Você está na página 1de 9

Journal of Educational Psychology

1989, Vol. 81, No. 3, 353-561

Copyright 1989 by the American Psychological Association, Inc.


0022-O663/89/SOO.75

Components Analysis of Cognitive Strategy Instruction: Effects on


Learning Disabled Students' Compositions and Self-Efficacy
Steve Graham and Karen R. Harris

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Department of Special Education


University of Maryland
The viability of self-instructional strategy training among learning disabled (LD) students exhibiting composition deficiencies was investigated. Furthermore, the theoretically proposed incremental effects of explicit self-regulation procedures were examined in terms of writing performance measures at posttest, maintenance, and generalization, and in terms of Ss self-efficacy. Ss
were 22 LD and 11 normally achieving students in the 5th and 6th grades. Results indicated that
self-instructional strategy training produced meaningful and lasting effects on Ss' composition
skills and a significantly heightened sense of self-efficacy. Explicit self-regulation procedures did
not significantly augment strategy-instruction effects for either writing performance or selfefficacy. Composition performance after instruction among LD Ss did not differ significantly in
terms of story grammar elements from that of a contrast group of normally achieving, competent
writers. However, normally achieving students' compositions were longer and received significantly higher quality ratings.

In 1980, Hobbs, Moguin, Tyroher, and Lahey commented


that relatively few cognitive-behavioral studies had focused
on academic skills. Since that time cognitive strategy instruction in academic areas has become a major focus of educational research with the mildly to moderately handicapped
and other inefficient learners (cf. Meichenbaum, 1983; Pressley & Levin, 1986). Cognitive-behavioral theorists have proposed that effective strategy instruction involves three major
components: strategies, knowledge about the use and significance of those strategies (metastrategy information), and selfregulation of strategic performance (cf. Brown, Campione, &
Day, 1981; Harris & Graham, 1985).
Multicomponent strategy instruction interventions based
on this theoretical view have proved efficacious in improving
performanceand frequently in obtaining generalization and
maintenanceamong young children and problem learners
(Harris, 1986a; Reeve & Brown, 1985). Pressley and Levin
(1986) indicated that such interventions have frequently resulted in performance among handicapped learners equivalent to that obtained by their nonhandicapped peers. Previous
self-instructional strategy-training studies have incorporated
strategy instruction (using a self-instructional format), metastrategy knowledge, and explicit self-regulation. This ap-

proach has been successful in improving written language and


mathematical problem-solving skills among learning disabled
(LD) students (Graham & Harris, 1989, in press; Graham,
Harris, & Sawyer, 1987; Harris & Graham, 1985).
Although multicomponent interventions are frequently effective, a major issue in cognitive strategy instruction research
is determining both the relative contributions of instructional
components and the variables responsible for change. Costbenefit relationships of components also need to be determined (Harris, 1985; Pressley, Forrest-Pressley, & ElliottFaust, in press). Component analyses studies remain rare,
however. Elliott-Faust and Pressley (1986) demonstrated that
multicomponent strategy instruction, including a complete
and effective strategy embedded in a self-instruction routine
combined with self-monitoring of strategy use and effectiveness, was superior to simply teaching children the task strategy. However, no studies were located that investigated the
three theoretically-based critical elements of multicomponent
strategy instruction.
It has been theorized that explicit instruction in self-regulation of strategic performance will result in incremental
effects and is critical to effective strategy deployment, production of new metastrategy information, independent strategy
use, and maintenance and generalization of effects (O'Leary
& Dubey, 1979; Pressley & Levin, 1986). Self-regulation
procedures have also been effective in increasing children's
self-efficacy (Bandura & Schunk, 1981). In addition, previous
research has established that LD students respond enthusiastically to self-regulation procedures and would highly recommend such procedures to other students experiencing learning
problems (Graham & Harris, 1989; Harris, 1986b; Harris &
Graham, 1985). It is not known, however, if explicit selfregulation procedures will produce incremental effects when
combined with an intervention that provides strategy instruction and metastrategy information. Thus, a major purpose of
the present investigation was to investigate possible incremen-

This study was supported by a grant from the University of


Maryland Graduate Research Board to Karen R. Harris; authorship
was determined alphabetically. Appreciation is expressed to the children, teachers, and principals of the Prince George's County schools
participating in this project. Gratitude is extended to the strategy
instructors, Linda Artman, Cassie Chapman, Barbara Danoff, Tricia
Marker, Beth North, and Andrea Rothman. Finally, we would like
to thank Michael Pressley and the anonymous reviewers for their
feedback on a draft of this article.
Correspondence concerning this article should be addressed to
Steve Graham, Department of Special Education, College of Education, University of Maryland, College Park, Maryland 20742.
353

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

354

STEVE GRAHAM AND KAREN R. HARRIS

tal effects that are due to instruction in self-regulation of


strategic performance.
A second major purpose of this investigation was to determine the viability of a cognitive strategy instruction approach
among LD students exhibiting composition deficiencies.
Composition skills represent a significant deficiency among
these, students. Merely requiring students, whether LD or
normally achieving, to write more frequently typically fails to
result in improved writing performance; practice alone does
not improve writing skills (Graham, 1982; Graham, Harris,
& Sawyer, 1988). Creative compositions written by LD students frequently lack even the most basic story elements
(Barenbaum, Newcomer, & Nodine, 1987; Graham & Harris,
1989; Nodine, Barenbaum, & Newcomer, 1985). Furthermore, judgments concerning the overall quality of LD students' writing are significantly related to the number and
quality of basic story elements included in their compositions
(MacArthur & Graham, 1987).
Stein and Glenn (1979) noted that the grammar or structure
of stories can be reduced to two major components: setting
and episode. The setting typically incorporates three elements:
introduction of the main character, a description of the locale,
and information about the time of the story. Episodes introduce and resolve the plot of the story and typically include
five elements: starter event (a precipitating event occurs), goal
(the main character reacts to the precipitating event and
formulates a goal), action (a planned effort to achieve the
goal), ending (the results of the action), and reaction (an
emphatic statement or a final response of a story character to
the consequences of the action). Short and Ryan (1984)
reported that story grammar training based on these elements
produced meaningful gains in reading comprehension among
poor readers. Short and Ryan's training emphasized comprehension monitoring, through the use of five questions about
settings and episodes, to select important aspects of story
information for further study. In the present study, the story
grammar strategy component of training emphasized the
generation of seven questions about setting and episode; these
seven questions include and expand upon the five questions
used by Short and Ryan. In the present investigation, however, these questions were used in a self-instructional strategytraining intervention designed to facilitate advanced planning
and the development of writing content.
The Present Investigation
Two self-instructional strategy training groups were included in the present study. Both groups received the story
grammar strategy instruction and instruction in the significance of this strategy. One group also received instruction in
explicit self-regulation. Self-regulation procedures included
criterion setting and self-monitoring (self-assessment and selfrecording). Instruction was conducted by preservice teachers
in the schools.
Multiple measures were used to assess instructional procedures and effects. The incorporation of story grammar schemata in LD students' compositions was assessed prior to and
immediately following instruction. To determine if changes
in schematic structure were associated with improvement in

