JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR 12,

675-682 (I 973)

Test Appropriate Strategies in Retention of Categorized Lists1

LARRY L. JACOBY
Iowa State University

Four categorized lists were presented for a single study and test trial. Form of retention
test (recognition, cued recall, or free recall) for Lists 1-3 was factorially combined with that
of List 4. Learning to learn was evident only for cued recall and improvement in that condition was primarily due to an increase in the number of items per category recalled. Effects in
Test 4 performance provided evidence that study strategy depended on the form of test
anticipated. Subjects anticipating a cued recall test apparently spent less time studying
category names and more time on the study of category instances than did subjects preparing
for free recall. Implications of test-appropriate study strategies for theories of memory are
considered.
Prior to a classroom test, students often
request information concerning the form of
test that is to be given. One gets the impression
that they plan to spend more time integrating
the material if they are told that they will
receive an essay rather than a short-answer or
recognition test. Several memory theorists
have shared students' intuitions with regard
to differences in test requirements. Kintsch
(1970) has suggested that the primary influence
of organization is on retrieval of presented
material; a process that is said to be important
for recall but involved in only a trivial manner
in recognition. Underwood (1972) has taken a
similar position by suggesting that the
attributes used for recognition might not
include associative attributes that are essential
for recall. Further, Underwood stated that
subjects may be able to influence the memory
composition of an item if they can anticipate
the form of the test. That is, subjects may learn
to encode information in a form that allows
maximal test performance.
The memory representation of an item can
be visualized as being hierarchical with information at higher levels being abstracted from
presentation of list items. The accessibility of
higher-level information might be necessary

to allow retrieval of information at lower

levels. For example, it might be necessary to
retrieve category names prior to the retrieval
of presented category instances. Thus, preparation for a test may necessitate the study of
information at several different levels of
abstraction. If subjects can anticipate the form
of the impending test, they are then free to
focus their study on required information that
the test will not provide. The result would be a
study strategy that is optimal for the particular
test form but might be quite inefficient for
tests of other types.
Most theorizing has centered around
differences in free recall and recognition tests.
However, a continuum with regard to the
number of search or retrieval cues provided by
a test can be visualized (Shriffrin, 1970).
Recognition and free recall would serve as
endpoints on this continuum with cued recall
falling between the two. A minimal number of
retrieval cues is provided by a free recall test.
Free recall of a categorized list requires that a
subject be prepared to retrieve both category
names and category instances. The necessity
of retrieving categories can be largely removed
by employing a cued recall test that provides
category names. I f a cued recall test is anticiz The author expresses appreciation to Eric Martell pated, subjects may be able to spend less time
for his assistance in the collection and scoring of data. studying category names and additional time
Copyright 1973 by Academic Press, Inc.
675
All rights of reproduction in any form reserved,
Printed in Great Britain

676

JACOBY

o n t h e s t u d y o f c a t e g o r y instances. T h e n u m b e r
o f r e t r i e v a l cues p r o v i d e d b y a r e c o g n i t i o n
test is n e a r m a x i m a l since a test i t e m m a y be
u s e d as a cue for r e t r i e v a l o f its o w n m e m o r y
representation. As a consequence, the optimal
s t u d y s t r a t e g y for a r e c o g n i t i o n test m i g h t be
e x p e c t e d to differ s u b s t a n t i a l l y f r o m t h a t f o r a
free recall test.
T h e logic o f t h e p r e s e n t e x p e r i m e n t was
similar t o t h a t o f i n v e s t i g a t i o n s o f l e a r n i n g to
l e a r n (e.g., P o s t m a n , 1969). T h e m e t h o d o f
test (free recall, c u e d recall, o r r e c o g n i t i o n )
was h e l d c o n s t a n t across t h r e e lists. T h i s
c o n s i s t e n c y o f test s h o u l d l e a d subjects to
e x p e c t t h e s a m e f o r m o f test f o l l o w i n g a f o u r t h
s t u d y list. T h e f o r m o f f o u r t h test a c t u a l l y
g i v e n was c o m b i n e d f a c t o r i a l l y w i t h t h a t o f
t h e first t h r e e lists. F o r e x a m p l e , free recall o f
t h e f o u r t h list f o l l o w e d free recall, c u e d recall,
o r r e c o g n i t i o n o f the first t h r e e lists. Differences in p e r f o r m a n c e o n the c o m m o n f o u r t h
test c a n be a t t r i b u t e d to strategies d e v e l o p e d
a c r o s s p r i o r lists. H i g h e s t p e r f o r m a n c e w h e n
test f o r m h a d b e e n h e l d c o n s t a n t across all
lists w o u l d i n d i c a t e t h a t s t u d y strategies
specific to t h e p a r t i c u l a r f o r m o f test h a d b e e n
developed.

