Table 1. Coding criteria for the efficiency of strategy awareness in retrospective interview
Partially adequate answer
Strategic response=Understanding how to apply a strategy in comprehending a text
d) In order to examine the relationship between strategy awareness and reading components, we created and tested two specific indices: i) The Cognitive Strategy Awareness Index (CSAI) was the composite score of students on 12 cognitive strategies, ranged from 0 to 24; it did not include the score on the ‘ambiguous’ cognitive strategy of skipping the difficult parts. The CSAI quantifies the awareness of cognitive strategies to an index, ii) The Metacognitive Strategy Awareness Index (MSAI) was the composite score of students on 10 metacognitive strategies, ranged from 0 to 20. The MSAI quantifies the awareness of metacognitive strategies to an index.
The second researcher (judge) rated all the transcribed data from the think-aloud protocols. The data rated for cognitive strategies and metacognitive strategies. After three months, the second researcher (judge) rated again 50% of the data (9 think-aloud protocols of poor readers and 9 of good ones). Intra-judge reliability averaged 96.4% for cognitive strategies and 94.3% for metacognitive strategies. The first researcher (judge) independently scored the other 50% of the data. Inter-coder reliability averaged 88.8% for cognitive strategies and 88.7% for metacognitive strategies. The two judges were not aware of participants’ reading scores when coding.
Awareness of cognitive strategies
Descriptive statistics (frequency and corresponding percentage) for each strategy are presented on Table 2; accordingly, the cognitive strategies were ranked from the highest percentage to the lowest. There were a total of 203 reports of cognitive strategy use.
As it is indicated on the table 2, good readers reported that they employ cognitive strategies much more frequently, almost twice, than poor readers. Furthermore, from a total of 13 cognitive strategies reported in the study, organisation strategies (underlining, using dictionary, using titles, and writing down) were the most favourite strategies among poor readers (an aggregate of 54.28%). Besides, there was no significant relationship between using dictionary strategy and reading group (χ2 = .55, df = 1, p > .05), while the chi-square statistic was not valid in the case of underlining strategy and writing down strategy since 50% of cells had expected frequencies less than 5. The only significant relationship was that good readers reported the strategy of using titles more frequent than poor readers (χ2 = 8.00, df = 1, p < .05) (see Table 3).
Strategies referred to more elaborate cognitive processes (guessing from the context, activating prior knowledge, imagery, keeping meaning in mind, summarising) were reported less frequently by poor readers (11.42% in the aggregate). As it is presented in the table 3, there were significant relationships between strategy of guessing from the context and reading group (χ2 = 18.84, df = 1, p = .000), between strategy of activating prior knowledge and reading group (χ2 = 16.20, df = 1, p = .000), between imagery strategy and reading group (χ2 = 9.75, df = 1, p < .01), and between strategy of keeping meaning in mind and reading group (χ2 = 12.50, df = 1, p = .001); while the chi-square statistic was not valid in the case of summarising strategy. Apart from ‘summarizing’, there were statistically significant relationships between awareness of ‘elaborate’ cognitive strategies and reading group, indicating that poor readers reported these strategies less frequently than poor readers did.
Similarly, poor readers seemed to be less aware of strategies related to using linguistic features of the text (using linguistic clues, using text markers) (an aggregate of 7.14%). As it is presented in the table 3, there were significant relationships between the strategy of using linguistic clues and reading group (χ2 = 19.31, df = 1, p = .000), and between the strategy of using text markers and reading group (χ2 = 5.60, df = 1, p < .05). These significant relationships indicated that poor readers reported strategies based on linguistic features less frequently than poor readers did.