story quality, pre- and posttest compositions were also evaluated in terms of overall quality. Furthermore, changes in
individual story grammar elements were examined to determine which aspects of narrative structure were affected by
instruction. Generalization across settings and short-term
maintenance were also assessed. In addition, a group of normally achieving, competent writers served as a contrast group
in posttest assessments, to determine if any differences in the
educational significance of the instruction existed between the
two groups. Treatment validity was determined by collecting
evidence on students' use of the story grammar strategy both
during instruction and independently. Teacher and student
comments collected throughout instruction were helpful in
further determining the social validity of the two intervention
approaches. Finally, LD students' self-efficacy for creative
composing was assessed before and after instruction.
In summary, the present study addressed five critical questions: (a) Do LD students benefit from self-instructional story
grammar strategy training (in either or both conditions)? (b)
Do explicit self-regulation procedures meaningfully augment
self-instructional strategy training combined with instruction
in the significance of the strategies? (c) Can validation of
treatment procedures be provided (i.e., can treatment validity
be established)? (d) Can evidence of educational validity be
established? (e) Did either or both of the intervention approaches meaningfully affect LD students' perceived selfefficacy?
Method

Subjects
LD. Subjects were 22fifth-and sixth-grade LD students receiving
resource room services in three elementary schools located in suburban, middle-class neighborhoods outside of Washington, DC. Eightysix percent of the students enrolled in these three schools were Black.
Because of problems of validity with the LD label and the heterogeneity of school-identified LD populations (Harris, 1986a), all of the
LD subjects selected for this study met the following stepwise criteria:
identification as LD by the school district, 1Q scores between 85 and
115 on an individually administered intelligence test, achievement
that was at least 2 years below grade level in one or more academic
areas, absence of any other handicapping condition, and interviews
with the resource room teachers indicating that significant composition problems were evident. IQ scores for the LD subjects were from
either the Wechsler Intelligence Scale for Children-Revised (WISCR; Wechsler, 1974) or the Slosson Intelligence Test (SIT, Slosson,
1971). All IQ scores were taken from school files and had been
administered by school psychologists for all of the subjects within the
past 3 years. Although WISC-R scares were preferable, they were
available for only 6 of the 22 LD subjects. Achievement scores for all
LD subjects were from the Peabody Individual Achievement Test
(Dunn & Markwardt, 1970), administered within the past year by
school personnel and taken from students1 files.
Subjects were randomly identified and evaluated on the basis of
these criteria until an equal number of fifth- and sixth-grade LD
students were selected. Fourteen of the subjects were boys and 8 were
girls. Seventeen of the subjects were Black, and 5 were White. Mean
IQ score was 95 (range = 85-113); mean chronological age was 11
years, 8 months (range = 10 years, 2 months-13 years, 7 months).
Further information on subject characteristics within experimental
conditions is presented in Table 1.

COMPONENTS ANALYSIS OF COGNITIVE STRATEGY INSTRUCTION

Table I
Subject Characteristics by Experimental Groups

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Group8
NA
SIST
SIST+SR
Variable
CA
142.18
137.64
133.60
M
SD
7.32
11.82
6.01
IQ
M
93.45
97.09

SD
8.03
6.61

Race
Black ()
9
8
8
White ()
2
3
3
Sex
Female (n)
4
4
5
Male (n)
7
7
6
Grade
5th (n)
5
6
5
6th (n)
6
5
6
Note. SIST = self-instructional strategy training; SIST+SR = selfinstructional strategy training plus self-regulation training; NA =
normally achieving; CA = chronological age.
a
n 11 for each group.
LD subjects were randomly assigned to one of the two self-instructional strategy-training conditions (self-instructional strategy training
or self-instructional strategy training plus self-regulation training).
Separate univariate analyses of variance (ANOVAS) were performed to
determine if there were any significant differences betwen the two
groups on important subject variables. No significant differences were
found on IQ, F (1, 20) = 1.36, p = .26, chronological age, F (1, 20)
= 1.18, p = .29, pretest self-efficacy, F(\, 20) - .38, p = .55, pretest
quality ratings, F (1, 20) = .02, p .90, and pretest story grammar
elements, F (1, 20) = .16, p = .70. Procedures for assessing selfefficacy, quality, and story grammar elements are described in the
next section.
Normally achieving. The normative comparison group consisted
of 11 randomly selected students who attended the same schools as
the LD subjects. IQ testing of normally achieving students was not
allowed by the school system. Although normal performance is
commonly defined at plus and minus one standard deviation from
the mean (equivalent to a percentile rank range of 16 to 84), we used
a more conservative definition of the lower end of the range. The
normally achieving subjects scored between the 29th and 81st percentiles (M 64.0) on the Cognitive Abilities Test (Thorndike &
Hagen, 1978) and above the 29th percentile on the following subtests
of the California Achievement Tests (1979): reading vocabulary (M
= 59.2), reading comprehension (M = 65.6), language mechanics (M
= 76.2), and language expression (M = 67.4). These students received
no special education services and were capable writers according to
their classroom teachers. Composition of this contrast group was
equivalent to that of the two LD training groups in terms of sex, race,
and grade (see Table 1).

