List categorization in Buschke's restricted reminding procedure

MISSING IMAGE

Material Information

Title:
List categorization in Buschke's restricted reminding procedure effects on encoding and retrieval processes in elderly adults' verbal memory
Physical Description:
ix, 144 leaves : ill. ; 29 cm.
Language:
English
Creator:
Cooley, Susan Gail, 1957-
Publication Date:

Subjects

Subjects / Keywords:
Memory -- Aged   ( mesh )
Aging   ( mesh )
Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- UF -- Clinical and Health Psychology   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1985.
Statement of Responsibility:
by Susan Gail Cooley.
General Note:
Typescript.
General Note:
Vita.
General Note:
Bibliography: leaves 137-143.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000524919
oclc - 16896104
notis - ACU6765
sobekcm - AA00006110_00001
System ID:
AA00006110:00001

Full Text











LIST CATEGORIZATION IN BUSCHKE'S RESTRICTED REMINDING
PROCEDURE: EFFECTS ON ENCODING AND RETRIEVAL
PROCESSES IN ELDERLY ADULTS' VERBAL MEMORY











By

SUSAN GAIL COOLEY


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA


1985














This dissertation is dedicated to my parents, Florence
and Warren Cooley, and my sister, Diane Cooley, with loving
gratitude for their patience, support, and good humor during
this lengthy project.














ACKNOWLEDGMENTS


There are a number of individuals whose contributions

I would like to acknowledge. Deep appreciation goes to my

chairman, Dr. Walter R. Cunningham, not only for his in-

sightful comments on these data but also for his continued

interest in and support of my personal and professional

growth. I would also like to thank Drs. Eileen B. Fennell,

Warren J. Rice, Hugh C. Davis, and Harold C. Riker for

serving as committee members, with particular thanks to

Eileen Fennell and Warren Rice for their comments on

earlier drafts of this paper.

Several people helped with technical aspects of the

data collection and analysis. I would like to thank Dr.

Stephan Sussman, of Santa Fe Community College in Gaines-

ville, Florida, for his generous assistance with young

adult recruitment and provision of testing space; research

assistant Elisa Kaye, for her cheerful transcription of

numerous audiotapes and detailed data scoring; and Nina

Dronkers, for statistical consultation and computer access

during analysis of the data. I would also like to thank

Rae Spivey for her expert typing of the final manuscript.

The contributions of several others were less direct

but no less valuable. In particular, I would like to thank


iii










Drs. Antonette Zeiss and Donald Lim, both of Palo Alto

Veterans Administrtion Medical Center, for their help in

keeping alive the spark of dissertation interest during my

internship year. In addition, Toni Zeiss continues to pro-

vide an important role model, as a shining example of that

most difficult combination, the true scientist-practitioner.

For their friendship and professional encouragement during

the difficult final phases of this project, I would also

like to express my warm appreciation to Nancy Hurley, Linda

Gonzales, Phil Ackerman, Karen Waters, Joe Bush, Ruth

Czirr, and Dana Jenkins. Very special thanks go to David

Cohen for his personal and material support of this project

and countless others.















TABLE OF CONTENTS


ACKNOWLEDGMENTS ......................................

ABSTRACT .............................................

INTRODUCTION .........................................

Changes in Memory Research Paradigms and
Terminologies .....................................
Secondary Memory Deficit in Older Adults ..........
Encoding vs. Retrieval Deficit in Secondary
Memory ............................................
Buschke's Restricted Reminding Procedure ..........
Present Study .....................................
Hypotheses ........................................
Preliminary Analysis ............................
Primary Analyses ................................
Secondary Analyses ..............................

METHOD ...............................................


Subjects ..........................................
Materials ......................... .............
Experimental Conditions and Procedures ............
Scoring ...........................................
Vocabulary Test .................................
Buschke Memory Task .............................
Summary of Design .................................
Data Analyses .....................................
Notes .............................................


RESULTS ..............................................

Preliminary Analysis of Vocabulary Scores .........
Vocabulary and Age ..............................
Vocabulary and Dependent Measures ...............
Primary Analyses ..................................
Hypothesis 1: Overall Recall ....................
Hypothesis 2: Secondary Memory Encoding .........
Hypothesis 3: Secondary Memory Retrieval ........
Hypothesis 4: Encoding and Retrieval Compared ...


Page









Page

Secondary Analyses ................................ 66
Hypothesis 5: Proportion of Words Retrieved
from Storage .................................... 66
Hypothesis 6: Consistent Secondary Memory
Retrieval ....................................... 69
Hypothesis 7: Encoding and Retrieval Across
Trials ....................................... ... 73
Encoding ...................................... 73
Retrieval ..................................... 76
Consistent retrieval .......................... 79
Hypothesis 8: Intrusions ........................ 79
Reported Memory Strategies ........................ 93

DISCUSSION ........................................... 99
Central Findings .................................. 100
Preliminary Analysis of Vocabulary Scores ....... 100
Primary Analyses ................................ 101
Overall recall ................................ 101
Secondary memory encoding and retrieval ....... 103
Secondary Analyses .............................. 105
Proportion of encoded words retrieved
from storage .................................. 105
Consistent retrieval from secondary memory .... 109
Patterns of encoding and retrieval across
trials ........................................ 110
Intrusion errors .............................. 110
Reported Memory Strategies ...................... 113
Methodological Limitations ........................ 114
Summary and Conclusions ........................... 118
Implications for Future Research and Clinical
Practice .......................................... 119

APPENDICES
A EXPERIMENTAL MATERIALS
Background Information Sheet .................... 125
Vocabulary Scoring Sheet ........................ 126
Sample Stimulus Word Card for Memory Task ....... 127
Related Word List Scoring Sheet ................. 128
Unrelated Word List Scoring Sheet ............... 129
Post-Test Inquiry ............................... 130

B INFORMED CONSENT FORM ............................. 132

C DESCRIPTION AND EXAMPLE OF COMPLETED BUSCHKE
MEMORY TASK SCORE SHEET ........................... 133

D GLOSSARY OF TERMS AND ABBREVIATIONS ............... 135

REFERENCES ........................................... 137

BIOGRAPHICAL SKETCH .................................. 144















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


LIST CATEGORIZATION IN BUSCHKE'S RESTRICTED REMINDING
PROCEDURE: EFFECTS ON ENCODING AND RETRIEVAL PROCESSES
IN ELDERLY ADULTS' VERBAL MEMORY

By

Susan Gail Cooley

May, 1985

Chairman: Walter R. Cunningham, Ph.D.
Major Department: Clinical Psychology

In the present study, several aspects of verbal re-

call memory in 52 young adults (range 17-34 years) and 52

community-dwelling older adults (range 61-80 years) were

investigated by analysis of variance procedures. Subjects

recalled either an unrelated list or a single-category

related list of 20 words in the multi-trial Buschke re-

stricted reminding procedure. The relationship between

subjects' verbal ability and memory was first explored.

In a series of primary analyses, effects of age and ex-

perimenter-generated list organization on recall memory

were determined, and age deficits in secondary memory en-

coding and retrieval were compared. A series of secondary

analyses involved examination of additional retrieval mea-

sures and patterns of encoding, retrieval, and intrusions


vii










across trials. Young and old subjects' post-test reports

of memory strategies were also compared. Major dependent

variables included both standard Buschke encoding and

retrieval scores (LTS, LTR) and an additional retrieval

measure derived from them (LTR/LTS).

A significant deficit in secondary verbal recall

memory was found for this group of well-educated, highly

verbal elderly adults. The age deficit in recall, however,

was eliminated when material to be remembered contained a

high degree of organization. Contrary to other investiga-

tors' emphasis on either encoding or retrieval problems in

the elderly, present results indicated significant age

deficits of approximately equal magnitudes in both of these

secondary memory stages. Also, facilitative effects of

list organization, for young and old, appeared confined to

the encoding stage, with no significant effect on retrieval

per se. Patterns of encoding, retrieval, and intrusions

across recall trials did not differ significantly between

the two age groups. However, learning intrusions along

with correct list words may have been more detrimental to

older adults' recall, due to their greater numbers of

intrusions. Reports of memory strategies among old and

young were highly similar. Age differences in recall may

thus reflect differences in how effectively strategies are

used, rather than in which ones are used. Finally, high

correlations among basic scores obtained from the Buschke


viii










procedure indicated need for further investigation of the

statistical and conceptual adequacy of Buschke's scoring

system.















INTRODUCTION


According to most stereotypes of human aging, increasing

forgetfulness is a striking and inevitable part of growing

older. In many cases, the subjective experiences of older

persons themselves confirm the stereotypes, as complaints

of memory problems are extremely common among older adults.

In the more controlled environment of the research labora-

tory, psychologists have attempted to specify, quantify,

explain, and even modify observed age-related memory defi-

cits. After several decades of investigation, researchers

have amassed a substantial body of objective data on memory

difficulties associated with both normal and abnormal aging.

Not surprisingly, the data reveal a more complex picture of

memory and aging than the stereotypes suggest.

The study to be described here was conducted in an

effort to shed light on a central issue which remains unre-

solved in research on normal aging and memory. It addressed

the basic question of the relative contributions of encoding

and retrieval difficulties to elderly adults' verbal second-

ary memory deficit. To provide the background context for

the present research, an overview of major findings from

the literature will be given before the specific nature and

results of this study are presented. First, however, some






-2-


of the potentially confusing changes in research terms and

models in this area will be briefly reviewed.


Changes in Memory Research Paradigms and Terminologies

Guided by changing theories and experimental models of

memory in general, research on memory and aging has wit-

nessed a succession of concepts, paradigms, and terminolo-

gies over the past several decades. Experimental approaches

have ranged from the stimulus-response (S-R) framework em-

phasized through the 1950s to later information-processing

models, two major variations of which include the concept

of multiple memory stores and the more recent levels-of-

processing approach. It will be assumed that the reader

has some familiarity with these basic models and terms, and

the descriptions and comments below will be necessarily

brief. For greater detail, one may consult general memory

texts (e.g., Klatzky, 1980; Wingfield, 1979) as well as

aging-related reviews (e.g., Botwinick, 1978; Cermak, 1980;

Craik, 1977; Hartley, Harker, & Walsh, 1980; Hultsch &

Pentz, 1980; Kay, 1959).

In the early S-R model, learning was viewed as the

formation and strengthening of associative bonds between a

particular stimulus and subsequent response. Repetition

was suggested as a major mechanism in the formation of the

bonds, which themselves formed the contents of memory.

Broadbent (1958) was among the first to discuss the

idea of separate stages of memory processing. The "modal










model" of this type (Murdock, 1967) involves the three

stages of (1) sensory stores, (2) short-term memory, and

(3) long-term memory. The first stage is typically de-

scribed as a.set of modality-specific sensory stores

(auditory, visual, etc.) in which sensory stimuli are held

very briefly in "raw" form, decaying rapidly unless further

processed. Stimuli that are attended to proceed to short-

term memory, which is viewed as a limited-capacity store in

which a few items are coded primarily acoustically and

maintained through active rehearsal. Through processes of

rehearsal and semantic coding, information can be passed on

to long-term memory, which is viewed as a permanent store

with unlimited capacity.

