delivered attribution retraining. Peter, McArthur, and Stasz (1988) examined the effects of attributions delivered either by a computer or by a human tutor. High school students practiced difficult algebra problems and, after each failure, were informed that they had failed due to either lack of effort or bad luck. The computer-delivered feedback was as effective as the human tutor in changing students' persistence and expectancies for success. The purpose of the present study was to examine the impact of computer-based attribution retraining for students with learning disabilities. Based on previous research regarding the relationship between attributions and school success, the attribution retraining provided in this study was designed to engender attributions to
ability and effort for successes and to lack of effort for failures. A publicdomain software program that delivered practice in multiplication computation was modified either to provide the attributions described above or to inform students of their progress. Three predictions were tested: First, students with learning disabilities who received attribution retraining should develop more adaptive attributions for success and failure in computer-based mathematics tasks. Second, attribution retraining should have a positive impact on students' persistence on multiplication problems. Finally, attribution retraining should have a positive impact on students' multiplication computation skills.
Method Materials CAI Program. The public-domain CAI program used in this study, Drill (Davis, no date), contains four sets of multiplication problems of increasing difficulty. Each 25-problem set presents 5 problems per screen. The opening screen of the program directs the student to choose a problem set and to enter the amount of time, per second, that should be allotted for solving each problem. Students in this study worked through the program in a series of levels. At the first level, students were directed to start with the easiest problem set and enter a 20-second-perproblem time limit. All students were told that their goal was to solve a problem set three times within the time limit with at least 80% accuracy. Upon reaching their goal, students were told to advance to the next level, in which they decreased the problem solution time by 5 seconds. When students had solved a problem set three times with 80% accuracy within a 5-second-perproblem limit, they were permitted to advance to the next most difficult problem set and allocate 20 seconds per problem. To help them visualize their progress, all potential levels of the program were represented on a bar graph. Students filled in a bar each
Downloaded from ldx.sagepub.com at UNIVERSITE DE MONTREAL on June 9, 2015
VOLUME 25, NUMBER 5, MAY 1992
time they completed a level (i.e., solved a problem set three times within a specified time limit with 80% accuracy). The program was modified to provide either attribution retraining or neutral (control condition) feedback after each screen of 5 problems and at the conclusion of each 25-problem set. In both conditions, after each 5-problem screen, the program displayed information about the student's accuracy (e.g., "You now have 15 correct out of 15 tries"). When attribution retraining students were meeting their goals, they received ability attribution feedback ("You really know these") after 5, 10, 15, and 20 problems, and effort attribution feedback ("You are really trying hard") after 25 problems. When attribution retraining students were not meeting their goals, they received effort attribution feedback ("You can get it if you keep trying" after 5, 10, 15, and 20 problems and "You can do better if you try harder" after 25 problems). When control group students met their goals, they received neutral feedback ("You are meeting your goal" after 5, 10, 15, and 25 problems and "You met your goal" after 25 problems). The control students also received neutral feedback when they were not meeting goals ("You are not meeting your goal" after 5,10,15, and 20 problems and "You did not meet your goal" after 25 problems). Initial Attributional Patterns. To equate the groups on initial attributional patterns, the Intellectual Achievement Responsibility (IAR) questionnaire (Crandall, Katkovsky, & Crandall, 1965) was administered to students at the start of the study. The IAR contains 34 item stems that depict either a positive or a negative achievement experience. For each item stem, students choose an alternative that represents a cause either controlled by the respondent or controlled by someone else in his or her environment. Students' total scores and their scores on the 19 items that specifically addressed attributions to effort were computed. Scores on the effort items were used as a blocking variable.
