ࡱ> %` ̎bjbj o̟̟/hhhh$nWnWnWPWL Xiv0Y~ZZZZZ$Z ZnopopopoZoru$whzvZZZZZvh"ZZ#vfffZ ZZnofZnofffZY 0 M{9nWzefRf 9v0ivfzzezfzfDZZfZZZZZvvfZZZivZZZZPnWnWhhh  THE APPLICATION OF RASCH MODELLING TO YES/NO VOCABULARY TESTS John Shillaw University of Tsukuba The study described in this paper examined the measurement properties of three Yes/No Vocabulary tests. The results of the study show that all three were unidimensional, reliable tests There were no major differences in students' scores on the tests whether all words, or real words only, were used for analysis. Further studies need to conducted, but there are promising indicators that Rasch modelling might be a useful method for producing more accurate Yes/No vocabulary tests. introduction Yes/No Tests Yes/No tests are based upon lexical decision tests that have been used extensively by cognitive psychologists attempting to model the mental lexicon (Taft 1991). These tests work by measuring subject's responses to real words and nonwords shown on a screen or computer display. Yes/No tests are similar in concept and are very easy-to-design and administer. The tests are composed of a number of real English words and a smaller number of nonwords, and subjects simply have to indicate whether or not they know the words. The score on the test is the proportion of real words claimed to be known adjusted for the proportion of nonwords identified as known. Anderson and Freebody (1983) compared the results of Yes/No tests with conventional multiple-choice items given to native-speaking children, and conclude that Yes/No vocabulary tests are better suited as a means of assessing their vocabulary knowledge. Their reasons for reaching this conclusion are: Yes/No tests are easier to construct; include more items; were easier for the children to answer, and the scores from the Yes/No tests are better predictors of reading ability than conventional vocabulary tests. Meara and Buxton (1987) used the same method with a group of EFL students and found that the scores from Yes/No tests were better at predicting students' grades on the First Certificate Examination than the scores from FCE multiple-choice vocabulary items. However, despite the reported effectiveness of Yes/No tests with both L1 and L2 subjects, there are several questions about the measurement properties of the tests. Firstly, the formula used to derive the scores can severely penalize students who incorrectly identify nonwords, and if the number of false alarms is high, it is possible for a subject to have a negative score on a test. Secondly, no attempt has been made to examine the characteristics of the items used in the tests to assess how much they contribute to determining a subjects score. The assumption appears to be that the items collectively make a good test, and no assessment has been made of the validity and reliability of the tests. Thirdly, because there is no statistical information on the items used in the tests, it has not been possible to create, with any degree of accuracy, Yes/No tests at a range of levels. So, although the Yes/No tests are described as vocabulary tests, they have never truly undergone any real scrutiny to determine whether they meet the measurement criteria that more conventional tests are required to meet. This study seeks to address the measurement questions mentioned above. Specifically, it aims to: 1. Examine the construct validity and reliability of three Yes/No tests. 2. Explore the use of Rasch scaling as a method of analysis to obtain information about the items in the tests and student responses to the items. 3. Examine whether Yes/No tests yield reliable results without including nonwords, i.e., using only a test of real words. 4. Assess whether pre- calibrated items can be used effectively to anchor items on other tests so as to be able to build up a bank of items that can be reliably used at different levels of ability. Rasch Scaling Rasch scaling is now widely used as the preferred method of analysis in mainstream testing. A full explanation of the theory underlying Rasch analysis is beyond the scope of this paper, but its merits have been described in detail within broad areas of test measurement (Lord, 1980; Wright and Stone, 1979; Hambleton, Swaminathan and Rogers, 1991) and within the field of EFL testing (Henning, 1987; Woods and Baker, 1985). The essential features of Rasch analysis are: 1. The difficulty of items and students' ability are measured on a common scale the logit scale which allows for a direct comparison to be made between the difficulty of an item and the probability of a student at any level of ability getting it correct. 2. Tests scores are independent of the restrictions of item difficulty and test population that limit classical test theory and analysis. 3. Items (or subjects) whose responses deviate from the population norm can be readily identified. 