108
916

 

1Tfiflus

This is to certify that the

thesis entitled

The Impact of Depressive Symptoms
on Explicit Memory Performance in the Elderly

presented by

Timothy L. Gannon

has been accepted towards fulfillment
of the requirements for

MJ. degree in Mil—1.093

the AM

Major professor

 

Date _Mamh_29_._]_994__

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

 

 

LIBRARY ‘
Mlchlgan State
Unlverslty

 

 

 

PLACE IN RETURN BOXto remove this checkout from your record.
w FINESneturn onor orbefore dat edue.

TOAVOD
‘ DATE DUE DATE DUE DATE DUE |
Ff

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l \

MSUle AnAtfinnat lvquction/E ueIO plponunlty Institution
WMma-ni

 

THE IMPACT OF DEPRESSIVE SYMPTOMS
ON EXPLICIT MEMORY PERFORMANCE
IN THE ELDERLY
By

Timothy Leo Gannon

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Psychology

1 994

ABSTRACT

THE IMPACT OF DEPRESSIVE SYMPTOMS
ON EXPLICIT MEMORY PERFORMANCE
IN THE ELDERLY

By

Timothy Leo Gannon

This study investigated whether depressive symptoms directly afl‘ect explicit
memory performance. Participants (g = 45) for this study were community dwelling
elderly (Mean age = 68) who were offered periodic assessments of their mood and
memory, in addition to, a 7-session workshop that focused on relaxation or cognitive
strategies for alleviating depression and memory problems. Level of depressive symptoms
were assessed with the Hamilton Depression Rating Scale, the Beck Depression
Inventory, and the Geriatric Depression Scale while total recall was assessed with the
Selective Reminding Test and the Logical Memory Test. These test scores were then
combined to produce a depression and memory factor. The average correlation between
these factors was -.35 (p<.05) within the same test period and -.38 (p<.05) across test
periods. A relationship between symptoms and memory performance was found.

Implications, as well as, limitations of these findings were discussed.

ACKNOWLEDGMENTS

I would like to take a moment to express my appreciation to the members of my
thesis committee, as well as, Dr. John Hunter. First, I would like to thank Dr. Ralph
Levine, who served as my statistical advisor. Dr. Levine ’spent an exceptional amount of
time and patience broadening my understanding of the theory and mechanics of the
statistical procedures used in this study. In addition, I would also like to thank Dr.
Hunter, my former statistics mentor, whose statistical software programs, expertise, and
availability also helped make this study possible. Next, I would like to thank Dr. John
Hurley for his insightful comments on my thesis proposal and final manuscript. Dr. Hurley
put considerable effort into increasing my understanding of what is required for a "journal
quality" manuscript. Finally, I would like to thank Dr. Norman Abeles, my mentor and '
committee chairperson. Through the effort of Dr. Abeles, I have always had funding to
ease the financial burden of supporting a family while in graduate school. In addition, Dr.
Abeles has always been readily accessible for consultations regarding clients, research, and
coursework issues. It has also been through his ongoing efforts that I have obtained
valuable experience in memory and mood research, neuropsychological and learning
characteristic assessments, and specific training in clinical neuropsychology. I am indebted

to each of you gentlemen.

iii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................ vi
INTRODUCTION ............................................................................................................. 1
Memory Training Studies ............................................................................................ 2
Meta-Analysis ............................................................................................................. 3
METHOD ......................................................................................................................... 7
Participants .................................................................................................................. 7
Test Instruments for Severity of Depression ................................................................. 7
Test Instruments for Explicit Memory .......................................................................... 10
Test Instrument for Mental Status ............................................................................... 12
Procedure ................................................................................................................... 13
RESULTS ........................................................................................................................ 14
Psychometric Properties .............................................................................................. 14
Stayers versus the Decliners ......................................................................................... 15
Workshop versus Controls ........................................................................................... 17
Causality ..................................................................................................................... 18
DISCUSSION ................................................................................................................... 2O
Psychometric Properties .............................................................................................. 20
Stayers versus Decliners ............................................................................................... 21
Workshop versus Controls ......................................................................... V ................. 21
Causality ..................................................................................................................... 22
Suggestions for Future Research ................................................................................. 23

iv

TABLE OF CONTENTS (CONT'D)

