Prior Knowledge Activation - American Psychological Association

4 downloads 70 Views 1MB Size Report
Oct 2, 2010 - are socialized to approach reading tasks within school contexts ... this model, text processing involves the transmission of meaning from the ..... animals, repair barns and fences, and do the many other chores necessary on a ...

Copyright 1998 by the American Psychological Association, Inc. 0022-0663/98/S3.00

Journal of Educational Psychology 1998, Vol. 90, No. 2,249-260

Prior Knowledge Activation: Inducing Engagement With Informational Texts Hiller A. Spires and Jan Donley

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.

North Carolina State University Failure to engage with informational texts is a problem frequently noted at the high school level, at which students are expected to read independently. As a means of addressing this issue, a prior knowledge activation strategy (PKA) was taught to ninth-grade students in which they were encouraged to make spontaneous connections between their personal knowledge and informational texts. Students who learned to use the PKA strategy consistently outperformed students in a main idea (MI) treatment group and those in a no-instruction control group on application-level comprehension questions but not literal-level questions. A second study replicated the operations of the first study, with the addition of an MI-PKA treatment designed to combine both strategies, Both the PKA and the MI-PKA combination groups performed higher on application-level comprehension questions and demonstrated more positive attitudes toward reading than the other groups.

Students frequently have difficulty reading and learning from informational texts (Applebee, Langer, & Mullis, 1989; Armbruster et al., 1990), a significant problem during the middle and secondary years when textbooks are a primary source of information. Some education critics suggest that the difficulty may be a function of how students are socialized to approach reading tasks within school contexts (Alvermann, 1986; Apple & Christian-Smith, 1991; Goodlad, 1984). In traditional school settings, students are encouraged to be collectors rather than creators of meaning during the reading process, in keeping with the longestablished "transmission model" of reading. According to this model, text processing involves the transmission of meaning from the author to the reader via the text, which serves as a repository of knowledge from which information is extracted and passed along unchanged (Smith, 1985; Straw & Sadowy, 1990). Students' own knowledge, attitudes, and experiences have a limited role in the reading process and may even be seen as interfering with comprehension. The assumptions of this transmission model of reading— that the text contains a static message and that the reader is a passive receptor of information—have clearly been challenged by more recent constructivist models (Spiro, 1980; Spivey, 1987,1990,1997). Spivey (1995) suggested that the reader draws on a number of knowledge sources (e.g., rhetorical knowledge, background knowledge and experiences, and cultural knowledge) to build meaning from a text. This active negotiation between the reader and the text results in a "constructed meaning" that is in direct contrast

to the traditional notion of "referential meaning" located within texts. Constructivist views are consistent with a wealth of cognitive research demonstrating that prior knowledge is a critical component of reading comprehension. For example, readers who possess high levels of knowledge consistently exhibit better comprehension and retention than readers with low levels of knowledge (e.g., Chiesi, Spilich, & Voss, 1979; Langer, 1981; Pearson, Hanson, & Gordon, 1979). Good and poor readers differ not only in the amount of knowledge they have available but also in how and whether they make use of their knowledge to facilitate comprehension. Good readers are more likely to use their prior knowledge throughout the reading process than less able readers (Bransford, Stein, Shelton, & Owings, 1981; Oakhill, 1984). When less capable readers are prompted to use their prior knowledge, however, their performance improves and more closely resembles that of good readers (Recht & Leslie, 1988). When research findings on prior knowledge activation are operationalized for practical classroom reading instruction, they often take the form of teacher-directed prereading strategies designed to help students activate or build background knowledge (Graves, Cooke, & Laberge, 1983; Langer, 1984; McCormick, 1989). These prereading strategies usually focus on helping students make connections between their existing domain knowledge and new information to be read in a text. Another body of literature demonstrates that inserting elaborative questions within factual expository material promotes recall (for a review, see Pressley, Wood, et al., 1992). In addition, asking students to generate questions during reading has proven beneficial for various kinds of learning. These types of interventions have proven successful with respect to student learning of text; however, they typically rely on external prompts (either directly from the teacher or embedded within the text) and may not help students independently activate their prior knowledge during reading (Woloshyn, Paivio, & Pressley, 1994). Although there is literature to support the effective-

Hiller A. Spires and Jan Donley, Department of Curriculum and Instruction, North Carolina State University. Correspondence concerning this article should be addressed to Hiller A. Spires, Department of Curriculum and Instruction, College of Education and Psychology, North Carolina State University, 602 Poe Hall, Campus Box 7801, Raleigh, North Carolina 27695-7801. Electronic mail may be sent to [email protected] 249

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.