METHOD

Materials
The 14 most frequently reported instances were
selected from each of 32 categories listed in the Battig
and Montague (1969) norms. Words that held an oddnumbered frequency rank in the norms were employed
as study items while those holding an even-numbered
rank served as new distractor items for the recognition
tests. Four 56-item study lists were formed with each
list containing seven instances each of eight different
categories; instances of a category were blocked during
study presentation.
Retention of each list was assessed by means of
either a free recall, cued recall, or recognition test. A
separate test booklet was prepared for each of the test
forms. The first page of each booklet was blank with the
exception of a statement that informed subjects that
they were not to turn that page until instructed to do so.
Instructions for the test that was to be given were
presented on the second page of each test booklet. Free
recall instructions informed subjects that they were to

write on the following page of the test booklet all of the

words they could remember from the list just presented.
Instructions in cued recall test booklets informed subjects that each of the following pages would contain
category names. They were instructed to write beneath
each category name words from that category that had
occurred in the study list just presented. Each of four
pages in the test booklet contained two category
names; categories were randomly assigned to test
position. Recognition instructions informed subjects
that each of the following pages would contain a
column of words and that they were to circle words
that had been presented in the study list. Each of four
pages contained a single column of 28 words. Study
and distractor items were from the same categories and
equal in number; items were randomly assigned to test
position with the restriction that two instances of the
same category could not occur in adjacent positions.

Design and Subjects

Form of retention test (free recall, cued recall, or
recognition) was held constant across the first three
lists studied and factorially combined with form of test
given on the fourth list (free recall, cued recall, or
recognition). The resulting design was a 3 x 3 factorial;
both factors were manipulated between subjects. Four
replications of this basic design were formed by rotating lists through study order so that each list served
equally often as the first, second, third, and fourth list
studied.
The subjects were 144 volunteers enrolled in an
introductory psychology course at Iowa State University; 16 subjects were assigned to each of the
experimental conditions. Since subjects were tested in
small groups ranging in size from two to five, it was
necessary to randomly assign groups rather than
individual subjects to experimental conditions.

Procedure
Study lists were videotaped and presented for a single
study trial at a rate of 2 sec per item. Subjects were
informed that they would see and be tested on four
lists, and that each list would contain 56 words
arranged in categories. At the end of each list, the
phrase "Begin Test" was presented. This was the signal
to open the test booklet, read instructions, and then
proceed to complete the test. The subjects had no means
of anticipating the form of test prior to reading test
booklet instructions. At the end of the 5-min. test
period, the experimenter instructed subjects to close
their test booklet and place it at the bottom of the pile
of test booklets in front of them. This procedure was
repeated until all four lists had been presented and
tested.

677

TEST-APPROPRIATE STRATEGIES
RESULTS AND DISCUSSION

Tests 1-3
Performance on Tests 1-3 was analyzed
separately for the three types of retention test.
All analyses included form of fourth test as a
factor. This factor did not approach significance in any of the analyses. Thus, there was no
evidence of differences among fourth-test
conditions prior to differential experimental
treatment.
Free and cued recall. Mean correct responses
and intrusion errors from Tests 1-3 are
presented in Table 1. Additional measures
TABLE 1
FREt (FR) AND CUED (CR) RECALL STATISTICS
FROMTESTS 1--3

