a factor study of the literal reading comprehension test

A FACTOR STUDY OF THE LITERAL READING COMPREHENSION
TEST AND THE INFERENTIAL READING COMPREHENSION TEST
Neila T. Pettit and Irvin W. Cockriel*
College of Education, University of Missouri, Columbia
Abstract: Literal and Inferential reading comprehension was assessed for 533
sixth grade students. The tests used are new instruments for measuring reading
comprehension. Each test provides subscale scores: six subscales for literal
reading comprehension and five subscales for inferential reading comprehension.
The total data were factor analyzed using principal components and maximum
likelihood procedures. The results indicate the two tests are measuring distinct
factors.
BACKGROUND
For many years broad definitions of reading comprehension have been
proposed by those educators working in the field of reading. The difficulty of
isolating and defining basic skills involved in reading comprehension has been
apparent. In fact, reading authorities are still trying to define what specific skills
contribute to what broad type of reading comprehension.
As early as 1919, W. S. Gray in the Eighteenth Yearbook of the National
Society for the Study of Education listed eight skills of comprehension (Gray,
1919).
Thorndike (1917a, 1917b, 1917c) reported the first analysis of comprehension after he studied errors students made about their reading. He noted the
underlying importance of word meanings when he wrote of the "over-potency" and
"under-potency" of certain words.
In 1949, Johnson discussed the factors of reading comprehension. She
concluded:
Comprehension in reading does not seem quite such a simple matter at
this point. Teaching for improved comprehension is a more complicated
job that teachers at one time thought it was . . . Comprehension is not a
skill, but a complex function . . . Comprehension is the result of many
component skills and abilities . . . Improvement in comprehension is
* Request for reprints should be sent to the senior author, College of Education, University of
Missouri, 102-107 Hill Hall, Columbia, Mo. 65201.
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
64
Journal of Reading Behavior 1974 VI, 1
dependent on improvement in its components . . . Currently accepted
tests of reading ability do not necessarily measure all of its components.
Robinson (1966) has expanded on Gray's earlier model of the reading
process. In this analysis of the reading process, comprehension includes:
(1) understanding the literal meaning of a writer;
(2) understanding the implied meaning of a writer;
(3) assessment of a writer's purpose, frame of reference, assumptions,
and generalizations;
(4) evaluation by the reader of the writer's ideas; and
(5) integration of information and ideas of a writer with the reader's
information and related experiences.
Cleland (1965) proposed a construct or model which he thought explained
the intellectual processes employed by a reader for comprehension of the language
of the writer or speaker. He listed six factors in comprehension: 1) perception; 2)
apperception; 3) abstraction; 4) appraisal; 5) ideation; and 6) application.
According to Cleland, the critical element in perception is the meaningful response
rather than simple recognition. With apperception, Cleland is referring to the
process of relating new material to one's background of experiences. These first two
factors of Cleland's model relate closely to factors referred to by other authors as
literal and inferential reading comprehension.
Clymer (1968) reported a taxonomy of reading comprehension designed by
Barrett. This taxonomy provides an orderly presentation of categories of reading
comprehension: 1) literal; 2) reorganization; 3) inferential; 4) evaluation; and 5)
appreciation. According to Barrett, literal comprehension focuses on ideas and
information explicitly stated in the reading selection and inferential comprehension
demands thinking and imagination that goes beyond the printed page. The first and
third categories of Barrett's taxonomy seem to require skills similar to those in
Cleland's perception and apperception factors.
A large number of researchers (Singer, 1965; Fagan, 1971; Simon, 1971;
Davis, 1972) have investigated aspects of reading comprehension and devised
models, theories, constructs, and taxonomies. There appears to be agreement
among these researchers that there is, indeed, some kind of hierarchy of reading
comprehension skills. Even though there is disagreement over the specific types of
comprehension skills, the majority of studies have found reading comprehension to
be composed of two broad categories at the very least: literal comprehension and
inferential comprehension.
However, even with such general agreement concerning the skills of reading
comprehension, there appears to be little evidence of tests to measure these two
skills. Yet, research indicates a need for teaching and testing specific comprehension
skills. Carroll (1927), in studying comprehension of detailed directions, concluded
that errors were frequent in sentences containing implied material.
