Upload
roger
View
214
Download
2
Embed Size (px)
Citation preview
This article was downloaded by: [UQ Library]On: 01 November 2014, At: 04:36Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number:1072954 Registered office: Mortimer House, 37-41 Mortimer Street,London W1T 3JH, UK
Reading PsychologyPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/urpy20
A COMPARISON OF FOURMETHODS OF DETERMININGINSTRUCTIONAL READINGLEVELSMichael A. Tulley a & Roger Farr ba Indiana University‐Kokomob Indiana UniversityPublished online: 10 Jul 2006.
To cite this article: Michael A. Tulley & Roger Farr (1987) A COMPARISON OFFOUR METHODS OF DETERMINING INSTRUCTIONAL READING LEVELS, ReadingPsychology, 8:3, 149-168, DOI: 10.1080/0270271870080302
To link to this article: http://dx.doi.org/10.1080/0270271870080302
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of allthe information (the “Content”) contained in the publications on ourplatform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy,completeness, or suitability for any purpose of the Content. Anyopinions and views expressed in this publication are the opinions andviews of the authors, and are not the views of or endorsed by Taylor& Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information.Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly inconnection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private studypurposes. Any substantial or systematic reproduction, redistribution,reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of accessand use can be found at http://www.tandfonline.com/page/terms-and-conditions
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
A COMPARISON OF FOUR METHODSOF DETERMINING INSTRUCTIONALREADING LEVELS
MICHAEL A. TULLEY, Indiana University-Kokomo
ROGER FARR, Indiana University
The purpose of this study was to examine the validity of estimates of
instructional reading level (IRL) produced by the Reading
Comprehension Test of the 1986 Metropolitan Achievement Tests
(MAT). Estimates of IRL provided by four methods were compared:
the MAT, basal reader placement tests, an informal reading inventory
derived from MAT reading comprehension test passages, and
classroom teachers' judgments.
Subjects tested were grade 2 (n=100), 4 (n=98) and 6 (n=91) students
from three elementary schools within the Kokomo, Indiana public
school district. Each subject was administered the MAT and a
placement test designed to accompany the basal reading series in use at
each grade level. Estimates of each student's IRL were also inventories
that were individually administered to 25 randomly selected students at
each grade level.
Three analyses of data were conducted: comparisons of percentages of
agreement among estimates of IRL, comparisons of the mean of each
estimate and of a composite IRL estimate, and comparisons of MAT
estimates with the average estimate produced by each of the other
Reading Psychology: An International Quarterly 8:149-168, 1987 149Copyright © 1987 by Hemisphere Publishing Corporation
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
150 M. A. TULLEY AND R. FARR
measures. Results showed a high percentage of agreement, at each
grade level, between the MAT estimates and those provided by each
study of similar design, in which estimates produced by the 1978
edition of the MAT were compared to estimates produced by other
commonly employed methods.
Based upon these results, the authors concluded that the 1986 edition of
the Metropolitan Achievement Tests provides for classroom
instructional reading levels, which can be used as an acceptable
alternative to traditional methods of determining IRL.
The purpose of this study was to compare estimates of instructional
reading levels (IRL) produced by four methods: the Reading
Comprehension Test of the 1986 edition of the Metropolitan
Achievement Tests (MAT), basal reading placement tests, informal
reading inventories, and classroom teachers' judgments.
Students' reading levels are often described by reading teachers and
specialists in terms of three levels: independent, instructional, and
frustration reading levels (Farr and Roser, 1970; Lapp and Flood,
1986). Generally, the instructional level is considered the level of "best
fit", or the level at which a student is most likely to benefit from
instruction. Reading material at this level is sufficiently difficult to
challenge students and yet require teacher assistance, but easy enough
to allow reading to occur. The independent level is considered the
highest level afwhich a student can read without assistance, and is
usually characterized by smooth, efficient reading. The frustration
level, on the other hand, is the level at which the material encountered
is too difficult, and is characterized by a drop in reading fluency and
understanding.