Mathematics Attribution Scale. Three items were developed to assess changes in those attributions targeted by the CAI program. For each item, students first were asked to imagine succeeding (or failing) at a computerbased mathematics activity. Then they were asked to rate the likelihood, on a 5-point scale, that (a) effort was responsible for success, (b) ability was responsible for success, and (c) lack of effort was responsible for failure. These three items were administered as a pre- and posttest measure. Persistence Measures. Two measures of persistence were used. First, as described above, students graphed the number of program levels they completed. The second measure was discretionary time for either continued multiplication practice or computer game play. During six sessions of the study, students were given the option of playing a computer game after completing 20 minutes of computer-based multiplication practice. The total number of minutes that each student used the game program was recorded. Computation Test. A multiplication computation test was developed as a pre- and posttest measure of skill attainment. It contained four sets of problems identical to those encountered in the CAI program. Students were directed to complete each problem set as quickly and accurately as possible and were timed. The number of digits they correctly computed per minute was counted and averaged across the four problem sets.
Sample The sample was drawn from resource room programs in five public schools in a large, urban school system that serves approximately 431,000 pupils. In this school system, "learning disability" is defined as a disorder in one or more of the basic psychological processes involved in understanding or using spoken or written language that is not caused by environmental disadvantage, mental retarda-
Downloaded from ldx.sagepub.com at UNIVERSITE DE MONTREAL on June 9, 2015
329 tion, or emotional disturbance. Students are placed in learning disability resource room programs when there is a 1- to 2-standard-deviation difference between their achievement and aptitude (as measured by standardized test scores) in one academic area. Resource room students spend more than 50% of the day in regular classes. The resource room teacher at each school was asked to select from her caseload seventh- and eighth-grade students with learning disabilities who were at least 1 year below grade level in mathematics achievement and could benefit from additional practice in multiplication computation. All students who met these criteria (N=37) were invited to participate in the study, and the 33 students who returned permission forms were included in the original sample. The final sample consisted of 29 students; 4 students were unable to complete the study due to schedule conflicts or sporadic attendance. Sixteen students were males and the mean age of the sample was 13.3 years (SD = 1.1). Fifteen (52%) were Caucasian, 9 (31%) were African-American, 3 (10%) were Hispanic, and 2 (3%) were Asian. Participating schools were located in lower middle to middle class neighborhoods. Average level of parental education for the sample was 12th grade. Students were 1 to 2 years below grade level in mathematics and 2 to 3 years below grade level in reading, as measured by the most recent administration of the Iowa Test of Basic Skills. Because IQ tests are not routinely administered in this school system, recent IQ data were unavailable for most of the sample. In an individual interview conducted before the study, students were asked to describe the locations in which they had used a computer and the activities in which they had engaged. All had previous computer experience and 17 had home computers. Over 50% of the sample had engaged in video game play, graphics, word processing, keyboarding practice, programming, mathematics and language arts CAI, and data-base activities.
330
Students were blocked on IAR effort score and then randomly assigned to either the attribution retraining or control condition. Gender and ethnic composition were similar for each condition and, as displayed in Table 1, the groups were comparable in total IAR score, chronological age/parental education level, and mathematics grade equivalent. Also, the groups were alike in the number of locations in which they had used computers and the types of computer activities in which they had engaged before the study.
Procedure The study was conducted in 30-minute sessions that took place in the student's resource room or the school's computer lab. Students worked in small groups of four to eight composed of other students who were participating in the same condition. On all pretest and posttest measures, research assistants read directions and, except for computation tests, individual items aloud. Students completed the IAR, mathematics attribution scale, and mathematics computation tests and then were introduced to the CAI program. Research assistants modeled its use and read aloud all feedback it provided. They also made sure that all students in each group could read and explain the program's feedback and scores. Students then were shown how to complete their progress graphs. Students worked independently at the computer for the next eight sessions. During the second session, research assistants demonstrated two computer games that the students could choose to play after 20 minutes of CAI practice. Research assistants monitored students throughout each session, informed them when 20 minutes had passed, reminded them they could play a game if they wished, and recorded the number of minutes each student engaged in the game. At the conclusion of every session, a research assistant reviewed each student's graph with him or her. If students had met their goal twice that day, the research assistant stated that they were
JOURNAL OF LEARNING DISABILITIES
TABLE 1 Demographic Variables Attribution retraining M SD IAR Effort Score IAR Total Score Age Parental Education Level Mathematics Grade Equivalent Reading Grade Equivalent Number of Locations of Computer Use Number of Types of Computer Use
14.1 24.9 13.6 12.9 6.3 5.1 6.9 4.6
2.5 4.0 .9 2.1 1.4 1.1 2.9 2.2
Control M SD 13.8 24.4 13.0 12.3 5.8 5.0 7.5 5.0
2.2 2.8 1.2 2.3 2.0 1.5 1.9 1.2
Note. IAR = Intellectual Achievement Responsibility questionnaire.
