TECHNICAL REPORT TPRI PROGRESS MONITORING FOR BEGINNING READERS (2006-2008 EDITION) Center for Academic and Reading Skills University of Texas-Houston Health Science Center And Texas Institute for Measurement, Evaluation, and Statistics University of Houston
2 TABLE OF CONTENTS LIST OF TABLES... 3 INTRODUCTION... 4 Overview of TPRI Progress Monitoring For Beginning Readers Kit...4 Components of the TPRI PMBR Kit...5 TPRI -PMBR Development Study...5 RELIABILITY OF TPRI PMBR... 7 VALIDITY OF TPRI PMBR... 8 SUMMARY RELIABILITY AND VALIDITY... 9 Equipercentile equating... 10
3 LIST OF TABLES Table 1. Demographic information for the TPRI -PMBR study... 12 Table 2. Internal consistency reliability* of the TPRI -PMBR forms... 1Error! Bookmark not defined. Table 3. Correlation of Word List with Test of Word Reading Efficiency...14 Table 4. Correlations of Story Fluency with Test of Word Reading Efficiency...15 Table 5. Equipercentile Equating Chart Grade 1... 16 Table 6. Equipercentile Equating Chart Grade 2... 18 Table 7. Equipercentile Equating Chart Grade 3... 22
4 INTRODUCTION The TPRI Progress Monitoring for Beginning Readers is designed to help teachers measure first, second, and third grade students progress in reading fluency. Proficient reading entails consolidation of the alphabetic principle and extraction of meaning at the word level. Fluency, or the ability to read connected text quickly and accurately, is an important component of proficiency. Fluent readers recognize words in text quickly and automatically, allowing them to focus attention on the meaning of the passage and on reading with expression. Research on poor readers in third grade and above shows that intensive intervention can remediate accuracy, but that fluency gains are more difficult to achieve (Torgesen et al., 2001; Torgesen, 2002). To develop fluency, children must be engaged in reading and need a strong sight word vocabulary. This vocabulary develops partly as a result of repeated exposures to words. If a child is delayed in access to print, or does not spend enough time reading, fluent reading skills may not emerge. Thus, it is important to prevent reading difficulties from occurring with effective classroom instruction and early intervention. By identifying students for whom fluency is an issue, and monitoring fluency as a marker for progress in reading development, early intervention is facilitated. The TPRI Fluency Kit is an individually administered oral reading fluency assessment that provides teachers with useful information to ensure that students receive targeted reading instruction that promotes fluency and ensures that progress can be frequently assessed. Overview of TPRI Progress Monitoring For Beginning Readers Kit The TPRI Progress Monitoring for Beginning Readers (PMBR) Kit is a research-based assessment tool designed to monitor student progress in Reading Fluency. An accurate and reliable source of information about student progress, the PMBR Kit is designed to monitor three
5 core aspects of student reading in grades 1, 2, and 3: 1) Reading Accuracy, 2) Reading Fluency, and 3) Reading Comprehension. Included in the fluency assessment is a timed Word List for Story Placement that assesses word-reading fluency and also places students into a Story at their Instructional or Independent Level. Also, reading accuracy and fluency of connected text are assessed by having students read short and engaging passages. Further, comprehension is assessed through the use of explicit and implicit comprehension questions. Questions of this nature enable a teacher to monitor students understanding of word meanings, ability to extract meaning from groups of words (e.g., clauses, sentences, and paragraphs), and ability to draw inferences. Components of the TPRI PMBR Kit In order to achieve these assessment goals, the TPRI PMBR Kit has three main components: 1. Word List for Story Placement: A list of 50 words selected for high accuracy and comparable difficulty to word-reading fluency. This list is identical for all grades and is used in the 6-week schedule only. 2. Reading Accuracy and Fluency Stories: Short passages written to conform to specific word-level features and story structure based on what is taught at different points at each grade level. 3. Reading Comprehension: Explicit and implicit questions designed to assess how well students understand the meaning of the text. For the 2-week schedule, two generic questions are provided as a tool to keep the focus on reading comprehension. TPRI -PMBR Development Study During the Spring of 2003, the Center for Academic and Reading Skills (CARS) at the University of Texas Health Science Center Houston and the Texas Institute for Measurement,