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 151
Estimates of students' reading levels are usually determined by the
extent to which students approach criteria of performance in word
recognition and comprehension. Although slight differences in
recommended criteria exist, the instructional level is usually
considered that level at which a student recognizes or correctly
identifies approximately 95-99% of the words, and also comprehends
approximately 70-90% of the material encountered in a particular
passage. A student's independent reading level is considered that level
at which a student recognizes approximately 99% or more of the
words, and comprehends 90% or more of the selection. The
frustration level is considered that level at which a student recognizes
90% or fewer of the words, and comprehends approximately 70% or
less (Alexander, 1983; Rupley and Blair, 1983; Ekwall and Shanker,
1985).
The importance of proper instructional placement cannot be
minimized; there is, perhaps, no single piece of information about
students' academic abilities as important as an accurate estimate of
instructional reading level. Unfortunately, none of the traditionally
available methods of determining IRL is highly precise, nor can it be
assumed that any method is more valid or accurate than any other.
Traditionally, only three methods have been used to determine
instructional reading level: informal reading inventories (IRIs), cloze
tests, and placement tests. Each of these methods is subject to some type
of limitation, however. An IRI, for example, while usually considered
the most valid of these methods, can be time-consuming to administer,
and often requires subjective analyses of student reading behavior.
Cloze passages can be easy to construct and administer, but their use
with primary age students has been seriously questioned. And, the
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
152 M. A. TULLEY AND R. FARR
requirement of an exact match between student responses and deleted
words has created a debate about the extent to which such tests truly
measure student understanding of material. Basal reader placement
tests are often more accurately described as either graded word lists,
modified IRIs, or some combination of the two. Although they are
usually relatively easy to administer (some are even group tests), their
primary purpose is to enable teachers to place students at an
appropriate level within a specific basal series. Thus, an IRL
determined using this type of instrument may have minimal
transferability to other materials encountered by students.
A fourth method, which has emerged only relatively recently, is the use
of standardized tests to determine instructional reading level. The
desirablity of the use of a standardized test is that such an instrument
may alleviate some of the limitations associated with the traditional
methods, such as reducing the time of administration, and achieving
greater objectivity and consistency in scoring. One of the few
standardized tests which yields IRLs is the Reading Comprehension
Test of the Metropolitan Achievement Tests (MAT). Studies which
have examined the validity of estimates produced by the 1978 edition of
the MAT have found this method of determining IRL to be generally
valid and accurate. There have been no similar examinations of the
1986 MAT, however, and this present study was initiated to investigate
the validity of estimates produced by this edition of the test.
The Metropolitan Achievement Tests
The grade levels of MAT reading passages generally match the
difficulty of the corresponding grade levels of most basal reading
series. By establishing criteria levels against which student
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 1 53
performance on reading comprehension test items is compared, the
MAT can yield a criterion-referenced score expressed as an IRL.
Detailed descriptions of the procedures used to control passage
difficulty and scoring criteria can be found in Content Validity of the
Reading Tests. Metropolitan Special Report No. 5 (1980), and in
summarized form in Smith and Beck (1980).
In determining instructional reading levels, the MAT employs criteria
similar to those used with informal reading inventories, cloze tests, and
basal reader placement tests. The cutoff for the instructional level is
that level at which a student is able to answer correctly a minimum of
60% of the questions pertaining to a particular passage, while
simultaneously answering 100% of the questions correctly on passages
below that level, and 0% correctly on passages above that level.
Because it is unrealistic to expect perfect performance, scoring
procedures allow for adjustments of less than 100% accuracy below
and more than 0% accuracy above the criterion score used to
determine the instructional reading level.
For the 1986 edition of the MAT, the level of difficulty of passages was
controlled using essentially the same procedures employed during
construction of the 1978 edition. Briefly, control of passage difficulty
was achieved in the following manner. The reading levels of passages
intended for kindergarten through grade 3 were controlled by
adhering to the Harris and Jacobsen (1972) formula for vocabulary
control. Passages at these levels were written so that at least 25% of the
vocabulary in each appeared on the Harris and Jacobsen list for that
intended level. Also, passages contained no more than 5% of their
vocabulary above the level for which they were intended, and no words
more than 2 grade levels above that of the passage.
