Thai is the standard language spoken officially and nationally by almost sixty-five million people, throughout every part of Thailand. The Bureau of Academic Affairs and Educational Standards (under the Office of the Prime Minister of Thailand) requires that fourth-grade students should have the ability to read words, rhyming words and short passages correctly and fluently. Therefore, this study recruited fourth-grade students.
The participants were 357 fourth-grade students (47.34% boys, 52.66% girls), with a mean age of 10.12 years (9.0–11.5 age range). They were recruited from nine public schools in Loei Province in north-eastern Thailand. This study applied a two-stage sampling
technique (Fraenkel & Wallen, 2006). In loei Province, there are 419 schools, 5499 grade 4 students (50.95% boys, 49.05% girls). The schools under Loei province was divided into three Primary Educational Service Areas (PESAs) based on the Geographical areas, and all the PESAs in Loei adopt the same educational curriculum. Therefore, the first stage, cluster sampling was selected according to the PESAs, and one of the three PESAs was chosen for the study. The schools in the chosen PESA were divided into three strata, based on the number of students in their school, which made the demographic distribution of students in schools within the same strata similar and conversely, different among the strata. Therefore, the stratified sampling technique was used in the second stage. The schools in each PESA were stratified into large, medium and small. Schools were then selected randomly from each
28
of these three strata, and all fourth-grade students in each selected school were tested. The table 1 shows the sample in this study. The initial sampling procedure resulted in 453 students recruited from 9 schools, however, during data collection, sample attrition occurred due to missing data, therefore, 96 students (21.19%) dropped out of the study and the final sample consisted of 357 participants (148 students from 2 large schools, 168 students from 5 medium schools and 41 students from 2 small schools). All students were native Thai speakers and had no known sensory or learning disabilities.
.
29 Table 1
Study sample.
School Room Boy Girl Total Time 1 Time 2
Drop out (%) Boy Girl Total Drop out (%) Boy Girl Total Phanokao
community
1 12 13 25 0.00 12 13 25 28.00 8 10 18
Phukradung community
2 29 49 78 4.00 27 48 75 21.79 22 39 61
Nhong-hin community
2 18 41 59 1.72 17 41 58 16.95 14 35 49
Na-go 1 18 15 33 3.13 18 14 32 21.21 14 12 26
Huy-Som 1 20 13 33 0.00 20 13 33 15.15 17 11 28
Nhongtana 2 29 16 45 9.76 26 15 41 28.89 18 14 32
Phakaw 1 16 14 30 0.00 16 14 30 23.33 12 11 23
Nhongkan community
2 18 23 41 5.13 18 21 39 19.51 15 18 33
Arawan 3 61 48 109 3.81 60 45 105 20.18 49 38 87
Total 15 221 232 453 3.42 214 224 438 21.19 169 188 357
30 3.3 Research Design
This study aimed to investigate the relationship between cognitive component and reading ability of grade 4 students in their reading development over time. To achieve this aim, a two-wave cross-lagged panel design was applied to assess possible causal directions using empirical data. This design was a method of dealing with longitudinal correlations between two variables to determine whether “the history of one variable (X) can predict the other variable (Y) after accounting for the history of (Y)” (Kenny & Harackiewicz, 1979).
This type of design allows for the examination of a causal association between two variables measured simultaneously over several time points. Each of the two variables in cross-lagged panel designs is regressed on itself at earlier time points (the auto regression effect) and also regressed on the second variable to test the influence of one variable on another over time.
This type of design is helpful in determining the directionality of the relationship between two variables.
The choice of time between observations for a cross-lagged panel model is important.