doing a good job. If they had not met two goals, the research assistant commented that the students had not met their goals that day. Research assistants provided no other instructions or reinforcement to students throughout the study. After each student completed eight computer sessions, the mathematics attribution scale and the multiplication computation test were readministered.
Fidelity of Implementation Several steps were taken to ensure that the procedures were implemented as described above. First, the five research assistants who conducted the sessions memorized a script for introducing the CAI program in each condition. They then role-played the introductory session among themselves, critiqued each other's performance, and proved to the author that they could satisfactorily deliver the session without errors. Twenty-five percent of the sessions at each school were observed by the author or the project coordinator, who ensured that research assistants were providing the proper feedback after each session without instructing or reinforcing students during the session. Finally, research assistants rotated regularly among schools and between conditions.
Results The primary dependent measures in this study were (a) attributions (pre-
and posttest mathematics attribution scale), (b) persistence (number of program levels completed and number of minutes spent on the game program), and (c) skill attainment (pre- and posttest computation tests).
Attributions Table 2 presents means and standard deviations for each of the three attribution items. These data were examined via three separate 2 (time of test) by 2 (groups) repeated measures ANOVAs. To control experiment-wise error, a Bonferroni adjustment was made and a was set at .02. There was a trend for students in both conditions to increasingly attribute successes to effort and failures to lack of effort. In addition, students in both groups were less likely to attribute successes to ability. However, preto posttest differences on these items were minimal and none were reliable, although the main effect for time of test on the attribution of success to ability item approached significance, F(l,27) = 2.9, p< .05.
Persistence As described above, students filled in a bar graph each time they completed a level of the program. Attribution retraining students completed an average of 16.0 (SD = 3.9) levels, whereas control students completed 11.6 (SD=3.3) levels. After testing and accepting the assumption of parallel slopes, F(l,24) = .23, an ANCOVA was
Downloaded from ldx.sagepub.com at UNIVERSITE DE MONTREAL on June 9, 2015
331
VOLUME 25, NUMBER 5, MAY 1992
TABLE 2 Pre- and Posttest Attribution and Computation Scores Attribution retraining M SD Attributions for success to ability Pretest Posttest Attributions for success to effort Pretest Posttest Attributions for failure to lack of effort Pretest Posttest Multiplication computation scores Pretest Posttest
computed on number of levels completed, with pretest multiplication score as the covariate. The attribution retraining and control groups were significantly different on this measure, F(l,25) = 9.6, p
The purpose of the present study was to examine the impact of attribution retraining, embedded within a mathematics computer-assisted instructional (CAI) program, on students' attributions, persistence, and mathematics computation. Twenty-nine school-identified students with learning disabilities from five urban schools participated in the study. The sample's mean age was 13.3 years. After blocking on initial attributional patterns, students were randomly assigned to a mathematics CAI program that provided either attribution retraining or neutral feedback. Students used their assigned program for eight 30-minute sessions. Results did not support the contention that attribution retraining would have a significant impact on students' attributions. However, students who participated in the attribution retraining condition completed significantly more levels of the program than their counterparts who received neutral feedback. Attribution retraining students also obtained significantly higher scores on a test of problems practiced during the CAI program. These results suggest that attribution retraining may be a desirable addition to the type of feedback typically provided by CAI programs. However, they also highlight the need for further research that examines the conditions under which specific attributions are most advantageous.