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
1 54 M. A. TULLEY AND R. FARR
Because adherence to these criteria created unnatural and cumbersome
discourse within passages intended for grades 4 through 6, criteria for
these levels were altered so that the total number of words at the
primer level were subtracted from the total number of words in the
passage, and only the non-primer total was considered in determining
the number of words at each level that appeared on the Harris and
Jacobsen list. A minimum of 30% of the vocabulary within each
passage appeared on the list for that intended grade level. The reading
levels of passages intended for grades 7 and above were controlled
through the use of the Dale-Chall formula, because the Harris and
Jacobsen list does not extend beyond grade 6. Emphasis at these levels
was placed upon creating passages of gradually increasing difficulty.
That is, passages at each grade level were clearly more difficult than
passages at preceding grade levels.
Previous Comparative Studies of IRL Estimates
At least three studies have been conducted in which various approaches
to determining instructional reading level have been compared. In
each, estimates produced by the MAT were included, as well as those
produced by the more traditional methods. The results of each study
show strong relationships among scores obtained from the various
methods. Smith and Beck (1980) conducted a study which involved
over 700 subjects from grades 1-4 and 6. Subjects were administered
the Reading Comprehension Test of the 1978 MAT, one of three
commonly available basal reader placement tests, and cloze tests
derived from MAT reading test passages. Additionally, randomly
selected subjects at each grade level were administered a
"standardized" IRL Results suggested a high degree of agreement
among the estimates produced by these four methods of determining an
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 155
IRL, and also showed that the IRL estimates of the MAT were closer to
both the cloze and the basal placement scores, than those scores were to
each other.
Bristow, Pikulski and Pelosi (1983) compared five methods of
estimating instructional reading level. Using a randomly selected
population of 24 pupils at each of three grade levels (2,4, and 6), these
researchers compared actual grade level placement in a basal series
with estimates of IRL derived from a commercially produced informal
reading inventory, a teacher-made IRI constructed from the basal
reader in use in the schools that participated in the study, the reading
subtest of the Wide Range Achievement Test (1978), and the Reading
Comprehension Test of the 1978 MAT. Results showed that except for
the WRAT, all estimates of instructional reading level were highly
correlated with each other for the entire population. Additionally, for
all subjects the MAT estimates were within one grade level of actual
basal level placement 89% of the time, and within one grade level of
the estimates provided by the teacher-made and the commercially
produced IRIs, 75% and 89% of the time, respectively.
Fair and Beck (1984) describe a study in which IRL estimates produced
by the 1978 MAT were compared with estimates determined using an
informal reading inventory derived from MAT Reading
Comprehension Test passages. Subjects were second and fourth grade
students (n=75 and 64, respectively). All subjects were administered
the MAT Reading Comprehension Test, and fifteen randomly selected
subjects at each grade level were administered the IRL The results,
although limited because of the relatively small number of subjects
selected from a single setting, nonetheless indicated agreement between
these two estimates. For approximately 50% of the subjects, scores
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
156 M. A. TULLEY AND Ft. FARR
from both estimates were at the same level, and another 45% differed
by only one grade level. The MAT was also found to yield a more
conservative estimate of pupils' IRLs. That is, MAT scores tended to
be lower in most cases than those provided by either basal reader
placement tests, or informal reading inventories.
Comparing Estimates of Instructional Reading Level
Grade 2,4, and 6 students were selected from three elementary schools
within the Kokomo, Indiana public school district. Kokomo is in many
respects a typical midwestern community. Located in the chiefly
agricultural north central portioiv of the state, this medium-sized city
has a predominantly white, middle class population of approximately
35,000, and a public school enrollment of just over 8,800 students. The
three participating schools were representative of the 11 elementary
schools within this district, in terms of student enrollment,
administrative structure, and socio-economic status.