The associations between variables change over time, any autoregressive or cross-lagged effect from the model is specific to the chosen lag. Thus the analyst assumes that a reasonable lag was used, meaning the time to collected data in cross-lagged panel design was not too long or too short and the time was appropriate to see the effect emerge (Selig, 2012). For cross-lagged panel technique, the analysis of three or more waves does have advantages, but when properly specified and using the most appropriate methods, analysis of two-wave data can still provide us with valid inferences about the effects of variable transitions on individual outcomes (Johnson, 2005). Of note is a previous study by Harlaar, Deater-Deckard,
Thompson, De Thorne, and Petrill (2011), which aimed to examine reading achievement and independent reading using two-wave cross-lagged panel design in one year with student age range from 10 to 11 years. The result of study found that reading achievement at age 10
31
significantly predicted independent reading at age 11. Finding of their study indicate that one-year and two-wave of cross-lagged panel design is sufficient to see the effect of reading development.
In the present study, the conceptual model of a cross-lagged panel design using two variables with measurement taken over two times point is presented in Figure 4. In Figure 4, the cognitive component and reading ability scores of students in fourth grade were measured at two times point. This design enabled the researcher to examine whether or not cognitive achievement at the first time point determines reading ability at the later time point.
Represented in Figure 4 are three types of relationships:
(1) Synchronous correlations between the two variables: These include the
unconditional zero order correlation (C1R1) between the cognitive component (C1) and the reading ability (R1), both at time-1, and the zero order correlation (C2R2) between the
cognitive component (C2) and the reading ability (R2), both at time-2 after controlling for the other effects in the model. Below is interpretation of some possible outcomes:
If C1R1is statistically significant, indicated the cross-sectional relationship between cognitive component at time-1 and reading ability at time-1.
If C2R2is statistically significant, indicated the cross-sectional relationship between cognitive component at time-2 and reading ability at time-2.
(2) Auto-regressive effects of the two variables: These include the effect (C1C2) of the cognitive component at time-1 (C1) on the cognitive component at time-2 (C2), and the effect (R1R2) of the reading ability at time-1 (R1) on the reading ability at time-2 (R2). Below is interpretation of some possible outcomes:
If C1C2 is statistically significant, then prior cognitive component at time-1 has an effect on subsequent cognitive component at time-2.
32
If R1R2 is statistically significant, then prior reading ability at time-1 has an effect on subsequent reading ability at time-2.
(3) Reciprocal effects between the two variables over time: These include the effect (represented by path coefficient C1R2 in Figure 4) of the cognitive component at time-1 (C1) on reading ability at time-2 (R2), and the effect (R1C2) of reading ability at time-1 (R1) on cognitive component at time-2 (C2). Below is interpretation of some possible outcomes:
If both C1R2 and R1C2 are statistically significant, then there is reciprocal effect of cognitive component and reading ability on each other by the other’s previous scores, after controlling for the effect of auto-regression.
If only C1R2 is statistically significant, but R1C2 is not significant, then prior cognitive component has an effect on subsequent reading ability, even after controlling for the effect of previous reading ability, but there is no evidence that previous reading ability would affect later cognitive component.
Conversely, If only R1C2 is statistically significant, but C1R2 is not
significant, then prior reading ability has an effect on subsequent cognitive component, even after controlling for the effect of previous cognitive component, but there is no evidence that prior cognitive component has an effect on later reading ability.
Figure 4. A cross-lagged panel design using two variables measured over two time points C1
33 3.4 Measurement Development
Measurement development of this study focused attention on the development of a valid measure of each of the underlying constructs. Table 2 provides the measurement
development procedure for the cognitive and reading ability measurement of reading in Thai.
Table 2
Scale Development Procedure.
Development Phase Scale Development Steps
Planning - Conducted a literature review of the competing theories of reading.
Construction - Determined and defined domains of various theories related to reading components.
- Create test items that are distinguishable by domain.
- Conducted expert reviews of all items for content and agreeability validation.
- Fix some items based on feedback from the expert reviews.
Validation - Assessed the reliability of the tests.
3.4.1 Planning. Based on the theoretical principles of subscales for reading reviewed in Chapter Two, the researcher identified two domains of reading include cognitive components (morphological awareness, decoding skills and rapid automatized naming) and reading abilities (reading comprehension and word recognition) and then created the items test for each component (see on Table 2).