250

SPIRES AND DONLEY

ness of student-generated questions (e.g., see Wong, 1985, and King, 1989,1992) during reading, the types of questions that are encouraged are typically specific to a particular domain of knowledge and may or may not tap students' personal knowledge. Whereas cognitive studies often focus on the role of domain knowledge, another body of constructivist literature, related to reader response theories, argues for the important role of readers' personal knowledge and experiences. Deemphasizing critical authority and teacher transmission of literary knowledge, reader response theories describe reading as an active as well as intensely personal process. From this perspective, prior knowledge includes not only the topic knowledge readers have accumulated but also their personal knowledge: the spontaneous and idiosyncratic associations with personal experience prompted by the text (Rosenblatt, 1978, 1983; Fish, 1980; Iser, 1978; Bleich, 1975). The function of personal knowledge is illustrated by Rosenblatt's transactional theory, which suggests that, within any reading experience, a reader's stance may fluctuate along a continuum between "aesthetic" and "efferent." A reader taking an aesthetic stance brings his or her own personal meaning to the text, in effect "living through" the textual event. Rosenblatt argues that, to optimally experience literary texts, students should take an aesthetic stance when reading, and reader response pedagogies were designed to encourage this type of personal engagement. In contrast, when adopting an efferent stance, the reader's "attention is centered predominantly on what is to be carried away or retained after the reading event" (Rosenblatt, 1989, p. 159). Meaning in informational texts is usually regarded as more obvious, more straightforward, and less open to idiosyncratic connections on the part of the reader. The text-driven, outcome-oriented efferent stance is typically assumed to be more appropriate for informational texts; therefore, by default, these texts appear to fall outside the purview of reader response theory. This dismissal of informational text as requiring less personal engagement on the part of the reader is unfortunate, not only because it is inconsistent with broader constructivist assumptions but also because it has led to a potentially valuable instructional approach being overlooked. In some instances, reader response approaches have promoted more higher level learning than traditional instruction in conjunction with the reading of literature (for a review, see Beach & Hynds, 1991). For example, Beach (1990) found that the amount and degree of students' autobiographical elaborations were highly correlated with the amount and degree of higher level interpretations of short stories, Petrosky's (1981) study produced a similar result with autobiographical elaborations of novels. When comparing aesthetic and efferent responses to literature, Many (1990, 1991) found that aesthetic responses included higher levels of understanding of literature in terms of students' use of inferences, analogies, and abstract generalizations. Similarly, Cox and Many (1992a, 1992b) found that aesthetic responses included creativity and variation, whereas efferent responses consisted of the labeling of literary elements with few elaborative references to the story.