Four estimates of students' instructional reading levels were obtained,
with data collection occurring in the same order for all subjects. The
order of data collection was: 1) administration of the MAT in a group
setting, 2) administration of basal reader placement tests (either group
or individual), 3) administration of an IRI constructed from the MAT
(administered to 25 randomly selected students at each grade level),
and 4) completion of a checklist in which teacher estimates of
instructional reading level were listed. Between 3 and 4 weeks were
required to complete the testing at each grade level. Tests were
administered by the authors, who were assisted by two Indiana
University students specifically trained to administer the tests utilized
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 157
during this study. Testing of students was conducted during the
morning, which coincided with that portion of the regular school day
typically devoted to reading instruction. Students were removed from
their regular classrooms and taken to a separate room for all individual
testing. Data collection occurred during the period from January 15 -
April 1, 1986.
Each of the four methods of determining estimates of instructional
reading level used in this study is described below.
Metropolitan Achievement Test. Students were administered the
Reading Comprehension Test of the 1986 MAT (Form L), in a group
setting, according to the directions in the test manual. Grade 2 subjects
were administered the Primary 2 level, which consists of 11 readings
passages ranging in difficulty from the primer level to grade 5. The 55
multiple choice items which accompany these passages measure student
ability to recognize detail and sequence, to infer meaning, cause and
effect, main idea, character analysis, and to draw conclusions.
Grade 4 subjects were administered the Elementary level, which
consists of 11 reading passages ranging in difficulty from grade 2
through grade 6. The 60 multiple choice items which accompany these
passages measure those same comprehension skills listed above. Grade
6 subjects were administered the Intermediate level, which consists of
11 reading passages ranging in difficulty from grade 3 through grades
9-10. The 60 multiple choice items which accompany these passages
measure, in addition to those comprehension skills listed above, student
ability to determine author's purpose and to distinguish fact from
opinion. Time allowed for completion was 40 minutes at each level.
Scoring service was provided by The Psychological Corporation,
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
158 M. A. TULLEY AND R. FARR
publisher of the MAT.
Basal reader placement test. Approximately one week after
administration of the MAT, students were administered the basal
reader placement test which accompanied whichever basal reading
series was in use at each grade level. Grade 2 and 4 classrooms in this
district were using the 1983 Houghton Mifflin basal reading series.
The placement instrument for the series is an individually administered
Informal Reading Inventory consisting of a series of graded word lists
which students read aloud, and graded passages which students read
silently. Comprehension is assessed through post-reading questions
asked by the examiner.
One modification in the intended use of this placement test occurred.
This instrument is primarily a means of determining if a given level of
the Houghton Mifflin series is appropriate for initial instructional
placement, and was not designed to yield independent, instructional,
and frustrational reading levels. For the purpose of this study,
however, a simple "yes/no" regarding the appropriateness of a given
level was insufficient. Therefore, students were asked to continue
reading passages until instructional and frustration levels could be
determined. Each inventory required between 10 and 20 minutes to
administer.
Grade 6 classrooms in this district were using the 1983 Macmillan basal
reading series. The Initial Placement Inventory used to place students
in that series, is comprised of a group administered silent reading
inventory consisting of graded reading paragraphs followed by written
multiple choice questions, and an individually administered oral
reading inventory consisting of graded paragraphs followed by
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 1 59
questions asked orally. The authors of these tests recommend that both
parts of the placement instrument be administered at separate settings,
but they also indicate that either the silent or the oral reading inventory
is sufficient to determine appropriate placement. Although it is
possible to record and to analyze oral reading miscues, for both the
silent and the oral reading inventories only the number of correct
responses to comprehension questions is used to determine reading
levels. Researchers could discern little substantive difference between
these two alternatives, other than format, and opted to administer only
the group silent reading inventory. Approximately 20 minutes was
required for students to complete the inventory.
MAT informal reading inventories. An IRI constructed from the
passages contained in each level of the MAT Reading Comprehension
Test was administered to 25 randomly selected students at each grade
level. IRIs were constructed by selecting one passage from each of the
various reading levels represented on each level of the test. Passages
were ordered from the lowest to the most difficult reading level, which
is the same order in which they appear on the test. Students read
passages orally, and miscues were noted by examiners. Questions asked
at the completion of the oral reading were derived from those same
questions that accompanied the MAT passage. Because IRI questions
were open-ended, rather than multiple choice, some modification in a
few of the items was necessary. In each instance, however, questions
were restated to correspond to the MAT question, and whenever
possible exact wording or phrasing from MAT questions was used as
the stem of the IRI question.