APPENDD( A Characteristics of the Meta-Analysis .......................................................... 25
APPENDIX B Psychometric Properties ............................................................................ 28
APPENDIX C Stayers versus Decliners ............................................................................ 33
APPENDIX D Workshop versus Controls ........................................................................ 38
APPENDD( E Causality ................................................................................................... 43
LIST OF REFERENCES ................................................................................................... 44

LIST OF TABLES

Table 1 Meta-Analysis Findings ........................................................................................ 6

Table 2 SRT Word Lists .................................................................................................. 12
Table 3 DEP and MEM Test-Factor Correlations ............................................................. 15
Table 4 Coefficient Alphas for the DEP and MEM Factors .............................................. 15
Table 5 Stability Analysis of the DEP and MEM Factors .................................................. 18
Table 6 Correlations Between DEP and MEM Factors ..................................................... 19
Table 7 Depression and Memory Measures ....................................................................... 27
Table 8 Item-Factor Correlations for Depressive Measures ............................................... 29
Table 9 Coefficient Alphas for Depression Measures ......................................................... 30
Table 10 Stayers Test-Factor Correlations ........................................................................ 31
Table 11 Stayers Coefficient Alphas for the DEP and MEM Factors ................................. 31
Table 12 Stayers Raw Score Means and SDS .................................................................... 32
Table 13 Comparison of Stayers and Decliners Means and SDS ......................................... 35
Table 14 Stayers and Decliners Correlation Coefficients .................................................... 36
Table 15 Stayers and Decliners Standard Deviation Ratios ................................................ 37
Table 16 Means and SDS for Control and Workshop Conditions ....................................... 40
Table 17 Treatment Corrleations for Control and Workshop Conditions ........................... 41
Table 18 Control Treatment Correlations .......................................................................... 41
Table 19 Workshop Treatment Correlations ...................................................................... 42
Table 20 Comparisons of Treatment Correlations .............................................................. 42
Table 21 Path Diagram ..................................................................................................... 43

vi

INTRODUCTION

Geriatric memory problems are one of the primary reasons for family referrals
(Heath, Grant, Kamps, & Margolin, 1991). Craik (1991) reported that age-related
declines in memory ability have been observed in explicit memory, working memory,
episodic memory, semantic memory, and prospective memory. Hypotheses that have
stated to account for these declines include differences in the ability to self-initiate
encoding and retrieval processes, fewer processing resources, additional attentional
demands competing for limited attentional resources, less mental energy, and brain
structures differentially susceptible to the aging process (Craik, 1991). Lowenthal et al.
(1967) observed that self-reports of declining memory increased from 31 percent for those
60 to 64 years of age to 65 percent for those over 75 years of age. However, since
complaints about memory deficits are not always correlated with actual memory
performance (N = 153; r = .05; Kahn, Zarit, Hilbert, & Niederehe, 1975) and memory
complaints have been positively associated with depression (N = 67, r = .25, West,
Boatwright, & Schleser, 1984; N = 120, r = .52, Shelton & Parsons, 1987; N = 144, r =
.49, Tun, Perlmutter, Russo, & Nathan, 1987), an individual's memory complaints may not
always be symptomatic of memory impairment.

