FUNDAMENTAL FACTORS OF

____ COMPREHENSiON IN READING _

FREDERICK B. DAVIS
Cooperotw Test Service of tna Amencon Council on Education·

A survey of the literature was made to structing the required number of addi-
determine the skills involved in reading tional items. The results also indicate need
comprehension that are deemed most for workbooks to aid in improving stu-
important by authorities. Multiple-choice dents' use of basic reading skills. The
test items were constructed to measure
study provides more detailed informa-
each of nine skills thus identified as basic. tion regarding the skills measured by the
The inrcrcorrclarions of the ninc skill Cooperative Reading Comprehen.Hon
scores were factored. each skill being Tests than has heretofore been provided
weighted in the initial matrix roughly in regarding the skills actually measured by
proportion to its importance in reading any other widely used reading test. Sta-
comprehension, as judged by authori- tistical techniques for estimating the re-
ties. The principal components were liability coefficients of indiVIdual scores
rather readily interpretable in terms of in principal-axes components, for deter-
the initial v:lriJblcs. Individual scores in mining whether component variances are
components I and 11arc sufliciendy reli- gre3ter than would be ~'ielded by chance.
able to warrant their usc for practical and for calculating the significance of the
purposes, and useful measures of other differences between successive compo-
components could be provided by con· nent variances are illustrated.

The application of techniques of The most important step in a study

factotial analysis to the investigation of that employs factorial procedures fot the
reading has been attempted sevetal times. investigation of reading comprehension
Fedet (11), Gans (12), and Langsam (23) is the selection of the tests the scores of
have published studies that employed which are to be factored. Unless these
Thurstone's centroid method, and un- tests provide reasonably valid measures
published studies have been made by Be- of the most important mental skills chat
dell and Pankaskie. So far as the writer have to be performed during the process
is aware, the study reported here is the of reading, the application of the most
first to make use of testS especially con- rigorous statistical pro'edur~s can not
structed to measure the mental skills in yield meaningful and significant results.
reading comprehension that are consid- The importance of this point can hardly
ered of greatest importance by authori- be overstated.
ties in the field... As the first step in the present study,
a careful survey' was made of the Iitera- ([)
·On le3ve for military service . ture to identify the comprehension skills J
... For a detailed presentation of the basic that are deemed most important by au-
data of thiS study, see (8).
thorities in the field of reading. A list of
From Psychoml'triito, 9(3),1944, pp. 185-
97. Reprinted with permission.
several hundred specific skills was com·

research
 236 READING COMPREHENSION

piled, many of them ovetlapping. This of five-choice objective teSt items were
list of skills was studied intensively by construered. All possible care was taken
the writer in order to group together those to obtain items that measured only one
that seemed to tequire the exercise of the rather than several of the nine skills.
same, or closely related, mental skills. The However, it was recognized that skill I
objeerive was to obtain several groups of {knowledge of word meanings) is basic
skills, each one of which would consti- to the measurement of all the other skills,
tute a cluster having relatively high since to read at all one has to reco.gnj.«
inrercorrelations and relatively low cor-
words and understand their meanings,
relations with other clusters of skills. and that some overlapping of skills 2-9
Nine groups of skills were sorted out is inevitable.
and labeled. For the purposes of this Since the final forms of the reading-
study, they are regarded as the nine skills comprehension tests used in this study
basic to comprehension in reading. In- were to be the published forms of Tests
cluded within them is the multitude of C1 and C2 of Form Q of the Coopera-
specific skills considered important by the tive Reading Comprehension Tests,
authorities consulted. These nine basic practical considerations [notably the reo
skills are as follows:
quirements of the procedure for obtain-
1. Knowledge of word meanings ing three equivalent "scales" in the tests
(6)J determined in some measure the
2. Ability to seleer the appropriate
number of items testing each basic skill
meaning for a word or phrase in the
that could be used. An eHort was made,
light of its particular contextual serting
however, to include the proportion of
3. Ability to follow the organization of items testing each one of skills 2-9 that
a passage and to identify antecedents conformed to the judgments of authori-
and references in it ties in the field of reading.
4. Ability to sereer the main thought of To obtain the intercorrelations of
a passage SCores in the nine basic reading skills se.
leered for measurement, 240 multiple-
Ability to answer questions that are choice items were administeted to a large
specifically answered in a passage numbet of freshmen in several reachers
Ability to answer questions that are colleges.· The students were told to mark
answered in a passage but not in the every item and were allowed an unlim-
words in which the question is asked ited amount of time. By this means, the
influence of speed of reading was reo
7. Ability to draw inferences from a
moved and the effects of mechanical dif.
passage about its contents
ficulties in word perception were
8. Ability to recognize the literary de- minimized. Of the 541 students tested,
vices used in a passage and to deter. 421 J.(tuall' answered every l[em, and,
mine its tone and mood when proof was obtaine t at t IS group
9 .. -\biliry to determine a wrIter's pur.
·Every freshman in all of the ttachers col.
pose, intent, and point of view, i.c.,
leges of the Sute ot Connecticut Jnd evtr~· frtsh-
to draw inferences about a writer man In two of the .\1Jssa.:husetts Surt Teachers
CoJltges ..:omprised the u.mple testtd. The (tH'
To provide a measure of each one
ing was done about a month atter the beginning
of these nine basic skills, a large number of the school year.