If the act of reading involves constructive processes, as many attest, then it is reasonable to expect reader response pedagogies to have some beneficial effects for informational texts in addition to literary texts. The shift in focus in contemporary education from fact gathering to constructed meaning suggests the need to broaden pedagogy to help students take advantage of the full continuum of response. Broadening the range of responses to informational text to include the aesthetic or personal seems particularly advantageous at secondary levels because of the unique developmental characteristics of adolescents: their essentially egocentric view of the world (Elkind, 1970) and corresponding preoccupation with issues that relate directly to their personal lives (e.g., relationships, love, and work). Giving students permission to bring their personal knowledge into the school context helps them establish the relevance of academic texts to their own interests and purposes and has motivational benefits for students in the middle and high school years (Donley, 1991; Marshall, 1989; Probst, 1984). Actually encouraging students to view informational reading through the lens of personal experience and exploration may help students construct a more lively and engaging relationship with the text, which in turn may help them attend to and possibly persist with the reading task. In sum, we believe that "resocializing" students to consider their own personal knowledge relevant to the learning experience is likely to have both a cognitive impact and an affective impact on learning. This assertion has been validated in conjunction with literary texts through a variety of descriptive and quasi-experimental studies. Conspicuously absent from the literature, however, are experimentally controlled studies attempting to apply reader response pedagogies to informational texts to determine whether cognitive and affective benefits might be achieved. We therefore designed a strategy for reading informational texts that invites students to acknowledge and build on the full range of their prior knowledge, from personal to domain specific. The strategy differs from those used in previous prior knowledge studies in that it gives the reader control over the conditions of prior knowledge activation (i.e., which knowledge and at what point in the text). The two experimental studies reported here tested the hypothesis that instruction in this prior knowledge activation strategy would enhance ninth-grade students' comprehension of informational as well as literary texts. This approach was compared with a more traditional text-based strategy in which the focus was on identifying important factual information during reading, similar to an efferent reading stance. Both treatment groups were compared with a control group answering study questions after reading.

Experiment 1 Method Participants The participants were 112 ninth graders who were enrolled in six different classes of a required social studies course in an urban high

PRIOR KNOWLEDGE ACTIVATION school in the Southeast. The classes were identified for participation through teacher volunteers. From the six classes, equal numbers of high, average, and low readers, as measured by the comprehension subtest of the California Achievement Test (CAT, 1992), were randomly assigned to three treatment groups through stratified random sampling.

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.

Materials Passages were selected from social studies textbooks in an attempt to provide some level of ecological validity of instructional and testing materials and to increase the opportunity for transfer to authentic academic reading tasks. The reading material used during instruction was limited to a chapter from the students' social studies textbook that had not yet been covered in class. Test materials consisted of three passages from a comparable ninth-grade social studies text (see Appendix for sample materials). Test passages were designed by reducing portions of chapters to approximately 1,200 words in length; all original subheadings and markers were kept intact. The social studies passages included topics on the family, ecology, and equal rights. In addition, "The Open Window," a short story by Saki (1958), was included in its entirety as a fourth passage to compare the relative effectiveness of strategies with literary texts. Although this type of strategy has already proven effective with literary texts, we decided to include a narrative passage as a point of comparison within the context of our particular study. Because narrative comparison was not the primary focus of our study, we decided to use only one short story so as not to overtax students during the testing phase. For each passage, a comprehension test was developed that included both literal and application-level questions. For each of the three social studies passages, 10 multiple-choice questions were developed to test literal comprehension. Multiple-choice questions were used to test literal comprehension because this format readily lends itself to tests of reliability. In addition, 10 application-level questions, 5 open-ended and 5 multiple choice, were developed to determine whether the reader could go beyond information in the passage to generate appropriate answers. For example, in one of the open-ended application questions for the passage on equal rights, students were asked "Do you think new laws are needed to ensure equal rights for all Americans? Why or why not?" The issue of additional laws was not addressed explicitly in the passage; therefore, students had to use their own background knowledge and ideas to formulate a response. For the narrative passage, 6 literal multiple-choice questions and 6 open-ended application-level questions were designed. No multiple-choice application questions were included. All passages and corresponding tests were critiqued by three social studies teachers and one reading specialist to establish content validity. Materials and procedures then were pilot tested on two sections of social studies classes that were not participating in the study. The Kuder-Richardson 20 formula was used to estimate internal consistency for the literal multiple-choice tests. This procedure yielded the following reliability coefficients; .79 (family), .85 (ecology), .80 (short story), and .83 (equal rights).