4. Items that have been pre-tested and calibrated can be used to anchor untested items so as to maintain a consistent scale. The application of Rasch scaling to Yes/No tests is no different to that of traditional language tests, and assumes that the scores from the test reflect a student's underlying competence/proficiency in vocabulary. One prerequisite for the application and interpretation of Rasch analysis is that the skill being measured is unidimensional and that the tests used are valid measures that adequately assess this latent trait. If any tests cannot be shown to be unidimensional, then Rasch analysis, strictly speaking, should not be used. A second prerequisite of Rasch analysis is that the target subjects should be representative of the broader target population and that the subjects should be a stable sample. Any marked deviation from the normed values of items will result in inaccurate measures and usually leads to misfits in items and subjects. Any misfitting items can be identified by large outfit statistics and need to examined for inconsistencies. If a student responds inconsistently to items whose difficulty levels are inconsistent with his/her estimated ability, they will be also flagged as misfitting and their scores have to be interpreted with caution. Method Instruments Meara (1992) has developed the Swansea Vocabulary Tests, which are a series of Yes/No tests at six difficulty levels. Three of the levels (1,2 and A) are based upon the word lists proposed by Nation (1986), whilst the other three are based upon Hindmarsh (1982). Each level comprises 20 tests, each test made up of 60 words: 40 real words and 20 nonwords. For the purpose of this study, three of the level 2 tests were used on the assumption that most of the subjects have a fairly basic vocabulary knowledge. In addition, Meara (personal communication) is of the opinion that the level 1 and 2 tests are the most reliable. Further information about the tests used is given in the procedure section. Subjects In the first part of the study, 7 classes of first-year undergraduate students (Group 1) were tested. The classes included students of differing levels of ability, and were broadly representative of the proficiency range within the total first-year undergraduate population. Six departments were represented: Physics (n=32), Medicine (n=27), Biology (n=26), Agriculture (n=33), Sports Science (n=29), and Social Planning (n=34). The seventh class was a small group (n=17) composed of the most advanced first-year students and a few second and third year students. In the classes were 3 students who were repeating courses they had failed in earlier years, and the total number of students was 201. For the second part of the study, the same 7 classes were reused (n=196) and 5 new classes (Group 2) were added. The second group were students of Physics (n=25), Medicine (n=22), Agriculture (n=38), Sports Science (n=30), and a second Advanced group of students from different departments (n=20). There were 4 students who were repeating classes, and a total of 139 students. Group 1 and Group 2 were matched by department, and, as closely as possible, by proficiency level. Procedure Stage 1 Two of the Swansea level 2 Yes/No tests, tests 203 and 207, were given to the Group 1 students. The tests are reproduced in Appendix 1. Both tests were administered consecutively in normal class time by the class teacher. Students were instructed to check the words on the test with Y or N as indicated in the test rubric and, when this was completed, they then re-entered the responses onto a magnetic card to be used for the computerized scoring. All Y responses were recoded as a, and N responses as b. The students also entered their department, class number and student identity number on the card. At the end of the test, all papers and cards were collected by the class teacher. Stage 2 All the magnetic cards were processed using a Sekonic 505 optical card reader and data files of student responses on the items were produced for each class. If there was any missing data from the cards, the test sheets were checked to try to recover the missing information. Errors were few and most missing information was retrieved. The data from each class was then combined into one file for each test and prepared for Rasch analysis. The two tests were analysed separately using the Bigsteps 2.1 program (Wright and Linacre; June, 1991). Both tests were analysed twice: firstly, with all items included, and secondly, an analysis of the real words only. Based on the analysis of all items from both tests, 15 items (10 real words and 5 nonwords) were selected for inclusion as anchor items in the third Yes/No test. Stage 3 The Swansea Yes/No test 218 was selected for the final testing stage. The 15 anchor words selected at stage 2 were inserted into the test and replaced a word of the same category, so that the ratio of real words to nonwords was maintained (see Appendix 2: real word anchors are underlined; nonword anchors are in bold italics). The test was then given to Group 1 and Group 2, and the administration and scoring was exactly the same as in stage 1 outlined above. Stage 4 The final stage involved a fairly complex series of analyses. Firstly, all the tests were analysed for unidimensionality as outlined in the following section. Then, the scores from test 218 were analysed using Bigsteps, each set of scores analysed 4 ways. 1) All words (real and nonwords), unanchored. 2) All words anchored against the values of the 15 anchor items computed from the analysis of tests 203 and 207, where all words were included for analysis. 3) The real words only, unanchored. 