research
FUNDAMENTAL FACTORS OF COMPREHENSION IN READING 237
RESEARCH/TABLE 1
Intercorrelauons. mc:ans. and standard deViations of raw scores In the nine basic reading skills. and
theIr reiaCionshlps With sex (.\1 = 421\

Skill 1 2 3 4 5 6 7 8 9 Sex· Mean a

1 .72 .41 .28 .52 .71 .68 .51 .68 .03 23.77 11.61
2 .34 .36 .53 .71 .68 .52 .68 -.07 12.70 3.25
3 .16 .34 .43 .42 .28 .41 -.01 4.20 1.73
4 .30 .36 .35 .29 .36 -.03 2.97 1.10
5 .64 .55 .45 .55 -.04 18.10 2.46
6 .76 .57 .76 -.01 25.67 5.67
7 .59 .68 .06 28.46 5.81
8 .58 -.05 6.75 1.86
9 -.05 15.19 4.07

• A posltlve coeffiCient In thiS column indicates that the men obr.llned a higher mean score than the
women.

consrirured a r restnrative sample of the stantially and significantly grearer than

entire: 541 students teste, e scores of zero.
onlv rhese 421 pupils were used in rhe Subjective judgment had forecast
factorial analySIS. In addmon ro the in- relatively high correlations between skill
(ercorrelations of the scores, the corre- 1 and each of skills 2-9. Inspection of
lations between sex and scores ~h of ResearduTable 1 in the light of ,he data
rhe nine skills were compured. As would
have been expected, the correlarions with RESEARCH/TABLE 2
sex were all insignificantly different from Reliablliry codficlcnrs of raw scores in each of
zero. This being so, there:was no need to [ne nine basic reading skills·
parrial uur rhe intluence of sex before
Number of
making a factorial analysis.
Skill 'II N Items
Research/Table 1 shows rhe inrer-
correlarions of the scores in the nine basic 1 .90 100 60
reading skills, and rheir relationships· with 2 .56 100 20
sex. 3 .44 100 9
The intercorrelations of the nine 4 .18 421 5
5
basic skills range from .16 to .76, the
6
.55
.77
100
100
"
42
values reflecting in parr their true rela-
7 .63 100 43
tionships and in parr the differences in
8 .64 100 10
their reliability. The reliability coeffi-
9 .71 100 27
cients of the scores in the nine skills are
shown in Research!T able 2. -The division of each test into two halves was
As would be expected in view of the accomplished in this case by arranging the items
in order of difficulty and asslgnmg alternate Items
widely different lengths of the tests used to tach "2If. It will be recalled thar speed had no
to measure the nine basic reading skills, influence on these scores. The reliability coeffi-
their reliability coefficiems differ consid- cient for skill 4 is b.1scd on 421 cases; the relia-
bility coeffiCients for the other ~kills uc based on
erablv. For even the least reliable, a representative sample of 100 cases drawn from
however. rh< r<liability coefficient is sub· the 421 available.