Instruments and Materials


Story grammar elements. We developed a scale for assessing the
schematic structure of written stories. The scale was designed so that
students' stories could be assessed in terms of the inclusion and
quality of the eight story grammar elements identified by Stein and
Glenn (1979): main character, locale, time, starter event, goal, action,
ending, and reaction.

355

For each story element, a score of 0 was assigned if the element


was not present in the story, and a score of 1 was assigned if the
element was included. Elements that were highly developed received
a score of 2. For the goal story element, a score of 3 was awarded if
two or more goals were present. Similarly, for the action story
element, scores of 3 or 4 were assigned if the actions or events
happened in a logical manner or if there was more than one welldefined episode (or both). Scores were determined for each element
separately; in addition, by totaling the scores for the individual
elements, a total story grammar element score was calculated (a total
of 19 points was possible).
All stories written by a subject were scored both by their instructor
and by a second trained examiner who was naive to the student's
group assignment and the purpose and design of the investigation.
For all scores (individual element scores and total score), the ratings
of the instructor and second examiner were averaged. Interobserver
reliability between the instructors and the second examiner for the
total story grammar element score was .80. Interrater agreement on
presence of the element for each of the eight story elements was as
follows: .95, for main character, .94, for locale; .96, for time; .64, for
starter event; .95, for goal; .93, for action; .84, for ending; and .96,
for reaction.
Evidence on the validity of the story grammar element scale was
obtained by correlating the total scores of 26 normally achieving
sixth-grade students with (a) their scores on the Thematic Maturity
subtest of the Test of Written Language (Hammill & Larsen, 1983)
and (b) a measure of written fluency (i.e., the number of words
written). Students' scores on the scale were moderately correlated
with thematic maturity (/ = .40) and number of words written (r =
.32). In addition, MacArthur and Graham (1987) reported that LD
students' total scores on this scale were significantly correlated with
measures of story length ( = .8O-.85) and story quality (rs = .71.86).
Holistic rating scale. A holistic rating scale was used to assess the
quality of subjects' stories following Graham's (1982) recommendations. Examiners were directed to read each story attentively but not
laboriously in order to obtain a general impression of quality. Examiners were told that aptness of word choice, grammar, organization, sentence structure, and imagination should all be taken into
account in forming a single judgment about the overall quality of the
writing sample and that no one factor should receive undue weight.
Compositions were then scored on a 7-point scale, with 1 representing
the lowest quality of writing, and 7 representing the highest quality.
Examiners were provided with a representative sample of a low,
medium, and high scoring composition to use as a guide or anchor
point in scoring the subjects* compositions. The high, medium, and
low compositions were obtained from a regular sixth-grade class; all
students in this class not receiving special services wrote a story, and
two trained graduate students selected the best, the middle, and the
poorest story on the basis of the scoring criteria noted above.
Two elementary-school teachers unfamiliar with the purpose and
design of the study independently assigned holistic quality scores to
the stories written by the LD students during the pretest and posttest
writing probes and to the stories written by the normally achieving
students. Prior to scoring the students1 stories, both examiners received considerable training and practice in using the holistic scoring
procedure. Interobserver reliability between the two examiners was
.76. For each story, the ratings of the two examiners were averaged.
Self-efficacy measure. An individually administered scale assessing self-efficacyor judgments of one's capability to perform given
activitiesfor creative writing was developed and administered according to procedures detailed by Bandura and Schunk (1981). The
efficacy scale ranged from 10 to 100 in 10-unit intervals; the higher
the scale value, the higher the perceived self-efficacy. Verbal descriptors occurred at the following points: 10 {not sure), 40 (maybe), 70

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

356

STEVE GRAHAM AND KAREN R. HARRIS

{pretty sure), 100 {real sure). Practice with the efficacy assessment
procedure was provided by having subjects judge their capability to
jump progressively longer distances, from a few inches to several
yards. Following this introduction, subjects were read 10 items probing self-efficacy for writing "a made-up story." The first 9 questions,
all introduced with the phrase "Can you write a story that,** were as
follows: (a) tells about the main character's feelings? (b) clearly tells
about the setting? (c) has a good beginning? (d) tells who the main
character is? (e) tells about several things that happen to the main
character? (f) tells when the story takes place? (g) tells where the story
took place? (h) tells what the main character wants to do? (i) has a
good ending? The 1 Oth question was (j) Can you write a good, creative
(made-up) story? Subjects were asked to be honest and to mark
privately the appropriate number on the scaleforeach item. Subjects
understood these procedures and did not experience difficulty completing their judgments. The summed magnitude scores divided by
the total number of questions provided the measure of strength of
self-efficacy.
Coefficient alpha for the self-efficacy scale was determined to be
.80 among 26 normally achieving sixth-grade students prior to this
study. In addition, the internal-consistency reliability (coefficient
alpha) of the self-efficacy measure among the 22 LD students in this
study at pretest was .76.
Writing stimuli. Four black-and-white pictures were used as writing stimuli for four different writing probes: pretest, posttest, generalization (administered in the student's resource room following
treatment), and maintenance (administered 2 weeks after the termination of treatment). Pictures were selected to be interesting to
elementary-school-age children, to be fairly easy to write about, and
to be similar in nature (showing only one character). The four pictures
were (a) a boy walking across a field, (b) a scuba diver in the water,
(c) an Indian on a horse, and (d) a man in a boat. The pictures were
randomly assigned to the pretest, posttest, generalization, and maintenance probes. Pictures were administered to all subjects in the same
order, because severalresearchers(cf. Englert & Thomas, 1987) have
demonstrated that similar writing stimuli produce similar results
(thus, randomly ordering stimuli for each subject appeared unnecessary). Additional, similar black-and-white pictures served as writing
stimuli for practice stories written during training.