4) The real words only, anchored against the values of the 10 items computed from the analysis of tests 203 and 207, where only the real words were included for analysis. Additionally, all the analyses detailed above were done separately for Group 1 and Group 2, and then for both groups combined. The following table summarises the methods of analysis used for the three tests. U = unanchored A = anchored TEST 203TEST 207TEST 218ALL WORDSREAL WORDSALL WORDSREAL WORDSALL WORDSREAL WORDSGROUPUAUAUAUAUAUA1((((((((2((((BOTH(((( Table seq Table \* arabic 1 Methods of Rasch analysis Analysis and discussion Evaluating unidimensionality and reliability There is no single, recognized test for test unidimensionality (Henning, Hudson and Turner, 1985; Bejar, 1980), however, conventionally, unidimensionality for a test can be claimed when one of two conditions is met from the results of factor analysis (Reckase, 1979). 1) The results of item level factor analysis show that the first factor accounts for at least 20% of the variance of the unrotated factor matrix. 2) The eigen value of the first factor is significantly higher than that of the next largest factor. It is important to note that the product-moment correlation matrix, which is usually used as input for most factor analysis, is not considered appropriate for the analysis of dichotomous test responses (Divigi, 1979). The preferred method is to do a factor analysis on the tetrachoric correlation matrix, which is the method that was used here. To test for unidimensionality and reliability, an item level factor analysis was done for all tests. Each test was analysed by, firstly, using all items, and then, using the real words only: the analysis of test 218 was further sub-divided by group. All analyses were done using the item factor analysis program Testfact 2.6 (Wilson, Wood & Gibbons, 1991). The results of the analyses are presented in ref B_Ref317868176 \* mergeformat Table 2 and show that, based on the above criteria, unidimensionality can be assumed for all of the analyses. ALL WORDSREAL WORDS ONLYTESTGROUP1st FACTORKR-201st FACTORKR-20203127.78%0.74325.49%0.848207129.86%0.68031.04%0.847218123.78%0.74732.17%0.848218222.75%0.74130.69%0.843218BOTH23.18%0.74330.90%0.845 Table seq Table \* arabic 2 Results of item factor analyses It is notable that the analyses where only the real words were used have a higher first factor variance than when all items are used; test 203 excepted. Further, the KR-20 coefficients are markedly higher for all the tests using the real words only. The differences can be explained by the fact that many of the nonwords have a very low or negative point-biserial correlation and the removal of these items from the analyses has led to the improved figures. This will be discussed further in the following section dealing with the Rasch analysis. Rasch analysis As mentioned earlier, the Rasch analyses were done using the Bigsteps 2.1 program. Whilst it is not the easiest program to use, it produces a wide range of information about the items in the test and the performance of the subjects. Most notably, part of the analysis output is the degree to which items or subjects fit the Rasch model and the effectiveness of the items and subjects as measured by the point-biserial correlation. The fit of the items and subjects is of major concern in this study because the effectiveness of the Yes/No tests largely rests upon there being few misfitting items and/or subjects. If either of the two parameters is markedly high, then the usefulness of Rasch modelling as a means of analyzing Yes/No tests is in doubt. According to McNamara (1990), the conventional criterion for assessing either an item or subject as misfitting is when the outfit of either exceeds (2. However, so little is actually known about the behaviour of items on tests of differing lengths, and with a variable number of subjects, it perhaps overly conservative to accept this criterion as nothing more than a useful rule of thumb. Nevertheless, mindful of this caveat, for the purpose of this study all misfitting items/persons are reported using the above criterion. The following table summarises the degree of misfit of items and subjects on the 3 tests under all analysis conditions. The figures show that, at a liberal estimate, on any one test no more than 15% of the items are designated as misfitting and only fractionally more than 7% of the subjects misfit. What is apparent from the table is that the difference is minimal between the number of misfitting subjects on the two forms of the tests. There is a greater difference, however, between the number of misfitting items on the test with all words and real words only. This can be accounted for by the fact that almost all the misfitting words on the test which included all words are nonwords. As noted earlier, the point biserial correlations for almost all of the nonwords were around 0, or slightly negative, and this would account for the high number of misfitting nonwords. In addition, U = unanchored A = anchored ALL WORDSREAL WORDSTESTGROUPU/AITEMSSUBJECTSITEMSSUBJECTS2031U6104122071U54132181U812492181A610492182U89372182A7938218BOTHU924418218BOTHA721417 Table seq Table \* arabic 3 Misfitting items and subjects almost all of the nonwords had very high facility values which would further explain the low point biserial correlations. Further to this point, most of the subjects who were identified