Procedure Individual participants from six different sections of the same social studies class were randomly assigned to one of three groups: prior knowledge activation strategy (PKA), main idea strategy

251

(MI), or a no-instruction control group. All participants took part in six 45-min instructional sessions followed by four 45-min testing sessions (three immediate testing sessions [Test 1, Test 2, and transfer test] and one delayed testing session 4 weeks later). Three researchers who served as instructors were randomly assigned to treatment groups. Instructional treatments. Both the PKA and MI groups received instruction in their respective strategies according to a model of explicit instruction (Pearson & Gallagher, 1983) in which responsibility for strategy completion is gradually shifted from teacher to student. This model of instruction, based on Vygotsky's (1978) developmental theory, has been used successfully to support other strategy instruction such as reciprocal teaching (Palincsar & Brown, 1984), transactional comprehension instruction (Pressley, Schuder, et al., 1992), and note-taking instruction (Spires, 1993). Vygotsky posited that the process of expanding cognition is best achieved through social interactions or "shared meaning" between an instructor and learner. Over time, the cognitive processes of the instructor are internalized by the learner and reappear in the learner's thinking. Wood, Bruner, and Ross (1976) referred to this same instructional process as "scaffolding" by connoting the idea of the instructor providing a support (i.e., scaffold) for the learner until the learner is capable of operating at a higher level without support. Metaphorically, the scaffold is removed and the learner functions independently at a new level of cognition. This is also similar to the well-established concept that higher cognitive functioning moves from the interpsychological (i.e., betweenpersons) plane to the intrapsychological (i.e., within-person) plane (Wertsch, 1978). Explicit instruction in this study involved the following components: (a) rationale and explanation of the strategy, (b) teacher modeling of the strategy, (c) teacher and student collaborative work with the strategy, (d) teacher and peer feedback on the use of the strategy, and (e) independent use of the strategy. The gradual shift of responsibility from the instructor to the student for strategy completion occurs between the steps of teacher modeling and independent use of the strategy. Both treatment groups followed the model of explicit instruction. In both groups, the teachers began instruction by providing an extensive rationale for why the strategy would be useful to the students and how it would help them comprehend written texts more successfully. Teachers then began modeling the use of the strategies and verbalizing their thinking as they executed their respective tasks. For example, the PKA instructors read aloud a portion of the text and then modeled oral elaborations by directly relating an idea in the text to personal knowledge they already possessed; the information could be related to personal thoughts and experiences of the reader or other subject domains. No initial judgments were made about the quality or appropriateness of the oral elaborations that the reader generated. The important goal was that of having the students make some type of connection. If the connection between the elaboration and the text was not readily apparent, the teacher would follow up with "Can you explain why that portion of the text reminded you of that specific experience or information?" This helped the student reflect on the quality of the connection that was being made. Often students would decide for themselves that there was not a strong connection and then elect to revise their elaboration. This type of prompting was based on research showing that students who activate prior knowledge by addressing "why" questions are better able to learn written materials (for a review, see Pressley, Wood, et al., 1992). The MI instructors read aloud portions of the text and then verbalized their thinking as they took notes in a split-page

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.

252

SPIRES AND DONLEY

note-taking format (see Spires & Stone, 1989). The modeling process included writing notes on an overhead screen while verbally explaining how and why certain information was targeted as main points and other information as corresponding details. In addition, the instructors provided rationales for why they chose to omit certain points from their notes and for why they combined others. The teachers in both instructional treatments then provided opportunities for guided practice in which teachers and students worked jointly to complete their respective tasks. During the guided practice phase, students also worked in small groups and then received feedback from both teachers and peers. Over the 6 days of instruction, teachers gradually released the responsibility for task completion to the student, and by the end of the instructional phase students were operating at an independent level with their respective strategies. For example, by the last 2 days of instruction, PKA students were writing their elaborations spontaneously and independently out to the side of the text, and MI students were writing main ideas and supporting details in a split-page note-taking format (see Tables 1 and 2 for examples). Control treatment. The control group worked with the same reading materials as the experimental groups for both instruction and testing. The instructional phase consisted of having students read a designated portion of the text and then answer corresponding questions at the end of the chapter. At the end of each session, the instructor provided students with feedback on correct answers. The rationale presented to the students was that engaging in sustained silent reading over a period of time would provide them with reading practice that, it was hoped, would affect their reading performance.