S
tudents with learning disabilities usually are identified and referred for special education services because of a specific academic deficit. However, research amply documents that these students also experience motivational deficits that further interfere with their academic performance or that may partially contribute to their learning handicaps (Deci & Chandler, 1986). One particularly important aspect of students' motivation for school tasks is their belief about the reasons they succeed and fail, or their causal attributions. Research has consistently demonstrated that students who attribute successes to their ability and effort and failures to their lack of effort tend to fare well on measures of academic success, including persistence (Andrews & Debus, 1978; Dweck
& Reppucci, 1973), expectancies for future success (Weiner, 1974, 1985), strategic behavior (Diener & Dweck, 1980), teacher ratings (Kistner, Osborne, & LeVerrier, 1988), and scores on achievement and criterion-referenced tests (Kistner et al., 1988; Newman & Stevenson, 1990). Research regarding the types of attributions made by students with learning disabilities for their academic successes and failures defies a concise synthesis. Researchers have studied different samples, have used varying instruments and procedures, and often have obtained equivocal results. In general, however, studies have shown that students with learning disabilities are more likely than their nondisabled peers to attribute their successes not to effort and ability, but to luck or the
JOURNAL OF LEARNING DISABILITIES VOLUME 25, NUMBER 5, MAY 1992 Downloaded from ldx.sagepub.comPAGES at UNIVERSITE DE MONTREAL on June 9, 2015 327-334
ease of a task (Jacobsen, Lowery, & DuCette, 1986; Pearl, 1982; Pearl, Bryan, & Donahue, 1980; Tollefson et al., 1982). Similarly, these students are more likely to attribute their failures to lack of ability rather than lack of effort (Diener & Dweck, 1978; Jacobsen et al., 1986; Palmer, Drummond, Tollison, & Zinkgraff, 1982; Thomas & Pashley, 1982). These attributional patterns have important implications for the academic progress of students with learning disabilities. Unless students believe that their successes are due to their abilities and efforts, they are unlikely to experience enhanced feelings of pride and selfesteem or feel confident about their chances for future success (Weiner et al., 1971). Furthermore, students who attribute their failures to lack of ability are more likely to have pessimistic expectations for future success, less likely to persist at academic tasks, and more likely to be debilitated by failure (Dweck & Reppucci, 1973; Graham, 1991; Peterson & Barrett, 1987; Weiner et al., 1971). These less-thanoptimal attributional patterns are of particular concern when one considers that students with learning disabilities must persevere in the face of failure to master many basic academic skills. Dweck (1975) demonstrated that maladaptive attributional patterns are amenable to change. She taught 12
328
JOURNAL OF LEARNING DISABILITIES
students, who were characterized as ' 'learned helpless" and whose performance deteriorated in the face of failure, to attribute their failures to lack of effort. Subsequently, these students were able to persist and even increase their efforts after failure experiences. Since Dweck's classic study, many researchers have replicated the general finding that students' attributions can be modified. Although most of these attribution retraining studies have focused on academic tasks, they also have addressed behaviors as diverse as littering (Miller, Brickman, & Bolen, 1975), interpersonal persuasion (Anderson, 1983), assembly work (Zoeller, Mahoney, & Weiner, 1983), and dropping out of school (Wilson & Linville, 1982,1985). Student with learning disabilities and nondisabled students who have participated in attribution retraining interventions have shown increased expectancies for future success, increased persistence, and enhanced performance on target tasks. In most studies, students also experience changes in attributions for tasks similar to those used in the study (Forsterling, 1985; Pearl, 1985). Changes in more general attributional patterns or dispositions have been more difficult to achieve (e.g., Borkowski, Weyhing, & Carr, 1988), leading researchers to conclude that attributions are domainspecific (Newman & Stevenson, 1990). Most attribution retraining studies have attempted to engender more frequent effort attributions among participants (Forsterling, 1985; Graham, 1991), but the studies have varied widely in their methods. Often, researchers have modeled adaptive attributions during success and failure at a target task (Dweck, 1975; Schunk, 1982, 1984; Thomas & Pashley, 1982). In other cases, peers have modeled adaptive attributions, often on videotape (Shelton, Anastopoulos, & Linden, 1985; Zoeller et al., 1983). Many researchers have required students to state the desired attributions and sometimes have reinforced them for doing so (Andrews & Debus, 1978; Thomas & Pashley, 1982). Other researchers have found that indirect