The multi-store model outlined above presents the

various memory stages as structurally separate stores. In-

vestigators within this framework have focused considerable

attention on the delineation of time and capacity parameters

of each stage. In contrast, Craik and Lockhart (1972) have

proposed a levels-of-processing model which focuses on what

the individual actually does with incoming stimuli, that

is, on the nature and extent of the individual's processing

of information. They suggest that stimuli can be processed

in three basic ways: in terms of physical, acoustic, and

semantic properties. These three types of processing are

said to form a dimension of "depth," with physical pro-

cessing the most shallow and semantic processing the

deepest. According to this model, better memory of an item









is associated with greater depth of processing as well as

greater elaboration or richness of associations at a

particular level. Although Craik and Lockhart (1972)

asserted that the major factor determining memory strength

is depth of processing, they also suggested that continued

processing or recirculation of information at any given

level also aids retention of the stimulus. Tulving and

Thomson (e.g., Thomson & Tulving, 1970; Tulving & Thomson,

1973) and others (e.g., Fisher & Craik, 1977) have indi-

cated the need for some modification of the original levels-

of-processing model. Their data provide the foundation for

the encoding specificity principle, in which it is posited

that memory performance is best when the type of retention

test (semantic, acoustic, structural) matches the type of

initial processing, even though semantic processing plus a

semantic test may result in the highest performance overall.

In an updated version of his model, Craik (1979) has re-

cently acknowledged the importance of such encoding/

retrieval cue compatibility along with depth of processing

for optimal memory performance.

In addition to the formulation of new models of memory,

there have been changes in the specific terms applied to

various memory stages and processes. In particular, within

the multi-store framework many investigators have now re-

placed "short-term" and "long-term" memory with "primary"

and "secondary" memory, respectively, following Waugh and

Norman's (1965) distinction. "Primary memory," sometimes










called "working" memory, refers to the limited amount of

material which can be actively focused on, that is, held

in conscious awareness by active rehearsal. Digit span is

one task typically used as a measure of primary memory.

The parallels with the "short-term" memory store are imme-

diately apparent. "Secondary memory," also known as "supra-

span" memory, refers to retention of information exceeding

primary memory capacity, that is, information beyond the

immediate span of apprehension. Like "long-term" memory,

secondary memory is viewed as a large-capacity, more perma-

nent store. The move toward these newer terms represents a

shift away from a focus on the uncertain time parameters

implied by "short-term" and "long-term" memory and an

emphasis instead on the capacities and processes of these

hypothesized separate memory systems. In the present

paper, the terms primary memory and secondary memory will

be used whenever possible, with meanings as described

above.


Secondary Memory Deficit in Older Adults

There have been contradictory findings and controver-

sial interpretations both within and between these different

experimental frameworks. However, there is now general

agreement that for normal, healthy individuals, greatest

age differences are found in secondary ("supraspan") memory

rather than in sensory memory or primary memory (Craik,

1977; Hartley et al., 1980; Smith & Fullerton, 1981). The










bulk of research to date has indicated minimal age deficits

in amount and rate of encoding at the sensory memory stage

(e.g., Cerella, Poon, & Fozard, 1982; Hines, Poon, Cerella,

& Fozard, 1982). Similarly, little or no age differences

in primary memory capacity are found, except when reorgani-

zation of the stimulus or division of attention is required

(Botwinick & Storandt, 1974; Craik, 1977; Raymond, 1971;

Smith, 1975). There is, however, some evidence of slightly

decreased speed of retrieval from primary memory with in-

creasing age (Anders, Fozard, & Lillyquist, 1972; Waugh,

Thomas, & Fozard, 1978).

In contrast to the results for sensory and primary

memory, large age differences are frequently found on mea-

sures of secondary memory. There is some evidence of de-

creased speed of retrieval from secondary memory in older

adults (Waugh et al., 1978). And a great deal of research

has demonstrated significant age deficits on measures of

amount of material retained in secondary memory. A major

question that remains, however, is whether the memory defi-

cit of older adults is primarily a problem in encoding

material into secondary memory storage or in retrieving

material from storage.


Encoding Versus Retrieval Deficit in Secondary Memory

Early attempts to specify the nature of age-related

memory deficits focused mainly on the retrieval stage.

Major support for the hypothesis that older adults'







-7-


secondary memory deficit is primarily a retrieval problem

has come from studies contrasting recognition and recall

memory performances. For example, Schonfield and Robertson

(1966) tested young and elderly subjects' recognition and

recall of a list of 24 words. Subjects ranged in age from

20 to 75 years old. No age differences in recognition were

found, although there was a consistent drop in recall

scores with increasing age. Since the older subjects

benefited disproportionately from the recognition procedure,

which is considered to minimize retrieval requirements

relative to free recall tasks, Schonfield and Robertson

concluded that poor memory performance in older adults

stems primarily from retrieval difficulties. Other re-

searchers using a variety of experimental paradigms have

reported similar results of significant age differences in

recall but not recognition, as well as better performance

overall on recognition rather than recall tasks (e.g.,

Rankin & Hyland, 1983; Shaps & Nilsson, 1980; West & Boat-

wright, 1983).

Some studies have shown that age decrements may occur

in recognition as well as recall. For example, using a 24-

item word list and subjects ranging in age from 19 to 75

years old, Erber (1974) reported significant age deficits

on both recognition and recall scores. Gordon and Clark

(1974a, 1974b) found poorer recognition performance in

elderly subjects relative to young subjects on tasks in-

volving retention of prose, lists of words, and lists of







-8-


nonsense syllables. Using signal detection methods in

their study of recognition memory, Harkins, Chapman, and

Eisdorfer (1979) similarly concluded that elderly subjects

were less accurate in discriminating between "old" and

"new" words than were young subjects.

Studies of recall and recognition have varied widely

in their materials, procedures, data analyses, etc., and

results are not always directly comparable. However, the

typical finding has been that age decrements in recognition

reported in such studies, if statistically significant, are

of small magnitude and are less severe than age differences

in recall memory (Botwinick & Storandt, 1974; Craik, 1977;

Erber, 1974; Gordon & Clark, 1974b). Regarding reports of

age deficits in both recognition and recall scores, Craik

(1977) has pointed out that the hypothesis that older per-

sons have particular difficulty with retrieval from storage

"depends simply on greater age losses in recall than in

recognition (and) would not be disconfirmed by the finding

of age decrements in recognition under some conditions"

(p. 401). Contrasting evidence has been presented by White

and Cunningham (1982) in a recent, thoughtful study. They

found statistically parallel age deficits in recall and

recognition when recognition scores were corrected for

guessing. These authors have emphasized the need to con-

sider error model assumptions and other methodological issues

involved in comparisons of recognition and recall scores.






-9-


Additional support for the retrieval hypothesis has

come from studies using cued recall, another method of

reducing memory retrieval demands. A number of investiga-

tors (e.g., Hultsch, 1975; Laurence, 1967b; Smith, 1977)

have reported that age differences in verbal memory are

greatly reduced, if not eliminated, when subjects are given

category labels as retrieval cues at time of testing. How-

ever, Drachman and Leavitt (1972) found no reduction of age

differences when the initial letters of list words were

used as retrieval cues in a cued-recall paradigm. On the

basis of his investigation of both semantic (category

labels) and structural (initial letters) cues, Smith (1977)

has argued for the relative efficacy of different types of

retrieval cues, with semantic cues more effective than

structural cues in reducing age differences in recall.

In Sm-ith's (1977) study, significant age deficits in

recall were found when there were (1) no cues at all; (2)

cues, structural or semantic, present at output (time of

testing) only; or (3) structural cues present at either

input only or both input and output. Age differences were

eliminated, however, when semantic cues were provided at

either input only or both input and output. Thus, Smith's

results suggest the importance of not only the semantic

type of cues but also the presence of such cues at input

(i.e., at time of encoding), for maximizing recall perfor-

mance of older adults.






-10-


Influenced in part by the levels-of-processing model

of memory, a number of investigators of memory and aging

have shifted their focus to the encoding stage. Support

for the hypothesis that encoding difficulty in older adults

is a primary explanation for age differences in secondary

memory is drawn from several sources. For example, there

is evidence that older adults are deficient in the spon-

taneous use of verbal and visual mediators during experi-

mental memory tasks. "Verbal mediator" typically refers

to a word, phrase, or sentence connecting the to-be-

remembered words (e.g., the phrase "candles on the cake"

linking stimulus words "candles" and "cake"). "Visual

mediator" refers to a mental image or picture of the ob-

jects represented by the stimulus words (e.g., mental image

of candles burning brightly on a birthday cake linking the

same stimulus word pair). The use of mediators as an en-

coding strategy has been shown to improve memory perfor-

mance, with visual mediators often more effective than

verbal mediators (Paivio, 1971).

Hulicka and Grossman (1967) found that with no special

instructions on a paired-associate task, older subjects'

spontaneous use of mediators was only about half as fre-

quent as that of young subjects. In addition, there is

evidence that when they do employ mediational techniques

spontaneously, older subjects tend to use verbal mediators

more often and visual mediators less often than do young

subjects (Hulicka & Grossman, 1967; Rowe & Schnore, 1971).






-11-


In a descriptive study of mnemonics spontaneously used by

elderly individuals, Camp, Markley, and Kramer (1983) also

found reports of using visual imagery relatively rare among

their elderly subjects. When instructions to use mediators

are given, however, older subjects benefit more than the

young, although they still do not reach the same level of

recall performance as young subjects (Canestrari, 1968;

Hulicka & Grossman, 1967).

Another aspect of encoding processes is organization,

i.e., the integration, categorization, or grouping of dif-

ferent items on a list. Results of a number of studies

have suggested that older adults are deficient in sponta-

neous organization of material to be remembered, although

they can use organizational aids or strategies provided for

them. For example, Hultsch (1974) examined the multi-trial

free recall performance of subjects ranging in age from 18

to 85 years. He obtained two measures of subjective organi-

zation, one indexing the degree to which subjects organized

the list words in serial order, and another indexing the

degree of concordance between a subject's order of recall

from one trial to another. Both measures of organization

showed consistent decreases with increasing age. Similar

results of significant age decrements in amount of organi-

zation have been reported by other investigators employing

similar as well as different free recall paradigms (e.g.,

Macht & Buschke, 1983; Smith, 1980).







-12-


There has been some question regarding the most appro-

priate measure of subjective recall, however (see Sternberg

& Tulving, 1977, for a review of available methods), and

not all measures have shown older adults to have less orga-

nization in their recall. Laurence (1966) used a different

measure than those of Hultsch (1974) and Smith (1980) and

found no significant differences in degree of subjective

organization between young and elderly adults despite a

difference in the two groups' free recall scores.

There is evidence that older adults can benefit from

experimenter-generated organization strategies. Hultsch

(1969) examined the free recall performance of subjects

aged 16 to 54 under conditions of (1) standard free-recall

instructions, (2) nonspecific instructions to organize list

words, and (3) specific instructions to organize words al-

phabetically. Results of this study are quite interesting.