Due to the essential role that memory plays in learning and functioning in day-to-
day life, memory impairment can have a devastating impact on the individual, family, and
society (e.g., health care system). Presently, researchers are attempting to operationalize
the characteristics of age-related memory decline further, as well as, develop effective
treatment regimens. For example, the National Institute of Mental Health (NIMH) has

1

2

introduced the term age-associated memory impairment (AAMI; Crook, Bartus, Ferris,
Whitehouse, Cohen, & Gershon, 1986) to characterize healthy individuals who are at least
50 years of age who have both objective declines in memory (i.e., _>_1 S_D_ below young
adult norms), as well as, subjective memory complaints (Crook, 1989). So far, research
on AAMI has focused on improving its inclusionary criteria and terminology (Smith,
Ivnik, Petersen, Malec, Kokmen, & Tangalos, 1991). For example, Blackford and LaRue
(1989) have suggested that AAMI be further subdivided into age-associated memory
decline (i.e., _>_1 SQ below young adult norms on 25% of the tests administered) and late-
life forgetfulness (i.e., 21 S_D below young adult norms on 50% or more of the tests
administered).
Memog Training Studies

Non-invasive treatment regimens for age-related memory decline have centered on
teaching the elderly memory training strategies. The typical paradigm used to assess their
effectiveness consists of (a) assessment on a memory task (e.g., word list, prose recall),
(b) instruction in one or more memory strategies (e.g., method of loci) across one or
multiple sessions, and (c) re-testing on an equivalent form of the original memory task.
Across 14 explicit memory training studies conducted with the elderly that used word list
tasks and/or prose recall to assess memory ability, all but one (Yesavage, Rose, & Spiegel,
1982) reported significant improvements in total recall when compared to pre-training
scores or to a control group (Hulicka & Grossman, 1967; Robertson-Tchabo, Hausman,
& Arenberg, 1976; Schmitt, Murphy, & Sanders, 1981; Zarit, Cole, & Guider, 1981;
Schafl‘er & Poon, 1982; Yesavage, Rose, & Spiegel, 1982; Rose & Yesavage, 1983;
Yesavage & Rose, 1983; Anschutz, Camp, Markley, & Kramer, 1985; Scogin, Storandt,
& Lott, 1985; Johnston & Gueldner, 1989; Rebok & Balcerak, 1989; Hill, Storandt, &
Simeone, 1990; Yesavage, Sheikh, Friedman, & Tanke, 1990). These training studies
used a wide range of techniques, with mnemonics (e.g., method of loci) being the most

common. Other techniques included word associations (i.e., mediators), afi‘ective

3

judgments, organizational relationships, relaxation training, concentration training, and
group discussions on memory issues (e.g., difference between vital and unnecessary
remembering; memory loss experiences). The designs used in some of these studies
enabled the authors to demonstrate that improvement in recall was not a result of practice
efi‘ects (Robertson-Tchabo et al. 1976; Yesavage & Rose, 1983; Hill et al. 1990),
motivational influences (Hill et al. 1990), or rehearsal length (Schmitt et al. 1981).

Although these studies provide evidence that improvements in explicit memory are
possible, their results were based on changes in group means. Schaffer and Poon (1982)
point out that, at the individual level, effects of memory training can be quite variable. In
order for the rehabilitation psychologist to effectively implement memory training
interventions among the elderly, factors that impede memory training need to be identified.
Johnson and Magaro (1987) have suggested that one such factor may be the degree of
depressive symptomatology. They have hypothesized that the degree of memory
impairment in an individual may be partly a result of the severity of the individual's
depressive symptomatology (e.g., greater depressive symptomatology results in greater
memory deficits). Here, the severity of depressive symptoms denotes a syndrome rather
than a nosological category. For example, depression as a syndrome indicates a group of
symptoms (e. g., depressed mood, diminished interest in activities, significant weight loss
or gain, fatigue) that may be primary to a diagnosis of Major Depression (i.e., a
nosological category) or be secondary to another psychopathological disorder (e. g.,
Schizophrenia). These disorders (e.g., Major Depression, Schizophrenia) are examples of
nosological categories that include the depressive syndrome as part of their diagnostic
criteria (DSM-III-R, 1987).
Meta-Analysis