research
 238 READING COMPREHENSION

RESEARCH/TABLE 3
Partial correl~tlon coe!ficients among skills 2-9. skill 1 being held consunt IN = 421)

Skill 3 4 5 6 7 8 9

2 .09 .23 .26 .40 .38 .:26 .37

3 .05 .16 .22 .22 .09 .20
4 .19 .23 .22 .17 .24
5 .45 .32 .26 .32
6 .53 .33 .53
7 .38 .40
8 .38

in Research/Table 2 reveals this to be so. These partial coefficients are shown in

h is apparent that skill 1 constitutes the Research/Table 3.
largest element common to all of the other Perhaps the most surprising feature
initial variables; hence, it may be of in- of the data in Research,Table 3 IS the
terest to study the inrercorrel:ltions of small size of the coefficients. After mak-
skills 2-9 when skill 1 is held constant. ing due allowance for the Jttenuation

RESEARCH/TABLE 4
Inltlal matrix of variances Jnd CQvan3nces·

Var;ahle x, x, xJ x. x, x. x. x, x,

x, 134. -0 F.OI 8.16 3.65 14.77 46.88 45.78 11.04 32.07

x, 10.56 1.94 1.29 4.22 13.03 12.90 3.17 8.93
x, 3.01 0.31 1.44 4.24 4.24 0.90 2.91
x. 1.l2 0.82 2.25 2.25 0.59 1.63
x, 6.05 8.93 7.85 2.07 5.53
x. 32.17 24.89 5.96 17.42
x. 33.75 6.33 16.00
x. 3.46 4.42
x, 16.54

·Vanances are shown In the di.1gonJ,i cells. The Kelley method would be equ311~· .lppllcable If the
scores In van:ables 1-9 were rtJ.nsformed into standard mcasurC'S. In thiS c.ase. the varIance 10 cach
diagonal cdl would be 1 anJ the ";".1\ .L!l.1n.;es woulJ be IJentlc",1 \\inn the Inrcn.:orrel.1uons shown ~n
Researchrrable 1. The resulnng matrix would undoubtedly present a more familiar appearance to
many nudenn. Each one of the baSIC reading skills would then have b~n weighted equally for
purposes of factorial analYSIS. However, authontles in the field of reading quite reasonably do not
ludge each one or the basiC slulls to be oi equal importance in the process of reading comprehenSIOn.
Of the many possible facronal Jnalyses :using different weights). that analYSISwhich appears to have
unique merit ISa prinCipal-axes solution based an a matrix of vanances and covanance'S In whi~h the
initial test varianc:cs are weighted to correspond with their relauve importance in the process of
readin~. as determined by the pooled rudgment of authonries. That is the rypc of facronal anOl1ySlS
thOlt It '4"JSIntenJeJ )huulJ be performed In the present study, but pracrtc~1 conslderaCions resuhed
In some modifications in the relative weights of the nine IOlrial variables.
for purposes of comparison. the Kelley method was used to perform a factorial lnalysls oi the
correlation matrix shown in ResearchITable 1 (C'xcluding sex) with unit variances in the diagonals. A
comp:)rl)on of the b.ctor IOJdings detlved from the rwo pnn":lpJ.l-ues analyses .nd from a centrOId
analysIS of the same data IS now In preparation.