34
3.4.2 Construction. Three experts reviewed all eight tests consist morphological structure test, morphological production test, phoneme isolation test, rapid word segmentation test, rapid colour naming test, rapid number naming test, reading comprehension and word recognition test to assess their content validity. The experts included two University lecturers with expertise in level Thai and one primary-school Thai language teacher. The objects of the test were given to the panel to evaluate for appropriateness of content, language and clarity of instructions. Table 3 shows the number of items and the total score of each test of reading ability and cognitive component which improve the content validity by expert judgment.
Table 3
Measurements.
Tests Number of items Scoring Total score Cognitive components
Morphological awareness test
- Morphological structure test 15 0/1 15
- Morphological production test 10 0/1 20
Decoding skills
- Phoneme isolation test 15 0/1 15
- Rapid word segmentation test 32 NS 142
RAN
- Rapid colour naming 50 RT -
- Rapid number naming 50 RT -
35 Table 3 (Continued)
Measurements.
Tests Number of items Scoring Total score Reading abilities
- Reading comprehension 35 0/1 35
- Word recognition 150 0/1 150
Note. 0/1 = incorrect response is scored as 0, correct response is scored as 1; RT = Response time; NS= number of words that were correctly separated.
3.4.3 Validation. In order to develop a scale for cognitive components and reading ability in this study, the log-transformed (Boscardin, Muthén, & Francis, 2008) and
Unidimensional Item Response Theory (UIRT) were applied to get proficiency estimates for each test.
For the morphological structure test, morphological production test, phoneme isolation test, reading comprehension and word recognition scale, instead of using the raw total scores of the ability scale of each test, this study used UIRT measure based on estimates of each person’s latent trait to represent those abilities with a mean of 0 and a standard deviation of 1.
UIRT models have a strong assumption that each test item is designed to measure some facet of the same underlying ability or a unique latent trait. It is necessary that a test intending to measure one certain trait should not be affected by other traits, especially when only the overall test scores are reported and used as an assessment criterion for various ability levels.
For unidimensional constructs, the Rasch model (or one-parameter logistic model: 1PLM) and the two-parameter logistic model (2PLM) are commonly used (Embretson & Reise, 2000). According to Thissen and Steinberg’s (1986) taxonomy classifies some IRT models designed to analyze such test items as “Binary Models”. The 1PLM and 2PLM belong to this
36
set. Suppose Xij represents the response of person j to item i, where Xij = 1 means the item i is answered correctly by person j, and Xij = 0 means the item i is answered incorrectly by person j. Then, these two models are expressed as the Equations (1) and (2), respectively.
𝑃 (𝑋𝑖𝑗 = 1⃒𝜃𝑗𝛽𝑖) = 1+exp(𝜃exp(𝜃𝑗−𝛽𝑖)
𝑗−𝛽𝑖) (1)
𝑃 (𝑋𝑖𝑗 = 1⃒𝜃𝑗, 𝛼𝑖, 𝛽𝑖) =1+exp[𝛼exp[𝛼𝑖(𝜃𝑗−𝛽𝑖)]
𝑖(𝜃𝑗−𝛽𝑖)] (2)
Where 𝜃𝑗 represents the ability parameter for examinee j; 𝛼𝑖 (item discrimination) and 𝛽𝑖 (item difficulty) refer to the parameters of item i. Assessment of measurement invariance across time involves checking that the item parameters 𝛼𝑖 and 𝛽𝑖 have not changed over time.
For this study, we assume that all items in a test are equally discriminating, therefore the unidimensional IRT using Rasch model (one-parameter logistic model: 1PLM) was apply, the ConQuest software (Adams, Wu, & Wilson, 2012) was used to estimate students’ abilities of morphological structure, morphological production, phoneme isolation, reading
comprehension and word recognition as measured by the respective scales presented in Table 3.