methods, in which students listen to or read attributions, are as successful as more direct methods (Fowler & Peterson, 1981). Attribution retraining research has been criticized for its overreliance on laboratory studies (Graham, 1991), and researchers have questioned whether it is feasible to implement attribution retraining in school settings (Pearl, 1985; Stipek & Weisz, 1981). Computer-based activities offer a promising but little-researched medium for the delivery of attribution retraining within classrooms. Computers are becoming more widely available in elementary and secondary schools, and students with learning handicaps often use computer-assisted instruction (CAI) to practice basic skills (Becker, 1990; Becker & Sterling, 1987; Cosden & Abernathy, 1990). Because computer software can analyze students' performance on an ongoing basis and provide highly individualized feedback, it may be an extremely effective vehicle for providing students with adaptive attributional feedback. Moreover, some studies have found that students perceive the computer's assessment of their performance to be fairer and more credible than teachers' (Okolo, Rieth, & Bahr, 1989). However, only one study could be found that examined the effect of computer-delivered attribution retraining. Peter, McArthur, and Stasz (1988) examined the effects of attributions delivered either by a computer or by a human tutor. High school students practiced difficult algebra problems and, after each failure, were informed that they had failed due to either lack of effort or bad luck. The computer-delivered feedback was as effective as the human tutor in changing students' persistence and expectancies for success. The purpose of the present study was to examine the impact of computer-based attribution retraining for students with learning disabilities. Based on previous research regarding the relationship between attributions and school success, the attribution retraining provided in this study was designed to engender attributions to
ability and effort for successes and to lack of effort for failures. A publicdomain software program that delivered practice in multiplication computation was modified either to provide the attributions described above or to inform students of their progress. Three predictions were tested: First, students with learning disabilities who received attribution retraining should develop more adaptive attributions for success and failure in computer-based mathematics tasks. Second, attribution retraining should have a positive impact on students' persistence on multiplication problems. Finally, attribution retraining should have a positive impact on students' multiplication computation skills.
Method Materials CAI Program. The public-domain CAI program used in this study, Drill (Davis, no date), contains four sets of multiplication problems of increasing difficulty. Each 25-problem set presents 5 problems per screen. The opening screen of the program directs the student to choose a problem set and to enter the amount of time, per second, that should be allotted for solving each problem. Students in this study worked through the program in a series of levels. At the first level, students were directed to start with the easiest problem set and enter a 20-second-perproblem time limit. All students were told that their goal was to solve a problem set three times within the time limit with at least 80% accuracy. Upon reaching their goal, students were told to advance to the next level, in which they decreased the problem solution time by 5 seconds. When students had solved a problem set three times with 80% accuracy within a 5-second-perproblem limit, they were permitted to advance to the next most difficult problem set and allocate 20 seconds per problem. To help them visualize their progress, all potential levels of the program were represented on a bar graph. Students filled in a bar each
Downloaded from ldx.sagepub.com at UNIVERSITE DE MONTREAL on June 9, 2015
VOLUME 25, NUMBER 5, MAY 1992
time they completed a level (i.e., solved a problem set three times within a specified time limit with 80% accuracy). The program was modified to provide either attribution retraining or neutral (control condition) feedback after each screen of 5 problems and at the conclusion of each 25-problem set. In both conditions, after each 5-problem screen, the program displayed information about the student's accuracy (e.g., "You now have 15 correct out of 15 tries"). When attribution retraining students were meeting their goals, they received ability attribution feedback ("You really know these") after 5, 10, 15, and 20 problems, and effort attribution feedback ("You are really trying hard") after 25 problems. When attribution retraining students were not meeting their goals, they received effort attribution feedback ("You can get it if you keep trying" after 5, 10, 15, and 20 problems and "You can do better if you try harder" after 25 problems). When control group students met their goals, they received neutral feedback ("You are meeting your goal" after 5, 10, 15, and 25 problems and "You met your goal" after 25 problems). The control students also received neutral feedback when they were not meeting goals ("You are not meeting your goal" after 5,10,15, and 20 problems and "You did not meet your goal" after 25 problems). Initial Attributional Patterns. To equate the groups on initial attributional patterns, the Intellectual Achievement Responsibility (IAR) questionnaire (Crandall, Katkovsky, & Crandall, 1965) was administered to students at the start of the study. The IAR contains 34 item stems that depict either a positive or a negative achievement experience. For each item stem, students choose an alternative that represents a cause either controlled by the respondent or controlled by someone else in his or her environment. Students' total scores and their scores on the 19 items that specifically addressed attributions to effort were computed. Scores on the effort items were used as a blocking variable.