~esecrch

1 _
 fUNDAMENTAl fACTORS Of COMPREHENSION IN READING
239

RESEARCH/TABLE 5
Coefficients ot each of the initial variables thar yield scores In the nine independent components
(factor loadings of skills 1-9 In components I-IX)

IV V VI VII VIII IX
Components I II III
5.282 3.828 3.306 2.321 1.956 1.006
Van;ance 192.1,70 22.824 8.657

V.mance
Skills 134.699
.006 -.016 .001 .011
1 .813 -.571 - .064 -.033 -.082
-.003 .971 -.019 -.017 -.028 -.076 10.563
2 .184 .124 -.005
.000 -.000 .000 .997 .000 - .004 3.009
3 .057 .054 -.001
.000 .067 .000 .000 .000 .996 1.220
4 .027 .048 -.000
-.003 -.Oll .970 -.014 - .024 - .012 6.050
5 .10i .149 .152
-.531 - .129 -.204 -.044 -.001 -.023 32.169
6 .341 .469 .567
.008 - .147 - .020 - .051 -.0.1 -.02ti 33.751
7 .336 .580 -.719
.141 -.000 .000 - .010 .981 -.007 3.456
8 .078 .105 - .001
.835 - .080 - .126 -.02:- - .166 - .013 16.540
9 .233 .253 .366

resulting ftom the comparatively low re- the most interesring that has been ob-
liability coefficients of some of the vari- tained by factorial techniques.
ables, it IS apparenr rhat reading The subjective judgment exercised in
.A( comprehension, as measured by the nine constructing the teSts of.the nine reading
1'" basic reading skills, is not a unitaty abil- skills is reflected in the surprising extent
ity. From the correlations it appears to which several of the tests appear to be
probable that a mental ability presenr to moderately "pure" factor measures. A
the greatest extenr in skills 6, 7. and 9 is word of caution must, however, be in-
second most importanr in producing the jected. Because some of the skills were
inrercorrelarions shown in Research! judged to be more important than others
Table 1. To explore this matter, a fac- in the reading process and because prac-
torial analysis was undertaken, using tical considerations governed to some
the method described by T. L. Kelley extent the number of items used to mea-
sure each of the nine reading skills, the
(22).'
The initial matrix of variances and standard deviations of the initial varia-
covariances used in the factorial analysis bles differed considerably. And, since the
is presented in ResearchfTable 4. initial matrix of variancl:s .lnd covari·
In Researchrr able 5 are presented ances used for the analysis reflected those
the ooefficients of each of the initial var- differences, the coefficients in Research!
iables (the nine basic reading skills) that Table 5 must be interpreted with due re-
yield the nine independent components gard for the magnitudes of the standard
obtained by factorial analysis. The de- deviations of the nine initial skills. Scores
sign shown in ResearchfT able 5 is one of in skill 1, for example, have a large Stan-
dard deviation in comparison with the
• For this study it was desirable (0 obtain standard deviations of scores in the other
the factor loadings of 111 significant components skills. So a small component loading in
rather than the loadings for only the twO or three skill 1 may be found to have more weight
largesr componentS; hence a fairly luge number
in a regression equation for obtaining
of subJects W3S tesred 3nJ Kelle~'·s method W.loS
selected as being most suitable for use.
scores in anyone of the components than