6 Evaluation, and Statistics (TIMES) at the University of Houston conducted a study with 1785 students from grades 1 to 3 to develop forms for regular progress monitoring of young readers fluency. Study examiners visited 96 classrooms at 9 schools around Houston Independent School District. The sample was roughly comparable in gender representation (48% female) and ethnically diverse across three grades of study. Table 1 shows the ethnicities represented in this study delineated by gender and also by grade. Students were asked to complete: (a) Sight Word Efficiency Subtest A from the Test of Word Reading Efficiency (Torgeson, Wagner, & Rashotte, 1999); (b) a timed sightword list developed by CARS and TIMES; (c) 2 randomly assigned stories; and (d) eight Reading Comprehension questions for each of the stories. Stories were randomly assigned within grade levels. Students unable to read the text were read the story by examiners and comprehension was administered as Listening Comprehension. On seven of the stories, five comprehension questions were administered. Reliability and validity information is presented on the 24 stories that comprise the TPRI -PMBR. Stories with 5 comprehension items are noted. After conducting the development study, the centers revised the forms and administration procedures of the TPRI -PMBR to facilitate effective classroom use. To this end, procedures for the two schedules (2-week and 6-week) described above were created. For more frequent progress monitoring (i.e., every 2 weeks), some elements of the assessment, such as the Word List for Story Placement, were eliminated in favor of a simpler and time-saving selection of passages. Also for more frequent progress monitoring less focus is placed upon reading comprehension. However, reading for meaning is still emphasized by asking students two generic comprehension questions, such as What is this story mainly about, although answers are not recorded. Thus, the current edition of the TPRI -PMBR uses the following procedural
7 heuristics: (1) every 2 weeks, students read the next story for 2 minutes and answer 2 questions, or (2) every 6 weeks, students place into a story using the word list, read the story, and answer 6 questions. Also, to facilitate meaningful comparisons across forms, an equipercentile equating procedure has been employed to develop a fluency conversion table. Equipercentile equating accounts for differences between stories when considering fluency rates and gives educators a relative assessment of fluency. The rationale and results of the equating are presented below. RELIABILITY OF TPRI PMBR There are three grade levels the in TPRI PMBR: Grade 1, Grade 2, and Grade 3 with eight stories within each grade. For each story eight (8) Reading Comprehension items were evaluated. However, on seven (7) of the stories, data were only collected on 5 comprehension items and have been noted in the tables. Additionally there is a 50-item Word List for Story Placement; the same items are used for all grades. Thus, there are 25 different forms (3 grades x 8 stories + 1 word list) evaluated for reliability. The teacher s manual provides explanations for when to administer each of these forms and the rationale underlying their development. For each form, a Kuder-Richardson 20 (KR-20) coefficient was computed. For dichotomously scored items, the KR-20 is analogous to Cronbach s alpha. The alpha coefficient ranges from 0 to 1.0 and was reported as the index of internal consistency. High alpha coefficients indicate that item variability is small relative to total test variability, or that all items perform similarly, and were all measuring the same construct. We evaluated the practical significance of the reliability coefficients as follows: Poor (0-.39), Adequate (.40-.59), Good (.60-.79) and Excellent (.80-1.0). These estimates of practical significance are arbitrary, but conventional, and provide a useful heuristic for interpreting the reliability data. Since there are
8 only 8 items for the comprehension component and somewhat small samples for individual stories, we expected a varied range of coefficients across the 24 story forms. Table 2 shows the reliabilities for the Word List for Story Placement and the 24 story forms. The Word List for Story Placement has excellent internal consistency reliability, α =.85. As expected, the 24 story forms range from α =.25 to α =.77. Using the conventions listed above, the majority (16 or 67%) of the stories show Adequate reliability, 5 stories show Good reliability, and 3 stories show Poor reliability. VALIDITY OF TPRI PMBR The TPRI -PMBR Word List for Story Placement from empirical work across several prior studies conducted by CARS and TIMES. Item Response Theory (IRT) was used to establish an item difficulty parameter for each item. Items on this list were chosen for their low difficulty levels for beginning readers. By parsing item-difficulty from student ability the word list is able to disentangle rapid word recognition from decoding. Stories in the TPRI - PMBR are linked to the Word List for Story Placement using multiple regression models. The upper quartile (i.e., 75th percentile) for student s fluency (words correct per minute, or WCPM) was identified for each story. Stories that had lower 75th percentile scores indicated a more difficult story because fewer WCPM indicates that students read such passages slower than those with higher WCPM scores. For each story a prediction model was run with fluency (WCPM) rates from the Word List for Story Placement predicted by fluency (WCPM) rates from each story. From these prediction models, a WCPM value from the Word List for Story Placement could be identified that corresponds to the 75th percentile score for each passage. Thus, for each story there is a value that represents the 75th percentile score and a corresponding value on the Word List for Story Placement. Subsequently, ranges of WCPM rates were created such that