as misfitting on the tests including all words were subjects who claimed not to know some of the easier real words. Few subjects' misfit was due to incorrectly claiming knowledge of nonwords. On the tests that included real words only, the misfitting items were words that it was anticipated students would not know, and the misfit may be due partially to guessing. However, there were some words that most students should have known and the misfit was caused by some of the more able students marking them as not known. Again, misfitting students were mostly those who indicated a lack of knowledge of some of the easier items. Overall, it would appear that both forms of the test, all words and real words, were successful tests in terms of the relatively low number of misfitting items and subjects, even using a rather conservative criterion of misfit. Had a more liberal criterion been adopted, identifying only the most noisy misfitting items and subjects, then the number of misfits would have been much lower. The question of what the criterion should be has to be left unresolved until further investigation can be made by cross-validation with other reliable measures. A comparison of the two forms of the test The results of the study so far suggest that either form of the Yes/No tests could be used as a form of assessment of vocabulary knowledge. There would, however, be obvious advantages if it could be shown that the test using only the real words was at least as effective as the test containing real and nonwords. To explore this matter, the scores from the two different forms of the tests, under all analysis conditions, were correlated, and the results of the correlations are tabulated below. All intercorrelations are reported in tables 5 to 7, but in fact only one set of correlations is important, the correlation between the tests including all words and the tests of real words only. All other anchored correlations are simply rescaled versions of the unanchored version and the intercorrelation with the unscaled version should be at or about unity. The figures reported in Table 4 are different because they are the intercorrelations between 2 unscaled tests. The following points pertain to Table 4 and the figures shown in Table 5 to 7. Table 4 shows that the intercorrelations on the two tests taken by Group 1 are very high, with a more than adequate level of agreement between the two tests. It should be noted that the correlation between the scores on the two tests including real words only is markedly higher than that between the tests including all words. This is undoubtedly due to the scores on the tests of all words being depressed by the marginal contribution of the nonwords to the test variance. As would be expected, the correlation is high between the scores on the test including all words and the test of real words only, reflecting, to some extent, a part-whole correlation. However, there is only a partial effect because when the real words only are analysed as a separate test, the scores on the items are rescaled according to the responses on these items only. In other words, the distribution of subjects' scores on subtest, and whole test, when analysed by Rasch analysis, will be different from correlating raw scores on a subtest with the scores on a whole test in the traditional way. Having said this, the variation in logit scores was not very high, and again this due to the low contribution by the nonwords to the test variance. 203 ALL203 REAL207 ALL203REAL.8900207ALL.7148.6689207REAL.7074.7967.8978 Table seq Table \* arabic 4 Correlations for scores on tests 203 and 207 218ALL UNANCHORED218ALL ANCHORED218REAL UNANCHORED218ALL ANCHORED.9999218REAL UNANCHORED.9311.9310218REAL ANCHORED.9200. 9200. 9858 Table seq Table \* arabic 5 Correlations for scores on test 218 All students 218ALL UNANCHORED218ALL ANCHORED218REAL UNANCHORED218ALL ANCHORED1.0000218REAL UNANCHORED.9329.9330218REAL ANCHORED.9322.9322.9999 Table seq Table \* arabic 6 Correlations for scores on test 218 Group 1 218ALL UNANCHORED218ALL ANCHORED218REAL UNANCHORED218ALL ANCHORED1.0000218REAL UNANCHORED.9281.9280218REAL ANCHORED.9268.92681.0000 Table seq Table \* arabic 7 Correlations for scores on test 218 Group 2 It would appear, therefore, that for this group of subjects that there is very little difference in the variance between test scores regardless of whether one uses the scores from the tests including all the words, or the tests including real words only. The only tangible difference was that the subjects' logit scores on the tests of real words only were lower than on the tests of all words. This simply reflected the high facility values of the nonwords on the tests of all words. Predictive validation Having established the reliability of the tests through Rasch analysis, the question remains whether the Yes/No tests actually have any relationship with proficiency in English. To determine this, the scores from the Yes/No tests were correlated with the total scores on the proficiency test taken by all subjects 3 months after the administration of the Yes/No tests. The correlations were between 0.42 and 0.48, and significant at p < 0. 