Testing Procedure and Scoring Testing began on Day 7, after the instructional component of the study had been completed. The dependent measures used for testing consisted of three immediate measures and one delayed measure. For the first two measures, students read informational passages and completed comprehension tests. The third measure was used in conjunction with a short story to assess the transfer effects of strategy training to narrative prose. The delayed measure, consisting of an informational passage and corresponding test, was administered 4 weeks after instruction to assess the degree to which students continued to benefit from the strategy instruction. For each of the four tests, students were instructed to use the reading strategies they had learned while reading a passage within a 20-min time limit, After students had finished reading, the instructor collected the passage and distributed the test. Students were allowed the remaining 30 min of the class period to complete the test. Those who finished early were encouraged to proofread their tests. Each test consisted of 10 literal-level questions, with the exception of the narrative test, which included only 6 questions. All literal questions were presented in a multiple-choice format. Application questions consisted of 5 open-ended and 5 multiplechoice items, with the exception of the narrative test, which had no multiple-choice questions and 6 open-ended questions. Answers to the open-ended questions were scored by two raters based on idea units described in Swarts, Flower, and Hayes (1984); raters used criteria that targeted both the quality and quantity of student elaborations. For each open-ended question, students could score between 0 and 3 points. Interrater reliability ranged from .80 to 1.00 for each of these questions (see Table 3 for an example of scoring).

Those students who missed more than 1 day of training or who were absent during any of the testing sessions were excluded from the study, which left a total of 79 participants for the data analysis. The high attrition rate was a result of students from all three groups missing part of the instruction because they were absent the entire school day or were called out of class for photograph sessions.

Results and Discussion A one-way multivariate analysis of covariance (MANCOVA) was conducted across treatment groups, with reading ability (i.e., CAT comprehension subtest score, with 100 points the maximum) as a covariate. The mean scores of the three groups (i.e., control group, M - 75.68, SD = 28.06; MI group, Af = 71.89.SZJ = 21.27;andPKAgroup,M - 66.91, SD = 29.46) represented average to slightly above-average Grade 9 scores relative to national norms. A covariate was used because despite random assignment of individual students to treatments, there was a marginally significant difference across groups on this variable (p < .09). Results are summarized in Tables 4 and 5. The Wilks's lambda statistic was used as the multivariate test of criteria. The overall value of this statistic showed a significant main effect for treatment group, F(16, 136) = 4.01,/? = .0001.

Literal-Level Comprehension Univariate F tests (with CAT scores as a covariate) yielded a significant main effect for group on literal comprehension for one of the four tests, Test 2, F(2,75) = 4.47, p = .01. Means and standard deviations are shown in Table 4. With the Kramer adjustment of Tukey's honestly significant difference (HSD) procedure on the adjusted means, the MI group outperformed both the PKA and control groups; there was no difference between the PKA and control groups. There was also a marginally significant effect on the literal component of the delayed measure, F(2, 75) - 2.70, p = .07. The MI group outperformed both the PKA and control groups, whereas no difference existed between the PKA and control groups.

Application-Level Comprehension Univariate F tests (with CAT scores as a covariate) yielded a significant effect on both the open-ended and multiple-choice formats for the application measures. Means and standard deviations are shown in Table 5. Significant results were observed on all four open-ended measures: Test 1,F(2,75) = 10.12,/> = .000; Test 2, F(2,75) = 13.35, p = .0001; transfer test, F(2, 75) = 4.69, p - .01; and delayed test, F(2, 75) = 8.47,/? = .005. With the Kramer adjustment of Tukey's HSD procedure on the adjusted means, the PKA group outperformed both the MI and control groups on all dependent measures. No differences were observed between the MI and control groups on Test 1, the transfer test, or the delayed test; however, the control group scored higher than the MI group on Test 2.

253

PRIOR KNOWLEDGE ACTIVATION

Table 1 Samples of Text and Corresponding Student Responses From the Prior Knowledge Activation Group Response

Text The family

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.

The family is a group of people who are united by ties of marriage, blood or adoption. The family provides the basic needs of food, clothing, shelter, and affection for all of its members. The family usually is the first group to which a person belongs. Infants are loved and cared for in their families. As they grow, children are taught certain skills, values, and traditions.