High-verbal subjects (based on vocabulary test scores)

showed no age deficit in recall under any of the experi-

mental conditions. Low-verbal subjects, however, did show

an age-related performance deficit under the standard and

nonspecific organization instructions but not under the

alphabetical organization instructions. Thus, it appears

that specific organization instructions can improve older

adults' recall, at least for those individuals with lower

verbal intelligence.

Laurence (1967a) provided another type of organization

aid in her study of free recall in young and elderly adults.







-13-


Subjects were tested for recall of 12-item lists of either

unrelated nouns or related nouns (members of a single

concept category). The use of categorized lists has been

shown to improve free recall performance of young adults

(e.g., Bower, 1970; Dallett, 1964; Ekstrand & Underwood,

1963; Underwood, 1983). Laurence found a similar effect of

improved recall with the categorized word list among her

elderly subjects. In addition, she found that the elderly

adults benefited more than the young adults from the use of

the related word list. That is, the age decrement in

recall performance found with the list of unrelated words

was eliminated when recall of the more highly organized

list of related words was tested.

Additional support for an encoding deficit explanation

of older adults' secondary memory difficulty comes from

investigations of depth of processing during memory tasks.

Eysenck (1974) tested young and old adults for free recall

following structural, acoustic, and semantic orienting

tasks (counting letters, generating rhymes, and generating

adjectives for stimulus words, respectively). A significant

age deficit in recall was found after the semantic orienting

task only, not after the nonsemantic orienting tasks. The

results suggested that older adults are deficient in the

use of deeper (more semantic) levels of encoding processes.

Subsequent studies, however, have produced conflicting

results. Rankin and Kausler (1979) employed a false recog-

nition paradigm to determine whether older adults were more






-14-


susceptible to false recognition of stimulus word rhymes

than synonyms, relative to young adults, a pattern which

the differential depth hypothesis would suggest. Contrary

to expectation, older subjects showed higher false recog-

nition rates for both rhymes and synonyms than did young

subjects. Rankin and Kausler concluded that the obtained

age differences in recognition performance level were indi-

cative of differences in extent of elaboration within a

processing level rather than in depth of processing. Using

a different false recognition paradigm in a more recent

study, however, Rankin and Hyland (1983) found no evidence

of age differences in either depth of processing or in

extensiveness of encoding within processing levels.

As indicated by the studies reviewed above, much of

the past research on memory and aging has emphasized either

encoding or retrieval processes as the source of secondary

memory deficits in older persons. Using a multitude of

paradigms and methods, researchers focusing on one or the

other stage have produced data that support their par-

ticular viewpoint. Some researchers, however, have acknowl-

edged that both stages may be affected by age (Craik, 1977).

In addition, there has been increasing recognition of the

essential interdependence between these two stages of memory.

It is obvious that in order to retrieve an item from memory,

it has to have been encoded in the first place. Difficulty

in retrieval may therefore reflect problems of initial en-

coding.






-15-


The importance of compatibility between encoding and

retrieval conditions or strategies has been emphasized by

Tulving and his associates (e.g., Tulving, 1979; Thomson

& Tulving, 1970), who have labeled the phenomenon the en-

coding specificity principle. West and Boatwright (1983)

have recently demonstrated the applicability of this prin-

ciple to the memory performance of old as well as young

adults. Using groups of young, middle-aged, and older

adults in a cued recall paradigm, West and Boatwright com-

bined semantic (category labels) and acoustic (rhymes) test

cues with semantic and acoustic orienting tasks. Semantic

processing, semantic cue test, and combined semantic

processing/semantic cue test conditions led to better

recall performance for all age groups. In addition, age

differences in recall found with acoustic processing and/or

acoustic cue test conditions were eliminated in the seman-

tic processing/semantic cue test condition. These results

provide support for the differential effectiveness of

"deeper" cue types as well as the importance, for all ages,

of matching encoding and retrieval cues.

This recent trend in thinking about the interdepen-

dence of encoding and retrieval is reflected in the fol-

lowing comments on older adults' memory problems:

One consistent story has been emerging over the
past several years of research in memory and
aging. The old recall less; and they encode less
effectively. Retrieval is so dependent upon suc-
cessful encoding that characterizing age deficits
as encoding or retrieval seems to be a matter of







-16-


preference. One could say that the major
locus of age deficits in memory is encoding;
old people encode less or encode less effec-
tively and, therefore, remember less. On the
other hand, one could say that the major locus
of age deficits in memory is retrieval; old
people retrieve less because they encode less
or less effectively. (Arenberg, 1980, p. 67)

In his recent review article, Smith (1980) also con-

cludes his comments on future research needs with a caution

against an "either/or" approach to encoding and retrieval

problems in old age. He calls instead for the use of

"paradigms and procedures in which the stages of

memory can be empirically separated, and yet simultaneously

examined" (p. 42).


Buschke's Restricted Reminding Procedure

One technique that was specifically designed in an

attempt to allow examination of both encoding and retrieval

operations in verbal memory is Buschke's restricted remind-

ing procedure (Buschke & Fuld, 1974). Although there are

several variations of this procedure, the basic format

involves the free recall of a list of related words (e.g.,

20 four-footed animals) over a series of 12 trials. Unlike

the conventional free recall procedure in which the entire

list is repeated before each recall trial, in this re-

stricted reminding procedure each word is re-presented

on successive trials only until it has been recalled once.

Thus, at the beginning of the second and all subsequent

trials, the examiner re-presents only those words which







-17-


have not yet been recalled at all, and the subject then

tries to recall all the words in the list. Each recall

trial is scored for a variety of measures, including over-

all recall (i.e., total number of list items recalled from

primary and secondary memory), number of items stored in

secondary memory, number of items retrieved from secondary

memory, and number of items consistently retrieved from

secondary memory on all subsequent trials.

A key feature of the restricted reminding technique is

the distinction that Buschke makes between that which has

been encoded into secondary memory and that which, on a

given trial, is retrieved from secondary memory. When an

item is recalled without presentation on that trial, it is

said to be retrieved from secondary memory. Retrieval

without presentation on that trial is believed to be from

secondary memory because interference from the presentation

and recall of other items prevents rehearsal and maintenance

of items in primary memory. A word recalled without presen-

tation on that trial is assumed to have been encoded into

secondary memory on (or before) the trial of its initial

recall (i.e., the last time that item was presented).

Thus, according to Buschke, on each trial the number of

items that have been encoded into secondary memory can be

compared with the number of items actually retrieved from

secondary memory.







-18-


It should be noted that these are operational defini-

tions that yield estimates, rather than exact determina-

tions, of amount of secondary memory encoding and retrieval

on a given trial. There are several reasons why scores for

this task should be viewed as estimates. For example, on

trials with presentations, a word from the recency portion

of the list (i.e., one of the last few items presented)

might be recalled from primary memory, secondary memory, or

both. The possible influence of primary memory on scoring

is considered minimal, however, for two reasons: (1) typi-

cally, no further presentations are required after the first

three or four trials (e.g., see Buschke, 1973; Buschke &

Fuld, 1974); and (2) words are not considered as part of

secondary memory unless they are recalled at least once

without presentation. It is also possible that a list word

later recalled without presentation may have been encoded

into secondary memory before the trial of its initial

recall (i.e., the trial from which it is counted as part of

secondary memory) but that retrieval failure prevented

demonstration of this earlier encoding. Furthermore, a

list word recalled once, only after presentation, but never

again may have been encoded into secondary memory but may

also reflect a retrieval failure. The measure of encoding

may therefore underestimate the amount of actual encoding

on a given trial.

Finally, a fundamental point to note for this task, as

for all recall tasks, is that recall performance is an







-19-


essential combination of encoding and retrieval. It is not

possible to demonstrate that a list word has been encoded

unless at some point the subject is able to retrieve it.

Measurement of encoding on this task, therefore, is predi-

cated on retrieval of list words. As with other recall

tasks, it is always possible that more encoding has oc-

curred than can be demonstrated, due to retrieval failure.

For these reasons, therefore, the measure of secondary

memory encoding obtained from the Buschke restricted re-

minding procedure should be considered a conservative,

albeit ultimately inexact, estimate of how much has been

encoded by a given trial, as defined by subsequent item

recall without presentation.

Another feature of the Buschke technique is the dis-

tinction made between consistent and random retrieval from

secondary memory. On a given trial, consistent retrieval

refers to those items that are retrieved without further

presentation on all subsequent trials. Examination of

subjects' consistent retrieval from secondary memory pro-

vides information on the nature and extent of their sub-

jective organization of the stimuli, that is, the extent to

which they are learning words as part of a "list" rather

than as single items (Buschke, 1973).

In providing separate measures of encoding and re-

trieval processes within a single verbal memory task, the

Buschke restricted reminding procedure represents a signi-

ficant departure from typical memory paradigms and measures






-20-


which either treat recall scores in a global fashion or

focus selectively on either encoding or retrieval processes.

Despite the qualification of terms required by its opera-

tional definitions of secondary memory encoding and re-

trieval, the Buschke procedure's design appears to have

great potential for allowing simultaneous examination of

encoding and retrieval processes. Thus, it seemed parti-

cularly well suited to address the unresolved issue of

encoding versus retrieval deficits in older adults' second-

ary memory and was therefore chosen for use in the present

study.

The Buschke restricted reminding procedure has been

used in a number of previous studies with younger popula-

tions, including normal adults (Buschke, 1974a, 1974b) and

children (Buschke, 1974b), alcoholic adults (Buschke &

Fuld, 1974; Todd, 1978/1979), Korsakoff patients (Fuld,

1976), and marijuana smokers (Miller, Cornett, & McFarland,

1978). Much of this work has been basically descriptive,

to demonstrate administration and scoring mechanics and

show typical performances of small samples from the various

populations. Data from Todd (1978/1979), however, indicate

acceptable reliability and validity of this procedure for

assessing adult memory dysfunction. To date, no studies

have appeared in which the restricted reminding procedure

has been used to examine the memory performance of elderly

adults.







-21-


A somewhat more widely used variation of the Buschke

procedure involves selective reminding (Buschke, 1973).

This procedure is very similar to restricted reminding,

except that on each trial the subject is reminded of all

list items not recalled on the immediately preceding trial.

Selective reminding has been employed with various popula-

tions, including normal adults (Buschke, 1973; Buschke &

Fuld, 1974) and children (Morgan, 1982), alcoholic adults

(Buschke & Fuld, 1974), and Huntington's disease patients

(Caine, Ebert, & Weingartner, 1977). Elderly adults have

also been included in a few studies with selective remind-

ing, for example an investigation of antidepressant medica-

tion effects on secondary memory in normal elderly (Bran-

connier, DeVitt, Cole, & Spera, 1982) and an investigation

of the diagnostic utility of various memory tests with

elderly SDAT (senile dementia of the Alzheimer's type)

patients (Branconnier, Cole, Spera, & DeVitt, 1982). Ober,

Koss, Friedland, and Delis (in press) also employed the

selective reminding procedure in their study of memory

functions of elderly patients with mild to moderate SDAT.

Using a modification of the selective reminding procedure

in which actual objects are presented, Fuld (1980) has

reported normative memory data for small samples of cogni-

tively intact community-dwelling and institutionalized

elderly adults.