00. Whilst these correlations are quite moderate, they are much better than the correlations between the scores on the proficiency test and the scores on the Yes/No tests using the high-threshold formula used in other studies. None of the scores based on the high-threshold formula, for any form of the Yes/No tests, achieved significance, with most correlations around zero, and some negative. Anchoring the tests The correlations reported in tables 5 to 7 clearly show that there was little variation, irrespective of whether the tests were anchored or not. However, using a program called RAQUEL (Rasch Anchor Quality Evaluation and Linking) revealed that only about half of the anchor items on any of the tests were in an acceptable range to be considered effective. This was assessed by considering the (2 fit of the distribution of responses between the criterion groups on the different tests (Group 1 vs. Group 1; Group 1 vs. Group 2, and Group 1 vs. both groups combined). These figures, once again, have to be examined with caution because the program assumes a level of homogeneity that was not exhibited in the scores for two groups. Further, both real and nonword anchor items were selected from a range of items that had highly variable point biserial and outfit properties. Had the anchor items been selected from only those items with good outfit statistics, the anchoring process undoubtedly would have been more accurate. Unfortunately, this would have precluded the use of almost every nonword. conclusion The results of this small study appear to show that the Yes/No tests used are more than adequate test instruments from the point of view of measurement criteria. The test scores from the Rasch analysis are superior to scores derived using the high-threshold formula when compared to scores on a standardized proficiency test. For this group of students, it would appear that essentially the same information about vocabulary knowledge can be assumed irrespective of whether scores from the tests of all words are used, or scores from the test using only the real words. In other words, on these tests and for these students, the presence of nonwords had little effect on their test performance. It is too early to conclude that tests made up only of real words can replace the usual Yes/No vocabulary test format, but there is some reason for optimism that it might be possible. Further studies with larger samples, with students with different L1 backgrounds, and students at different levels of proficiency will be necessary before any firm conclusions can be drawn. Probably the most exciting prospect arising from this study is the possibility of creating a whole range of tests by selectively sampling words across a range of frequencies. Once a core sample of words has been calibrated, new tests can be produced very quickly, anchored by using some of the calibrated items. The new tests can be administered to representative, sample groups within a given population so that a large number of words can be calibrated at one time. There is also great potential for compiling tests that can be easily administered by computer on the same principles as current Computer Adaptive Tests (CAT). The CAT item bank would consist of a large number of calibrated items at a range of difficulty levels and a student would probably only need to respond to something like 25 30 words before a final level could be determined. Acknowledgments. I would like to thank all the teachers and students who took part in the administration of the three tests during a very busy time in the academic year. My thanks also to Dr. Neil Jones of UCLES for his advice on using Bigsteps and RAQUEL. References Anderson, R.C. and Freebody, P. 1983: Reading comprehension and the assessment and acquisition of word knowledge. Advances in Reading Language Research 2, 231-56. Bejar, I. (1980). A procedure for investigating the unidimensionality of achievement tests based on item parameter estimates. Journal of Educational Measurement 17(4), 283-96. Divigi, D. R. (1979) Calculation of the tetrachoric correlation. Psychometrika 44, 169-172. Hambleton, R. K., Swaminathan, H. & Rogers, H. J. (1991). Fundamentals of Item Response Theory. Newbury Park: Sage. Henning, G. (1987). A Guide to Language Testing. Cambridge, Mass.: Newbury House. Henning, G., Hudson, T. & Turner, J. (1985). Item response theory and the assumption of unidimensionality for language tests. Language Testing 2, 2 141-154. Hindmarsh, R. (1982) Cambridge English Lexicon. Cambridge: Cambridge University Press. Lord, F. M. (1980). Applications of item response theory to practical testing problems. Hillsdale NJ: Lawrence Erlbaum. McNamara, T. F. (1990) Item Response Theory and the validation of an ESP test for health professionals. Language Testing 7,1, 52-75. Meara, P. (1992). EFL Vocabulary Tests. Centre for Applied Language Studies, University of Wales, Swansea. Meara, P. and Buxton, B. (1987). An alternative to multiple choice vocabulary tests. Language Testing 4, 2, 142-151. Nation, I. S. P. (1986). Word Lists (revised edition). Wellington: Victoria University English Language Centre. Reckase, M. D. (1979) Unifactor latent trait models applied to multifactor tests: results and implications. Journal of Educational Statistics 4(3), 207-30. Taft, M. (1991). Reading and the Mental Lexicon. Hillsdale NJ: Lawrence Erlbaum. Wilson, D. T., Wood, R., & Gibbons, R. (1991). Testfact. Chicago, IL.: Scientific Software