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
Pettit, Cockriel
65
Dewey (1935) tried to clarify the relation between ability to secure facts and
ability to do inferential thinking. He felt these two factors represented two types of
comprehension skills. He found product-moment correlation coefficients between
ability to obtain fact and inferential thinking ranging from .38 to .65. The results of
Dewey's study indicated a need for more careful training in inferential reading.
Dewey concluded that it should not be assumed that tests that measure factual
recall adequately measure understanding. Three years later, Imus (1938) and his
associates, were among the early researchers to question the value of the reading
comprehension tests in use at that time. Data from a study on comprehension
maturity tests showed that reading1 for information and reading for inference are
relatively independent skills (Feder, 1938).
Further, skills actually measured by reading comprehension tests have been
studied in detail. The Davis (1944) study, using a factor analysis technique,
provided more detailed information regarding the skills measured by the Cooperative Reading Comprehension Tests than had before been provided regarding skills
actually measured by any other widely used reading test. Alshan (1964) completed
a factor analysis of the Davis Reading Test to determine if 1) in a typical test of
reading comprehension, the items define a number of distinct factors of reading
ability and 2) those items written to measure one specific skill consistently had
loadings on the same factor. The five orthogonal factors obtained in this
sophisticated statistical treatment proved to be unidentifiable. In other words, only
one factor, "reading comprehension," was apparent; and those items written to
measure a specific skill did not seem to be identifiable as distinct skills.
While several researchers (Carroll, 1927; Richards, 1929; and Albright, 1927)
have studied comprehension from errors made by readers, most of the detailed and
specific errors found in these studies fit into broad groups of skills of reading
comprehension. In these studies, as others, both the literal and inferential
dimensions of comprehension are apparent.
More recently eighi skills of comprehension have been measured, including
skills dealing with literal or explicitly stated comprehension and skills calling for
inferential comprehension (Davis, 1968). This study showed "that part of the
variance of these eight comprehension skills is unique; therefore, teaching one of
them cannot be counted on to cause improvement in others." (Davis, 1968) Thus,
the research provides evidence of the uniqueness of certain reading comprehension
skills but no evidence of tests actually measuring these specific skills.
Pettit (1970) tested the effects of reading for given purposes on literal and
inferential comprehension. Setting of purpose for reading was provided for the
experimental groups by instructions to read to find answers to questions calling for
stated or implied meanings, along with three practice exercises. She found such
direct instruction did not significantly affect achievement on measures of either
literal or inferential reading comprehension. However, she did find achievement to
vary considerably within these dimensions of reading. Also, she noticed a need for
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
66
Journal of Reading Behavior 1974 VI, 1
reading comprehension tests with subscales to measure specific comprehension
tasks to help improve the teaching of these skills.
Comprehension studies have utilized a variety of techniques in attempts to
determine specific skills in comprehension. Multiple-regression studies (Holmes,
1948, 1954; Singer, 1965; and Holmes and Singer, 1966) have provided data on
substrata factors involved in reading comprehension. Other studies (Harris, 1948;
Flanagan, 1959; and Shaycroft, 1964) have researched procedures to estimate
unique nonchance variance of reading tests. Davis (1968) reported the estimated
unique nonchance variance in each of eight fundamental skills of comprehension in
reading. Since most studies reported appear to have skills dealing with literal and
inferential reading comprehension, no one specific theory or model was adhered to
in the present study. Rather, the authors felt the two dimensions studied and their
subscales would be a part of most models and taxonomies.
Although reading skills have positive correlations with one another, no study
was found showing positive transfer of learning from one skill to another. Until
further evidence is forthcoming, it seems necessary to test and teach specific skills
without reliance on transfer of learning between skills. Agreement among educators
and researchers indicates the need to test and teach a minimum of two general
comprehension factors, literal and inferential, and to devise test items which will
reveal specific skills within these general factors.
METHOD
The sample used in this study was drawn from the population of sixth grade
students in midwestern public schools. There were 533 students who completed all
of the tests; and they were the subjects for this study.
The instruments used were the Inferential Reading Comprehension Test
(IRC) and the Literal Reading Comprehension Test (LRC). These tests were
developed experimentally as an integral part of one of the present authors'
dissertation research (Pettit, 1970). From a review of the literature, those reading
tasks which constitute inferential comprehension and those which constitute literal
comprehension were identified. Original reading passages were written. Topics
chosen were assumed to hold the interest of upper elementary level students. The
passages involved material already familiar to the subjects. Questions pertaining to
each passage were then written. Five college students in the field of elementary
education reviewed the passages. The reading difficulty of each passage was
determined using the Dale-Chall formula for predicting readability. The passages
produced readability scores ranging from the 4.23 grade level to the 8.50 grade
level. Validity of the tests was determined by a five member panel of educators
with expertise in the field of reading. The experimental version of these tests
consisted of 32 reading passages and 146 multiple-choice questions.