-22-


Present Study

In the present study, the Buschke restricted reminding

procedure was used to address the question of the relative

contributions of encoding and retrieval difficulties to

elderly adults' verbal secondary memory deficit. According

to Buschke, the design of this procedure allows separation

and simultaneous measurement of encoding and retrieval

processes (Buschke, 1973; Buschke & Fuld, 1974). An iso-

lated focus on one or the other stage, typical of previous

research in this area, is thus avoided. The present study

also extended use of this procedure to an elderly, community-

dwelling sample; this is a group whose memory performance

on this task had not previously been examined. Restricted

rather than selective reminding was chosen as it allows

measurement of encoding and retrieval without any confound-

ing by continuing presentation of list items (which can be

viewed as additional encoding and/or retrieval cues) and

thus provides a more stringent estimate of initial encoding

and subsequent retrieval (Buschke & Fuld, 1974).

Degree of organization of the list of words to be

recalled was manipulated, and effects on Buschke encoding

and retrieval measures for community-dwelling elderly and

young adults were compared. Young and old adults recalled

either an unrelated list or a single-category related list

of words in the Buschke restricted reminding procedure. In

light of previous research on the facilitative effect on

recall performance achieved with more highly organized word






-23-


lists (e.g., Bower, 1970; Dallett, 1964; Ekstrand & Under-

wood, 1963; Underwood, 1983), as well as the particular

benefit reported for older adults (Laurence, 1967a), it was

expected that higher list organization, in the form of the

related list of words, would result in greater overall re-

call in both age groups, with a disproportionate effect on

the older group. Beyond this expectation, a major question

was whether such effects of list organization would be due

primarily to its facilitation of encoding processes or re-

trieval processes. That is, would high list organization

allow more items to be encoded into secondary memory or

more items to be retrieved from secondary memory?

Support for the hypothesis of an encoding deficit in

older adults would be indicated by a disproportionate in-

crease in the number of items encoded into secondary memory

for the old as compared to young adults. Alternatively,

support for a retrieval deficit in older adults would be

shown by a disproportionate increase in the number of cor-

rect retrievals from secondary memory for the old compared

to the young group. The ideal case would be one in which

there would be an interaction between age and list organi-

zation for only the encoding or retrieval measure, thus

indicating a particular age-related deficit in that stage

alone. In light of previous research suggesting some age

decrement in each stage, however, it was likely that both

encoding and retrieval measures would show an interaction






-24-


between age and list organization. If so, the magnitudes

of such interactions could then be compared, to determine

the relative deficits in these two stages of older adults'

verbal memory.

The primary analyses therefore focused on the effects

of list organization on the absolute numbers of items en-

coded and retrieved by old and young subjects. In addition,

a number of secondary analyses were conducted. First, be-

cause one might argue that the amount of retrieval is

dependent on the amount of initial encoding, an additional

retrieval measure was included for analysis. This was the

proportion (averaged across all trials) of a subject's

encoded words that were retrieved. It was expected that

analysis of this variable, which takes into account the

amount of initial encoding by a subject, would parallel

that of the primary retrieval measure.

Second, effects of list organization and age on trial-

by-trial changes in encoding and retrieval were also inves-

tigated. It was expected that an age deficit in rate of

list learning, if any, would be less in the high list

organization condition. Further support for the encoding

or retrieval deficit hypothesis would be demonstrated by a

disproportionate effect of list organization on the older

subjects' across-trials rate of encoding or retrieval,

respectively, relative to the rates of young subjects.

Third, consistent retrieval from secondary memory was

also examined, both in terms of absolute amount and increase






-25-


over trials. Based on previous research indications of

less subjective organization in older adults' verbal memory

recall (e.g., Hultsch, 1974; Macht & Buschke, 1983; Smith,

1980), it was expected that relative to the young, the

older group of subjects would show an overall smaller

amount of consistent retrieval and/or a slower rate of its

increase over successive trials. Such age differences in

consistent retrieval might be reduced in the high list

organization condition, however.

In a final secondary analysis, effects of age and list

organization on number of intrusions (i.e., responses of

words not on the list) were also examined. Little sys-

tematic attention has been given to intrusion errors among

normal elderly adults. Poon and Walsh-Sweeney (1981) re-

ported significantly more intralist intrusions for elderly

compared with young adults on a paired-associate task;

absolute numbers of such errors for each group were not

reported, however. Employing the Buschke selective remind-

ing recall task, Ober et al. (in press) found very low

rates of extralist intrusions for their normal elderly and

mild to moderate SDAT patients, with no significant differ-

ences among the groups. Fuld and others, however, have

found somewhat higher rates of word intrusions among SDAT

patients in the context of a more extended test battery and

have argued for the clinical utility of response intrusions

as a diagnostic sign of SDAT (Fuld, 1983; Fuld, Katzman,

Davies, & Terry, 1982). In their work, Fuld and her







-26-


associates specifically defined intrusions as the inappro-

priate recurrence of a response from a preceding test item,

test, or procedure. This operational definition of intru-

sions is a more specialized one than what would typically

be considered an intrusion on a recall task (i.e., any

incorrect word response, not necessarily a repetition from

a previous task).

Given the paucity of data on number of extralist in-

trusions of normal elderly adults, particularly for the

Buschke restricted reminding recall procedure, it seemed

reasonable to test the null hypothesis of no age differ-

ences on this variable in the present study. However,

there was an additional rationale for hypothesizing no age

differences in number of intrusions. That is, there are

two age-related factors whose influence on intrusions, if

any, might tend to cancel each other. These potential

factors are (1) encoding and/or retrieval difficulty per

se, and (2) cautiousness.

Intrusion errors may reflect failure of one or more of

the processes by which stimulus words are initially marked

as to-be-remembered list words and later located and

identified during recall (Anderson & Bower, 1972). For

example, a subject may think of associated words during

list presentation and erroneously encode those along with

the actual list words. Or, during the retrieval search

process an associated word may be located and incorrectly

identified as a list word. To the extent that older adults







-27-


have encoding and/or retrieval problems, therefore, they

may be expected to have more intrusions, relative to young

adults, particularly in the categorized list condition

(e.g., intrusion of related but incorrect category members).

Intrusion errors may also reflect guessing by subjects.

Relative to young adults, older adults have been found to

be more cautious in a number of laboratory situations and

to make more errors of omission on a variety of verbal

learning and recall tasks (Botwinick, 1978). A similar

cautiousness, or unwillingness to guess if not completely

sure a word was on the stimulus list, might therefore be

expected for the old subjects in this study. They would

thus be likely to make fewer intrusions, relative to the

young subjects.

Operation of both of these factors among the old sub-

jects could result in a cancelling out of each other's

effect on number of intrusions. The net result, therefore,

might be no age difference in number of intrusions. For

this reason also, therefore, it seemed appropriate to test

the hypothesis of no age differences on this variable.

A measure of subjects' estimated verbal intelligence,

that is, a vocabulary test score, was also obtained. Re-

sults of several studies have suggested the possibility of

verbal ability as a mediating factor in verbal memory task

performance. Hultsch (1969) found age-related recall

deficits and greatest benefit from specific organization

instructions among his low-verbal elderly subjects. Bowles







-28-


and Poon (1982) similarly reported a recognition memory

deficit only for their older group with lower verbal

ability. Lower vocabulary level has also been associated

with older adults' poorer recall and recognition of prose

materials, compared with those of young adults (Gordon &

Clark, 1974b; Taub, 1979). In some cases, an overall

positive relation between vocabulary level and memory

performance, regardless of age, has also been found (e.g.,

Gordon & Clark, 1974b; Hultsch, 1969). Thus, it was antici-

pated that verbal ability might be related to recall memory

performance, particularly for the older group of subjects,

in the present study. In a preliminary analysis, the

relationship between young and old subjects' vocabulary and

memory scores was explored to determine the possible impor-

tance of verbal ability as a mediating factor in addition

to age and list organization in subsequent analyses of

memory scores.

Finally, information was obtained from subjects re-

garding their memory strategies during the experimental

task and at home. Post-hoc analysis of this information,

obtained following the conclusion of the memory task, was

basically descriptive; no specific predictions of age dif-

ferences were made.







-29-


Hypotheses

To summarize, the specific hypotheses of the present

study were as follows:

Preliminary Analysis

1) Vocabulary score would be positively correlated

with dependent measures of overall recall,

secondary memory encoding and retrieval, and

consistent retrieval from secondary memory.

Also, this correlation would be greater for the

old than the young group of subjects.

Primary Analyses

1) Overall recall scores would show main effects of

age and list organization as well as an age by

list interaction. Specifically, recall would be

greater for young versus old and related versus

unrelated list groups, and age deficits would be

less in the related list condition.

2) Secondary memory encoding scores would show main

effects of age and list as well as an age by list

interaction. More list words would be encoded by

young versus old and in the related versus unre-

lated list condition, and age deficits would be

less in the related list condition.

3) Secondary memory retrieval scores would show

main effects of age and list as well as an age by







-30-


list interaction. More list words would be

retrieved by young versus old and in the related

versus unrelated list condition, and age deficits

would be less in the related list condition.

4) Relatively greater age deficits would be found

in either secondary memory encoding or retrieval

scores. The direction of this relatively greater

age difference was not specified a priori, as it

represented an unresolved issue that the study

was intended to help clarify.


Secondary

5)


Analyses

Analysis of the additional retrieval measure,

proportion of words encoded into secondary memory

that are retrieved, would parallel that of the

primary retrieval measure. That is, a greater

proportion of encoded words would be retrieved by

young versus old and in the related versus un-

related list condition, and age deficits would be

less in the related list condition.


6) Consistent retrieval scores would show main

effects of age and list as well as an age by list

interaction. Specifically, more list words would

be retrieved consistently by young versus old and

in the related versus unrelated list condition,

and age deficits would be less in the related

list condition.







-31-


7) Age by list by trial interactions would be found

for secondary memory encoding, retrieval, and

consistent retrieval scores. Specifically, old

adults' scores on each of these measures across

trials would increase more slowly than those of

young adults, but such age deficits would be less

in the related list condition.

8) There would be no difference between the two age

groups in mean number of intrusions.
















METHOD


Subjects

There were 104 subjects, including 52 young adults

(Young group) and 52 older adults (Old group). There were

equal numbers of males and females in each age group as

well.

Young adult volunteers were introductory psychology

students at Santa Fe Community College in Gainesville,

Florida, who received extra course credit for their parti-

cipation in this study. The mean age of the Young sample

was 21.4 years (range 17-34 years). Mean level of education

was 13.4 years (range 12-16 years).

Older adult volunteers were recruited mainly from

groups of local retired educators and federal employees who

had previously expressed willingness to be contacted for

participation in research at the University of Florida in

Gainesville. Individuals in these groups were contacted by

letter and telephone, and those aged 60 or older and living

in the community (i.e., non-institutionalized) were asked

to participate. A few subjects were recruited from spouses

and friends of those individuals initially contacted. The

mean age of the Old sample was 70.2 years (range 61-80


-32-







-33-


years). Mean level of education was 16.6 years (range

12-20 years).