General Procedures
Six senior-level undergraduate students who had extensive field
experience and were majoring in special education served as instructors; each instructor worked with an equivalent number of students
from both of the instructional groups. Instructors were blind to the
specific theoretical issues examined by the study so as not to predispose them to either approach. The instructors were introduced to all
subjects as "special writing teachers" who would be working at their
school. Instructors received considerable training and practice in the
application of the instructional procedures, until they were able to
role-play the lessons without error. Detailed lesson plans containing
step-by-step procedures for each phase of instruction were provided
in a notebook. Each step of the daily lesson plan was checked off as
it was completed; we monitored these notebooks closely and talked
with each instructor frequently throughout the study to ensure standard training procedures across instructors.
Instructors worked with small groups of two or three students at
their schools; instructors met with each group for approximately 45
min, 2 to 3 days a week, for 2 to 3 weeks. Because instruction was
criterion-based (students were required to master each lesson before
proceeding to the next), the number of sessions and amount of time
required varied slightly for each instructional group. The number of
sessions required for the strategy instruction groups ranged from five

to seven, with instructional time ranging from 5 to 6 hr in total. For


the strategy instruction plus self-regulation groups, the number of
sessions ranged from five to eight; total instructional time ranged
from 4 to 6.5 hr. Thus, the self-regulation components were easily
incorporated into the instructional procedures and did not necessitate
either extra sessions or longer instructional times. All of the subjects
mastered each of the instructional steps, and none of the subjects
found the instruction to be particularly complicated, perhaps because
of the criterion-based nature of instruction.
The two pre- and posttests (writing probe and self-efficacy measure)
were administered individually by the writing instructors on different
days (to avoid carryover effects) at the subject's school, as was the
maintenance writing probe. The short-term maintenance probe was
administered 2 weeks after the termination of instruction; a second,
longer term maintenance probe was not possible because of the
conclusion of the school year. The generalization-across-settings writing probe was collected in the resource room by the resource room
teacher within 1 week after instruction. The writing probe for normally achieving subjects was administered by an instructor congruent
with the time of posttesting for LD subjects. At no time during any
of these sessions were subjects provided feedback on their performance. When the pretest, posttest, maintenance, and generalization
writing probes were administered, the following standardized instructions were given:
Look at this picture and write a story to go with it. Use everything
that you have learned about writing stories to help you. Please
remember that I cannot help you write the story. However, if
you do not know how to spell a word, I will write it out for you.

Instructional Procedures
The self-instructional strategy training steps and components were
developed and validated previously (cf. Graham & Harris, in press;
Graham, Harris, & Sawyer, 1987; Harris & Graham, 1985). In both
conditions, Instruction emphasized the student's role as an active
collaborator and emphasized interactive learning between teacher
and students, with responsibility for recruiting and applying strategies
gradually placed upon the student. Principles of interactional scaffolding and Socratic dialogue were incorporated; instructors were
enthusiastic and responsive to each child and provided individually
tailored feedback. Strategies were explicitly and overtly modeled in
context; the goal and significance of the strategies were also made
clear. Finally, all instruction was criterion-based rather than timebased, and previously taught skills/strategies were routinely reviewed.
The strategy instruction and the strategy instruction plus selfregulation groups received the same instructional program except
that the strategy instruction plus self-regulation groups received instruction and practice in (a) self-monitoring their performance (recording and graphing the number and kind of story grammar elements contained in their practice stories) and (b) criterion setting
(establishing goals for the number of elements to be included in
subsequent practice stories). Instructional steps for the two groups
were as follows (all steps were conducted in groups of two to three
students).1
Step 1: ^retraining. Pretraining focused on defining, identifying,
and generating story grammar elements using a mnemonic; this

1
Five lesson plans were developed to incorporate the seven training
steps. Lesson Plan 1 covered Training Steps 1 and 2; Lesson Plan 2
covered Training Steps 3, 4, and 5; Lesson Plan 3 covered Training
Steps 5 and 6; Lesson Plans 4 and 5 covered Training Step 7. Lesson
plans were typically completed in one or two training sessions. Copies
of these lesson plans are available from the authors.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

COMPONENTS ANALYSIS OF COGNITIVE STRATEGY INSTRUCTION


mnemonic was then used in the writing strategy taught later. A small
chart provided the mnemonic for the seven story grammar questions:
W-W-W; What = 2; How = 2. The questions were (a) Who is the
main character, who else is in the story? (b) When does the story take
place? (c) Where does the story take place? (d) What does the main
character want to do? (e) What happens when he or she tries to do
it? (f) How does the story end? and (g) How does the main character
feel? After discussing the meaning of each element, students practiced
until they could recite the mnemonic and its meaning from memory.
Students then identified story grammar elements in existing stories
and generated story elements while looking at a picture.
Step 2: Review current performance level and training rationale. Once students had mastered the story grammar elements, the
instructor and students discussed students' pretest performance in
terms of their ability to incorporate the elements into their compositions. The instructor and students also discussed the goal of instruction (to write better stories), why this is important, and how inclusion
and expansion of the story grammar elements improve a story.
Subjects in the strategy instruction plus self-regulation condition
received a graph depicting their performance on the pretest, and
graphing procedures were explained.
Step 3: Describe the learning strategy. A small chart was used to
introduce and discuss afive-stepwriting strategy. The five steps were
(a) look at the picture; (b) let your mind be free; (c) write down the
story part reminder (W-W-W, What = 2, How = 2); (d) write down
story-part ideas for each part; and (e) write your story: use good parts
and make sense. The instructor also modeled and then discussed with
the students three creativity self-statements helpful in thinking of
good story parts (e.g., "Take my time, good parts will come to me";
"Let my mind be free, think of new, fun ideas"; and "What ideas do
I see in this picture?"). Students then generated two or three of thenown, preferred self-statements, recorded them on paper, and practiced
using these self-statements to generate story parts.
Step 4; Model the strategy and self-instructions. The mnemonic
andfive-stepstrategy charts, the list of creativity self-statements, and
a new stimulus picture were set out. The instructor modeled the use
of the writing strategy by writing a story while "thinking out loud."
Consistent with Meichenbaum's (1977) self-instructional training
guidelines, the instructor modeled four additional types of self-instructions while composing: problem definition (i.e., "What is it I
have to do?"), planning (i.e., "Now I'd better write down my storyparts reminder"), self-evaluation (i.e., "Am I using all my parts so
far?"), and self-reinforcement (i.e., "Good, I like these parts!"). A
modeling script was provided in the lesson plans. After the modeling,
instructor and students discussed the importance of what we say to
ourselves while we work. Students generated and recorded their own
examples of the four types of self-instructions.
Step 5: Mastery of strategy steps. Students were required to
practice the five-step writing strategy until it was memorized. Paraphrasing was allowed as long as meaning remained intact.
Step 6: Controlled practice. The instructor and students conjointly planned one story following the first four steps of the five-step
strategy, each student then wrote the story independently. The mnemonic andfive-stepstrategy charts, as well as the student-generated
self-instruction lists, were available as prompts. Although the instructor directed and monitored the process, she did not write the stories.
In both conditions, the teacher and students reviewed each student's
story as a group; if any of the story elements were missing, the group
discussed how and where they could be added. For students in the
strategy instruction plus self-regulation groups, the students and
instructor initially set a goal to include all of the story grammar
elements. After the stories were written, each student in these groups
and the instructor independently counted the number of story elements included in the story, compared counts, graphed the number
on the student's chart, and compared performance to the criterion.