research
 READING COMPREHENSION
240
would be expected from an inspection of hension and that any program of reme-
ResearchfTable 5 alone.' dial teaching designed to improve the
A srudy of the values in Research! abiliry of srudents to understand what
Table 5 (making due allowance for the they read must include provision for vo- ,
magnirudes of the standard deviations of cabulary building. When one combines I
the initial variables) reveals that the nine the evidence that word knowledge is so
components are rather readily identifia- important an element in reading with the
ble in terms of the original nine reading fact that the development of an individ-
skills. Component I is dearly word ual's vocabulary is in large measure
knowledge (skill 1). Its positive loadings dependenr on his interests and his back- \
in each of the nine basic reading skills ground of experience, the relatively low
reflect the fact that to tead at all it is correlations between reading tests in dif-
necessary to recognize words and to re- ferent subject-matter fields are under-
call their meanings. standable.· There IS, however, no
It is dear that word knowledge plays necessiry to conclude that all of the fun-
a very important part in reading compre- damental factors of comprehension in
reading are not involved in reading ma-
• Readers who are most familiar with the terials in various subject-matter fields.
centrOid med'lOd of factorial analysis have some-
Component II has been termed a
umes questioned this statement. A principal-axes
itnalysis makes it possible to obuin very readily
measure of reasoning in teading. It has
a giHn tndi ..".dual"s s~ore in anyone of the com- its highest positive loadings in the r.,...o
ponents for which regression coefficients (or fac- reading skills that demand ability to in-
ror loadings) have b«n determined. For example. fer meanings and to weave together sev·
IndiVIdual scores in component I may be ob- era I statements. It may seem puzzling at
tained from the follo ....IOS regression equation:
first that this component should have a
Cl = ,813'.'(1) ..... 184(.'(1) + .OS-IX)} strong negative loading in skill 1 (word
+ .027IX.1 + .107(X.1 knowledge), but consideration of the
+ .J4I(X.I + .336,X.1
+ .0781X,1 + .23JIX.i. psychological meaning of components I
and II indicates that this should be ex-
In rhi'S equation. variables 6 and 7 have
pected_ The explanation undoubtedlv lies
nearly Identical regression codticlents. but we
know that the standard deviation of ,,·an.ble 6 is in the fact that individuals who know
5.67 wntle that of vari;ablc 7 is 5.81. Thereiore. accurately the meanings of a great many
variable 7 will have 3 slightly greater weIght In words are thereby given a head start
Jetermtntng.n indiVidual's ..core 10 .:omponent 1 toward getting the meaning of what thev
than will variable 6 despite the tact that the fac·
read. ~refore. if we are to meJ.sure
tor 1030dings of v:uiable1 6 and -; tn .:omponent I
reasoning In reading independently of
are 3o1most the same.
A Simple and convenlent :lid 10 interpreting word knowledge, we must give mdmd-
the regression coefficients with proper regud for Uils who are defiCient 10 word knowl-
the Sizes at the standard deVIations of the IOtrlal eage a "handicap" and then see how well
varIables IS to construct a table containing each
they reason when Jhey are Elacea on e~
regression coefficient multiplied by the 3ppropn·
ate standard deviation of an mitial variable. For
terms with their fellows in word knowl-
example. the factor loading of skill 1 in compo- edge. Component II appatently mea-
nent I (.813) would be multiplied by the st3nd3td 'wes the ability to see the relationships
devlauon of sk1l11 (11.61). Yielding 9.4; the fac-
of ideas.
tor loading of skill 2 in component I (.184\ would
be multiplied by the st3nd.Jrd deVIation 01 skdl 2
I '\.2 ~ . ~',dJing .6; and so on. • For d3t3 on tim pOint see (5).