Using a 4-point scale of Excellent, Good, Fair, or Poor

health, 58% of the total sample of subjects (48% of the

Young group and 69% of the Old group) rated themselves as

in good health. Thirty-six percent of the sample (48% of

the Young and 25% of the Old) rated themselves as in excel-

lent health. Thus, 95% of the total sample (96% of the

Young, 94% of the Old) reported themselves in good or

excellent health.


Materials

A Background Information Sheet (see Appendix A) was

developed for subjects to record personal demographic

information including age, sex, years of education, occupa-

tion (former occupation, if retired), major illnesses,

chronic medical conditions, and current medications. Also

included was the 4-point self-rated scale of overall health,

as described above.

The Wechsler Adult Intelligence Scale--Revised (WAIS-R)

Vocabulary subtest (Wechsler, 1981) was used to obtain an

estimate of verbal intelligence for each subject. A copy of

the form used for recording and scoring subjects' responses

is given in Appendix A. A 4" x 6" white card with the list

of vocabulary words printed on it was also provided for the

subjects' visual inspection during this task.






-34-


Two different word lists were employed for the memory

task (see Table 1). The Related List contained 20 related

nouns from a single concept category. These were the 20

four-footed animal names from the long form of the Buschke

restricted reminding recall procedure (Buschke, 1973). The

Unrelated List contained 20 unrelated nouns which the pres-

ent investigator matched with the 20 related nouns for ini-

tial letter, word length, and frequency of usage in the

English language (Carroll et al., 1971; Thorndike & Lorge,

1944). Each stimulus word was printed in 1/2" high black

letters on an individual 3" x 5" white card. Two decks of

stimulus cards were then constructed, one containing the

Related List and one containing the Unrelated List of words.

See Appendix A for an example of these stimulus word cards.

An examiner scoring sheet was also developed for use with

each of the two word lists (see Appendix A).

In addition, a Post-Test Inquiry form was developed,

on which the examiner recorded subjects' reports of (1)

memory strategies during the experiment and in daily life,

(2) previous attempts to improve memory, and (3) previous

participation in memory research (see Appendix A).


Experimental Conditions and Procedures

Half of each age group (Young, Old) was assigned to

each of the two word list conditions, Low Organization

(Unrelated List) and High Organization (Related List).

Subject assignment to list condition was random, with







-35-


TABLE 1

Word Lists Employed in the Buschke Memory Task


Unrelated List


Related List

Dog
Fox
Horse
Buffalo
Lion
Rhinoceros
Elephant
Antelope
Bear
Lamb
Rat
Raccoon
Sheep
Llama
Goat
Cheetah
Squirrel
Beaver
Donkey
Turtle


Door
Frog
Hole
Battery
Lamp
Rectangle
Envelope
Abdomen
Ball
Lunch
Rug
Ranger
Sword
Lapel
Ghost
Chestnut
Spider
Banner
Daisy
Tractor






-36-


groups balanced for sex. Thus, there were four separate

groups of 26 subjects, with 13 males and 13 females in

each: Young/Low List Organization; Young/High List Organiza-

tion; Old/Low List Organization; Old/High List Organization.

All subjects were tested individually in a single

session each. Subjects in the Young group were tested in

an office on the Santa Fe Community College campus. Sub-

jects in the Old group were tested in an office on the

University of Florida campus. All testing was conducted by

the same examiner (i.e., the present investigator).

Subjects first read and signed an Informed Consent

Form (see Appendix B) which briefly explained the purpose

and nature of the study. They then completed the Background

Information'Sheet. The examiner assisted subjects if they

had questions or difficulties with individual items.

Next the examiner administered the WAIS-R Vocabulary

subtest. Standard instructions, administration procedures,

and scoring criteria were employed, as described in the

WAIS-R test manual (Wechsler, 1981).

After a brief rest break if desired by the subject,

the Buschke restricted reminding verbal recall task was

begun. The task was presented with the following instruc-

tions by the examiner:

Now I'm going to read you a list of words. As
I say each word, I'll show it to you on one of
these cards. It's a long list, but I want you to
listen carefully and just try to remember as many
words as you can. After I've read them, I'm
going to ask you to tell me all the words you







-37-


can remember, in any order. We're going to do
this 12 times. O.K., now listen carefully and
try to remember as many words as you can.

Depending on the subject's assigned list condition,

either the Unrelated or the Related word list was employed.

On the first trial the examiner presented all 20 words both

auditorially and visually.1 Words were read aloud to the

subject at a rate of one item every 4 seconds.2 As each

word was read, it was simultaneously displayed on a 3" x 5"

white card which remained on view until the next stimulus

word was read (4 seconds). On all subsequent trials, words

were presented auditorially only, at the same 4-second rate.

After all 20 items had been presented on Trial 1, the

examiner immediately began the recall period with these

instructions:

Now tell me as many of the words as you can,
in any order.

Several prompts were used to encourage subjects to extend

their recall. For all subjects, after the first 20-second

pause in responding, the examiner said:

Try a little longer. See if you can think of
some more.

After each additional response, the examiner prompted with

one of the following comments:

How about one more? or Try to get one more.

The recall period was terminated after the subject had

failed to respond for 20 seconds following one of these

examiner prompts.







-38-


Each subsequent trial began with the re-presentation

of only those words that had not yet been recalled at all.

The examiner prefaced such trials as follows:

Now I'll remind you of the ones you haven't said
yet at all.

After the reminders were given, the examiner again imme-

diately asked subjects to recall as many words as possible

from the entire list:

Now tell me all the words you can, from the
whole list.

When all list items had been recalled once and there were

no further re-presentations, trials began with the following

instructions from the examiner:

Now start over, and tell me all the words you
can, from the whole list.

There were 12 recall trials in all. Recall periods on

Trials 2-12 were terminated as described above for Trial 1.

Subjects were not told the total number of words in the

list nor whether their responses were correct or incorrect.


Scoring

Vocabulary Test. The WAIS-R Vocabulary subtest was

scored according to the response criteria in the WAIS-R

test manual. Raw scores, with a maximum possible range of

0-70, were used in the data analyses.

Buschke Memory Task. During the Buschke restricted

reminding memory task, the examiner recorded all correct







-39-


list word responses with a simple hatch-mark in the appro-

priate row and column of the score sheet. Intrusions were

likewise indicated by a hatch-mark in the "Intrusions"

section of the score sheet. In addition, each subject's

performance on this task was audio-taped. The tapes were

later transcribed by a research assistant, who recorded

the order of responses and the actual names of intruded

(non-list) words.

On each trial of the Buschke procedure, each subject's

performance was scored for the following measures:

1) Number of correct recalls (from primary and

secondary memory).

2) Number of list words encoded in secondary memory.

3) Number of correct retrievals from secondary

memory.

4) Proportion of words encoded in secondary memory

that were retrieved.

5) Number of words recalled consistently on all sub-

sequent trials.

6) Number of intrusions (i.e., responses of words

not on the list).

In addition, the following measures summed across all

12 trials were computed for each subject:

1) Total number of correct recalls (RECALL).

2) Total number of words encoded into secondary

memory (LTS).4






-40-


3) Total number of correct retrievals from secondary

memory (LTR).

4) Total number of words retrieved consistently from

secondary memory (CLTR).

5) Total number of intrusions (INTRUSIONS).

Each subject's average proportion of encoded words

that were retrieved (LTR/LTS) was also computed from the

trial data.

Scoring for the trial measures of correct recalls,

words encoded in secondary memory, words retrieved from

secondary memory, and words recalled consistently followed

directions provided by Buschke and Fuld (1974).

RECALL (total number of correct recalls) was computed

by summing the number of correct recalls across all 12

trials. Possible scores on this measure range from 0 to

240.

LTS (total number of words encoded into secondary

memory) was computed by summing the number of words re-

called at least twice over the 12 recall trials. Possible

scores range from 0 to 20.

LTR (total number of correct retrievals from secondary

memory) was computed by summing the total number of correct

recalls for all words recalled at least twice. Possible

scores on this measure range from 0 to 240.

CLTR (total number of words retrieved consistently

from secondary memory) was computed by summing the number






-41-


of words which, from any given trial, were then correctly

recalled on every subsequent trial. Possible scores range

from 0 to 20.

INTRUSIONS (total number of intrusions) was computed

by summing the total number of incorrect responses (words

not on the list) across the 12 trials. There was no upper

limit on this measure.

An example of a completed score sheet is given in

Appendix C.


Summary of Design

The experimental design of the present study is sum-

marized in Table 2, indicating the independent and dependent

variables which were investigated.


Data Analyses

Means and standard deviations were obtained for WAIS-R

Vocabulary scores and for all dependent measures. Relation-

ships between Vocabulary score and the dependent measures

were investigated in each age group by means of Pearson

correlation coefficients. Effects of age and list condition

on each of the primary dependent variables RECALL, LTS, and

LTR were tested by separate 2(Age) x 2(List) analyses of

variance (ANOVAs). The partial omega squared measure of

effect size (w2 partial) was used to determine the amount

of variance in LTS and LTR accounted for by each effect in

the model. Effects of age and list condition on each of







-42-


TABLE 2

Summary of Experimental Design


Independent Variables: AGE (Young/Old)
LIST ORGANIZATION (Low/High)

Additional Subject Variable: Vocabulary Score

Dependent Variables:

a. Primary

1. Total correct recall (RECALL)
2. Total number of words encoded into secondary
memory (LTS)
3. Total number of words retrieved from secondary
memory (LTR)

b. Secondary

1. Average proportion of words encoded into
secondary memory that are retrieved (LTR/LTS)

2. Cumulative curve of number of words encoded in
secondary memory (LTS over trials)

3. Trial-by-trial curve of number of retrievals
from secondary memory (LTR over trials)

4. Total number of words retrieved consistently
from secondary memory (CLTR)

5. Cumulative curve of number of words consis-
tently retrieved from secondary memory (CLTR
over trials)


6. Total number of intrusions (INTRUSIONS)







-43-


the secondary dependent variables LTR/LTS, CLTR, and INTRU-

SIONS were also determined by separate 2(Age) x 2(List)

ANOVAs. The patterns of age and list groups' performances

on LTS, LTR, and CLTR across trials were examined by means

of separate 2(Age) x 2(List) x 12(Trial) ANOVAs on these

dependent measures.

Several less formal, post-hoc analyses were conducted

with data obtained from the Post-Test Inquiry. Subjects'

reports of memory strategies used during the experiment and

in their daily lives were categorized, and the frequency

with which various strategies were used in each age group

was determined. In addition, age group differences in

reports of using the same memory strategies in daily life

as during the experiment were investigated by chi-square

(x2) analysis.

Notes

1. The standard procedure for the Buschke restricted re-
minding task involves only auditory presentation of
list words. Visual presentation was added on the
first trial in the present study to ensure correct
initial perception of all stimulus words in the event
of any mild hearing difficulty among subjects, parti-
cularly among the older individuals.