Inc. Woods, A., & Baker, R. (1985). Item Response Theory. Language Testing 2, 2, 119-140. Wright B. D. & Linacre, M. (1991) Bigsteps 2.1. Chicago: Mesa Press. Wright, B. D. & Stone, M. H. (1979). Best Test Design. Chicago: Mesa Press. Notes APPENDIX 1 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ LEVEL 2 Test 203 _______________________________________________________________________ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Write your student id. number here ___________________________ What you have to do: Read through the list of words carefully. For each word: if you know what it means, write Y (for YES) in the box if you don't know what it means, or if you aren't sure, write N (for NO) in the box. _____________________________________________________________________ 1 [ ] conduct 2 [ ] dominate 3 [ ] perform 4 [ ] organism 5 [ ] ideal 6 [ ] court 7 [ ] leave out 8 [ ] growth 9 [ ] crowded 10 [ ] restificate 11 [ ] antile 12 [ ] magic 13 [ ] determine 14 [ ] spring 15 [ ] garrick 16 [ ] fraction 17 [ ] logalation 18 [ ] acquire 19 [ ] reflect 20 [ ] beam 21 [ ] aspect 22 [ ] column 23 [ ] kellett 24 [ ] separation 25 [ ] punishment 26 [ ] entertain 27 [ ] sink 28 [ ] fumicant 29 [ ] rescue 30 [ ] ruin 31 [ ] skelding 32 [ ] advertise 33 [ ] mascarate 34 [ ] mollet 35 [ ] angle 36 [ ] webbert 37 [ ] uniform 38 [ ] physical 39 [ ] inspect 40 [ ] dyslaxative 41 [ ] cement 42 [ ] correctivate 43 [ ] portman 44 [ ] progress 45 [ ] transmit 46 [ ] external 47 [ ] primality 48 [ ] beautitude 49 [ ] worrall 50 [ ] technique 51 [ ] exchange 52 [ ] cordle 53 [ ] challinor 54 [ ] hardly 55 [ ] keable 56 [ ] mount 57 [ ] volt 58 [ ] mean 59 [ ] pruden 60 [ ] bubble ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ LEVEL 2 Test 207 _______________________________________________________________________ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Write your student id. number here ___________________________ What you have to do: Read through the list of words carefully. For each word: if you know what it means, write Y (for YES) in the box if you don't know what it means, or if you aren't sure, write N (for NO) in the box. _____________________________________________________________________ 1 [ ] sheep 2 [ ] barmion 3 [ ] possumate 4 [ ] helpful 5 [ ] prevent 6 [ ] economic 7 [ ] considerable 8 [ ] lang 9 [ ] possible 10 [ ] allaway 11 [ ] influence 12 [ ] freeze 13 [ ] neutration 14 [ ] neutral 15 [ ] defend 16 [ ] geography 17 [ ] tebbit 18 [ ] oaten 19 [ ] career 20 [ ] acquince 21 [ ] laboratory 22 [ ] extend 23 [ ] affair 24 [ ] over 25 [ ] loveday 26 [ ] besides 27 [ ] contribute 28 [ ] compel 29 [ ] brimble 30 [ ] cadle 31 [ ] site 32 [ ] fluctual 33 [ ] rate 34 [ ] gasson 35 [ ] act 36 [ ] mine 37 [ ] ideal 38 [ ] doubtly 39 [ ] coloniate 40 [ ] catalogue 41 [ ] unit 42 [ ] rickard 43 [ ] dimension 44 [ ] stand for 45 [ ] vibrate 46 [ ] observe 47 [ ] crime 48 [ ] putbrace 49 [ ] angle 50 [ ] salary 51 [ ] export 52 [ ] lock 53 [ ] report 54 [ ] ashill 55 [ ] suppose 56 [ ] attract 57 [ ] laminastic 58 [ ] tradition 59 [ ] revolution 60 [ ] solitist APPENDIX 2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ LEVEL 2 Test 218 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Write your student id. number here ___________________________ What you have to do: Read through the list of words carefully. For each word: if you know what it means, write Y (for YES) in the box if you don't know what it means, or if you aren't sure, write N (for NO) in the box. _____________________________________________________________________ 1 [ ] yandle 2 [ ] inner 3 [ ] bulbicate 4 [ ] defend 5 [ ] fluid 6 [ ] pimlott 7 [ ] point 8 [ ] attract 9 [ ] operation 10 [ ] factor 11 [ ] observe 12 [ ] discipline 13 [ ] progress 14 [ ] overend 15 [ ] fluctual 16 [ ] mean 17 [ ] aspect 18 [ ] barnish 19 [ ] laboratory 20 [ ] set 21 [ ] hopeful 22 [ ] decaphage 23 [ ] steep 24 [ ] uniform 25 [ ] sample 26 [ ] style 27 [ ] split 28 [ ] concerned with 29 [ ] dyslaxative 30 [ ] cement 31 [ ] revise 32 [ ] ackrill 33 [ ] exist 34 [ ] personal 35 [ ] chain 36 [ ] manomize 37 [ ] sparling 38 [ ] horobin 39 [ ] normal 40 [ ] abstemptious 41 [ ] upset 42 [ ] alternative 43 [ ] entire 44 [ ] advertise 45 [ ] brick 46 [ ] blind 47 [ ] radius 48 [ ] gorman 49 [ ] murrow 50 [ ] threat 51 [ ] deceive 52 [ ] harmonical 53 [ ] doubtly 54 [ ] transmit 55 [ ] dimension 56 [ ] mascarate 57 [ ] rebel 58 [ ] barmion 59 [ ] annual 60 [ ] encopulate    page 9  It was not possible to match classes from Biology and Social Planning because of examination schedules.  Testfact does not calculate the eigen values of the factors, however, it can be assumed that the first factor had a significantly higher eigen value than the second, because in all analyses the first factor had a variance more than double that of the second factor.  Outfit is a measure of the degree of distance a subject's score or an item's value deviates from the predicted value. The mean is 0 and the standard deviation is 1, so outfit can therefore be interpreted as a standard normal distribution.  This test is administered to all first-year students at the end of one year's course of study. The test consisted of two subtests of listening skills (10 and 14 items respectively); a subtest of vocabulary (10 items); a test of reading comprehension (13 items) and 10 items each of grammar and error recognition. All items are in MCQ format and the overall reliability (KR-20) of the test that the subjects took was 0.84.  