A pilot study of 59 sixth grade students in an urban Missouri city established
the reliability of the test. Using the Kuder-Richardson formula for reliability, the
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
Pettit, Cockriel
67
test scores produced results of .92 for the Test of Literal Comprehension and .92
for the Test of Inferential Comprehension (Pettit, 1970).
Revised versions of these tests are titled the Literal Reading Comprehension
Test (LRC) and the Inferential Reading Comprehension Test (IRC). The revised
tests contain the same reading passages as the experimental test. The revision was
designed to balance the types of items in the tests and to create subscales within
each test. The LRC, designed to measure a reader's ability to read for stated ideas,
contains six subscales: 1) recognition of stated detail, 2) recognition of stated main
ideas, 3) recognition of stated sequences, 4) recognition of stated comparison, 5)
recognition of stated cause and effect, and 6) recognition of stated character traits.
The IRC, designed to measure a reader's ability to read for implied meanings,
contains five subscales: 1) recognition of implied details, 2) recognition of implied
main ideas, 3) recognition of implied word meanings, 4) recognition of implied
outcome, and 5) recognition of implied character traits. There was no need for
re-evaluation of validity by the panel of reading authorities or for a re-evaluation of
the readability levels since there was no change in the original reading passages
themselves.
The LRC contains 18 reading passages and 101 multiple choice items. The
IRC contains 14 reading passages and 73 multiple choice items. No subscale
contains fewer than ten multiple choice items.
Sample questions from each test and their subscales follow:
Inferential Reading Comprehension Test
Detail - At what time of the day did Tom arrive at the vacant house?
a)
morning
b)
noon
c)
afternoon d)
evening
Main Idea — A good title for this selection could be:
a)
A Summer's Day
c)
The Master Takes a Ride
b)
An Owner and His Pet
d)
The Three Pets
Word Meaning - The "seceded states" refers to:
a)
The Union States
c)
The Mexican States
b)
The Confederate States
d)
The Caribbean States
Character Traits - From this selection, one can assume the tall man was:
a)
A general of this country
c)
Father of a soldier hero
b)
A well-known writer
d)
President of this country
Outcomes — The "Purpose" with which the animals trotted down the road is probably:
a)
to get their exercise
c)
to follow their master
b)
to guard the gate
d)
to search for food
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
68
Journal of Reading Behavior 1974 VI, 1
Literal Reading Comprehension Test
Detail — The Gila monster grows to be:
a)
two feet long
b)
ten feet long
c)
d)
Main Idea - What's the best title for this story?
a)
Eastern was a Tough School
c)
b)
Benny Wins the Game
d)
Sequence — The dinosaurs disappeared:
a)
During the 1st Age of Fire
b)
During the last Age of Ice
c)
d)
greenish in appearance
brown and slender in appearance
Benny, the Dodger
Eastern Jr. High vs. Southern
Jr. High
During the 1st Age of Ice
During the great Age of Water
Comparison — Select the group below in which all belong to the amphibean family:
a)
lizards, newts, salamanders
c)
frogs, toads, lizards
b)
frogs, toads, salamanders
d)
newts, salamanders, snakes
Cause & Effect — What caused the change in Roman culture to Gothic style
architecture:
a)
Modern times began
c)
Rome was invaded from the north
b)
People were tired of old
d)
None of the above
buildings
Character Traits — According to this selection, which of the below is not true of the
toad?
a)
They have rough skins
c)
They hibernate in winter
b)
They can leap up to 6 feet
d)
They are usually only seen at night
The estimates of reliability for each subscale are presented in Table 1. The
reliabilities were computed using the Kuder-Richardson formula 21.
RESULTS
The means and standard deviations of the eleven subskills are presented in
Table 2.