reseorch

1 _
FUNDAMENTAl FACTORS OF COMPREHENSION IN READING
241

Component III is not so readily in- component consists of about 77% of the
terpretable as most of the others, but it original variance of skill 3.
is clear that individuals who obtain high Component VIII measures specific
scores in this component focus their at· knowledge of literaty devices and tech-
(ention on a writer's explicit statements niques, and probably reflects the influ-
a~ost to the exclusion of their implica- ence of training in English more than the
rions. Component IV measures chidly the other components do. Component IX is
ability to loemifY"awrlter's mtent u/- composed largely of ability to select the
pose, or pomt 0 vIew (skill 9). Individ- main thought of a passage; it may be
uals wno obtam hIgh scores in this considered a measure of ability in the
component are less concerned with what synthesis of meaning. The variance of
component IX comprises approximatel v
a writer says than wlth why he says it.
Such individuals should presumably be 82 % of the original variance of skill 4
better able to detect bias and propa- (ability to select the main thought of a
ganda than individuals who obrain low passage). Students who make high scores
scores in this component. Component V in component IX are presumably those
is composed principally of ability to fig- who would be most capable of writing
ure out from the context the meaning of adequare summaries and precis of what
an unfamiliar word or to detennine which they read.
one of several known meanings of a word Of the nine components described,
is most appropriate in its particular con· all except components'll, Ill, and IV can,
textual setting (skill 2). It is reasonable for practical purposes, probably be
that it should be essentially unrelated to measured satisfactorily by means of raw
skill 1, which measures memoty for iso- scores in one of the nine basic reading
lated word meanings. The slight negative skills sel<cted initially. Components V
loadings of skills 6 and 7 in component through IX accounr for only a small
V may result from the fact that the latter fraction of the total variance, but their
measures deductive reasoning, while skills variances are significantly different.' A
6 and 7 measure inductive processes. number of the skills considered most im-
Judging by its vety high loading in portant by authorities in the field of
skill 5, component VI seems to be largely
·Thc writer ls indebted to Professor T. L.
a measure of ability to grasp the detailed Kelley for the deYelopment of a precise test of the
statements in a passage. It is probably a variance ratios of components obt:llned by hiS
fairly direct measure of the ability to get iterarive process. This reSf is descnbed In the Jr·
what 1. A. Richards has called "the lit- ride by Pro;cssor Kelley that lmmedlatdy follows.
The differences between tne variances of
eral sense m<aning" of a passage. Skill 5
successive components ue all slgn!licant at tne
was originally intended to measure this one.per.cent level with tne exception of tne dif-
abiliry and the results of the analysis sug- fetences between tne variances of components V
gest that this ability is more than a name; and VI. and VII and VIII; those differences are
it appears to be a real psychological en- signi6cant approximately at the nve-pc:r<ent level.
It snould be noted tnat tne vanance ratio
4

tity distinct from other mental skills in-

rest of the sisni6cance of the diffCKfta bC'tWC"Cn
volved in reading. Component VII seems component variances il pcenninc:d by the Kelle,
to be a measure principally of skill 3 metnod but is not permlned by otner metnods of
(ability to follow the organization of a factorial analysis tnu,are frequently employed.
passage and to identify antecedents and Whetner tne variance of component IX (tne
smallen component) IS signilicantiy greater than
references in it). The variance of this
research
 242 READING COMPREHENSION

RESEARCHITABLE6 RESEARCHITABLE7
V.1nance ranos of successive componentS Reliabiliry coeffiCIents. means, and St.1nd3rd
deViations of the SIX independent components
Degrees of haVing reli3biliry coeificlents su~s{Jntlally
Component Freedom VarIance F greater than zero

406 192.170 } Standard

8.180 Component Mean DeViation
'u
II 399 11.814 }
2.663 I .94 46.30 13.87
II .48 14.14 4.78
III 403 8.657 } III .18 .81 2.94
1.622 IV .17 -.62 2.30
IV 399 5.282 } VII .33 .27 1.53
1.387 VlII .29 .70 1.40
V 401 3.828 }
1.158
5), rhe reliability coefficients of scores in
VI 401 3.306 } the nine components have been derer·
1.428
mined empiricaJly, using the same sam-
VII 403
1.317} 1.18! ple of 100 cases for which odd and even
scores in each variable were obtained in
VIII 400 l.9;6 } computing the reliability coefficients of
1.944 rhe nine inirial variables.
IX 400 1.006
Inspection of Researchff able 7 re-
veals rhat only components I and II are
.reading include independenr elements thar measured wirh sufficient reliability to
should be raughr separately. ir is nor warrant the:ir use for practical purposes.
enough to assign learning exercises in However, when rhe significance of rhe
reading rhar consist of passages followed reliability coefficienr or each one of ,he
by facrual qucstions to be answered. Such nine components is tested; it becomes
exercises will not necessarily call rhe stu- evident rhat useful measures of ar least
dent's attention to the separate and es- three additional components could cer-
sentially unrelated reading skills [har he rainly be provided by constructing rhe
uught to master or give him sufficient required number of addirional irems of
practice in each one of rhem. the appropriate types. Since several of
Because individual scores in each of rhe components may be satisfactorily
[he independent components defined measured, for practical purposes, by raw
above can readily be csrimared by using scores in appropriate types of test items,
appropnate regression equations \Cf. construction of a large number of rhe
ante, foornore following Researchffable indicated types of irems has already been
starred. It is believed rhat rhese may be
usefulfor instructional as well as for
would be YIelded by chance may be determined
by nonng whether the rehahillty coerficlent of
-The st.1ndard error of a split·half re!i.1DII-
component IX IS slgnaJicantly greater than zero,
iry coeffiCIent, corrected by the SpeJ.rman-Bro,,·n
This IS not established by the data, It is highly
formub.. mar be obtained by using Shen's formula.
likely, however, rhat the vanance of che next
largest component is sigmficandy greater than 2(1 - rll)
would be YIelded by chance. <T'll :II 'w'N