In Rasch analysis, item fit indexes are reported for individual items by Mean Square error (MNSQ). The MNSQ statistic is sensitive to response patterns of persons whose ability estimates match an item’s difficulty estimate. Over fit indicates that the observations contain less variance than is predicted by the model; under fit indicates more variance in the
observations than is predicted by the model (e.g., the presence of idiosyncratic groups) (Wilson, 2005). The weighted and un-weighted MNSQ differ in that the weighted MNSQ weighs persons performing closer to the item value more heavily; therefore, persons whose ability is more closely matched to the items’ difficulty level will be weighted more heavily
37
than those who are not (Bond & Fox, 2001). Bond and Fox (2001) recommend that Rasch modellers pay more attention to the weighted MNSQ. According to Adams and Khoo (1996), items with adequate fit will have weighted MNSQ between .75 and 1.33. To justify item fit in this study the ranging of MNSQ of Adams and Khoo (1996) was applied.
Given that the ability of the morphological structure test, morphological production test, phoneme isolation test, reading comprehension and word recognition were estimated using UIRT measure, the Rasch reliability (EAP/PV) was applied to determine the reliability of those tests. ConQuest software (Adams, Wu, & Wilson, 2012) was used to generate the Rasch reliability (EAP/PV) of the scales, which indicates the extent to which the observed total variance is accounted for by the model variance. It indicated “how well the sample of subjects had spread the items along the measure of the test” (Fisher, 1992). The EAP/PV reliability is analogous to Cronbach’s alpha and can be interpreted similarly where the minimum acceptable cut-off level for Cronbach’s alpha is 0.50 (Portney & Watkins, 2000).
For the rapid word segmentation test, rapid colour naming test and rapid number naming test, which are the speed tests, instead of using the raw score of each test, the log-transformed scores was applied. Those three scale score were expect to have normal distribution, thus consistent with the previous study of Boscardin, Muthén and Francis (2008), the raw scores from the three tests were log-transformed scores to approximate a normal distribution before the analyses. The log-transformed was described by Boscardin, Muthén, and Francis (2008) using the following formula:
𝐿𝑇𝑆 = 𝑙𝑜𝑔2{ 𝐶𝑂𝑅 (𝑇𝑇𝑆
𝑇𝐼𝑅 + 0.1) }
38 Where;
LTS : Log-transformed score COR : Correct response TIR : Time response TTS : Total score
Since the rapid colour naming and rapid number naming tests were used to collect data repeatedly at two time points to measure each characteristic. The consistence of these
measurements over time was used to determine the reliability of these two tests; in other words, the test-retest reliabilities of the rapid colour naming and the rapid number naming test were reported in this study. Reliability of the rapid word segmentation test was measured in terms of its internal consistency. That is, Cronbach's alpha was applied to determine its reliability. Table 4 presented the methods used to determine the validity and reliability of all tests in this study.
Table 4
Validation of each test.
Tests Ability score Reliability
Cognitive components Morphological awareness
Morphological structure test UIRT EAP/PV
Morphological production test UIRT EAP/PV
39 Table 4 (Continued)
Validation of each test.
Tests Ability score Reliability
Decoding skills
Phoneme isolation test UIRT EAP/PV
Rapid word segmentation test Log-transformed Cronbach's alpha RAN
Rapid colour naming Log-transformed Test-retest Rapid number naming Log-transformed Test-retest Reading abilities
Reading comprehension test UIRT EAP/PV
Word recognition test UIRT EAP/PV
3.5 Measurements
The measurements of this study consist of two parts: a cognitive component tests and a reading ability tests in context of Thai. Six tests of the cognitive components and two tests of the reading abilities tests were developed based on the fourth-grade Thai curriculum guidelines issued by the Ministry of Education. The following Table 4 shows the tests used in this study.
3.5.1 Cognitive component tests
3.5.1.1 Morphological awareness. The original test was created by McBride-Chang, Wagner, Muse, Chow and Shu (2005) to measure the morphological awareness of children in kindergarten and second grade. In order to measure the morphological awareness of the fourth-grade student participants in this study, a modified version was used. The
40
modifications made the test more appropriate for Thai language and the age group of the participants. Two sub-tests were used to assess the morphological awareness domain: a morphological structure test and a morpheme production test.