The intercorrelations of the scores for the 11 basic comprehension skills is
shown in Table 3. Thus, correlations between skills range from .40 to .74. It can
clearly be seen from Table 3 that the literal subscales are correlated higher with
each other than with the inferential scales and vice versa. The intercorrelations
between the literal and inferential scales range from .40 to .58. These intercorrelations tend to confirm the independence of the scales, yet reflect positive
correlations, suggesting that all reading skills tend to be correlated.
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
Table 1
Estimates of Reliability for the 11 Subscales
Scale
KR-21
Literal Detail
Literal Main Ideas
Literal Sequence
Literal Comparison
Literal Cause and Effect
Literal Character Traits
Inferential Detail
Inferential Main Ideas
Inferential Word Meaning
Inferential Character Traits
Inferential Outcomes
.74
.61
. .69
.61
.51
.73
.61
.58
.48
.75
.65
Table 2
Means and Standard Deviations for the 11 Subscales
of the Literal Reading Comprehension Test
and the Inferential Reading Comprehension Test*
Scale
Literal Details
Literal Main Ideas
Literal Sequence
Literal Comparison
Literal Cause & Effect
Literal Character Trait
Inferential Details
Inferential Main Ideas
Inferential Word Meaning
Inferential Character Traits
Inferential Outcome
Mean
Standard
Deviation
8.30
6.05
7.13
5.15
5.38
7.25
5.66
5.48
4.99
7.80
5.61
2.96
2.18
2.28
2.16
1.90
2.69
2.28
2.24
1.89
3.21
1.88
* N=533 sixth graders, male and female
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
Table 3
Pearson Product Moment Intercorrelations of the 11 Subscales
Scale
Literal Details
Literal Main Ideas
Literal Sequence
Literal Comparisons
Literal Cause & Effect
Literal Character Trait
Inferential Details
Inferential Main Ideas
Inferential Word Meaning
Inferential Character Trait
Inferential Outcome
1
2
3
4
5
6
7
8
9
10
11
.69
.66
.64
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
.63
.61
.59
.65
.62
.61
.58
.68
.65
.65
.64
.64
.52
.48
.45
.46
.47
.51
.51
.50
.45
.49
.48
.53
.64
.49
.43
.40
.43
.46
.50
.63
.58
10
11
.58
.57
.51
.52
.53
.60
.70
.72
.64
.51
.47
.43
.43
.46
.49
.64
.64
.61
.74
Pettit, Cockriel
71
A principal components factor analysis (Nie, Bent, and Hull, 1970) was one
procedure employed to analyze data. Option PA 2 was used, which replaces the
main diagonal elements of the correlation matrix with communality estimates.
Initial estimates of the communalities are given by the squared multiple correlation
between a given variable and the remaining variables in the matrix.
From the 11 variables, two principal components emerged which accounted
for 71.3 percent of the variance. The factor loadings of the varimax rotated factors
are shown in Table 4. The six scales with heavier loadings on Factor One are 1)
recognition of explicitly stated details, 2) recognition of explicitly stated main
ideas, 3) recognition of explicitly stated sequences, 4) recognition of explicitly
stated comparisons, 5) recognition of explicitly stated cause and effect, and 6)
recognition of explicitly stated character traits.
The five scales loading on Factor Two are 1) recognition of implied details, 2)
recognition of implied main ideas, 3) recognition of implied word meanings, 4)
recognition of implied outcomes, and 5) recognition of implied character traits.
The subscales from the LRC for form Factor One and the subscales from the
IRC form Factor Two. All of the factor loadings of the 11 scales are positive, and
all 11 load moderately on the second factor.
Clearly two distinct factors are identifiable. Factor One would be called
Literal Comprehension and Factor Two would be called Inferential Comprehension.
It should be noted that the factor loadings suggest that each variable is measuring
more than one theoretical dimension. For example, Recognition of Literal Details
loads decisively on Factor One (.75), but it also loads moderately on Factor Two
(.36).
As there are a variety of factor analysis techniques, it becomes a serious
question as to which technique is most appropriate under a given set of conditions.
Spearritt (1972) has recently conducted a refactorization of reading data from a
study by Davis (1968) using new maximum likelihood procedures and reports
convincing evidence for the superiority of these procedures.
A computer program developed by Joreskog and Thillo (1971) for factor
analysis by maximum likelihood was also used to analyze the data.* The
convergence criterion value was set at .0005 and the program was started assuming
one factor. For maximum likelihood solution, a test of significance for the number
of factors is possible. For one factor, the chi square value was 552.99 with 44
degrees of freedom. Tucker's reliability coefficient was .838. The chi-square value is
significant at the .01 level, so the hypothesis that only one factor exists is rejected.