reseorcn
FUNDAMENTAL FACTORS OF COMPREHENSION IN READING 243
measurement purposes when they are teading. The results indicate that thete is \
employedin combination with other need for reliable tests to measure several
workbook materials. of the nine basic skills that have been I
Since useful measures of compo- defined and for workbooks to aid in im-
nents I and II are already available, a proving students' abilities in them. The
profile chart for making a graphic record need for correlating scores in existing
of scores in these mo'O components has reading tests with scores in several of the
been prepared and is described in con- principal components seems obvious.
siderable derail elsewhere (9). And, not least, the study provides more
The correlations of components I and detailed information regarding the skills
II with the Q and L scores derived from measured by the Cooperative Reading
the American Council on Education Psy- Comprehension Tests than has hereto-
chological Examination and with the to- fore been provided regarding the skills
tal score on the Nelson-Denny Reading actUally measured by any ocher widely
Test have also been reported in the liter- used reading cest.•
ature (9, 370-371). It is hoped that the Finally, it is hoped thac the data pre-
relationships between components I and sented will draw attention ro che impor-
II and other well·known reading tests can tanCe of che mental skills invoh'ed in
be obtained, for if components I and II reading and act as a stimulus to further
are regatded as fundamental abilities in research in che fundamental facrors of
reading it is of paramount importance to comprehension.
dteermine the exrent to which the read-
ing tests now commonly used in high
schools and colleges actUally measure • Frederick B. D.1Vls. et al.. Th~ Coopera-
each of these abilities. tille Reading ComprehensIon Tesu. Lower .md
The study reported here has ex- Higher Ltllds. Forms Q. R, S. and T. Eight sep-
plored one means of investigating the arate 40-minutc reading tests are now aV:lIlabJe
and .are distributed by the Coopcratl\le T CSt Serv-
psychological nature of reading ability. ice, 15 Amsterdam Avenue, New York. KY., a
It has suggested a means of determining nonprofit agency of the Amcric.an Council on
the validity of tests of comptehension in Education.

REFERENCES
1. Adler, M. J. Ho .....to re3d a book, Ne.....Yorlc: 5. The Cooperanvc General AchIevement Tests
Simon IS.:. S..:huster, 1940. (RC'\'lsedSCrles): Informanon concerning [heIr
construction, Interpretation, and usc. Ne .....
2. Alderman, G. H. Improving comprehension York: Cooperative Test Service. 1940.
In re3dmg. j. ~dtlc. R~s.. 19~6, 13, 11-21.
6. The Cooperative Reading COmpreflenSIOn
3. Rerry, B. T. ImprOVing freshman reading Tests: Information concerning theIr con-
habIts. Engl. J.t College: EJltlun, 1931,20, struction, Intcrpreta[Jon • .1nd usc. Ne.....York:
8240-828. Cooperative Test ServIce, 1940.
7. The Coopcrari~e Reading Comprehension
4. Curoll. R. P. An experimental srudy of com- Tests, Lo.....er and Higher levels; Forms Qt
.prehension in reading. Ne.....York: Teachers R, S, and T. Ne .....York: Cooperative Test
Coilege, Columbia Unlversiry, t 927. Service.

reterences