Mathematics Attribution Scale. Three items were developed to assess changes in those attributions targeted by the CAI program. For each item, students first were asked to imagine succeeding (or failing) at a computerbased mathematics activity. Then they were asked to rate the likelihood, on a 5-point scale, that (a) effort was responsible for success, (b) ability was responsible for success, and (c) lack of effort was responsible for failure. These three items were administered as a pre- and posttest measure. Persistence Measures. Two measures of persistence were used. First, as described above, students graphed the number of program levels they completed. The second measure was discretionary time for either continued multiplication practice or computer game play. During six sessions of the study, students were given the option of playing a computer game after completing 20 minutes of computer-based multiplication practice. The total number of minutes that each student used the game program was recorded. Computation Test. A multiplication computation test was developed as a pre- and posttest measure of skill attainment. It contained four sets of problems identical to those encountered in the CAI program. Students were directed to complete each problem set as quickly and accurately as possible and were timed. The number of digits they correctly computed per minute was counted and averaged across the four problem sets.
Sample The sample was drawn from resource room programs in five public schools in a large, urban school system that serves approximately 431,000 pupils. In this school system, "learning disability" is defined as a disorder in one or more of the basic psychological processes involved in understanding or using spoken or written language that is not caused by environmental disadvantage, mental retarda-
Downloaded from ldx.sagepub.com at UNIVERSITE DE MONTREAL on June 9, 2015
329 tion, or emotional disturbance. Students are placed in learning disability resource room programs when there is a 1- to 2-standard-deviation difference between their achievement and aptitude (as measured by standardized test scores) in one academic area. Resource room students spend more than 50% of the day in regular classes. The resource room teacher at each school was asked to select from her caseload seventh- and eighth-grade students with learning disabilities who were at least 1 year below grade level in mathematics achievement and could benefit from additional practice in multiplication computation. All students who met these criteria (N=37) were invited to participate in the study, and the 33 students who returned permission forms were included in the original sample. The final sample consisted of 29 students; 4 students were unable to complete the study due to schedule conflicts or sporadic attendance. Sixteen students were males and the mean age of the sample was 13.3 years (SD = 1.1). Fifteen (52%) were Caucasian, 9 (31%) were African-American, 3 (10%) were Hispanic, and 2 (3%) were Asian. Participating schools were located in lower middle to middle class neighborhoods. Average level of parental education for the sample was 12th grade. Students were 1 to 2 years below grade level in mathematics and 2 to 3 years below grade level in reading, as measured by the most recent administration of the Iowa Test of Basic Skills. Because IQ tests are not routinely administered in this school system, recent IQ data were unavailable for most of the sample. In an individual interview conducted before the study, students were asked to describe the locations in which they had used a computer and the activities in which they had engaged. All had previous computer experience and 17 had home computers. Over 50% of the sample had engaged in video game play, graphics, word processing, keyboarding practice, programming, mathematics and language arts CAI, and data-base activities.
330
Students were blocked on IAR effort score and then randomly assigned to either the attribution retraining or control condition. Gender and ethnic composition were similar for each condition and, as displayed in Table 1, the groups were comparable in total IAR score, chronological age/parental education level, and mathematics grade equivalent. Also, the groups were alike in the number of locations in which they had used computers and the types of computer activities in which they had engaged before the study.