Test number
Statistic

Correct responses
FR
CR

20.9
29.2

22.1
34.6

21.5
36.2

Category intrusions
FR
CR

1.1
2.1

1.5
2.0

1.7
1.7

Categories recalled
FR
CR

5.7
7.8

5.5
7.9

5.4
8.0

Items per category

FR
CR

3.6
3.7

4.1
4.4

4.0
4.6

The number of correct free recall responses

did not change significantly across successive
tests. In contrast, there was a substantial
increase in cued-recall correct responses,
F(2, 94) = 42.40, p < .001. The frequency of
intrusion errors did not vary significantly as a
function of test number in either the free or
cued recall condition.
Category recall was near perfect on all cued
recall tests. The number of categories recalled
was lower in the free recall condition and
showed a slight, nonsignificant decline across
successive tests. Mean IPC recall from Tests
1-3 is shown in the bottom rows of Table 1.
An analysis of these data revealed that IPC
recall increased across tests, F(2, 188) = 36.52,
p < .001, and was higher for cued than for free
recall, F(1, 94) = 7.99,p < .01. The interaction
of test form and number was also significant,
F(2, 188) = 6.40, p < .01. Additional analyses
revealed that the increase in IPC recall was
significant in both the free, F(2, 9 4 ) = 8.51,
and the cued, F(2, 9 4 ) = 34.44, recall conditions, p < .001 for both.
Recognition. Mean correct recognitions
(hits) and errors (false alarms) from Tests 1-3
are shown in Table 2. Both the frequency of hits
TABLE 2
RECOGNITION STATISTICS FROMTESTS 1-3

Test number
Statistic

included in Table 1 were employed to separate

influences on category recall from those on
recall of category instances. Category recall
was defined as the number of categories from
which at least one word was recalled. The items
per category measure (IPC) was defined as the
ratio of the number of words recalled to the
number of categories recalled. The definition
of measures was identical for the free and cued
recall conditions. Both the category and IPC
measures have been used by other investigators
(e.g., Tulving & Pearlstone, 1966).

Hits
False alarms

42.29
7.85

42.94
8.06

43.19
8.06

and false alarms remained quite stable across

tests. A difference score was computed for each
subject by subtracting the number of false
alarms from hits; the signal detection model
was employed to obtain d' as a second measure
of recognition. The effect of test number did
not approach significance, F < 1, in the analysis of either measure.

678

JACOBY

The present recall results can be compared

with those of Tulving and Pearlstone (1966).
Tulving and Pearlstone exposed subjects to a
single study and test trial of a categorized list
and found that cued recall produced higher
performance than did free recall. Further
analyses revealed that the cued recall advantage was totally due to a larger number of
categories recalled; IPC recall did not differ
for the two types of tests. The first test trial
results of the present investigation were in
agreement with the results reported by
Tulving and Pearlstone. As shown in Table 1,
the cued condition produced a higher level of
word and category recall than did the free
recall condition. On the first test, mean IPC
recall was nearly identical in the two test
conditions so that the word recall advantage
of the cued condition can be totally attributed
to differences in category recall on later tests,
however, cued recall held an advantage in both
IPC and category recall. There were apparently
both storage and retrieval differences between
the free and cued recall conditions on the
later tests. When a cued-recall test was anticipated, subjects were able to spend additional
time studying category instances.
Variation in number of correct responses
across Tests 1-3 might be demanded as evidence of strategy development. I f so, there is
clear evidence of strategy development only
for the cued recall condition. The free recall
condition showed learning to learn in IPC
recall but a corresponding decrease in
category recall; the result was that the total
number of correct free recall responses
remained relatively stable across tests. There
was no evidence of learning to learn across
successive recognition tests.
Test 4
The influence of preceding test form on
fourth test performance provides an additional
means of assessing strategy development. If
strategies appropriate to a particular test have
been developed, fourth test performance
should be highest when that test is of the same

form as the three preceding tests. Differences

among conditions in fourth test category and
IPC recall provide information concerning the
nature of strategies that have been developed.
Mean fourth test correct responses and
errors are presented in Table 3 for each combination of test conditions. Category and IPC
means are also included in Table 3 for conditions engaging in free or cued recall as a fourth
TABLE 3
FREE RECALL (FR), CUEDRECALL (CR), AND
RECOGNITION STATISTICSFROM TEST 4