9 each range contained the cutoff for a specific story. Thus, a student s WCPM on the Word List for Story Placement can be linked to 75th percentile performance on the story indicated on the placement criteria. Performance on these items is then used to place students into stories of appropriate difficulty. Correlations between the Word List for Story Placement and the Sight Word Efficiency Subtest A from the Test of Word Reading Efficiency (Torgeson, Wagner, & Rashotte, 1999) for the three different grade levels are presented in Table 3. At all three grade levels the Word List for Story Placement correlates highly (>.80) with the Test of Word Reading Efficiency. Fluency estimates from each of the 24 stories in the PMBR were also correlated with the TOWRE Sight Word Efficiency Subtest A and are presented in Table 4. All correlations are significantly greater than zero at p <.01 and range from.43 to.90. Twenty of the 24 stories (83%) were correlated greater than.70, three (13%) between.50 and.70, and one story correlated.43 with TOWRE. SUMMARY RELIABILITY AND VALIDITY To summarize, the 25 forms of the TPRI -PMBR show varied internal consistency reliability and excellent validity. The majority of forms (22/25 or 88%) show Adequate or Good reliability using the heuristics above. Similarly, all but 2 of the forms show excellent validity, correlating with the TOWRE around r =.70 or above. While these demonstrate minimally adequate properties, refinement is necessary. Further developmental work is being undertaken to improve the reliability, validity, and usability of future editions of the TPRI - PMBR.
10 EQUIPERCENTILE EQUATING In order to equate fluency scores (WCPM) across the different stories, we used an equipercentile equating method rather than linear equating. Linear equating would adjust scores for differences across forms in the mean and standard deviation, such that the resultant scale scores for different forms will have the same mean and standard deviation. However, linear equating assumes that the same adjustment should be made throughout the distribution of WCPM scores in order to equate fluency scores taken from different stories. In contrast, equipercentile equating is more general in that it allows that the relation between the two forms may change across the range of scores on the base form. For example, it could be that Chirp is more difficult over one score range than the passage to be equated, but easier over another score range. Alternatively, it could be that the difference between forms is not consistent throughout the range of WCPM scores such that one adjustment is needed for the lowest scores and another adjustment is needed for scores near the center, and still another adjustment for scores at or near the top. Equipercentile equating can take these differences into account. To calculate the equiperentile rank for a target score, the cumulative percentage of the scores below the target score is added to one half the percentage of cases who obtained the target score. For example, if 70% of test takers achieve a score 19 or less on a given test and 5% of individuals earn a score of 20, then the equipercentile rank of 20 is 72.5. The reason that the equipercentile ranking is not equal to the cumulative percent at a given score draws from psychological measurement theory (Blommers & Forsyth, 1977). Through the conversion of raw scores to equipercentile ranks, students scores on nonequivalent forms can be compared. However, when comparing scores across a large number of tests it may be more useful to determine the raw scores that correspond to specific integer
11 percentiles. These conversions utilize a similar rationale to the raw score conversions to equipercentile ranks described above, but determine the WCPM score on a given form that is associated with specific percentiles. This latter approach facilitates the creation of conversion tables that can handle multiple forms at the same time. Tables 5-7 provide equipercentile conversions for the eight fluency stories per grade back to a common scale score where the scale score gives the percentile rank for the student. That is, the columns of Tables 5-7 give scores for each of the eight stories per grade associated with a particular integer percentile rank. To use the table, a teacher would simply locate the title of the story that the child had been asked to read. The teacher would then read down the column until she/he found the WCPM score for the specific child or, more commonly, found the two rows that bracket the student s score. The teacher would then read across the rows to the column marked percentile and find the entry in the percentile column associated with the row (or set of rows) for the student s score. To get the most precise score, linear interpolation could be used. However, teachers could also round up or down with only slight loss of accuracy.