2. Buschke's original description of the restricted and
selective reminding procedures (Buschke, 1973) involved
presentation of 20 list words at a 2-second rate over
12 trials. Subsequently, however, Buschke and other
investigators have used various numbers of list items
and trials and various presentation rates. Although
the 2-second presentation rate has been most typical
in studies with young adults, Miller et al. (1978)
reported a 3-second rate in their study of young
adults, and Fuld (1980) used a 5-second rate in her
study of elderly adults. In some preliminary pilot







-44-


testing conducted before the present study was begun,
a few young and old subjects were tested with a 2-,
4-, or 5-second rate of presentation. The 4-second
rate was selected for the present study as it appeared
to provide adequate time for subjects to engage in
various encoding strategies without resulting in
either a floor or ceiling effect on scores.

3. Buschke (1973) and Buschke and Fuld (1974) have empha-
sized the importance of providing subjects with suffi-
cient time and encouragement to achieve maximum recall.

4. Buschke's original acronyms of LTS, LTR, and CLTR
(referring to long-term storage, long-term retrieval,
and consistent long-term retrieval, respectively) have
been retained to facilitate comparison of the present
results with those of other studies in which the
Buschke procedure is employed. The equivalent terms
to which these acronyms refer in the present study
are secondary memory encoding, secondary memory re-
trieval, and consistent secondary memory retrieval,
respectively.

5. Partial omega squared was used as the measure of
amount of variance accounted for by a given effect.
This measure is computed as follows:


2 SSeffect dfeffect x (ISerror)l
S partial = -- ------------
SSeffect + (N dfeffect x MSerror]


Partial omega squared measures the strength of an ef-
fect independent of the strengths of other effects in
a factorial design (i.e., with the other factors'
effects partialed out). In contrast, the usual omega
squared measure of strength for a particular effect
depends on the strengths of other effects in the de-
sign (Maxwell, Camp, & Arvey, 1981). In the present
study, partial omega squared was considered more
appropriate than the usual omega squared measure be-
cause various effect sizes were to be compared with
each other. In particular, a measure of the Age x
List interaction effect sizes, independent of the main
Age and List factor effects, was desired for the com-
parison of Age x List interactions on the encoding and
retrieval measures.















RESULTS


Preliminary Analysis of Vocabulary Scores

Vocabulary and Age

WAIS-R Vocabulary score means, standard deviations,

and minimum and maximum values for the two age groups are

presented in Table 3. A t-test of mean differences revealed

that the mean Vocabulary score of the Old group (M = 59.5)

was significantly higher than that of the Young group (M =

46.5), t(102) = 5.12, p<.005. This difference is consistent

with the generally higher level of education of the Old

group in this study.

Vocabulary and Dependent Measures

It was expected that higher verbal ability, indicated

by higher WAIS-R Vocabulary scores, would be correlated

with better performance on the dependent measures of over-

all recall, secondary memory encoding and retrieval, and

consistent retrieval from secondary memory. It was further

expected that this relationship would be stronger in the

Old group than in the Young group. To test this hypothe-

sis, Pearson correlation coefficients were calculated

between VOCABULARY and RECALL, LTS, LTR, LTR/LTS, and CLTR

in the Total sample and in the two Age groups and List

groups separately. Matrices depicting the relationships


-45-







-46-


TABLE 3

Means, Standard Deviations, Minimum and Maximum Values of
WAIS-R Vocabulary Scores for Young and Old Age Groups


Mean

Standard Deviation

Minimum Value

Maximum Value


AGE GROUP

Younga Olda

46.50 59.50

9.48 6.50

28 41

64 68


an=52.

bPossible scores range from 0 to 70.







-47-


among these variables by Age and List groups are presented

in Tables 4-7. Correlations among the dependent measures

are also included in the tables for descriptive purposes.

Contrary to the prediction, Vocabulary score was not

significantly related to any of the dependent variables in

either of the two Age groups or the two List groups. There-

fore, Vocabulary was not included as a variable in any fur-

ther analyses.


Primary Analyses

Hypothesis 1: Overall Recall

Means, standard deviations, and minimum and maximum

values of RECALL scores for both Age and List groups are

given in Table 8.

Hypothesis 1 predicted both Age and List main effects

as well as an Age x List interaction on this measure. A

2(Age) x 2(List) ANOVA on RECALL was performed, with results

as presented in Table 9.

As predicted there were significant main effects of Age

(F ,100 = 13.34, p<.001, 2 partial = .11) and List (F1,100
= 27.84, p<.0001, w2 partial = .20), and a significant Age x

List interaction (F1,100 = 6.76, p<.02, w2 partial = .05).

Tests of simple effects revealed the following: Old sub-

jects performed significantly better in the Related List

condition than in the Unrelated List condition (Related List

M = 175.0, Unrelated List M = 127.2, p<.0001). Young sub-

jects, however, did not differ significantly in the two List







-48-


TABLE 4

Correlation Matrix for Vocabulary and Memory
Scores in Young Group (n=52)



VOCABa RECALL LTSa LTR LTR/LTS CLTR



.10 ---
(.245)

.07 .90*
(.318)

.09 .99* .92*
(.252)

.02 .56* .17 .53*
(.453) (.114)

.09 .92* .83* .92* .55*


(.269)


probability (p) values are
the associated correlation


given in parentheses below
coefficients.


*p<.001.


VOCAB

RECALL


LTS


LTR


LTR/LTS


CLTR







-49-


TABLE 5

Correlation Matrix for Vocabulary and Memory
Scores in Old Group (n=52)


VOCABa


.12
(.206)

.12
(.202)

.11
(.211)

.10
(.243)

.13
(.181)


RECALL


.92*


.99*


.56*


.95*


LTSa


LTR


LTR/LTS CLTR


.93*


.22
(.054)

.85*


.54*


.95*


.62*


aprobability (p) values are
the associated correlation


given in parentheses below
coefficients.


*p<.001.


VOCAB

RECALL


LTS


LTR


LTR/LTS


CLTR






-50-


TABLE 6

Correlation Matrix for Vocabulary and Memory
Scores in Unrelated List Condition (n=52)


LTR


.69*


.97*


LTR/LTS CLTR


.70*


probability (p) values are
the associated correlation


given in parentheses below
coefficients.


*p<.001.


VOCABa


VOCAB

RECALL


LTS


LTR


LTR/LTS


CLTR


RECALL


.94*


.99*


.71*


.97*


LTS


.95*


.47*


.92*


- .20
(.078)

-.18
(.106)

-.20
(.077)

-.16
(.129)

-.19
(.087)







-51-


TABLE 7

Correlation Matrix for Vocabulary and Memory
Scores in Related List Condition (n=52)


VOCABa


-.08
(.295)

.03
(.410)

-.07
(.305)

-.15
(.140)

-.06
(.341)


RECALL


.77*


.99*


.74*


.87*


LTSa


LTR


LTR/LTS CLTR


.80*


.16
(.122)

.65*


.71*


.86*


.70*


probability (p) values are
the associated correlation


given in parentheses below
coefficients.


*p<.001.


VOCAB

RECALL


LTS


LTR


LTR/LTS


CLTR







-52-


TABLE 8

Means, Standard Deviations, Minimum and Maximum Values
of RECALL Scores for Age and List Groups


Mean

Standard Deviation

Minimum Value

Maximum Value


AGE GROUPa
Old Young

151.08 173.23

40.24 30.09

65 82

222 229


LIST GROUPa
Unrelated Related

146.15 178.15

40.73 24.36

65 126

229 225


an=52 for each group.

bPossble scores range from 0 to 240.
Possible scores range from 0 to 240.







-53-


TABLE 9

2(Age) x 2(List) ANOVA on RECALL Scores


Source

Age

List

Age x List

Error


SS

12760.60

26624.00

6465.38

95645.54


df

1

1

1

100


MS

12760.60

26624.00

6465.38

956.46


F

13.34

27.84

6.76


p

.0004

.0000

.0107


2 partial


.11

.20

.05







-54-


conditions (Related List M = 181.3, Unrelated List M =

165.1, p>.06). Furthermore, performance of the two Age

groups differed significantly only in the Unrelated List

condition (Young M = 165.1, Old M = 127.2, p<.0001), not in

the Related List condition (Young M = 181.3, Old M = 175.0,

E>.4). Thus, the age decrement in overall RECALL found in
the Unrelated List condition was eliminated in the Related

List condition. These results are displayed graphically in

Figure 1.

Hypothesis 1 was thus supported insofar as the pre-

dicted main effects and interaction were found. Analysis

of the interaction showed the need for some qualification,

however. Whereas the Young as a whole recalled more than

the Old, this age difference was accounted for by subjects

in the Unrelated List condition only. There was no signifi-

cant age difference in the Related List condition. In addi-

tion, RECALL was significantly better in the Related versus

Unrelated List condition for the Old only; the Young showed

no significant difference in RECALL between the two List

conditions.

Hypothesis 2: Secondary Memory Encoding

Means, standard deviations, and minimum and maximum

values of LTS scores for both Age and List groups are given

in Table 10.

Hypothesis 2 predicted Age and List main effects and

an Age x List interaction. A 2(Age) x 2(List) ANOVA was







-55-


Key:

S--- = YOUNG group
A- --A = OLD group


........... ......./ ..


Related


LIST CONDITION


Figure 1. Age x List interaction on RECALL scores.








-56-


TABLE 10

Means, Standard Deviations, Minimum and Maximum Values
of LTS Scores for Age and List Groups


Meanb

Standard Deviation

Minimum Value

Maximum Value


AGE GROUPa
Old Young

15.44 16.73

3.74 2.61

7 8

20 20


LIST GROUPa
Unrelated Related

14.25 17.92

3.43 1.74

7 12


n=52 for each group.

Possible scores range from 0 to 20.







-57-


performed, with results as presented in Table 11. The

findings were as predicted, with significant effects of Age

(F i = 6.62, p<.02, wa partial = .05), List (F ,,00 =

53.78, p<.0001, w2 partial = .34), and an Age x List inter-
2
action (F, 100 = 9.20, p<.004, 2 partial = .07).

Tests of simple effects revealed the following: Both

Young and Old groups encoded significantly more list words

in the Related List condition than in the Unrelated condi-

tion (Young Unrelated List M = 15.6, Related List M =

17.8, p<.01; Old Unrelated List M = 12.8, Related List M =

18.0, p<.0001). However, the two Age groups differed

significantly only in the Unrelated List condition (Young M

= 15.6, Old M = 12.8, p<.0001), not in the Related List

condition (Young M = 17.8, Old M = 18.0, p>.7). As shown

in Figure 2, in the Related List condition the age decre-

ment in LTS was completely eliminated.

Thus, Hypothesis 2 was supported, in that the predicted

main effects and interaction were obtained. Analysis of

the Age x List interaction, however, indicated that the age

deficit in encoding scores was attributable to the Unrelated

List condition only; the Old did not differ significantly

from the Young in number of words encoded in the Related

List condition.

Hypothesis 3: Secondary Memory Retrieval

Means, standard deviations, and minimum and maximum

values of LTR scores for both Age and List groups are

given in Table 12.







-58-


TABLE 11

2(Age) x 2(List) ANOVA on LTS Scores


Source

Age

List

Age x List

Error


SS

43.16

350.78

60.01

652.27


df

1

1

1

100


MS

43.16

350.78

60.01

6.52


F

6.62

53.78

9.20


p

.0116

.0000

.0031


2
w2 partial

.05

.34

.07







-59-


Key:

- 0 = YOUNG group
A- --A = OLD group


Unrelated


Related


LIST CONDITION


Age x List interaction on LTS (encoding] scores.