In order to examine the hypothesis of Johnson and Magaro (1987), a meta-analysis
was first conducted (Hunter & Schmidt, 1990). The criteria used for this procedure are

described in detail in Appendix A. A meta-analysis was chosen over a narrative review

4

due to the inherent biases in summarizing the results of a group of studies in a narrative
format (Smith, Glass, & Miller, 1980). A search of the literature resulted in 9 studies from
which 13 effect-sizes were calculated (Gibson, 1981; Zarit, 1982, O'Hara, Hinrichs,
Wallace, Lemke, & Kohout, 1986; Hart, Kwentus, Taylor, & Hamer, 1987; Hart,
Kwentus, Taylor, & Harkins, 1987; Poitrenaud, Moy, Girousse, Wolrnark, & Piette, 1989;
Feehan, Knight, & Partridge, 1991; King, Caine, Conwell, & Cox, 1991; Lichtenberg,

. Manning, & Turkheimer, 1992). The additional effect-sizes were due to the inclusion of
multiple groups or multiples measures of depression and memory within some of the
studies. With the exception of Zarit (1982) and Lichtenberg et al. (1992), in which
subjects were not assessed for depression (i.e., nosological category), and one group in
O'Hara et al. (1986) that did not meet the criteria for depression, the remaining studies
used individuals who met the criteria for depression as defined in the Diagnostic and
Statistical Manual of Mental Disorders III, (DSM-III; American Psychiatric Association,
1980), its revision (DSM-III-R; American Psychiatric Association, 1987), or the Research
and Diagnostic Criteria (RDC; Spitzer, Endicott, & Robins, 197 8). Other than Zarit
(1982), who did not use a control group, each of these studies compared groups with
elevated depressive symptoms with groups of healthy elderly free of psychiatric diagnoses
and elevated levels of depressive symptoms on a word list or prose recall task.

For each of these studies, "d," an effect-size statistic and "r," the point-biserial
correlation, were computed (Hunter & Schmidt, 1990; pp. 272-273). The d statistic
expresses the difference between the depressed and non-depressed groups in standard
deviation units while the point biserial correlation describes the amount of relationship that
exists between the group (i.e., depressed versus control) and the level of explicit memory.
For more detailed characterizations of these measures and the depression and memory
tests used in these studies, please consult Appendix A. Due to the small number of
studies, the effects of moderator variables could not be assessed. Hence, the overall

findings are limited to the weighted average and weighted standard deviation for the d and

5

r statistics (Table 1). As can be seen, the point biserial correlations across each of these
studies suggest a negative relationship (Ea_vg = -.360) between group and explicit memory
ability. In other words, greater memory ability was associated with the healthy control
groups while the depressed groups had relatively poorer explicit memory. This finding
does appear to be congruent with the relationship hypothesized by Johnson and Magaro
(1987). However, since the majority of these studies used depressed patients, another
interpretation of these findings is that memory impairment is related to some other facet of
depression (i.e., nosological category) and that the relationship with the severity of
depressive symptoms is merely spurious. For example, O'Hara et al.’s (1986) group that
did not meet the criteria for depression yet reported high levels of depressive symptoms
had a very low correlation with memory recall (1 = -.111). Such a low correlation does
not suggest a relationship between depressive symptom severity and memory impairment.
In addition, these nine studies did not address the issue of causality which is hypothesized
to exist (Johnson & Magaro, 1987). Specifically, depressive symptoms are hypothesized
to directly impair memory ability. Hence, the purpose Of this study is to determine if the
severity of depressive symptoms directly affects explicit memory performance as measured
through total recall scores on word list and prose recall tasks. Based on the hypothesis of
Johnson and Magaro (1987), path coefficients from the depression factor to the memory

factor should be the largest.