357

Step 7: Independent performance. Students independently composed two stories using the five-step strategy and self-instructional
statements. Positive and corrective feedback was provided as needed;
as in Step 6, the instructor and student reviewed each story (in both
conditions) and discussed how missing elements (if any) could be
included. Transition to covert self-instruction was encouraged. Students were allowed to use the charts and self-instruction list only for
the first of the two stories written. Students in the strategy instruction
plus self-regulation group followed the same goal-setting and selfmonitoring procedures described in Step 6.
Throughout the instructional sessions, students were asked to share
what they were learning with their teachers and parents. Students
discussed with their instructors how what they were learning could
be used in their resource and regular classrooms; spontaneous comments made by students during these discussions were recorded by
the instructors. Both the instructors and the investigators discussed
the interventions with the students and the resource room teachers
and recorded their comments. It should also be noted that any paper
that students wrote on during the writing probe sessions (pre, post,
maintenance, and generalization) was collected, because this could
provide evidence concerning strategy usage (i.e., Were the seven story
grammar questions and their responses written out prior to story
generation?).

Results
Story Elements: Total Score
A 2 (instructional groups) x 4 (trials) repeated measures
ANOVA design was used to examine the effects of the two

instructional conditions (strategy instruction vs. strategy instruction plus self-regulation) on LD subjects' total scores on
the story grammar element scale and to determine if these
scores differed significantly at pretest, posttest, generalization,
and maintenance. The F ratio for trials was significant, ^ 3 ,
60) = 20.85, p < .01, MSe = 2.80; however, the main effect
for instructional groups and the interaction effect were not
statistically significant. Thus, the stories written by LD students across the four writing probes differed in terms of their
story grammar structure, although the addition of explicit
self-regulation components did not significantly affect performance on this measure. Means and standard deviations
for total scores are presented in Table 2.
Multiple comparisons were made among the means by
using the Newman-Keuls test. At the .01 level, total story
grammar element scores at posttest, generalization, and maintenance were significantly higher than pretest scores. No other
significant differences were noted.
Comparisons with normal writers. Effectiveness of the
instructional procedures in terms of story grammar elements
was further examined by comparing the total story elements
scores of the two instructional groups on the posttest writing
probe (see Table 2) and the contrast group of normally
achieving students {M = 8.70). Total story grammar element
scores for the LD students at posttest ranged from 7 to 11.5
(the pretest range was 2.5 to 9), and scores for the normally
achieving subjects ranged from 5 to 12.5. A one-way ANOVA
for total score on the story grammar element scale yielded a
nonsignificant effect for group. In contrast, preplanned orthogonal comparisons at the .01 level revealed that prior to

358

STEVE GRAHAM AND KAREN R. HARRIS

Table 2
Means and Standard Deviations of Total Story Grammar
Elements Score as a Function of Training Group and
Writing Probe

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Group
Condition
Pretest
M
SD
Posttest
M
SD
Generalization
M

SD

SIST"

SIST+SR-

Combmedb

6.05
2.32

5.68
1.95

5.86
2.10

9.82
1.01

9.23
1.06

9.52
1.05

9.86
2.52

7.95
2.39

8.91
2.59

Maintenance
M
9.05
8.36
8.70
SD
1.29
2.06
1.72
Note. SIST = self-instructional strategy training; SIST+SR = selfinstructional strategy training plus self-regulation.

instruction the performance of the normally achieving students was significantly higher than the average pretest performance of the two instructional groups of LD students (M
= 5.86). As noted previously, the pretest performance of the
two groups did not differ significantly.
Neither the LD nor the normally achieving students scored
near the maximum possible on total story grammar elements
(to avoid ceiling effects, the scale was designed so that 19
points were possible only for unusual, exemplary stories).
However, the change in the writing of the LD students was
clearly educationally significant. Nodine et al. (1985) indicated that the minimal essential components of a creative
story are a main character, goal, action, and ending. Although
only 36% of the LD subjects' stories at pretest met this criteria,
91 %, 73%, and 73% of their stories met this criteria at posttest,
maintenance, and generalization, respectively. Among the
normally achieving subjects, 83% of stories met this criterion.
Furthermore, none of the LD subjects evidenced any debilitative effects from training. Total story grammar element
scores improved from pre- to posttest for 20 of the 22 LD
subjects, but scores for the other 2 subjects, who had attained
the two highest scores on the pretest, remained unchanged
from pre- to posttest. Finally, 86% of LD students' posttest
stories contained seven of the eight story grammar elements
possible; only 36% of pretest writing probes met this criteria.