14 -


12 -


10


Figure 2.







-60-


TABLE 12

Means, Standard Deviations, Minimum and Maximum Values
of LTR Scores for Age and List Groups


Mean

Standard Deviation

Minimum Value

Maximum Value


AGE GROUPa
Old Young

146.58 170.02

43.62 32.44

52 70

222 229


LIST GROUPa
Unrelated Related

140.52 176.08

43.90 25.75

52 118

229 225


an=52 for each group.

Possible scores range from 0 to 240.







-61-


Hypothesis 3 predicted Age and List main effects as

well as an Age x List interaction on this measure of re-

trieval from secondary memory. A 2(Age) x 2(List) ANOVA

was performed, with results as presented in Table 13. As

predicted, there were significant effects of Age (F1,100

12.97, E<.001, 2 partial = .10), List (F1,100 = 29.85,

p<.0001, w2 partial = .22), and the Age x List interaction

(F, 00 = 6.99, p<.01, 2 partial = .05).

Tests of simple effects revealed the following: Both

Young and Old groups retrieved significantly more list

words in the Related List condition than in the Unrelated

List condition (Young Unrelated List M = 160.8, Related

List M = 179.2, p<.05; Old Unrelated List M = 120.2, Re-

lated List M = 173.0, p<.0001). However, the two Age

groups differed significantly only in the Unrelated List

condition (Young M = 160.8, Old M = 120.2, p<.0001), not in

the Related List condition (Young M = 179.2, Old M = 173.0,

p>.4). Thus, the age decrement in LTR was eliminated in

the Related List condition, as shown in Figure 3.

Hypothesis 3 was therefore supported in the same

manner as was Hypothesis 2: The predicted main effects and

interaction were found; however, the Age x List interaction

was such that the age difference was attributable to the

Unrelated List condition only. In the Related List condi-

tion, Old and Young did not differ significantly in number

of words retrieved.







-62-


TABLE 13

2(Age) x 2(List) ANOVA on LTR Scores


Source SS df MS F partial

Age 14288.10 1 14288.10 12.97 .0005 .10

List 32873.10 1 32873.10 29.85 .0000 .22

Age x List 7702.16 1 7702.16 6.99 .0095 .05

Error 110138.42 100 1101.38 --- --- ---







-63-


Key:

S--- = YOUNG group
A- -A = OLD group


Unrelated


Related


LIST CONDITION


Figure 3. Age x List interaction on LTR (retrieval) scores.


200-


190-


180-


170-


160-


150-


140-


130-


120-


110-


100-


/


! !










Hypothesis 4: Encoding and Retrieval Compared

Both encoding (LTS) and retrieval (LTR) scores showed

main effects of Age and List and an Age x List interaction.

For both LTS and LTR, List condition accounted for the

greatest amount of variance (w2 partial = .34 and .22 for

LTS and LTR, respectively). Age accounted for 5% of the

variance in LTS scores and 10% of the variance in LTR

scores. Most importantly, the Age x List interaction ac-

counted for 7% of the variance in LTS and 5% of the vari-

ance in LTR scores. Furthermore, tests of simple effects

showed that the LTS age decrement found in the Unrelated

List condition was completely eliminated in the Related

List condition. For LTR, the results were essentially the

same, with a statistically significant age difference in

the Unrelated List condition only.

The results thus far provide support for both an en-

coding and a retrieval deficit in older adults when material

to be recalled has a low level of organization. With the

percent of variance accounted for ( 2 partial) by the Age x

List interaction (i.e., the Age x List effect size indepen-

dent of any other effects) as an index of the magnitude of

the age deficit, the present data indicate encoding and

retrieval deficits of virtually equal magnitudes (w2 par-

tial = .07 and .05 for LTS and LTR, respectively). Thus,

at this point Hypothesis 4, predicting a greater age de-

ficit in either encoding or retrieval, was not directly

supported.







-65-


It is important to note that scores on the LTS and LTR

measures were found to be highly correlated in both age

groups (r = .92 and .93 between the two measures in the

Young and Old group, respectively). Furthermore, both of

these measures were found to be highly correlated with the

overall RECALL measure (in the Young group, r = .90 between

RECALL and LTS, r = .99 between RECALL and LTR; in the Old

group, r = .92 between RECALL and LTS, r = .99 between RE-

CALL and LTR). Partial correlations between LTS and LTR,

with the effect of RECALL removed, remained significant

(r = .47 and .35 between the two measures in the Young and

Old group, respectively; p<.05 in each case).

The actual independence of the LTS and LTR measures,

therefore, seems highly questionable. Both LTS and LTR

scores may in fact measure the same process rather than two

separate memory processes. Alternatively, they may reflect

separate but highly interdependent processes. In either

case, the finding of essentially equal-magnitude age defi-

cits (i.e., Age x List interactions) on the two measures may

simply be an artifact of the high correlation between the

two scores.

Thus, the use of LTS and LTR scores for a direct com-

parison of age deficits in encoding and retrieval may not

be most appropriate. In addition to these primary analyses,

however, several secondary analyses were conducted in an

effort to shed further light on the encoding and retrieval

processes of the Young and Old subjects.







-66-


Secondary Analyses


Hypothesis 5: Proportion of Words Retrieved from Storage

LTR/LTS represents an additional retrieval measure, one

which takes into account, or attempts to control for, the

amount of initial encoding by subjects. It is the average

proportion (across all 12 trials) of encoded list words

that were retrieved on a given trial. Means, standard de-

viations, and minimum and maximum values of LTR/LTS scores

for both Age and List groups are given in Table 14. It may

also be noted that the correlation between encoding (LTS)

and this retrieval measure was non-significant in both age

groups (r = .17 and .22 between LTS and LTR/LTS in the Young

and Old group, respectively, p>.05; see Tables 4 and 5).

Hypothesis 5 predicted main effects of Age and List as

well as an Age x List interaction on this measure. A 2(Age)

x 2(List) ANOVA was performed, with results as presented in

Table 15.

Hypothesis 5 was only partially supported. Contrary

to prediction, neither the List nor the Age x List effect

was significant. However, there was a statistically signi-

ficant main effect of Age (F1,100 = 10.83, p<.002, 2 par-

tial = .09). Old subjects retrieved a somewhat smaller

average proportion of their encoded list words than did

Young subjects, regardless of List condition (Old M = .83,

Young M = .88). In addition, performance on this retrieval

measure did not differ between the two List conditions for







-67-


TABLE 14

Means, Standard Deviations, Minimum and Maximum Values
of LTR/LTS Scores for Age and List Groups


AGE GROUPa,b
Old Young


Standard Deviation


Minimum Value

Maximum Value


.08

.66

.99


.88

.06

.74

.99


LIST GROUPa,b


Unrelated

.86

.08


.66

.99


a
n=52 for each group.

All scores expressed as percentages. Possible values
range from 0.00 to 1.00.


Mean


Related

.85

.07







-68-


2(Age) x 2(List)


TABLE 15

ANOVA on


LTR/LTS Scores


Source

Age

List

Age x List

Error


SS

.05678

.00432

.00270

.52424


df

1

1

1

100


MS

.05678

.00432

.00270

.00524


F

10.83

.82

.52


p

.0014

.3664

.4746


2
w2 partial

.09


--


--







-69-


either the Young or Old group. These results are displayed

graphically in Figure 4.

Thus, even when amount of initial encoding was taken

into account, Old adults still showed a retrieval deficit

in comparison with Young adults, regardless of the degree

of organization of the list to be remembered.

Hypothesis 6: Consistent Secondary Memory Retrieval

Means, standard deviations, and minimum and maximum

values of CLTR scores for both Age and List groups are

given in Table 16.

Hypothesis 6 predicted main effects of Age and List,

and an Age x List interaction. A 2(Age) x 2(List) ANOVA

was performed, with results as presented in Table 17. As

predicted, there was a significant Age effect (F,100 =

11.48, p<.002, w2 partial = .09), List effect (F,100 =

14.07, p<.0005, w2 partial = .11), and an Age x List inter-

action (F ,100 = 4.23, p<.05, w2 partial = .03).

Tests of simple effects revealed the following: Old

adults showed a significantly greater amount of consistent

retrieval in the Related List compared to the Unrelated

List condition (Unrelated List M = 10.4, Related List M =

14.1, p<.0001). Young adults, however, did not differ in

amount of consistent retrieval in the two List conditions

(Unrelated List M = 13.9, Related List M = 15.0, p>.2).

Furthermore, the two Age groups differed significantly

in the Unrelated List condition only (Young M = 13.9, Old






-70-


Key:

- = YOUNG group
A-- --A = OLD group


Unrelated Related


LIST CONDITION


Figure 4. Results of 2 x 2 ANOVA on LTR/LTS scores.


A- -------







-71-


TABLE 16

Means, Standard Deviations, Minimum and Maximum Values
of CLTR Scores for Age and List Groups


Meanb

Standard Deviation

Minimum Value

Maximum Value


AGE GROUPa
Old Young

12.27 14.42

3.83 3.11


LIST GROUPa
Unrelated Related

12.15 14.54

4.09 2.66

4 8

20 20


an=52 for each group.

Possible scores range from 0 to 20.







-72-


TABLE 17

2(Age) x 2(List) ANOVA on CLTR Scores


Source

Age

List

Age x List

Error


SS

120.615

147.846

44.462

1050.620


df

1

1

1

100


MS

120.615

147.846

44.462

10.506


F

11.48

14.07

4.23


p

.0010

.0003

.0423


w partial

.09

.11

.03


--







-73-


M = 10.4, p<.0002), not in the Related List condition

(Young M = 15.0, Old M = 14.1, p>.3). These results are

displayed graphically in Figure 5.

Thus, Hypothesis 6 was supported in that the predicted

main effects and interaction were found. Analysis of the

interaction, however, showed that the Age difference was

attributable to the Unrelated List condition only and that

the List difference was attributable to the Old group only.

Hypothesis 7: Encoding and Retrieval Across Trials

The trial-by-trial performance of subjects was exa-

mined for three variables: LTS, LTR, and CLTR. For each

variable, a 2(Age) x 2(List) x 12(Trial) ANOVA was per-

formed, with Trial as a within-subject variable. In each

case an Age x List x Trial interaction was predicted (i.e.,

a different List x Trial interaction for the two Age

groups). Results for each variable will be presented

separately.

Encoding. Results of the 2(Age) x 2(List) x 12(Trial)

ANOVA on LTS scores are presented in Table 18. There were

significant main effects of Age (F,1200 = 87.12, p<.0002),

List (F ,1200 = 646.72, p<.0002), Trial (F l ,1200 = 44.20,

p<.0002), and an Age x List interaction (1,1200 = 112.76,

p<.0002). Contrary to prediction, however, there was no

Age x List x Trial interaction. Neither was there an Age x

Trial nor a List x Trial interaction. Hypothesis 7 was

therefore not supported for this variable. The pattern of







-74-


Key:

- = YOUNG group
A- --A = OLD group


Unrelated


Related


LIST CONDITION




Figure 5. Age x List interaction on CLTR (consistent
retrieval) scores.