This program was developed by Dr. Neil Jones of the University of Cambridge Local Examinations Syndicate \ ]  ?@[\z%{%$'%'e'l'((((9):)U)V)\)])k)l)%*,*~++,&,l-s-F/M/P0Q00ʭʟʑʑʑʑʑʑʟʟʟʟhoh85OJQJaJhoh86OJQJaJ%jhoh80JOJQJUaJh,OJQJaJhoh8OJQJaJhoOJQJaJhAYhoOJQJaJ!jhAYhAYOJQJUaJ4DQghN O \ ] j   $$$da$gdo$$dgdo$$dgdo$$d]^gdo $$dgdo$$$da$gdo/Gˎ#_?@[\$$$Qd^`Qa$gdo$$dgdo$$$SWd^S`Wa$gdo$$$^Qd^^`Qa$gdo $$$da$gdo"""z%{%Z'['e'm'$*%*-*~++k-l-t-E/F/N/ $$$da$gdo$$dgdo$$dgdoN/P0Q001C11122222222$$$d$Ifa$gdo $$$da$gdo0111112222_3a3b3d3e3g3h3j3k3m3n3o3p3q3r3s3t33333333333333333333333333䔂#hohoOJQJaJmHnHu!jhoh8OJQJUaJ jhoh8OJQJaJhoh85OJQJaJhoh,OJQJaJh8OJQJaJh,OJQJaJhoh8OJQJaJhoh8>*OJQJaJ3223 333*343?3~~~~~~~$$$d$Ifa$gdonkd$$Ifl4\k8%k   / 4 lalf4?3@3F3H3J3L3N3P3WEEEEEE$$$d$Ifa$gdokd$$Ifl4֞k= 8 %k            d  4 lalf4P3R3T3V3X3Z3\3^3_3a3c3d3f3g3i3j3l3m3o3q3s3u3v3x3y3z3{3FfFfK $$$d$Ifa$gdo{3|3}3~3333333333333333333333FfFf$$$d$Ifa$gdo3333333%4455/606779999999$$$d$Ifa$gdo $$$da$gdo$$dgdogd,$$dgdo$$dgdo33333/6067788#9$9F9G9N9O999999999:::::::::::e=q=@@@ö♌~~l#hohoOJQJaJmHnHuhoh85OJQJaJhoh,OJQJaJh8OJQJaJ%jhoh80JOJQJUaJhohoOJQJaJ!jhoh8OJQJUaJhoh86OJQJaJhoh8OJQJaJ h,h,h,OJQJaJ(9999999:~~~~~~$$$d$Ifa$gdonkd$$Ifl4\X `%X X   G   4 lalf4::: ::::$:jXXXXXX$$$d$Ifa$gdokd $$Ifl4ֈX `%X   X  X X X   4 lalf4$:%:):+:2:8:?:E:jXXXXXX$$$d$Ifa$gdokd"$$Ifl4ֈX `%X XXXX 4 lalf4E:F:J:L:S:Y:`:f:jXXXXXX$$$d$Ifa$gdokd#$$Ifl4ֈX `%X XXXX 4 lalf4f:g:k:m:t:z:::jXXXXXX$$$d$Ifa$gdokd$$$Ifl4ֈX `%X XXXX 4 lalf4::::::::jXXXXXX$$$d$Ifa$gdokd&$$Ifl4ֈX `%X XXXX 4 lalf4:::::==(=j[[MMM>$$dgdo $$$da$gdo$$dgdokdI'$$Ifl4ֈX `%X  X X X X   4 lalf4(=@@3B4BBBEEEEEEEE$$$d$Ifa$gdo $$$da$gd, $$$da$gdo@@@@3B4BBBBEEEEFFFFFFFFFFFFsHsNtNPPuUUUUUUUUUƯtpcccthoh,)OJQJaJh,)#hohoOJQJaJmHnHu!jhoh8OJQJUaJh,)OJQJaJhoh85OJQJaJhoh,OJQJaJh8OJQJaJh,OJQJaJ%jhoh80JOJQJUaJhoh8OJQJaJ jhoh8OJQJaJ&EEEEEF FFF}kkkkkkk$$$d$Ifa$gdokd{($$Ifl4rrz 4&    .   4 laf4FF F"F$F&FWA///$$$d$Ifa$gdo$$$d$If^a$gdokdm)$$Ifl4֞rz  4&            4 laf4&F)F+F.F/F3FE/$$$d$If^a$gdokd*$$Ifl4֞rz  4&  4 laf4$$$d$Ifa$gdo3F5F7F9F;F=F?F@FEkd+$$Ifl4֞rz  4&  4 laf4$$$d$Ifa$gdo@FDFFFHFJFMFOFQF$$$d$Ifa$gdo$$$d$If^a$gdoQFRFVFXFZF\FWA///$$$d$Ifa$gdo$$$d$If^a$gdokd-$$Ifl4֞rz  4&         4 laf4\F_FaFcFdFhFE/$$$d$If^a$gdokdG.$$Ifl4֞rz  4&  4 laf4$$$d$Ifa$gdohFjFlFnFpFrFtFuFEkd/$$Ifl4֞rz  4&         4 laf4$$$d$Ifa$gdouFyF{F}FFFFF$$$d$Ifa$gdo$$$d$If^a$gdoFFFFFFWA///$$$d$Ifa$gdo$$$d$If^a$gdokd0$$Ifl4֞rz  4&         4 laf4FFFFFFE/$$$d$If^a$gdokd1$$Ifl4֞rz  4&  4 laf4$$$d$Ifa$gdoFFFFFFFFEkd3$$Ifl4֞rz  4&         4 laf4$$$d$Ifa$gdoFFFFrHsH+JVLWLLtNuNPPtUuUvU~UUU$$$d$Ifa$gdo$$dgdo $$$da$gdo $$$da$gd,)gd,)$$dgdoUUUUUUv`J4$$$ d$Ifa$gdo$$$ zd$Ifa$gdo$$$ d$Ifa$gdo$$$d$Ifa$gdovkd?4$$If84\e <&      t&4 8af4yt,)UUUUXF0$$$ d$Ifa$gdo$$$d$Ifa$gdokd35$$If84\e <&      (t&4 8af4p(yt,)UUUU,kd6$$If84\e <&      (t&4 8af4p(yt,)$$$ d$Ifa$gdo$$$ zd$Ifa$gdoUUUUUU4vkd7$$If84\e <&          t&4 8af4yt,)$$$ d$Ifa$gdo$$$ zd$Ifa$gdo$$$ d$Ifa$gdo$$$d$Ifa$gdoUU"V#V$V%V7VGVZV[Vgnkd8$$If84\e <&      4 8af4$$$d$Ifa$gdodgdo$$dgdo U"V#V$VVVVVVVVVWWWWWWWWWWWXXXXXXXXXX \ \\\^^^___ ____#_$_+_,_hoh85OJQJaJhoh86OJQJaJ%jhoh80JOJQJUaJ#hohoOJQJaJmHnHu!jhoh8OJQJUaJh,)OJQJaJhoh,)OJQJaJh8OJQJaJhoh8OJQJaJ2[VkVqVrVsV$$$ zd$Ifa$gdo$$$ d$Ifa$gdo$$$d$Ifa$gdosVtVVV`N8$$$ d$Ifa$gdo$$$d$Ifa$gdokd9$$If84\e <&      (4 8af4p(VVVVV8kd;$$If84\e <&      (4 8af4p($$$d$Ifa$gdo$$$ zd$Ifa$gdoVVVVVVWN??$$dgdonkd3<$$If84\e <&          4 8af4$$$ d$Ifa$gdo$$$ zd$Ifa$gdo$$$ d$Ifa$gdoWWWW'W7WJWKW[Wvd$$$d$Ifa$gdonkd=$$If84\e <&      4 8af4$$$d$Ifa$gdodgdo[WbWcWdWeWxW~W8kd>$$If84\e <&      (4 8af4p($$$d$Ifa$gdo$$$ d$Ifa$gdo~WWWWW8kd^?$$If84\e <&      (4 8af4p($$$d$Ifa$gdo$$$ zd$Ifa$gdoWWWWWWWN??$$dgdonkd@$$If84\e <&          4 8af4$$$ d$Ifa$gdo$$$ zd$Ifa$gdo$$$ d$Ifa$gdoWWWX X3X4XDXKXvdN$$$ d$Ifa$gdo$$$d$Ifa$gdonkdqA$$If84\e <&      4 8af4$$$d$Ifa$gdodgdoKXLXMXNX4kd_B$$If84\e <&      (4 8af4p($$$ d$Ifa$gdo$$$ zd$Ifa$gdoNXaXgXmXnX$$$ d$Ifa$gdo$$$ zd$Ifa$gdo$$$ d$Ifa$gdo$$$d$Ifa$gdonXoXXX`N8$$$ d$Ifa$gdo$$$d$Ifa$gdokdC$$If84\e <&      (4 8af4p(XXXXXXXZZdUUGGG $$$da$gdo$$dgdonkdD$$If84\e <&          4 8af4$$$ d$Ifa$gdo$$$ zd$Ifa$gdoZZ0^1^E^bbbccff5j6jGj8k9k:k;k=k>kIkJk$$dgdU$$$da$gdo $$$da$gdo$$dgdo,____ccff6jFj#k,k0k6k9k:k;k=k>kHkJkkkkkrlllllllmmm?mcm󱧝si\i\hohUOJQJaJh8OJQJaJhUhUOJQJaJhUh856hUh85 hoh8hUhUOJQJaJhp2OJQJaJh_>5OJQJaJhoh86OJQJaJh,)OJQJaJhoh8EH OJQJaJ jchoh8OJQJaJhoh8OJQJaJ$Jkkkllmmzm{mmmpnqnnnFoGooo>p?ppp+q,qqqQd^`Qgdo$$dgdUcmymzm{mmmmmmQnbnonpnqnnnnnnn#oEoFoGoooooooo=p>p?pppppppp*q+q,qqqqqqqqr r!rSr[rrrrrrrrrss's(s)sOs_swsxsyshUOJQJaJhoh86OJQJaJhohUOJQJaJh8OJQJaJhoh8OJQJaJIq r!rrrrr(s)swsxssssss4t|tttt u # p@ P dgdo$ # p@ P da$gdoQd^`Qgdoysss|tuu3u4uSu]uwuuuu({|||||||} }-}.}ǂтӂ|45T^x?E ;CEQU r}ˆˈĴĴĴĢĴĴĴĢĴĴĢĴ"hp2h856OJQJ^JaJhp2h8>*OJQJ^JaJhp2h8OJQJ^JaJhp2h85OJQJ^JaJhoh85OJQJaJhp2hU5OJQJaJ@ u(uJuuuuu/v0vpvqvvvvv7w8wzw{wwwxxBxCxxx # p@ P dgdoxxx y yMyNyyyyyz zczdzzzzz'{p{{{||Y|n| # p@ P dgdon|o||||"}D}}}}}~~W~X~~~~~bc # p@ P dgdo)*ij12tu<=Ƃ҂ӂ$ # p@ P da$gdo # p@ P dgdoӂ4|}у҃ )K23uvDE # p@ P dgdoʆˆ NOч҇WX߈!"de # p@ P dgdỏՉ/02389;<ABCDFGH^_Ɏʎˎ̎ùÏphp2hU6CJOJQJaJhp2hUCJOJQJaJ)jhp2hU0JCJOJQJUaJhUhAYCJmHnHu hUCJjhUCJUhvL2jhvL2U"hp2h856OJQJ^JaJhp2h8>*OJQJ^JaJhp2h8OJQJ^JaJ#/124578:;EFG^ʎˎ̎$a$ # p@ P dgdo7 0P@PBP. A!n"n#n$n% 4P@PBP. A!n"n#n$n% Ddo vT  C 0Alognostics4b L/# :IT%yoD(nL/# :IT%yoPNG  IHDR*˟PLTE3f3333f333ff3fffff3f3f̙3f3333f3333333333f3333333f3f33ff3f3f3f3333f3333333f3̙333333f333ff3ffffff3f33f3ff3f3f3ffff3fffffffffff3fffffff3fff̙ffff3fffff3f̙3333f33̙3ff3ffff̙f3f̙3f̙̙3f̙3f3333f333ff3fffff̙̙3̙f̙̙̙3f̙3f3f3333f333ff3fffff3f3f̙3fDbPtRNS%bbKGDH cmPPJCmp0712HsGIDATXG[v Dyj4 .z!r_k?