Table 1 Meta-Analysis Findings

 

 

Author(s) Depressed Control

Sarnple a M. .82 n M. 5.2 d t
1. Gibson (1981)

I& O 20 25.65 6.16 20 34.70 6.63 -l.41 -.59
2. Zarit (1982)

CD? 79 -.69 -.33

CD? 79 -.41 -.21

CD? 79 -.64 -.31

CD? 79 -.47 -.23
3. O'Hara et al. (1986)

CD 22 5.00 2.09 25 6.08 1.85 -.55 -.27

C 23 5.57 2.78 25 6.08 1.85 -.22 -.11
4. Hart, Kwentus, Taylor, and Hamer (1987)

O 14 70.70 16.50 16 83.60 12.40 -.89 -.42
5. Hart, Kwentus, Taylor, and Harkins (1987)

? 10 12.20 1.50 14 17.70 1.40 -3.81 -.90
6. Poitrenaud et al. (1989)

I 24 14.29 3.67 33 19.03 4.31 -1.17 -.52
7. Feehan et al. (1991)

O 10 5.50 2.43 10 7.20 2.46 -.70 -.34
8. King et al. (1991)

I 23 6.40 1.50 23 7.50 1.40 -.76 -.36
9. Lichtenberg et al. (1992)

CD? 16 7.30 3.50 19 11.50 2.20 -1.63 -.71

Weighted Mean: -.84 —.36
Weighted SD: -.68 -. 18

?: Patients diagnosed depressed, did not specify if inpatient or outpatient

C: Community Dwelling, did not meet criteria for depression

CD: Community Dwelling, met depression criteria in epidemiological study, never treated
CD?: Community Dwelling, were not assessed for depression

I/O: Inpatient/Outpatient diagnosed depressed

METHOD

Participants

The participants were a subset of community-dwelling elderly (N_=208) recruited
through newspaper and radio ads, circulars, and presentations at local senior citizen
centers. Each participant was offered multiple assessments of their mood and memory, as
well as, a 7-session workshop that taught primarily relaxation or cognitive strategies for
alleviating depression and memory problems. Forty-five participants (i.e., "stayers") who
agreed to be assessed across three occasions and had a Mini-Mental State Examination
(F olstein, Folstein, & McHugh, 1975) score of 24 or better were included in the present
study. These participants were composed of both men (p = 13) and women (p = 32)
whose mean age was 68 (S_D = 10y, Range = 52 to 92y) and who had, on average, 13
years of schooling (M = 13y, SD = 3y, Range = 5 to 20y).

Test Instruments for Severity of Depression

Since multiple measures of depression could more fiilly capture any depressive
symptomatology being experienced by the participants (Kendall et al., 1987), three
instruments were chosen. The first instrument was an observer-rating scale, the Hamilton
Depression Rating Scale (Hamilton, 1960). The remaining two, the Beck Depression
Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) and the Geriatric
Depression Scale (Yesavage et al., 1983), were self-report measures. Together, these
three measures more fully address the DSM-III-R criteria for depression and criteria

characteristic of depression in the elderly population (Weiss, Nagel, & Aronson, 1986).