Preceding test
Statistic for
test 4

FR

CR

Recognition

Correct responses
FR
CR
Recognition

20.8
29.6
41.9

20.1
36.8
47.9

20.9
27.9
43.5

Category intrusions
FR
CR
Recognition~ "

.8
2.7
10.0

1.3
3.0
9.2

1.0
3.2
9.0

Categories entered
FR
CR

5.2
7.8

4.4
8.0

5.1
7.9

Words per category

FR
CR

4.0
3.8

4.6
4.6

4.1
3.5

"These are number of false alarms.

test. As was the case for Tests 1-3, category
recall was defined as the number of categories
from which at least one word was recalled
while IPC recall was defined as the ratio of the
number of words recalled to the number of
categories recalled.
Free and cued recall. With free recall as a
fourth test, the effect of preceding test form
did not approach significance in either the
analysis of correct responses or that of intrusion errors, F < 1. The frequency of free recall
correct responses and errors on the fourth test
were nearly identical to those on the first free

TEST-APPROPRIATE STRATEGIES

recall test. When the fourth test was one of cued

recall, the influence of preceding test form on
correct responses was significant, F(2, 4 5 ) =
9.45, p < .01. Correct cued recall responses
were more frequent after cued recall on Tests
1-3 than after either free recall or recognition.
The number of cued recall intrusions did not
differ significantly among conditions.
Category recall was near perfect for all
conditions when recall was cued on the fourth
test. Free recall as a fourth test produced a
lower level of category recall, and differences
among preceding test conditions. Although the
effect was not significant, F(2, 4 5 ) = 1.27,
p > .10, free recall of categories was numerically lower after cued recall than after either
recognition or free recall
Differences among preceding test conditions
should be most pronounced for categories
represented in the early portion of a study list.
Owing to their recency, recall of categories
represented near the end of a list might not
reflect differences in study strategy. A further
analysis related free recall of categories to the
study list position (input block) of category
instances. Category recall as a function of
input block is presented in Figure 1 for con-

679

ditions that engaged in either free or cued

recall on Tests 1-3. The category recall curve
from the first free recall test is also presented
in Figure 1 for purposes of comparison. A plot
of number of words recalled from each input
block (not conditionalized on category recall)
was nearly identical to that shown in Figure 1
so that the curves are descriptive of word as
well as category recall probability
Fewer categories were free recalled from
Blocks 1--4 after cued than after free recall of
the first three lists, F(1, 45) = 6.03, p < .025;
recall from later blocks was nearly identical
for the two conditions Comparisons with the
curve from the first free recall test suggest that
changes in study strategy developed in the cued
but not in the free recall condition. Subjects
anticipating a cued recall test apparently spent
less time on the study of categories.
The influence of study strategy on the effect
of position within a category was the topic of
an additional analysis Items free recalled on
the fourth test were classified by position
within an input block for conditions that had
engaged in either free or cued recall on Tests
1-3. The analysis was conducted only on
items that had occurred within Blocks 2-7;
items from Block 1 and 8 were eliminated from
the analysis due to the possibility of primacy
b-----~TEST I FREE RECALL
and recency effects
~ - - ~ F R E E RECALL
o----oOUEO RECALL
Results of the above analysis revealed a
1.00
marginally significant, F(6, 180) = 2.10, p <
.05, main effect of position within a category.
.90
=..,
The first-presented instances of a category
.8o
were recalled with a higher probability than
were later-presented instance (.37, .36, .32, .35,
__d .~o
\
31,.26, and.29). The interaction of preceding
.6o
\
test condition and position within a category
a.
~
did not approach significance, F(6, 180) = 1.35,
~, .50
p > .10. Thus, preparation for cued recall
.40
influenced the recall of categories but did not
alter the effect of position within a category.
30
Free and cued recall IPC means from Test 4
.~o ~ ~ ~ ~ ; ~ ~
are presented in the last rows of Table 3. With
either type of fourth test, IPC recall was
INPUT BLOOK
FIG. 1. Category recall probablity on free recall highest when Tests 1-3 were cued, F(2, 90) =
test 4 as a function of prior test form and input block. 12.16, p < .001. Recognition and free-recall