Fact is that some people just don't live in a fairy-tale life like that Many Americans don't care one way or the other about their families, (personal knowledge)

The changing family The people who settled America believed in strong family ties The Pilgrims settled in America. They got their large families and the importance of a good family life. Many families, for together for a big meal and started Thanksgiving, (subject example, were large and included grandparents and other relaknowledge) tives in the same household. Today most families are much smaller. Their lifestyles have changed. Yet despite these changes, Americans continue to believe in the importance of the family. The traditional family How different was the colonial family from the family of today? The first census of the United States, taken in 1790, showed that the average family had about four children. Many families were even larger. At that time, our country was largely rural, which means that most American people lived and worked on farms. Each child was a welcome addition to the family, for there was plenty of work to be done on the farm. Older boys worked with their father. They learned how to plow the soil, plant seeds, and harvest the crops. They also learned to care for the animals, repair barns and fences, and do the many other chores necessary on a farm. The mother taught the daughters to sew and cook, make soap, and do the other household chores that kept the family going.

This made me think of Little House on the Prairie because that's how it goes there. The boys work on the farm and the girls work inside, (personal-subject knowledge)

The move to cities During the 1800's, American life began to change fairly rapidly. One hundred years ago, seven of every ten Americans lived on farms or in rural areas. Today only one in four Americans lives in a rural area. This change came about because of the remarkable progress in science and technology that took place during the past hundred years. Americans soon found use for the new discoveries and inventions. These led to the building of large factories in many urban, or city areas. The factories needed many workers. At the same time, the development of better farm machinery meant that fewer people were needed to work on the farms. Farm families began to move to urban areas to seek jobs in the factories. This movement of Americans away from the farms to the cities resulted in changes in family life.

The light bulb and electricity were some of these new inventions, (subject knowledge)

The family was once the main influence in the lives of children. Many other influences also have become important for children today. Schools have taken on part of a child's education that was once thought only the job of the family. Television, too, has become an important influence in the lives of the young. Two other trends in the American way of life have been putting stress on the family. One is the increase in the divorce rate. The other trend is the increase in the number of families in which both parents work. This brings up the problem of who is to take care of the children.

My friend Julie's recently divorced. This has put a lot of emotional stress on her. (personal knowledge)

Note. Text excerpts are from American Civics (Constitution edition), by W. H. Hartley and W. S. Vincent, 1987, Orlando, FL: Harcourt Brace. Copyright 1987 by Harcourt Brace. Adapted with permission of the publisher.

254

SPIRES AND DONLEY

Table 2 Sample of Split-Page Notes From a Student in the Main Idea Group Main idea Institutions Families Changing family

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.

Traditional family

Move to cities

City family

Corresponding details Organized, customary ways of doing things Group of people united by marriage, blood or adoption Families used to be large & include relatives Now families are much smaller In 1790 average family had four kids Boys worked outside and girls worked inside Farm family was basic work unit in colony Move began in 1800's New inventions led to factories Machinery replaced farm workers Families moved to cities to work in factories Smaller than past families Kids spend more time in school Vi the women between 16 and 64 work outside the home Elderly no longer live with kids Life overall easier today because of education and prosperity

Significant results also were observed on the multiplechoice questions in Test 1,F(2, 75) = 14.09,/?= .0001; Test 2, F(2, 75) = 12.86, p = 0.01; and the delayed test, F(2, 75) = 2.64, p - .07. With the Kramer adjustment of Tukey's HSD procedure on the adjusted means, the PKA group outperformed the MI and control groups on Tests 1 and 2. The source of the marginally significant difference found for the delayed measure indicated that both the PKA and MI groups outperformed the control group. (Note that there was no multiple-choice test for the transfer measure.) In summary, explicit instruction in how to activate prior knowledge during reading positively affected students' performance on application-level questions in Experiment 1. On the application measure, the PKA group consistently outperformed the MI and the control groups, on both the open-ended and multiple-choice questions. The only exception was the delayed multiple-choice measure, on which the PKA group outperformed the control group but not the MI group. The PKA strategy seemed to transfer to narrative prose even though the training was conducted on informational prose. Consistent with previous research (see Beach & Hynds, 1991), it may be that the nature of the narrative format encourages the reader to take an aesthetic stance, which lends itself well to the type of personal engagement that subsequently results in application-level thinking. Having students engage in prior knowledge activation during reading provides practice with thinking beyond the text. This type of thinking then appears to carry over into question answering that requires the student to go beyond the text. Although purely speculative, it is possible that students felt positive about being allowed to bring their