(1) Morphological structure test. The morphological structured test measured the ability to produce words by using different morphemes. This test comprised 15 items, each consisted of a sentence structure in which an inflected word was embedded. The students were asked to use a sentence to make a new word that fit in with the scenario provided. All items consisted of real words. For example:
First sentence: งานที่ครูให้นักเรียนน ากลับไปท าที่บ้านเรียกว่า ‘การบ้าน’: Work that a teacher asks a student to do at home, we call 'homework'.
Second sentence: งานที่ครูให้นักเรียนท าที่โรงเรียนเรียกว่า ‘การโรงเรียน’: Work that a teacher asks a student to do in school, we call (schoolwork).
The score was based on the number of correct responses.
(2) Morphological production test. In this test, two-syllable (two-morpheme)Thai words were orally presented by the experimenter. Within each two-morpheme word, one morpheme was specified as a target. The students were then asked to provide two new words by using the target morpheme. One word should have the same meaning as that of the
original word, and the other word should have a different meaning; both morphemes were identical in pronunciation.
For example, given the word 'ห้องครัว'/'hong-kour' ('kitchen'), the students were asked to give a new word by using the target morpheme 'ห้อง', in which 'ห้อง' had the same meaning as that of the original word 'ห้องครัว'. One acceptable answer would be 'ห้องอาหาร'/'hong-arhan' ('kitchen with dining table').
41
Next, the students were asked to give a new word that included the target morpheme 'ห้อง'/'hong' but with a different meaning from that of the original word 'ห้องครัว'. One acceptable answer would be 'ห้องขัง'/'hong-kang'(jail).
This test comprised 10 items, each consisted of real words. The total possible score was 20 points, with two points for each item. The score was based on the number of correct responses. The test was administered individually.
3.5.1.2 Decoding skills. Two sub-tests were used to assess the decoding domain: a phoneme isolation test and a rapid orthographic word chain test.
(1) Phoneme isolation test. This test was developed according to the original test of phoneme isolation created by Hulme, Caravolas, Malkova, and Brigstocke (2005). In this test, each item required reading four words aloud, and the students were asked to check which word used a different initial consonant sound from the other three and then to mark this on the answer sheet. For example, four words were presented: (1)
ข้าว
: Kaw, (2)เข้ม
:Kaem, (3)
เข็น
: kaen and (4)ร้าว
: raw, and the students were instructed to select the wordwhich had an initial consonant different from the other three. The correct response is (4) since its initial consonant differs in sound from those of the others (r:
รอ
), which all sound the same(k:
ขอ
). The phoneme isolation test had 15 items in total, and the score was based on thenumber of correct responses.
(2) Rapid word segmentation test. The rapid word segmentation test of this study was developed based on the original test of the orthographic segmentation task created by Bråten, Lie, Andreassen, and Olaussen (2009). In this test, the items were un-spaced, printed words presented in rows, starting from two words and going up to nine words (totalling 32 items).
42
The students were instructed to identify the words by drawing lines to separate all words contained within the un-spaced, printed string of words. For example, ก าไรคงคาค าราม consisted of three words that students had to identify by drawing two lines to separate the words: ก าไร (profit)/คงคา(river)/ค าราม(roar). The test score was based on the total number of words that were correctly separated within 60 seconds.
3.5.1.3 Rapid automatized naming. This test was adopted from the RAN/RAS test (Wolf & Denckla, 2005). There were two sub-tests: rapid colour naming (RCN) and rapid number naming (RNN). Two measures of speeded naming, which measured the naming of objects, were administered individually to the children.
(1) RCN. The test consisted of five colours (red, black, yellow, blue and green) in a row, recurring 10 times in random sequences. The test involved naming the five colours as quickly as possible. To ensure the students’ familiarity with the names of these five colours, practice items were used. The students were asked to name each colour two times, and the score was based on the average time spent in completing the test. The test was administered
(1) RCN. The test consisted of five colours (red, black, yellow, blue and green) in a row, recurring 10 times in random sequences. The test involved naming the five colours as quickly as possible. To ensure the students’ familiarity with the names of these five colours, practice items were used. The students were asked to name each colour two times, and the score was based on the average time spent in completing the test. The test was administered