Procedure The study was conducted in 30-minute sessions that took place in the student's resource room or the school's computer lab. Students worked in small groups of four to eight composed of other students who were participating in the same condition. On all pretest and posttest measures, research assistants read directions and, except for computation tests, individual items aloud. Students completed the IAR, mathematics attribution scale, and mathematics computation tests and then were introduced to the CAI program. Research assistants modeled its use and read aloud all feedback it provided. They also made sure that all students in each group could read and explain the program's feedback and scores. Students then were shown how to complete their progress graphs. Students worked independently at the computer for the next eight sessions. During the second session, research assistants demonstrated two computer games that the students could choose to play after 20 minutes of CAI practice. Research assistants monitored students throughout each session, informed them when 20 minutes had passed, reminded them they could play a game if they wished, and recorded the number of minutes each student engaged in the game. At the conclusion of every session, a research assistant reviewed each student's graph with him or her. If students had met their goal twice that day, the research assistant stated that they were
JOURNAL OF LEARNING DISABILITIES
TABLE 1 Demographic Variables Attribution retraining M SD IAR Effort Score IAR Total Score Age Parental Education Level Mathematics Grade Equivalent Reading Grade Equivalent Number of Locations of Computer Use Number of Types of Computer Use
14.1 24.9 13.6 12.9 6.3 5.1 6.9 4.6
2.5 4.0 .9 2.1 1.4 1.1 2.9 2.2
Control M SD 13.8 24.4 13.0 12.3 5.8 5.0 7.5 5.0
2.2 2.8 1.2 2.3 2.0 1.5 1.9 1.2
Note. IAR = Intellectual Achievement Responsibility questionnaire.
doing a good job. If they had not met two goals, the research assistant commented that the students had not met their goals that day. Research assistants provided no other instructions or reinforcement to students throughout the study. After each student completed eight computer sessions, the mathematics attribution scale and the multiplication computation test were readministered.
Fidelity of Implementation Several steps were taken to ensure that the procedures were implemented as described above. First, the five research assistants who conducted the sessions memorized a script for introducing the CAI program in each condition. They then role-played the introductory session among themselves, critiqued each other's performance, and proved to the author that they could satisfactorily deliver the session without errors. Twenty-five percent of the sessions at each school were observed by the author or the project coordinator, who ensured that research assistants were providing the proper feedback after each session without instructing or reinforcing students during the session. Finally, research assistants rotated regularly among schools and between conditions.
Results The primary dependent measures in this study were (a) attributions (pre-
and posttest mathematics attribution scale), (b) persistence (number of program levels completed and number of minutes spent on the game program), and (c) skill attainment (pre- and posttest computation tests).
Attributions Table 2 presents means and standard deviations for each of the three attribution items. These data were examined via three separate 2 (time of test) by 2 (groups) repeated measures ANOVAs. To control experiment-wise error, a Bonferroni adjustment was made and a was set at .02. There was a trend for students in both conditions to increasingly attribute successes to effort and failures to lack of effort. In addition, students in both groups were less likely to attribute successes to ability. However, preto posttest differences on these items were minimal and none were reliable, although the main effect for time of test on the attribution of success to ability item approached significance, F(l,27) = 2.9, p< .05.
Persistence As described above, students filled in a bar graph each time they completed a level of the program. Attribution retraining students completed an average of 16.0 (SD = 3.9) levels, whereas control students completed 11.6 (SD=3.3) levels. After testing and accepting the assumption of parallel slopes, F(l,24) = .23, an ANCOVA was
Downloaded from ldx.sagepub.com at UNIVERSITE DE MONTREAL on June 9, 2015
331
VOLUME 25, NUMBER 5, MAY 1992
TABLE 2 Pre- and Posttest Attribution and Computation Scores Attribution retraining M SD Attributions for success to ability Pretest Posttest Attributions for success to effort Pretest Posttest Attributions for failure to lack of effort Pretest Posttest Multiplication computation scores Pretest Posttest
computed on number of levels completed, with pretest multiplication score as the covariate. The attribution retraining and control groups were significantly different on this measure, F(l,25) = 9.6, p