Individual Element Scores


The scores for each of the separate story grammar elements
were analyzed using a one-way multivariate analysis of variance with the four trials as a factor. Because previous analyses
indicated no significant differences between the two groups
of LD students on total story grammar elements score, effects
of group membership were not analyzed. Results were evaluated based on Rao's approximation of WUks's lambda. A
multivariate effect for trials was obtained, F (24, 163) = 3.05,
p < .01. Separate univariate analyses revealed that significant

trial effects were obtained for the following story grammar


elements: time, F(3, 63) = 9.07, p < .01, MSe = .20; starter
event F(3, 63) = 3.94, p < .05, MSe = .24; goal, F(3, 63) =
11.52, p < .01, MSe = .44; action, F(3, 63) = 4.41, p < .01,
MSe = .55; ending F(3y 63) = 3.80, p < .05, MSe - .24; and
reaction, F(3t 63) = 7.95, p < .01, MS, = .15. Subsequent
post hoc analyses using the Newman-Keuls test indicated that
at the .01 level, posttest, generalization, and maintenance
scores were greater than pretest scores for the story grammar
elements of time, goals, and reaction. Identical results at the
.05 level were obtained for the ending story grammar element
For the action story grammar element, posttest and maintenance scores were significantly greater than pretest scores at
the .01 level, whereas, at the .05 level, generalization scores
were greater than pretest scores. For the starter event, only
the posttest score was significantly greater than the pretest
scores (p < .01). No other significant differences were noted.
Thus, with the exception of main character and locale, all
of the instructed narrative structures improved significantly
following training; maintenance and generalization of these
effects were obtained for all but one element, starter event.
Failure to obtain significant improvements in main character
and locale scores was due to LD students' high prior abilities
in generating these elements. Scores for each of the story
grammar elements as well as the percentage of students incorporating each element into their stories are presented in
Table 3.

Quality Ratings
A 2 (instructional groups) x 2 (trials) repeated measures
design was used to examine the effects of strategy
instruction, and strategy instruction plus self-regulation on
the quality of LD students' compositions and to determine if
the quality of their pre- and posttest stories differed significantly. Only the Fratio for trials was significant, F(l,20) =
6.15, p < .05 (MSe = 1.07); the main effect for instructional
groups and the interaction effect were not significant. The
average quality ratings on LD students' pretest and posttest
compositions were 2.14 (SD = 1.25) and 2.91 (SD = 1.27),
respectively.
Story quality was analyzed further by comparing the posttest ratings of the two instructional groups of LD students
and the contrast group of normally achieving students (M =
4.82). A one-way ANOVA for quality ratings yielded a significant effect for group, F{2, 30) = 6.18, p < .01, MS* = 2.18.
Preplanned orthogonal comparisons revealed that, at the .01
level, the performance of the normally achieving students was
significantly higher than the average posttest performance of
the two instructional groups of LD students (M = 2.91), As
noted earlier, the posttest performance of the two groups did
not differ significantly. Therefore, we next inspected the
length of stories. The mean length of stories written by normally achieving subjects was 160 words, whereas the mean
length of posttest stories among LD subjects was 83 words
(the pretest mean for the LD students was 85 words). Thus,
length appears to be a viable explanation for the higher quality
rating for normally achieving students' writing.
ANOVA

COMPONENTS ANALYSIS OF COGNITIVE STRATEGY INSTRUCTION

359

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 3
Learning Disabled Students' Mean Scores and Percentage of Stories Incorporating Each
Story Grammar Element
Pretest
Posttest
Generalization
Maintenance
Story grammar
element
M
%
M
%
M
%
M
%
Main character
1.02
100
.95
95
.93
91
.98
95
Locale
1.02
91
.95
91
1.14
100
1.00
100
Time
.64
73
1.25
95
1.23
91
1.14
91
Starter event
.27
36
.77
82
.59
64
.48
64
Goal
.73
50
1.82
100
1.39
77
1.64
86
Action
1.25
86
1.93
95
1.77
86
1.98
95
Ending
.59
64
1.02
91
1.00
91
.95
86
Reaction
.36
36
.77
77
.91
82
.70
68
Note. These values are based on the 22 students from the self-instructional strategy training and selfinstructional strategy training plus self-regulation training groups.

Strategy Usage

Writing Performance

Evidence on strategy usage provided validation of instructional manipulations. Examination of the papers that students
wrote on during the writing probes revealed that the story
grammar mnemonic and responses were generated by 91%,
91%, and 100% of the subjects during the posttest, generalization, and maintenance probes, respectively.

Self-Efficacy
A 2 (instructional groups) x 2 (trials) ANOVA with repeated
measures design was used to examine the effects of strategy
instruction and strategy instruction plus self-regulation on LD
students' self-efficacy scores and to determine if pre- and
posttest scores differed significantly. Although the F ratio for
trials was significant, F{\t 20) = 43.5, p < .01, MSe = 47.02,
the main effect for instructional groups and the interaction
effect were not significant. LD students' average self-efficacy
scores on the pretest and the posttest were 75.2 (SD = 16.4)
and 88.8 (SD = 10.2), respectively.

Discussion
Results of the present investigation are meaningful in terms
of each of the five research questions addressed. First, selfinstruction and strategy instruction plus self-regulation on LD
LD students' composition skills, an area of significant deficit
among these students. Second, explicit self-regulation procedures, although theoretically advocated, did not create incremental effects in terms of either LD subjects' self-efficacy at
posttest or their composition performance at posttest, shortterm maintenance, or generalization across settings. Third,
validation of instructional manipulations and confirmation
of mediating responses were obtained. Fourth, educational
validity of self-instructional strategy training was established.
Fifth, self-efficacy improved significantly among subjects in
both intervention conditions. These results are discussed further in terms of writing performance, incremental effects of
explicit self-regulation, and self-efficacy.