12 Table 1. Demographic information for the TPRI -PMBR study. Grade Gender Ethnicity Unreported African- American Caucasian Hispanic Other Total Unreported 11 0 0 0 1 12 Grade 1 Female 2 90 136 41 37 306 Male 2 101 138 45 43 329 Total Grade 1 15 191 274 86 81 647 Unreported 2 0 0 0 0 2 Grade 2 Female 2 94 138 44 28 306 Male 1 74 148 52 25 300 Total Grade 2 5 168 286 96 53 608 Unreported 4 0 1 1 0 6 Grade 3 Female 0 75 101 48 21 245 Male 0 95 120 46 18 279 Total Grade 3 4 170 222 95 39 530 Total 24 529 782 277 173 1785
13 Table 2. Internal consistency reliability* of the TPRI -PMBR forms. Alpha N Word List for Story Placement (50 items).85 1448 Reading Comprehension by Grade by Story Grade 1 1 My Cat Sam (8 items).57 94 Grade 1 2 Bud Gets the Ball (8 items).45 99 Grade 1 3 Fun in the Sand (8 items).56 98 Grade 1 4 Running (8 items).70 93 Grade 1 5 Big Dog (8 items).53 83 Grade 1 6 Playing House (5 items).59 86 Grade 1 7 The Goat (8 items).66 74 Grade 1 8 Chirp (5 items).33 82 Grade 2 1 Surprise (8 items).25 91 Grade 2 2 How to Make a Telephone(8 items).51 90 Grade 2 3 Bird s Nest (5 items).52 96 Grade 2 4 Bucking Horse (5 items).55 100 Grade 2 5 Are Bats Birds? (8 items).57 95 Grade 2 6 Tree Homes (5 items).36 98 Grade 2 7 Fire Trucks (8 items).77 91 Grade 2 8 Little Hawk (5 items).58 94 Grade 3 1 Animals (8 items).67 74 Grade 3 2 Michelangelo (5 items).41 72 Grade 3 3 Frogs and Toads (8 items).55 79 Grade 3 4 Catching Grasshoppers (8 items).41 94 Grade 3 5 Lakes and Streams (8 items).54 94 Grade 3 6 Trees (8 items).59 73 Grade 3 7 Swimming Pool (8 items).75 82 Grade 3 8 Robins (8 items).48 84 * Reliabilities reported are Kuder-Richardson 20 (KR20) coefficients for internal consistency of dichotomous items, analogous to coefficient alpha. Additional questions have been added and are currently under evaluation.
14 Table 3. Correlation of Word List with Test of Word Reading Efficiency. Word List for Story Placement & Grade TOWRE Sight Word Efficiency Form A All Grades.90 1.93 2.82 3.85 All correlations are significant at p <.01.
15 Table 4. Correlations of Story Fluency with Test of Word Reading Efficiency. Story Word List for Story Placement & TOWRE Sight Word Efficiency Form A Grade 1 1 My Cat Sam.90 Grade 1 2 Bud Gets the Ball.85 Grade 1 3 Fun in the Sand.87 Grade 1 4 Running.82 Grade 1 5 Big Dog.59 Grade 1 6 Playing House.74 Grade 1 7 The Goat.79 Grade 1 8 Chirp.77 Grade 2 1 Surprise.69 Grade 2 2 How to Make a Telephone.69 Grade 2 3 Bird s Nest.80 Grade 2 4 Bucking Horse.76 Grade 2 5 Are Bats Birds?.78 Grade 2 6 Tree Homes.79 Grade 2 7 Fire Trucks.80 Grade 2 8 Little Hawk.75 Grade 3 1 Animals.73 Grade 3 2 Michelangelo.77 Grade 3 3 Frogs and Toads.82 Grade 3 4 Catching Grasshoppers.75 Grade 3 5 Lakes and Streams.43 Grade 3 6 Trees.84 Grade 3 7 Swimming Pool.72 Grade 3 8 Robins.78 All correlations are significant at p <.01.
Table 5. Equipercentile Equating Grade 1
Table 5. Equipercentile Equating Grade 1 Cont. 17
Table 6. Equipercentile Equating Grade 2 18
Table 6. Equipercentile Equating Grade 2 19
Table 6. Equipercentile Equating Grade 2 20
Table 6. Equipercentile Equating Grade 2 21
Table 7. Equipercentile Equating Grade 3 22
Table 7. Equipercentile Equating Grade 3 23
Table 7. Equipercentile Equating Grade 3 24
Table 7. Equipercentile Equating Grade 3 25