680

JACOBY

conditions tended to produce higher IPC

recall when the fourth test was free rather than
cued; preceding cued recall produced nearly
identical IPC performance on free and cued
recall tests. Improvement in cued IPC recall
on the fourth as compared to the first test was
evident only for the condition that engaged in
cued recall on all tests. All conditions produced
higher IPC recall on the fourth free recall test
than was evident on the first test of that
type.
Recognition. Mean recognition hits and false
alarms are also presented in Table 3. The
number of hits after cued recall was significantly larger than that after either recognition,
/7(1, 45)=4.97, p < . 0 5 , or free recall,
F(1, 45) = 8.59, p < .01 ; the number of false
alarms did not differ significantly among
conditions. Only the difference between
preceding free and cued recall was significant
when recognition was corrected for guessing
by subtracting false alarms from hits, F(1,45) =
4.36, p < .05, or by employing d' scores,
F(1, 45) = 4.76, p < .05. Further analyses
failed to reveal any significant effects or interactions involving input block in either the
probability of a hit or false alarm.

DISCUSSION
The existence of test-appropriate study strategies is of considerable theoretical importance.
Several theorists (e.g., Postman, 1963) have
attempted to account for all memory effects
by postulating variation along a single hypothetical dimension. For example, a "strength"
theory postulates that test performance is a
function of the memory strength of to-beremembered items (Postman, 1963). The
memory requirements of recognition and
recall tests have been said to differ only with
regard to the degree of memory strength
necessary to allow a correct response. Presumably, cued recall would require a memorystrength intermediate to those required for
free recall and recognition. A strength theory

would predict that, within limits imposed by

ceiling effects, any variation capable of
increasing performance level on one type of
test would also enhance performance on tests
of all other types. The existence of testappropriate study strategies is incompatible
with strength theory or any other unidimensional theory of memory.
In an influential paper, Tulving and Pearlstone (1966) made a distinction between the
availability and the accessibility of items in
memory. The superiority of cued over free
recall was given as evidence that information
sufficient to recall additional items was available in memory but not accessible during a
free recall test. Given the distinction between
availability and accessibility, a major problem
is determining what factors influence accessibility. If category names are used as retrieval
cues for category instances in free recall, what
factors influence the retrievability of category
names ?
The answer to the above question might be
that category names are abstracted and studied
independently of category instances. That is,
relationships among items might be abstracted
and studied separately so as to increase their
retention as retrieval cues for presented items.
The memory representation of an item can be
viewed as being hierarchical with information
at higher levels being abstracted from
presented items. Retrieval may be constrained
by this hierarchy so that higher-level information must be accessible prior to the retrieval of
lower-level information. Inaccessibility of
information in memory would then imply that
other information, at a higher level in the
hierarchy, was not available at the time of test.
Study might be distributed among activities
concerned with construction of the hierarchy
and other activities designed to allow retention
of points in the hierarchy once they have been
constructed. For example, additional study
of a category name after it has been abstracted
might be necessary to allow its later retention.
Retention of information at any level in the
hierarchy would then be relatively independent