For 2 factors the chi-square value is 33.158 with 34 degrees of freedom. Tucker's
reliability coefficient is 1.00. The hypothesis that 2 factors exist is accepted. The
* The authors are indebted to Educational Testing Service for providing the maximum
likelihood computer programs to analyze portions of data for this study. Appreciation is also
expressed to Don George and Robert Nester for their help in establishing the maximum
likelihood computer programs for operation on the University of Missouri computer.
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
Table 4
Varimax Rotated Principal Component Matrix
Scale
Factor 1
Factor 2
.747
.739
.748
.686
.694
.734
.329
.354
.313
.396
.296
.366
.331
.269
.327
.337
.383
.733
.714
.679
.793
.760
Literal Details
Literal Main Ideas
Literal Sequence
Literal Comparisons
Literal Cause & Effect
Literal Character Triat
Inferential Details
Inferential Main Ideas
Inferential Word Meaning
Inferential Character Trait
Inferential Outcome
Table 5
Maximum Likelihood Factor SolutionVarimax-Rotated Factor Loadings
Sub scales
Factor 1
Factor 2
1
2
3
4
5
6
7
8
9
10
11
0.758
0.737
0.749
0.688
0.696
0.732
0.334
0.350
0.317
0.394
0.293
0.365
0.337
0.271
0.325
0.334
0.384
0.724
0.719
0.671
0.798
0.767
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
Pettit, Cockriel
73
resultant varimax factor loadings for 2 factors from the maximum likelihood
procedure are shown in Table 5.
The factor loadings of the 11 subscales from the two procedures are very
similar, and the results from the maximum likelihood procedure tend to agree with
the results from the principal component analysis.
DISCUSSION
Reading comprehension is composed of two primary dimensions, literal and
inferential. It appears from the present data that the six subscales written to
measure literal reading comprehension did, in fact, load positively on Factor
One-literal comprehension. Additionally, those five subscales designed to measure
inferential reading comprehension loaded positively on that factor. Thus, these two
instruments appear to measure those skills intended to be measured.
The fact that subscales of Factor One were found to have moderate loadings
on Factor Two is not too surprising. It tends to confirm the general opinion of other
educators that a hierarchy of reading comprehension skills exists and that all
reading skills tend to be correlated. It seems logical to assume that a reader must
obtain the literal meaning of a reading selection before he can understand implied
meanings.
The maximum likelihood factor analysis supports the conclusion that literal
reading comprehension and inferential reading comprehension are separately
distinguishable dimensions in reading. Further, it provides evidence that these tests
measure more than one basic ability. Spearritt (1972) has said, " . . . although
certain comprehension skills can be differentiated, present types of reading
comprehension tests, as distinct from word knowledge tests, largely measure one
basic ability . . . " .
It is believed by the writers that the LRC and IRC are instruments capable of
discrimination of specific types of reading comprehension skills. Part of the
variance of these reading comprehension skills appears to be unique. However, since
transfer between reading skills cannot be assured without direct teaching and direct
measurement, the LRC and the IRC should prove to be valuable measurements for
diagnostic teaching purposes.
It is hoped that teachers, by analyzing the weak answers can diagnose the
cause of a limitation within one or the other dimension of reading comprehension
and thus provide help to students. Barrett's taxonomy has provided specific
requirements for students that might be of practical importance to teachers. In fact,
the present study adds supportive evidence for two dimensions of Barrett's
Taxonomy of Reading Comprehension.
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
74
Journal of Reading Behavior 1974 VI, 1
REFERENCES
ALBRIGHT, B. F. Typical reading disabilities of college entrants. Unpublished
master's thesis. University of Southern California, 1927.
ALSHAN, L. M. A factor analytic study of items in the measurement of
fundamental factors of reading comprehension. (Doctoral dissertation,
Columbia University) New York, New York: University Microfilms, 1964,
No. 65-4756.
CARROLL, R. P. An experimental study of comprehension in reading. New York:
Bureau of Publishing, Teachers College, Columbia University, 1927.
CLELAND, D. L. A construct of comprehension. Proceedings of the International
Reading Association, 1965, 10, 59-64.
CLYMER, T. What is reading?: some current concepts. 67th Yearbook of National
Society for the Study of Education, 1968, 7-29.