personal knowledge to bear on reading tasks that typically are perceived as impersonal. As a result, students may have felt more personally invested in the reading task. Obviously, in an academic context, the successful reader is expected to comprehend the factual information in the text as well as think beyond the text on an application level. The students in the PKA group did not have an advantage on the literal comprehension measures. In fact, the MI group had a slight advantage in terms of this type of performance, although the results were statistically significant only for Test 2. We therefore concluded that a strategy combining the tasks of the MI and PKA groups might yield a more well-rounded comprehension experience for the reader. In addition, we decided that our research design might be strengthened and have more direct application for the classroom by having teachers rather than researchers conduct the instructional treatments. Experiment 2 A second study was conducted to investigate the effects of the PKA strategy when combined with the MI strategy. This study was a replication of Experiment 1 with the following exceptions: (a) A fourth group that combined PKA and MI strategy instruction was added; (b) regular classroom teachers conducted the instruction and testing instead of researchers; and (c) students' attitudes toward reading were assessed to provide an understanding of the broader effects of the strategy instruction.

Table 3 Sample Scoring for Open-Ended Application Question Score

Student response

0 1

No, I don't think so. No, because if the mother is not willing to take care of its kid personally it should not have had the kid. Yes, I think business should be required to provide child care centers. It is extremely difficult to find a good daycare where you can be sure your child will be properly taken care of and if the business provided it, it would make it much easier for the parents. Plus the child and parent would be near each other in case of an emergency. No, overall, I don't feel that business should be required to have child care centers because it would be too expensive for them. If they can afford it, however, I think it is a good idea because the parent will be closer to their child in case the child needs them and they will not have to waste gas to drive to day care to pick up the kid. I do not trunk they should be required to because it is not their responsibility to find a way to take care of someone's child. The parent should do that or should stay at home for a while and take care of their own lid. Actually, it is better for the kid and society if the parents stay home with their children.

2

3

Note. The question was as follows: "Do you think that businesses should be required to provide child care centers for the children of their employees? Why or why not?"

255

PRIOR KNOWLEDGE ACTIVATION

Table 4 Means and Standard Deviations for Literal Comprehension Measures: Experiment 1 Test 1

Test 2

Delayed test

Group

M

SD

Adjusted M

M

SD

Adjusted M

M

SD

Adjusted M

M

SD

Adjusted M

Control (n = 30) Main idea (n = 25) Prior knowledge (n =24)

8.30 8.72 8.58

1.62 1.21 1.69

8.42 8.66 8.49

8.67 9.48 8.63

1.60 0.71 1.76

8.79 9.42 8.53

4.60 4.68 4.63

1.28 1.65 1.66

4.71 4.63 4.54

7.00 7.92 7.21

1.78 1.38 1.64

7.12 7.86 7.11

Note.

Ten points were possible for Test 1, Test 2, and the delayed test. Six points were possible for the transfer test.

Method 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.

Transfer test

Participants and Materials The sample for this study consisted of 170 ninth-grade students who were enrolled in required social studies classes at the same urban high school the following year. Equal numbers of high, average, and low readers, based on their performance on the reading comprehension subtest of the CAT were assigned to each of die four treatment groups through the process of random stratified sampling. Those students who missed more than 1 of the 6 days of training or who were absent during any of the testing sessions were excluded from the final data set. This left a total of 161 participants for the final analysis. The same instructional and testing materials used in Experiment 1 were used in Experiment 2, with the addition of a reading attitude survey.