The administration of the IRI followed by approximately 2 weeks the
administration of the MAT. Researchers deliberately spaced the
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
160 M. A. TULLEY AND R. FARR
administration of the MAT and the IRI to allow as much time as
possible between these two instruments, to minimize whatever effect
might exist because of prior exposure to these passages. Each IRI
required approximately 20 minutes to administer, and instructional
reading levels were determined using those criteria for word
recognition and comprehension typically employed with this type of
instrument.
Teacher estimates of IRL. Each classroom teacher whose students
participated in the study was asked to complete a checklist, in which
they indicated their judgment of each of their students' instructional
reading level. Written directions accompanying the checklist included
a definition of an IRL which emphasized that it should be thought of as
the level at which the student is best able to achieve success, with
assistance, in reading instruction. Teachers were encouraged not to be
influenced by any factor other than each student's reading ability, such
as effort, interest, attitude, general intelligence, behavior, or
attendance.
The checklist consisted of an alphabetical listing of all students, and a
series of columns labeled by grade level from primer through grade 8.
Teachers were asked to place a check in the column across from each
student's name which corresponded to that student's estimated IRL.
Checklists required approximately 30 minutes to complete.
Results
A total of 289 students were tested, 100 at the second grade, 98 at the
fourth grade, and 91 at the sixth grade. Three analyses of data were
conducted: comparisons of percentages of agreement among estimates
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 161
of IRL, comparisons of the mean of each estimate and of a composite
IRL estimate, and comparisons of MAT estimates with the average
estimate produced by each of the other measures. A discussion of each
follows.
Percentage of agreement among estimates. Table 1 presents by
grade the percentage of agreement among each of the four methods of
estimating instructional reading level. Percentages listed in each
column indicate the extent to which each pair of estimates were the
same, or within 1, 2, 3, or 4 grade levels of one another. Of the 100
grade 2 students tested, for example, the estimates of IRL produced by
the MAT and those produced by the Houghton Mifflin basal placement
test were the same in 40% of the cases, within one grade level of one
another in 90% of the cases, and within 2 grade levels of one another in
100% of the cases.
Also at the second grade, estimates of IRL produced by the MAT and
those provided by teachers were either the same or within one grade
level of one another in 97% of the cases. MAT estimates and those
produced by an informal reading inventory derived from the MAT
(n=25) were either in exact agreement or within one grade level of one
another in 96% of the cases. Similarly high percentages of agreement
were found among all pairs of estimates at this grade level. In only one
instance of paired comparisons did estimates differ by more than two
grade levels.
Among the fourth grade subjects, estimates of instructional reading
level produced by the MAT were in exact agreement or within one
grade level of estimates produced by the Houghton Mifflin basal
placement test in 89% of the cases, and within two grade levels of one
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
162 M. A. TULLEY AND R. FARR
TABLE 1
Percentage of Agreement Within Various RangesAmong Estimates of Instructional Reading Level
Range ofAgreement
Exact
Within:+or-1 level
+or-2 levels
+or-3 levels
+or -4 levels
Exact
Within:+oM level
+or-2 levels
+or-3 levels
+or-4 levels
Exact
Within:+oM level
+or-2 levels
+or-3 levels
+or-4 levels
1986 MAT&
BasalPlacement
40%
90%
100%
-
-
31%
89%
98%
100%
-
22%
70%
91%
100%
-
1986 MAT&
TeacherEstimate
57%
97%
100%
-
-
25%
83%
98%
100%
-
26%
79%
93%
100%
-
Comparison
1986 MAT&
MATIRI*
GRADE2
36%
96%
100%
-
-
GRADE4
36%
76%
96%
100%
-
GRADE6
24%
52%
96%
100%
-
BasalPlacement
&TeacherEstimate
35%
95%
99%
100%
-
29%
74%
95%
100%
-
30%
72%
95%
100%
-
BasalPlacement
&MAT IRI*
44%
96%
100%
-
-
24%
72%
96%
100%
-
20%
64%
92%
96%
100%
TeacherEstimate
&MAT IRI*
52%
96%
100%
-
-
28%
80%
-
96%
100%
20%
88%
92%
100%
-
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 163
another in 97% of the cases. MAT estimates and those provided by
teachers were in exact agreement or within one grade level of one
another in 83% of the cases, and within two grade levels of one another
in 98% of the cases. MAT estimates and those produced by an informal
reading inventory derived from the MAT were in exact agreement or
within two grade levels of one another in 96% of the cases.