681

TEST-APPROPRIATE STRATEGIES

of that at other levels, and a direct function of

the quantity of study it had received.
The present result revealed that retention of
information at different levels of abstraction
was influenced by study strategy. Subjects
anticipating a cued test were able to free recall
fewer categories but more instances of each
recalled category than were subjects that
anticipated a free recall test. Within the
framework presented above, these results can
be interpreted as evidence that subjects
preparing for a cued test spent more time
studying category instances and less time
studying category names. Cued recall and
recognition results from the fourth test support
the claim that less time was spent studying
category instances when a free recall test was
anticipated; both fourth test recognition and
cued recall performance were lower when
subjects anticipated free rather than cued
recall. Thus, retention of category names and
retention of category instances were found to
be relatively independent, and influenced b'y
study strategy.
It was earlier stated that a cued recall test
eliminates the necessity of retrieving categories.
The problem one faces with a statement of
this type is identical to that encountered by
theories of recognition memory. Recognition
apparently depends on the similarity of the
encoded version of a test item and its study
counterpart (Jacoby & Hendricks, in press;
Tulving & Thomson, 1971). Correspondingly,
the cue effectiveness of a category name would
be expected to depend on the similarity of its
encoding to the category representation
stored during study. Martin (1972) has voiced
concern with an identical problem in pairedassociate learning by noting that the encoding
of a stimulus might vary between presentations. The problem is even more pronounced
in the cued recall case when the category name
is not presented during study but must be
abstracted from category instances. The
category name may be used as a cue for
retrieval of a representation encoded during
study that is only similar, not identical to the

category name. One advantage of repeated

testing might be that subjects learn to increase
the similarity of representations encoded
during study to the cues that will be offered
by the test.
Study strategies that were specific to either
free recall or recognition were apparently not
developed in the present investigation. When
subjects are unable to anticipate the form of
test, they might behave in a conservative
fashion and prepare for free recall. A preexperimental strategy appropriate for free
recall might then have been employed from the
outset; both category names and instances
may have been studied beginning with the
first list presented. Subjects receiving free
recall tests would have no reason to modify
their preexperimental strategy. It is surprising,
however, that the initial strategy was not
modified appreciably by subjects receiving
recognition tests. A potential explanation is
that the recognition test did not provide unambiguous category information (Jacoby,
1972). There was no assurance that all recognition test items were from a category presented
during study, and there was also the danger
of categorizing a test item differently from the
way in which it had been categorized during
study. Thus, category information may have
been studied and used to aid recognition
performance. Additional research is needed
to clarify differences in memory requirements
of free recall and recognition tests.

REFERENCES
BATTIG, W. F., & MONTAGUE,W. E. Category norms
for verbal items in 56 categories: A replication
and extension of the Connecticut category norms.
Journal of Experimental Psychology, 1969, 80
(3, Pt. 2).
JACOB'g, L. L. Effects of organization on recognition
memory. Journal of Experimental Psychology,
1972, 92, 325-331.
JACOB'g,L. L., & HENDRICKS,R. L. Recognition effects
of study organization and test context. Journal of
Experimental Psychology, in press.

682

JACOBY

KINTSCH, W. Models for free recall and recognition.

In D. A. Norman (Ed.), Models of human memory.
New York: Academic Press, 1970.
MARTIN,E. Stimulus encoding in learning and transfer.
In A. W. Melton & E. Martin (Eds.), Coding
processes in human memory. Washington, D.C. :
Winston, 1972.
POSTMAN,L. One-trial learning. In C. N. Cofer & B. S.
Musgrave (Eds.), Verbal behavior and learning.
New York: McGraw-Hill, 1963.
POSTMAN, L. Experimental analysis of learning to
learn. In G. H. Bower & J. T. Spence (Eds.), The
psychology of learning and motivation, Vol. 3.
New York: Academic Press, 1969.
SHIFFRIN,R. M. Memory search. In D. A. Norman
(Ed.), Models of human memory, New York:
Academic Press, 1970.

TULVING,E., & PEARLSTONE,Z. Availability versus

accessibility of information in memory for words.
Journal of Verbal Learning and Verbal Behavior,
1966, 5, 381-391.
TULXaNG, E., & TrlOMSON,D. W. Retrieval processes
in recognition memory: Effects of associative
context. Journal of Experimental Psyehology, 1971,
87, 116-125.
UNDERWOOO, B. J. Are we overloading memory ? In
A. W. Melton &E. Martin (Eds.), Codingproeesses
in human memory. Washington, D.C. : Winston,
1972.

(Received May 18, 1973)