Procedure Students were randomly assigned to one of four treatment groups: PKA, MI, MI-PKA, or a no-instruction control group. Four instructors, regular ninth-grade social studies teachers, were randomly assigned to treatment groups. A design limitation of the study is that even though students and instructors were randomly assigned to treatment groups, we were unable to control for teacher effects. Instructors received 4 hr of training in their respective treatments, as well as scripted lesson plans to follow for both instruction and testing sessions. Although we did not share the specific hypotheses of the study with the teachers, the teachers understandably made their own judgments about the value of the various instructional treatments. There are limitations of the study to the degree that any of the teacher judgments inadvertently were passed on to the participants. We met periodically with the instructors to answer questions and address problems that arose during the instructional sessions. In addition, we made unannounced visits to the classrooms to observe instruction and to ensure that established procedures were being followed. Instructional procedures were identical to the first study with the addition of the MI-PKA group. The instructors for this group simply combined the explicit instruction procedures for the MI and PKA groups so that students used the split-page note-taking format to record main ideas and supporting details, as well as writing reader-generated elaborations in the margins of the text. Because the total instructional time remained constant across all groups, the MI-PKA group received half the amount of explicit training on each strategy of the other treatment groups. Testing materials and procedures were the same as in Experiment 1, with the addition of the Rhody Secondary Reading Attitude Assessment (Rhody, 1978), a 25-item survey designed to assess students' general attitudes toward reading. The survey includes statements such as "You think reading is boring," to which students respond using a 5-point Likert scale. Point values were

assigned to each response ranging from 1 (strongly disagree) to 5 (strongly agree) creating a range of possible scores from 25 to 125. All other dependent measures were scored in the same manner as in Experiment 1. Open-ended application questions were scored by two raters according to the same criteria used in Experiment 1. Interrater reliability ranged from .82 to .92 for each of the questions. Results and Discussion A one-way MANCOVA was conducted across treatment groups, with reading ability as a covariate (i.e., control group, M= 56.71, SD = 28.59; MI group, M = 63.28, SD = 27.86; PKA group, M = 67.02, SD = 26.21; and MI-PKA group, M = 72.00, SD = 24.52). Despite random assignment, there was a marginally significant difference between groups (p < .07) on CAT scores after attrition. Results for these analyses can be found in Tables 6 and 7. Wilks's lambda was used as the multivariate test of criteria. There was a significant main effect for treatment, F(33, 389) = 7.43, p = .001. Literal-Level Comprehension Univariate F tests (using CAT scores as a covariate) yielded a significant difference only for literal-level comprehension across groups, F(3, 142) = 8.11, p = .001. Means and standard deviations are shown in Table 6. With the Kramer adjustment of Tukey's HSD procedure on the adjusted means, all three experimental treatment groups outperformed the control group on the delayed measure. In addition, there was a nonsignificant trend for the MI-PKA group to outperform the other groups on Test 2 and the transfer test. Application-Level Comprehension Univariate F tests (using CAT scores as a covariate) and the Kramer adjustment of Tukey's HSD procedure on the adjusted means yielded significant differences on both the open-ended and multiple-choice formats for the applicationlevel questions. Means and standard deviations can be found in Table 7. For the open-ended questions, both the PKA and MI-PKA groups significantly outperformed the MI and control groups on Test 1, F(3, 142) = 7.12, p = .001, and Test 2, F(3,142) = 4.38, p = .006, For the transfer test, the PKA group outperformed the control and MI groups, and the MI-PKA group outperformed the MI group, F(3, 142) =

256

SPIRES AND DONLEY

5.69, /» = .001. For the delayed measure, the PKA group outperformed the control and MI groups, and the MI-PKA group outperformed the control group, F(3, 142) = 5.01,

•nmri

Significant results also were observed for multiple-choice questions in Test 1,F(3,142) = 18.46, p = .001;Test2,F(3, 142) = 33.77, p = .001; and the delayed test, F(3, 142) = 30.48, p = .001. With the Kramer adjustment of Tukey's HSD procedure on the adjusted means, all three experimental groups outperformed the control group, and the PKA and MI-PKA groups outperformed the MI group. (Note that there was no multiple-choice component for the transfer test.)

f

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.

SO