²vtGGT]~#mlaBG[!VImk@#{- 뤱G\0]1 = ӷ#Ey?CH˹if4*A*pe86!{l!9<Ⳅt!È08F|? Сy"s )2ZXfbB}#J|%nD2?$Qlnj8tXQk'u+B/RO$$Ifl!vh5X5X5X5X5X5#vX#v:V l45X5/  / / /  /  / / / /  / / / /  4alf4L$$Ifl!vh5X5X5X5X5X5#vX#v:V l45X5/  / / / /  / / / / / /  / / /  4alf40$$Ifl!vh5X5X5X5X5X5#vX#v:V l45X5/  / /  / / / / / /  / / /  4alf40$$Ifl!vh5X5X5X5X5X5#vX#v:V l45X5/  / /  / / / / /  / / /  /  4alf4$$If!vh5555 5. #v#v #v. :V l455 5. /  /  / /  / /  / 4af48$$If!vh5555555#v#v:V l455/  /  /  / / / / /  / / /  /  4af48$$If!vh5555555#v#v:V l455/  / / / / / / / /  / / /  4af48$$If!vh5555555#v#v:V l455/  /  / / / / /  / / / / /  4af4*$$If!vh5555555#v#v:V l455/  / / / / / / /  /  /  / 4af48$$If!vh5555555#v#v:V l455/  /  / / / / /  / / / / /  4af4*$$If!vh5555555#v#v:V l455/  / / / / / / /  /  /  / 4af4*$$If!vh5555555#v#v:V l455/  /  / / / / /  / / /  /  4af48$$If!vh5555555#v#v:V l455/  / / / / / / / /  / / /  4af4*$$If!vh5555555#v#v:V l455/  /  / / / / /  / / /  /  4af4$$If!vh5 5 5 5 #v :V 84t&5 / /  / /  / /  /  / / 44 8f4yt,)a$$If!vh5 5 5 5 #v :V 84 (t&5 / /  / / / / / /  /  / 44 8f4p(yt,)7$$If!vh5 5 5 5 #v :V 84 (t&5 /  / / / / / /  44 8f4p(yt,)$$If!vh5 5 5 5 #v :V 84t&5 /  /  / / / / /  /  44 8f4yt,)$$If!vh5 5 5 5 #v :V 845 / /  / /  / /  /  / / 44 8f4[$$If!vh5 5 5 5 #v :V 84 (5 / /  / / / / / /  /  / 44 8f4p(1$$If!vh5 5 5 5 #v :V 84 (5 /  / / / / / /  44 8f4p($$If!vh5 5 5 5 #v :V 845 /  /  / / / / /  /  44 8f4$$If!vh5 5 5 5 #v :V 845 / /  / /  / /  /  / / 44 8f4[$$If!vh5 5 5 5 #v :V 84 (5 / /  / / / / / /  /  / 44 8f4p(1$$If!vh5 5 5 5 #v :V 84 (5 /  / / / / / /  44 8f4p($$If!vh5 5 5 5 #v :V 845 /  /  / / / / /  /  44 8f4$$If!vh5 5 5 5 #v :V 845 / /  / /  / /  /  / / 44 8f4[$$If!vh5 5 5 5 #v :V 84 (5 / /  / / / / / /  /  / 44 8f4p(1$$If!vh5 5 5 5 #v :V 84 (5 /  / / / / / /  44 8f4p($$If!vh5 5 5 5 #v :V 845 /  /  / / / / /  /  44 8f4X@X Normal$$dh5$7$8$9DH$CJ_HmH sH tH H@H Heading 1 x<@&5;OJQJ>@> Heading 2 <<@&5B@B Heading 3 <@&5CJDA@D Default Paragraph FontVi@V  Table Normal :V 44 la (k@(No List 8+@8  Endnote TextCJ@&@ Footnote ReferenceH*>*@> Endnote ReferenceH*4 @"4 Footer  !4@24 Header  !6"@6 Caption xx5@>@@ Title$<a$ 5OJQJ<O< Abstract]^$18 T+W̆kylk̆9DQghNO\]j    #_?@[\z{Z[em$"%"-"~##k%l%t%E'F'N'P(Q(()C)))*********+ +++*+4+?+@+F+H+J+L+N+P+R+T+V+X+Z+\+^+_+a+c+d+f+g+i+j+l+m+o+q+s+u+v+x+y+z+{+|+}+~++++++++++++++++++++++++++++%,4--/.0.//1111111111111222 2222$2%2)2+22282?2E2F2J2L2S2Y2`2f2g2k2m2t2z2222222222222255(5883:4:::============> >>>> >">$>&>)>+>.>/>3>5>7>9>;>=>?>@>D>F>H>J>M>O>Q>R>V>X>Z>\>_>a>c>d>h>j>l>n>p>r>t>u>y>{>}>>>>>>>>>>>>>>>>>>>>>>>>>r@s@+BVDWDDtFuFHHtMuMvM~MMMMMMMMMMMMMMMMMMMM"N#N$N%N7NGNZN[NkNqNrNsNtNNNNNNNNNNNNOOOO'O7OJOKO[ObOcOdOeOxO~OOOOOOOOOOOOOP P3P4PDPKPLPMPNPaPgPmPnPoPPPPPPPPPRRR0V1VEVZZZ[[^^5b6bGb8c9c:c;c=c>cIcJcccddeeze{eeepfqfffFgGggg>h?hhh+i,iii j!jjjjj(k)kwkxkkkkkk4l|llll m(mJmmmmm/n0npnqnnnnn7o8ozo{oooppBpCppppp q qMqNqqqqqr rcrdrrrrr'spsssttYtntotttt"uDuuuuuvvWvXvvvvvwwbwcwwwww)x*xixjxxxxx1y2ytyuyyyyy>>MMMNNNOOOPPP̆       !B_Ref3178681762͆2͆(IPQY%-DG , 1 ,57Bmp(-TYTY$)v{+1NV["b"###$$"$b$j$%!%L&T&&&''''(("(*(?(G(i(q(**++,,R,c,,,,,"-)-----./D/O///0 000|11,444W4_45 555B5G5e5m5n6s666o7y77777888899I:S:;';;;.<8<<<<< ==7=?=====>>??X?`???h@p@@@AABB^ChCEEzFFGGGG&H7HBHJHHHaJiJKKPLXL\LaLMMUM]MMMMMMM%N+N7N=NGNNN[NaNtN{NNNOO'O-O7O>OKOQOeOlOOOOPPP P'P4P:PNPUPoPvPR"RRR S%SVW=YDYYYZZb[g[,]4]c c#c+cTcWc]ceccc'd8dddddddeee"e*f;fqfzfgg,i3iDiMiSj[jk knnnn/o6o[oeoooOpWpppppppq qZqeqqqqqqrrr,r3rprvrrrrr ssuuuu7v;vdvkvvvvw@wHwww xx#x(xMxUxvx|xxxxx)y0yyy5z;zwzzzz|}(}1}m}t}$~+~;~C~~~~~r} "*9@dp .4q{́Ձ$.//11224578:;FGՂڂ>C͆9c9c:c:c;c;c.//11224578:;D͆  6,)vL2p2_>Zn,8oAYU*******+ +++*+4+?+@+F+H+J+L+N+P+R+T+V+X+Z+\+^+_+a+c+d+f+g+i+j+l+m+o+q+s+u+v+x+y+z+{+|+}+~++++++++++++++++++++++11111111111222 2222$2%2)2+22282?2E2F2J2L2S2Y2`2f2g2k2m2t2z22222222222==========> >>>> >">$>&>)>+>.>/>3>5>7>9>;>=>?>@>D>F>H>J>M>O>Q>R>V>X>Z>\>_>a>c>d>h>j>l>n>p>r>t>u>y>{>}>>>>>>>>>>>>>>>>>>>>>>uMvM~MMMMMMMMMMMMMMMMMMM$N%N7NGNZN[NkNqNrNsNtNNNNNNNNNNNOO'O7OJOKO[ObOcOdOeOxO~OOOOOOOOOOOP P3P4PDPKPLPMPNPaPgPmPnPoPPPPPP/147:FGʆ͆Q*@HP LaserJet 1200 Series PCLNe00:HP LaserJet 1200 Series PCLHP LaserJet 1200 Series PCLHP LaserJet 1200 Series PCL4C 4dXXA4DINU"4J4HP LaserJet 1200 Series PCL4C 4dXXA4DINU"4J4l\̆`@UnknownG: Times New Roman5Symbol3& : Arial9Garamond?5 : Courier New"V zŦŦE&nnBnnB!V xx42V HX ?Zn2Y/N tests and rasch analysis John ShillawUniversity of Wales SwanseaOh+'0 8D d p |  Y/N tests and rasch analysisJohn ShillawNormalUniversity of Wales Swansea4Microsoft Office Word@dN@L@\B3@d^h9nn՜.+,D՜.+,\ hp  Unknown OrganizationB' Y/N tests and rasch analysis Title@@4_AdHocReviewCycleID_EmailSubject _AuthorEmail_AuthorEmailDisplayName_ReviewingToolsShownOnce&fqyour missing article!shillaw@nanzan-u.ac.jpJohn Shillaw  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root Entry FW|{9Data E1TablezWordDocumentoSummaryInformation(DocumentSummaryInformation8CompObjq  FMicrosoft Office Word Document MSWordDocWord.Document.89q