DAVIS, F. B. Fundamental factors of comprehension in reading. Psychometrika,
1944, 9, 185-97.
DAVIS, F. B. Research in comprehension in reading. Reading Research Quarterly,
1968, 3, 499-545.
DAVIS, F. B. Psychometric research on comprehension in reading. Reading
Research Quarterly, 1972, 7, 628-678.
DEWEY, J. C. The acquisition of facts as a measure of reading comprehension,
Elementary School Journal, 1935, 35, 346-348.
FAGAN, W. T. Transformation and comprehension, The Reading Teacher, 1971,
25, 169-175.
FEDER, D. D. Comprehension maturity tests—a new technique in mental
measurement. Journal of Educational Psychology, 1938, 29, 597-606.
FLANAGAN, J. C. Flanagan aptitude classification tests: Technical report. Chicago
Science Research Associates, 1959.
GRAY, W. S. Principles of method in teaching reading as derived from scientific
investigation. In National Society for the Study of Education, Eighteenth
Yearbook, Part II. Bloomington, Ill.: Public School Publishing Co., 1919, pp.
26-51.
HARRIS, C. W. Measurement of comprehension in literature, The School Review,
1948, 56, 280-289; 332-342.
HOLMES, J. A. Factors underlying major reading disabilities at the college level.
Unpublished doctoral dissertation, University of California, Berkeley, 1948.
HOLMES, J. A. Factors underlying major reading disabilities at the college level.
Genetic Psychology Monographs, 1954, 49-3-95.
HOLMES, J. A., & SINGER, H. Speed and power of reading in high school.
Washington D. C.: U. S. Government Printing Office, 1966. (Cooperative
Research Monograph, No. 14).
IMUS, H. A. et al. An evaluation of visual factors in reading. Hanover, New
Hampshire: Dartmouth College Pub., 1938.
JOHNSON, M. S. Factors in reading comprehension. Ed. Admin. and Supervision,
1949, 35, 385-406.
JORESKOG, K. & THILLO, M. New rapid algorithms for factor analysis by
unweighted least square generalized least square, and maximum likelihood.
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016
Pettit, Cockriel
75
Research Memorandum 71-5. Princeton, New Jersey: Educational Testing
Service, 1971.
NIE, N. DALE, BENT & HULL, C. H. Statistical Package for the Social Sciences.
New York: McGraw-Hill, 1970.
PETTIT, N. T. Effects of Reading for Given Purposes on Literal and Inferential
Comprehension. (Doctoral dissertation, University of Missouri) Columbia,
Missouri: University Microfilms, 1970.
RICHARDS, I. A. Practical criticism. New York: Harcourt Brace, 1929.
ROBINSON, H. M. The major aspects of reading. In H. A. Robinson (Ed.) Reading:
Seventy-five years of progress. Chicago: University of Chicago Press, 1966,
pp. 22-32. (Supplementary Educational Monographs, No. 96).
SHAYCOFT, M. F. What the tests measure. In J. C. Flanagan, F. B. Davis, et al.
(Eds.) The American high school student. Pittsburgh: Project TALENT
Office, University of Pittsburgh, 1964.
SIMON, H. D. Reading comprehension: The need for a new perspective, Reading
Research Quarterly, 1971, 7, 338-364.
SINGER, H. Substrata-factor theory of reading, research and evaluation of
critiques, Proceedings of the International Reading Association, 1965, 10,
325-331.
SPEARRITT, D. Identification of subskills of reading comprehension by maximum
likelihood factor analysis, Reading Research Quarterly, 1972, 8, 92-111.
THORNDIKE, E. L. The psychology of thinking in the case of reading.
Psychological Review, 1917, 24, 220-234 (a)
THORNDIKE, E. L. Reading as reasoning: A study of mistakes in paragraph
reading, Journal of Educational Psychology, 1917, 8, 323-332. (b)
THORNDIKE, E. L. The understanding of sentences. Elementary School Journal,
1917, 18, 98-114 (c)
TEST REFERENCES
PETRIEL READING COMPREHENSION TEST. Neila T. Pettit and Irvin W.
Cockriel. Lucas Brothers Publishing Company, 909 Lowery Street, Columbia,
Missouri 65201, 1973.
Downloaded from jlr.sagepub.com at PENNSYLVANIA STATE UNIV on May 16, 2016