8

The Hamilton Rating Scale for Depression (H_A_M), is a 17-item scale that was
originally designed to measure the severity of depression in individuals already diagnosed
depressed. However, since its inception, it has been extensively used to assess depressive
symptoms in other populations as well (Hedlund & Vieweg, 1979). Test-retest stability
coefficients for the HAM ranged from .61 (Mean interval = 15 days; Lyons, Strain,
Hammer, Ackennan, & Fulop, 1989) to .70 (3 week interval; Maier, 1990) while
Cronbach's (1951) alpha was .76 (Rehm & O'Hara, 1985). Sensitivity to change, which is
a scale's ability to detect changes (e.g., in depressive symptoms) in individuals across time,
was .65 for the HAM (Maier, 1990). Correlations of the HAM and psychiatrist's ratings
of global severity of depressed patients ranged from a mean of .88 (range = .84 to .90) for
three studies reviewed by Hedlund and Vieweg (1979) to .67 (Maier, 1990). Stukenberg,
Dura, and Kiecolt-Glaser (1990) observed that the area under the ROC (i.e., Receiver
Operating Characteristics) curve, which represents the probability of a measure correctly
categorizing subjects (e.g., depressed versus nondepressed), was .85 for the HAM. In a
sample of psychiatric inpatients (Lichtenberg, Steiner, Marcopulos, & Tabscott, 1992), the
HAM's sensitivity and specificity was 9% and 92%, respectively. These indices were
detemrined by comparing those subjects who fell above and below a cut-off criterion.
Sensitivity is the proportion of individuals correctly diagnosed depressed (i.e., proportion
of true positives) while specificity is the proportion of individuals correctly diagnosed
afi‘ectively healthy (i.e., proportion of true negatives; Masur et al. 1989). The authors
mentioned that the low sensitivity of 9% is at odds with what is reported in the literature
and suggested that it may have been due to their study's large number of demented
subjects.

Concurrent validity for the HAM was determined based on its correlation with
other measures of depression. The HAM had an average correlation of .77 (range = .73
to .80) with the Beck Depression Inventory across two studies that examined non

psychiatric samples (Beck, Steer, & Garbin, 1988). This correlation with the Beck

9

Depression Inventory was lower, however, in a sample of mixed psychiatric and medical
inpatients (r = .52; Fitzgibbon, Cella, & Sweeney, 1988) and across seven studies
assessing psychiatric inpatients (£a_vg = .58; range = .31 to .82; Hedlund & Vieweg, 1979).
The HAM also had a correlation of .60 with the Brief Symptom Inventory Depression
scale (Derogatis & Spencer, 1982) in a sample of community-dwelling elderly
(Stukenberg, Dura, & Kiecolt-Glaser, 1990). In a review of seven factor analyses,
Hedlund and Vieweg (1979) concluded that the first of two factors generally evident
across the studies appeared to measure the severity of depressive symptoms. However, in
a more recent study, O'Brien and Glaudin (1988) observed four factors that they labeled
somatic complaints, anorexia, sleep disturbance, and agitation/retardation. The 23-item
version of the HAM was used for the present study. Scores on this version can range
from O to 75. Zero denotes a lack of depressive symptoms while 75 suggests severe
depression.

The Beck Depression Inventory (IE1) is a 21-item, multiple-choice inventory that
rates the intensity of depressive symptoms, especially those of a somatic nature. Based on
a psychometric review of the BDI (Beck, Steer, and Garbin, 1988), Cronbach's (1951)
alpha had a mean of .81 (range = .73 to .92) based on 15 samples of non psychiatric
subjects while test-retest stability coefficients had a mean of .76 (1 hour to 4 month
interval; range = .60 to .90) based on 5 samples of non psychiatric subjects. Correlations
of the BDI and clinical ratings of depression for three samples of non psychiatric patients,
had a mean of .60 (range = .55 to .73). The concurrent validity of the BDI, when
correlated with the Geriatric Depression Scale, was .85 with medical outpatients (Norris,
Gallagher, Wilson, & Winograd, 1987). The value with the HAM across two non
psychiatric samples, however, was slightly lower ([818 = .77, range = .73 to .80; Beck et
al., 1988). In a review by Beck et al. (1988), the authors concluded that the most recent

factor studies suggest one depressive factor which measures depressive severity. Scores

10

on the BDI can range form 0, which indicates that depressive statements were not
endorsed, to a maximum value of 36 which indicates severe depression.