Self-instructional strategy training had significant and


meaningful effects on both the schematic structure of LD
students' compositions and the quality of their stories. The
power of this instructional approach is further highlighted by
the relative brevity of training. Although instruction was
criterion-based, total instructional time ranged from 4 hr to
6.5 hr. Although the positive effects of instruction were maintained over a 2-week period, further studies should investigate
long-term maintenance. Previous work, however, would indicate that even with explicit self-regulation, booster sessions
are necessary for long-term maintenance (Harris & Graham,
1985;Meichenbaum, 1983).
Similarly, although generalization from the instructional
setting to the special education classroom was obtained, generalization to other settings (e.g., the regular classroom) remains to be investigated. Although previous research indicates
that generalization from one writing genre to another is
difficult to obtain (Graham & Harris, in press), some anecdotal evidence for generalization was obtained in this study.
Several subjects reported using or adapting the strategy (or
both) for other types of classroom writing assignments. One
subject reported using the story parts to help her in English
class when they read stories, and another subject reported
using the strategy to write outlines. Several subjects also
mentioned writing stories at home. Finally, evidence of social
validity was found in terms of the strongly positive evaluations
of instruction provided by students, writing instructors, and
resource room teachers. Both resource room teachers and
regular classroom teachers indicated that they intended to
continue strategy instruction.
Results of the present study also suggest that, although
instruction was efficacious, room for improvement in LD
students' compositions existed. Although there was no significant difference between LD students' posttest stories and
stories written by normally achieving students in terms of
inclusion of the story grammar elements and the quality
ratings of LD students' creative essays improved significantly
from pre- to posttest, LD students' posttest quality ratings
were significantly lower than those of normally achieving
students. Length appeared to be an important factor in this
difference because LD students' compositions were approxi-

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

360

STEVE GRAHAM AND KAREN R. HARRIS

mately half the length of those of their normally achieving


peers. Thus, further instruction could emphasize writing
longer stories (e.g., adding more detail) while maintaining
schematic integrity.
A control group was not included in this study because it
was not allowed by the school system for ethical reasons, and
a wait-list control group was not possible because of the
approaching end of the school year. Practice effects might
therefore be suggested as a counterexplanation ofthe improvement in LD subjects* compositions. This is unlikely, however.
Significant composition deficits are well-substantiated among
LD students, and practice alone is unlikely to result in improvement (Barenbaum et aL, 1987; Graham & Harris, in
press; Nodine et al., 1985). Furthermore, in another study
using the same pictorial writing prompts, Graham et al. (1988)
found that LD subjects' writing performance actually decreased slightly in a practice-only condition. In addition, no
significant differences existed between the instructional
groups on important subject variables such as IQ and pretest
performance, thus strengthening experimental control. Finally, evidence was obtained that subjects used the strategy
both during and after instruction, thus strengthening the
argument that instruction mediated performance.

Incremental Effects
Interestingly, no incremental effects due to explicit selfregulation of strategic performance were found in terms of
the schematic structure of LD students' compositions, the
quality of stories, the educational significance of their improvement, or their judgments (self-efficacy) concerning their
capability to write stories. It should be noted that we did not
compare the presence of self-regulation with the absence of
self-regulation. Rather, the comparison made was between
strategy instruction combined with instruction in and use of
explicit self-regulation procedures (goal setting, self-assessment, and self-recording) and strategy instruction alone. It is
likely, however, that the strategy instruction included implicit
self-regulation information and may thus have induced less
formal self-regulation procedures. Theorists have noted that
a self-regulation function is inherent in strategy interventions
such as self-instructional training (Meichenbaum, 1983; Pressley et al., in press). Indeed, it would be difficult to conceive
of any meaningful strategy instruction and acquisition procedures in which self-regulation is neither implicit nor
induced. It has been argued, however, that explicit selfregulation instruction will significantly augment strategic performance and is particularly important in obtaining generalization and maintenance (cf. Brown et al., 1981; Pressley et
al., in press; Pressley & Levin, 1986).
Clearly, both replication of the results of this study and
further components analysis research are needed. Future studies, however, might profitably focus on the characteristics as
well as the components of instruction. Noteworthy here is
that instruction focused on a meaningful academic task,
emphasized interactive learning, was provided by preservice
teachers in the students' schools, and was criterion-based
rather than time-based. Any or all of these characteristics may
be related to the lack of incremental effects due to explicit

self-regulation in this study. Furthermore, future research


should explore the possibility that explicit self-regulation enhances regulation and effects of succeeding strategies taught
or acquired. Finally, the inclusion of explicit self-regulation
procedures was in no way "costly," and students frequently
mentioned "the graphs" as one of their favorite instructional
components (consistent with our earlier work; cf. Graham &
Harris, 1989). Explicit self-regulation appears to have strong
social validity.

Self-Efficacy
A major issue in the present investigation was the effects of
the two-strategy intervention conditions on subjects' selfefficacy. Self-efficacy is postulated to have a causal effect on
performance as it mediates choice of activities, expenditure
of effort, and persistence in the face of difficulty. Furthermore,
explicit, proximal goal-setting procedures, such as those used
in the present investigation, have been effective in increasing
children's self-efficacy (Bandura & Schunk, 1981). Thus, augmental effects of explicit instruction in goal setting and selfmonitoring appeared likely. Such effects were not realized;
self-efficacy increased significantly among subjects in both
conditions.
Equally as notable as the lack of augmental effects was the
relatively high level of self-efficacy among LD students at
pretest. The ability to assess one's own capabilities, and particularly the ability to know that one has a problem, is an
important metacognitive skill (Brown et al., 1981; Harris,
Graham, & Freeman, 1988). A growing body of research
indicates that young children and problem learners experience
significant difficulties with predicting or assessing their performance (cf. Brown et al., 1981; Harris et al., 1988). In the
present study, LD students consistently overestimated their
composition abilities. Bandura and Schunk (1981) noted the
same problem of overestimation among children making
judgments of mathematical self-efficacy; such overestimation
was attributed to misperceptions of task demands, faulty selfknowledge, and selective attention to mastered, as opposed to
unmastered, task elements. Unreatistically high pretask expectancies may also be due to comprehension deficiencies,
use of a self-protective coping strategy, or a developmental
delay in the ability to match task demands to ability level
(Harris et al., 1988). Thus, further research on pretask expectancies may be important in understanding problem learners.
Although the need for further research is evident, the present
study adds to a growing body of research that indicates that
sound strategy instruction produces meaningful results.