At the sixth grade level, estimates of instructional reading level
produced by the MAT were in exact agreement or within one grade
level of estimates produced by the Houghton Mifflin basal placement
test in 69% of the cases, and within two grade levels of one another in
90% of the cases. MAT estimates and those provided by teachers were
in exact agreement or within one grade level of one another in 77% of
the cases, and within two grade levels of one another in 91% of the
cases. MAT estimates and those produced by an informal reading
inventory derived from the MAT were in exact agreement or within
one grade level of one another in 52% of the cases, and within two
grade levels of one another in 96% of the cases.
Comparison of mean estimates. The mean estimate produced by
each method of determining instructional reading level was computed
for each subject at each grade level. Also computed was a composite
IRL, arrived at by averaging for each student estimates provided by
teachers, basal placement tests, and IRIs derived from the MAT. Mean
composite IRLs were then computed, and this score compared to the
IRL estimate produced by the MAT.
As shown in Table 2, the mean estimate produced by the MAT was the
highest of the four mean estimates, among both grade 2 and grade 6
subjects, and the second highest estimate among grade 4 subjects.
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
164 M. A. TULLEY AND R. FARR
TABLE 2
MEANS AND STANDARD DEVIATIONS FOR EACH ESTIMATE OF INSTRUCTIONALREADING LEVEL, AND FOR COMPOSITE IRL
ESTIMATE
TEACHER ESTIMATE
BASAL PLACEMENT
MATIRI(N=25)
1986 MAT
COMPOSITE IRL*
MAT MINUS COMPOSITE"
XS.D.
XS.D.
XS.D.
XS.D.
XS.D.
XS.D.
GRADE2
2.24.85
2.311.01
2.04.96
2.371.14
2.26.86
.11
.63
GRADE4
4.011.09
4.161.04
4.521.39
4.471.01
4.11.93
.36
.81
GRADE6
5.931.55
5.941.40
5.481.81
6.671.61
5.911.35
.76
.95
•Composite IRL is the average of the estimates provided by basal reader placement test;informal reading inventory derived from the MAT and teacher judgment."Difference between the composite IRL and the 1986 MAT estimate of IRL.
However, the range of mean estimates at each grade level was
relatively small. Among second grade subjects, mean estimates ranged
from 2.04 to 2.37, a difference of less than half a grade level. Among
fourth grade subjects the range was approximately half of a grade
level, and among sixth grade subjects the range was just over one grade
level. MAT estimates compared favorably to composite IRL means,
with a difference of only a fraction of a grade level among second
grade subjects, less than half a grade level among the fourth graders,
and less than one grade level among the sixth graders.
MAT estimates and the average of other estimates. For each
subject, estimates provided by teachers, basal reader placement tests,
and an informal reading inventory derived from the MAT were
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 165
averaged, and that score was compared to the estimate of IRL
produced by the MAT. Table 3 presents by grade level a comparison,
within specific ranges, of these averages and MAT estimates. MAT
estimates were found to be in relatively high agreement with the
average of the other estimates, at all three grade levels. Estimates were
either exact or within one grade level of one another in 91% of the
cases among the second grade subjects, in 83% of the cases among the
fourth grade subjects, and in 68% of the cases among the sixth grade
subjects. In only 1% of the cases among the fourth grade subjects,and in
only 2% of the cases among the sixth grade subjects,did MAT estimates
and the average of the other estimates differ by more than two grade
levels.