20 1


i i


. .







-75-


TABLE 18

2(Age) x 2(List) x 12(Trial) ANOVA on LTS Scores



Source df SS F p

Age 1 629.00 87.12 .0001

List 1 4669.39 646.72 .0001

Trial 11 3510.27 44.20 .0001

Age x List 1 814.15 112.76 .0001

Age x Trial 11 10.42 .13 .9995

List x Trial 11 52.19 .66 .7809

Age x List x Trial 11 5.15 .06 1.0000

Error 1200 8664.08 ---






-76-


LTS scores across trials was not significantly different

for Young and Old subjects. The pattern across trials did

not differ significantly for the two List conditions,

either.

Follow-up tests of Trial mean differences were con-

ducted for the full sample (N=104). The pre-planned set of

all possible pair-wise comparisons was tested by means of

Fisher's least significant difference (LSD) procedure.

Results indicated significant increases in LTS scores on

Trials 2 and 3, after which performance leveled off.

Figure 6 displays these results.

Retrieval. Results of the 2(Age) x 2(List) x 12(Trial)

ANOVA on LTR scores are presented in Table 19. There were

significant main effects of Age (F1200 = 123.19, p<.0002),

List (F ,1200 = 297.81, p<.0002), Trial (F11,1200 = 12.16,

p<.0002), and an Age x List interaction (F1,1200 = 71.62,

p<.0002). There was no Age x List x Trial interaction, no

Age x Trial interaction, nor a List x Trial interaction,

however. Thus, the pattern of LTR scores across trials was

not significantly different for Old versus Young adults nor

for Unrelated versus Related List conditions. Again,

Hypothesis 7 was not supported.

Fisher's LSD procedure was again used to test the pre-

planned set of all possible pair-wise comparisons of trial

means for the combined sample (N=104). Results indicated

significant increases in LTR scores on Trials 2 and 3.






-77-


Key:
S-- = YOUNG/Related List
A---A = OLD/Related List
o----o = YOUNG/Unrelated List
-- -6 = OLD/Unrelated List


/


0 1 I I I I I I I I I
1 2 3 4 5 6 7 8 9 10 11 12
TRIAL



Figure 6. LTS (encoding) trial means for the four
experimental groups.


1_ -- ---L- -A -6- -A- A- --A-
P-






-78-


TABLE 19

2(Age) x 2(List) x 12(Trial) ANOVA on LTR Scores


Source


Age

List

Trial

Age x List

Age x Trial

List x Trial

Age x List x Trial

Error


df

1

1

11

1

11

11

11

1200


SS

1161.55

2808.00

1261.63

675.26

17.93

128.21

25.26

11314.46


F

123.19

297.81

12.16

71.62

.17

1.24

.24


p

.0001

.0001

.0001

.0001

.9985

.2570

.9937






-79-


Although LTR scores continued to climb slightly after Trial

3, the changes were not statistically significant. Figure

7 displays these results.

Consistent retrieval. Results of the 2(Age) x 2(List)

x 12(Trial) ANOVA on CLTR scores are presented in Table 20.

Again, there were significant main effects of Age (F1,1200

= 114.70, p<.0002), List (F1,1200 = 64.59, p<.0002), Trial

(F11,1200 = 57.06, p<.0002), and an Age x List interaction

(F,1200 = 68.70, p<.0002). However, there was no Age x
List x Trial interaction, no Age x Trial interaction, nor a

List x Trial interaction. For this variable also, Hypothe-

sis 7 was not supported.

Because the pattern of CLTR scores across trials was

not significantly different for the two Age groups nor for

the two List groups, follow-up tests of Trial means were

again performed using the combined sample (N=104). Results

of Fisher's LSD test of the pre-planned set of all possible

pair-wise comparisons revealed significant increases in

CLTR scores on Trials 2 and 3. After Trial 3, scores

continued to rise at a regular but slower rate through the

end of the task. Figure 8 displays these results.

Hypothesis 8: Intrusions

Means, standard deviations, and minimum and maximum

values of INTRUSIONS for both Age and List groups are given

in Table 21. Values in Table 21 represent sample distribu-

tions after Winsorization (Winer, 1971) of one extreme

outlier (105 intrusions) in the Old/Related List condition.






-80-


Key:

-- = YOUNG/Related List
A- -A = OLD/Related List
-----o = YOUNG/Unrelated List
LA---A = OLD/Unrelated List


I LII I I I


2 3 4 5 6 7 8 9 10 11 12

TRIAL


Figure 7. LTR (retrieval) trial means for the four
experimental groups.


16

14

12

10 -

8-


/


0 '--
1


..A- -a.- -6.- --- -L .- --A- --A







-81-


TABLE 20

2(Age) x 2(List) x 12(Trial) ANOVA on CLTR Scores


Source df


Age

List

Trial

Age x List

Age x Trial

List x Trial

Age x List x Trial

Error


1

1

11

1

11

11

11

1200


SS

1300.54

732.32

7117.25

779.00

9.59

75.74

13.40

13606.31


F

114.70

64.59

57.06

68.70

.08

.61

.11


p

.0001

.0001

.0001

.0001

1.0000

.8248

.9998






-82-


Key:
S- = YOUNG/Related List
A---A = OLD/Related List
0---o = YOUNG/Unrelated List
A---a = OLD/Unrelated List


A


0 I I I i I I I- I I I
1 2 3 4 5 6 7 8 9 10 11 12

TRIAL



Figure 8. CLTR (consistent retrieval trial means for the
four experimental groups.


P- rlr







-83-


TABLE 21

Means, Standard Deviations, Minimum and Maximum Values
of INTRUSIONS Scores for Age and List Groups


AGE GROUPa
Oldb Young


LIST GROUPa
Unrelated Relatedb


Meanb


Standard Deviation


12.06

15.33


6.71

7.76


9.67

12.23


Minimum Value

Maximum Value


a
n=52 for each group.

Values represent distribution after Winsorization of
outlier in the Old/Related List condition.


9.10

12.65






-84-


Hypothesis 8 predicted no mean difference in INTRU-

SIONS between Age groups. A 2(Age) x 2(List) ANOVA was

performed, with results as presented in Table 22. Contrary

to prediction, there was a significant Age effect (F,100
2
4.94, p<.05, w2 partial = .04). There was no List nor Age

x List effect, however. Regardless of List condition, Old

adults made more intrusion errors than did Young adults

(Old M = 12.06, Young M = 6.71). Hypothesis 8 was there-

fore not supported. Figure 9 displays these results.

Several follow-up analyses of the intrusions data were

also conducted. First, the number of different intrusions

was calculated for each subject (i.e., not counting repeti-

tions of a given intruded word). Means, standard devia-

tions, and minimum and maximum values of this variable for

both Age and List groups are presented in Table 23.

A 2(Age) x 2(List) ANOVA was performed, with results

as presented in Table 24. There was a significant main

effect of Age (F1,100 = 4.35, p<.05, w2 partial = .03).

Neither the main effect of List nor the Age x List interac-

tion was significant, however. Regardless of List condi-

tion, Old adults made a greater number of different intru-

sions than did Young adults (Old M = 3.23, Young M = 1.65).

Second, proportions of Young and Old subjects who made

various numbers of intrusions (total INTRUSIONS and Differ-

ent Intrusions) were calculated. These results are pre-

sented in Figures 10 and 11. The proportions of Young and







-85-


TABLE 22

2(Age) x 2(List) ANOVA on INTRUSIONS Scores


Source

Age

List

Age x List

Error


SS

743.12

8.65

.62

15054.23


df

1

1

1

100


MS

743.12

8.65

.62

150.54


F p

4.94 .0286

.06 .8110

.00 .9491


2
w2 partial

.04







-86-


Key:

----- = YOUNG group
A- -A = OLD group


A --A


Related


Unrelated


LIST CONDITION


Figure 9. Results of 2 x 2 ANOVA on INTRUSIONS scores.


I -






-87-


TABLE 23

Means, Standard Deviations, Minimum and Maximum Values
of Different Intrusions for Age and List Groups


AGE GROUPa


Mean


Standard Deviation


Old

3.23

5.10


Young

1.65

1.79


LIST GROUPa


Unrelated Related


2.56

3.86


2.33

3.94


Minimum Value

Maximum Value


a
n=52 for each group.







-88-


TABLE 24

2(Age) x 2(List) ANOVA on Different Intrusions Scores


Source

Age

List

Age x List

Error


SS

64.65

1.38

.15

1487.46


df


1

1

1

100


MS

64.65

1.38

.15

14.87


F

4.35

.09

.01


p

.0396

.7609

.9192


w) partial

.03


----


--


--








-89-


3 ;




O 0 (D


II II 0
So Co






d ; 0-4
d0 0








n C) o lo
^wv + oO :





: z1 n 0 U

000
> *4-H U








7:1 0 4-J
Mu 0 oO a .-
e -i 3o u
-- (D z -*H OU
O 0 r c
0 *Her3DC)
cr > E
C 3 +-4 CN



I 0 c

S-F- O 0 -.0




Z1 C0
U1 .



c O O0
I n -H C









CD D 0
CD I^WVWWw C) C) C
LLO 0\ 0 C) I-) C) 0 +


SV ( -
on P a (%) U C )
(N (N (S ^












-- ------- (/ --4








bOl
*i-



















3
0
5-4 3
M 0
O


o O


II II

O --s


il5sss



'--1
-NXXX
>-i L-,-_^_~





I-I


, \\vvw


aldueS jo (%) 0u031ad


-90-


4-
cz








rJ1 0
cz





on

C>





0 u


4- C) *







o
0 m -i

i ,= -H


34 0
Oj C,














C3
0 C *












0)
0 <4-J 4-
4-0 0
u0 II


ed 0y

>- 10




&3+- 4-

03 G (=
4- ( <
MC ^ 1


i- o Ln c Lo c u-c o


e--
LnTiC~







-91-


Old subjects making no intrusions at all were highly simi-

lar. Among those subjects who made intrusions, however,

there were differences between the two Age groups. For

example, for the Young who made intrusions, the modal

number of total INTRUSIONS was 1 (5 subjects), and the

modal number of Different Intrusions was also 1 (15 sub-

jects). For the Old who made intrusions, the modal number

of total INTRUSIONS was 11 (4 subjects), and the modal

number of Different Intrusions was 2 (13 subjects). Another

major difference between the two Age groups' performances

was that the Old group showed a much greater range of

scores for both total INTRUSIONS and Different Intrusions.

In each case, a substantial proportion of Old subjects

scored beyond the maximum value obtained by the Young

subjects.

Third, intrusions were categorized as Consistent,

Repeated, or Isolated. Consistent Intrusions were defined

as those which, from a given trial, were then repeated on

every subsequent trial (comparable to the CLTR measure for

correct recalls). Repeated Intrusions were defined as

those said on at least two trials by a subject, even if not

on successive trials. Thus, Consistent and Repeated Intru-

sions are overlapping measures, with Consistent Intrusions

a more restricted set of responses (i.e., Consistent Intru-

sions represent a subset of the more general category of

Repeated Intrusions). Finally, Isolated Intrusions were