References
Bandura, A., & Schunk, D. (1981). Cultivating competence, selfefficacy, and intrinsic interest through proximal self-motivation.
Journal of Personality and Social Psychology, 41, 586-598.
Barenbaum, E., Newcomer, P., & Nodine, B. (1987). Children's
ability to write stories as a function of variation in task, age, and
developmental level. Learning Disability Quarterly, 10, 175-188.
Brown, A. L., Campione? J. C , & Day, J. D. (1981). Learning to

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

COMPONENTS ANALYSIS OF COGNITIVE STRATEGY INSTRUCTION


learn: On training students to learn from texts. Educational Researcher, 10, 14-21.
The California Achievement Tests: Technical bulletin No. 1 (1979).
Monterey, CA: CTB/McGraw-HUL
Dunn, L. M., & Markwardt, F. C. (1970). Peabody Individual
Achievement Test. Circle Pines, MN: American Guidance Service.
Elliott-Faust, D. J., & Pressley, M. (1986). How to teach comparison
processing to increase children's short- and long-term listening
comprehension monitoring. Journal of Educational Psychology,
78, 27-33.
Englert, C. S., & Thomas, C. C. (1987). Sensitivity to text structure
in reading and writing: A comparison between learning disabled
and non-learning disabled students. Learning Disability Quarterly,
10, 93-105.
Graham, S. (1982). Composition research and practice: A unified
approach. Focus on Exceptional Children, 14, 1-16.
Graham, S., & Harris, K. R. (1989). Cognitive training: Implications
for written language. In J. Hughes & R. Hall (Eds.), Cognitive
behavioral psychology in the schools: A comprehensive handbook
(pp. 247-279). New York: Guilford Press.
Graham, S., & Harris, K. R. (in press). Improving learning disabled
students' skills at composing essays: Self-instructional strategy
training. Exceptional Children.
Graham, S., Harris, K. R., & Sawyer, R. (1987). Composition instruction with learning disabled students: Self-instructional strategy
training. Focus on Exceptional Children, 20, 1-11.
Graham, S., Harris, K. R., & Sawyer, R. (1988). [Improving learning
disabled students' composition skills with story grammar strategy
training: A second components analysis of self-instructional strategy training]. Unpublished raw data.
Hammill, D,, & Larsen, S. (1983). Test of written language. Austin,
TX: Pro-Ed.
Harris, K. R. (1985). Conceptual, methodological, and clinical issues
in cognitive-behavioral assessment. Journal of Abnormal Child
Psychology, 13, 373-390.
Harris, K. R. (1986a). The effects of cognitive-behavior modification
on private speech and task performance during problem solving
among learning disabled and normally achieving children. Journal
ofAbnormal Child Psychology, 14, 63-77.
Harris, K. R. (1986b). Self-monitoring of attentional behavior versus
self-monitoring of productivity: Effects on on-task behavior and
academic response rate among learning disabled children. Journal
ofApplied Behavior Analysis, 19, 417-423.
Harris, K. R., & Graham, S. (1985). Improving learning disabled
students' composition skills: Self-control strategy training. Learning
Disability Quarterly, 8, 27-36.
Harris, K. R., Graham, S., & Freeman, S. (1988). The effects of
strategy training and study conditions on metamemory among

361

learning disabled students. Exceptional Children, 54, 332-338.


Hobbs, S. A., Moguin, L. E., Tyroler, U.f & Lahey, B. B. (1980).
Cognitive behavior therapy with children: Has clinical utility been
demonstrated? Psychological Bulletin, 87, 147-165.
MacArthur, C, & Graham, S. (1987). Learning disabled students'
composing with three methods; Handwriting, dictation, and word
processing. Journal of Special Education, 21(3), 22-42.
Meiehenbaum, D. (1977). Cognitive-behavior modification: An integrative approach. New York: Plenum Press.
Meiehenbaum, D. (1983). Teaching thinking: A cognitive-behavioral
approach. Interdisciplinary voices in learning disabilities and remedial education (pp. 1-28), Austin, TX: Pro-Ed.
Nodine, B., Barenbaum, E., & Newcomer, P. (1985). Story composition by learning disabled, reading disabled, and normal children.
Learning Disability Quarterly,, 8, 167-181.
O'Leary, S. G., & Dubey, D. R. (1979). Applications of self-control
procedures by children: A review. Journal of Applied Behavior
Analysis, 12, 449-465.
Pressley, M., Forrest-Pressley, D., & Elliott-Faust, D. J. (in press).
How to study strategy instructional enrichment: Illustrations from
research on children's prose memory and comprehension. In F.
Weinert & M. Perlmutter (Eds.), Memory development: Universal
changes and individual development. Hillsdale, NJ: Erlbaum.
Pressley, M., & Levin, J. R. (1986). Elaborative learning strategies for
the inefficient learner. In S. J. Ceci (Ed), Handbook of cognitive,
social, and neuropsychological aspects of learning disabilities (pp.
175-211). Hillsdale, NJ: Erlbaum.
Reeve, R. A., & Brown, A. L. (1985). Metacognition reconsidered:
Implications for intervention research. Journal of Abnormal Child
Psychology, 13, 343-356.
Short, E. J., & Ryan, E. B. (1984). Metacognitive differences between
skilled and less skilled readers: Remediating deficits through story
grammar and attribution training. Journal of Educational Psychology, 76, 225-235.
Slosson, R. L. (1971). Slosson Intelligence Test. East Aurora, NY:
Slosson Educational Publications.
Stein, N. L,, & Glenn, C. C. (1979). An analysis of story comprehension in elementary school children. In R. O. Freedle (Ed.), Advances
in discourse processes: Vol. 2. New directions in discourse processing. Norwood, NJ: Ablex.
Thoradike, R., & Hagen, E. (1978). Cognitive Abilities Test. Boston:
Houghton Mifflin.
Wechsler, D. (1974). Manual for the Wechsler Intelligence Scale for
Children-Revised. New York: Psychological Corporation.

Received September 2, 1987


Revision received March 10, 1989
Accepted March 20, 1989

Você também pode gostar