TABLE 3
COMPARISON OF 1986 MAT ESTIMATES AND AVERAGE OF IRI, BASAL PLACEMENTAND TEACHER ESTIMATES
1986 MAT AND AVERAGE
EXACT
WITHIN HALF GRADE(+ OR -.01 TO .50)
WITHIN HALF TO ONE GRADE(+OR-.51 TO 1.0)
GRADE 2
22%
69%
91%
GRADE 4
16%
46%
83%
GRADE(
9%
36%
68%
WITHIN ONE TO TWO GRADES 100% 98% 99%(+OR-1.1 TO 2.0)
WITHIN TWO TO THREE GRADES - 100% 100%(+ OR -2.1 TO 3.0)
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
166 M. A. TULLEY AND R. FARR
Conclusions
The purpose of this study was to compare estimates of instructional
reading level produced by four methods. The results suggest that
because of high percentages of agreement among these four methods,
each will yield similar information about students' IRLs.
Any method of determining IRLs is of value only if teachers have
confidence that it will enable them to efficiently and accurately place
students in materials at levels appropriate for optimum instructional
benefit. Because of this, teachers should not rely solely upon one
method of determining IRL, and decisions about students' reading
levels should not be based upon information gained from a single
assessment. The 1986 edition of the MAT, when added to a teacher's
repertoire of assessment techniques, appears able to provide an
efficient and accurate alternative method of determining IRLs, which
can assist with initial placement decisions, and is also useful for
confirming or corroborating information about student abilities
gained from other sources.
Further, there appear to be at least three advantages associated with the
use of the MAT which may not accrue when other methods are
employed. The most noteworthy, perhaps, from the classroom
teacher's perspective, is that the MAT can be more economical in
terms of time and labor. The MAT is administered in a group setting,
and thus may be particularly advantageous for those teachers who must
make placement decisions about a large number of students. Because
MAT estimates of IRL are produced in conjunction with other testing
purposes, use of the MAT may also reduce the amount of testing that
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
DETERMINING INSTRUCTIONAL READING LEVELS 167
must be conducted by teachers. And, because the MAT is a
standardized test, estimates meet standards of consistency and
objectivity which probably exceed those of the more traditional,
informal methods of determining IRL.
References
Alexander, J. (1983). Teaching Reading in the Elementary School (2nd
Edition). Boston: Little, Brown and Company.
Bristow, P.S., Pikulski, J.J., & Pelosi, P.L. (1983). A comparison of
five estimates of reading instructional level. The Reading Tea-
cher, 37, 273-279.
Content Validity of the Reading Tests. Metropolitan Special Report No.
5. (1980). New York: The Psychological Corporation.
Eckwall, E.E., & Shanker, J. (1985). Teaching Reading in the Elemen-
tary School. Columbus, OH: Charles E. Merrill.
Farr, R., & Beck, M. (1984). Validating the instructional reading level
score of the metropolitan achievement tests. Journal of Research
and Development in Education, 17, 55-64.
Farr, R., & Roser, N. (1979). Teaching a child to read. New York:
Harcourt, Brace, Jovanovich.
Harris, A.J., & Jacobsen, M.D. (1972). Basic Elementary Reading
Vocabulary. New York: Macmillan.
Lapp, D., & Flood, J. (1986). Teaching students to read. New York:
Macmillan Publishing Company.
Rupley, W.H., & Blair, T.R. (1983). Reading diagnosis and remedia-
tion: Classroom and clinic. (2nd Edition). Boston: Houghton
Mifflin Co.
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014
168 M. A. TULLEY AND R. FARR
Smith, W., & Beck, M.D. (1980). Determining instructional reading
levels with the 1978 metropolitan achievement tests. The Read-
ing Teacher, 34, 313-319.
Wide Range Achievement Test, 1978 Edition. (1978). Wilmington,
Delaware: Jastak Associates.
Dow
nloa
ded
by [
UQ
Lib
rary
] at
04:
36 0
1 N
ovem
ber
2014