The Geriatric Depression Scale (_G_D_S_) is composed of 30 yes-no questions that
primarily examine mood and psychological symptoms. Test-retest stability coefficients of
.98 (Mean interval = 15 days; Lyons, Strain, Hammer, Ackerrnan, & Fulop, 1989) and .85
(1 month interval; Parrnalee, Lawton, & Katz, 1989) have been reported. The mean value
for Cronbach's (1951) alpha was .91 based on studies by Yesavage et al. (1 = .94; 1983),
Parmalee et al. (1 = .91; 1989), and Salamero and Marcos (r = .87, 1992). A study by
Harper, Kotik-Harper, and Kirby (1990) yielded a sensitivity of 85% for the GDS in a
sample of depressed elderly, while in a study by Koenig, Meador, Cohen, and Blazer
(1988), a sensitivity of 92% and a specificity of 89% were obtained. Concurrent validities
of .62 and .81 were obtained when the GDS was correlated with the HAM at two
different assessment periods with medical inpatients (Lyons, Strain, Hammer, Ackerman,
& Fulop, 1989). This value was slightly higher (r = .83) when medical outpatients were
assessed (Norris, Gallagher, Wilson, & Winograd, 1987). Correlations were also slightly
higher between the GDS and BDI ([ = .85; Norris et al., 1987). Two factor analyses of
the GDS found six (Parmalee et al., 1989) and nine-factor (Salamero & Marcos, 1992)
solutions although both studies mentioned that the GDS was unidimensional (i.e., general
depression factor) based on Cattell's (1966) scree criterion. Scores on the GDS can range
from a score of 0, which indicates that no depressive statements were endorsed, to a
maximum value of 30 which indicates severe depression.

Test Instruments for Bigglicit Memory

Logical Memory (LM) Form I of the Wechsler Memory Scale (W echsler & Stone,
1945) served as a measure of prose recall. It has an average test-retest stability of .79
(time interval not mentioned; Bowden & Bell, 1992) and was found to be significantly
correlated (; = -.375) with the Halstead-Reitan Average Impairment Scale in a sample of

brain-damaged subjects (Russell, 1975). LM is one of four tests (i.e., Storandt Brief

11

Dementia Battery) used in a discriminant fimction for detecting Alzheimer's disease
(Storandt, Botwinick, Danziger, Berg, and Hughes, 1984). This function had a sensitivity
of 98% on a cross-validation sample. Across several factor analyses reviewed in Erickson
and Scott (1977) and Prigatano (1978), LM consistently loaded on the memory factor.
Total recall scores for LM can range from 0 to 22. The higher the score, the greater
number of information units are recflled by the participant.

The Selective Reminding Test (Sig, Buschke, 1973) served as the word list task
with lists of equivalent words (Masur et al., 1989) being used for each of the three
assessments (Table 2). Each list contained 12 unrelated words (Hannay & Levin, 1985).
Test-retest stability has been found to range from .73 (6 month interval; Ruff, Quayhagen,
& Light, 1988) to .89 (2 hour interval; Masur et al., 1989). In a study by Masur et al.
(1989), the SRT displayed a sensitivity and specificity in the detection of patients with
mild Alzheimer's disease of 86 and 99%, respectively. However, in a second study
(Masur, Fuld, Blau, Crystal, & Aronson,1990) with patients diagnosed with dementia
based on DSM-III-R criteria, the SRT obtained a sensitivity of 47% and a specificity of
86%. Total recall scores were based on the sum of words recalled across 6 trials and
could range from 0 to a maximum value of 72. The higher the number, the more words
were recalled by the participant across the trials. This was chosen as a measure of long
term explicit memory over other word list tasks since it controls for the influence of short
term memory (Buschke, 1973; Buschke & Fuld, 1974).

To provide a measure of subjective memory complaints, two self-report items of
memory problems were created. The first item (M113) asked, "Are you currently
experiencing any problems with your memory?" while the second item (M) stated,
"If (you answered) yes (to the 1st item), please tell us what kinds of problems you are
experiencing. Remembering names ( ), faces ( ), date, month, or year ( ), appointments

( ), where you put things ( ), what you went into a room to do/get ( ), to take medication

( ), other ( )." The stability coefiicient was .77 for the first item and .66 for the second

item. These coefficients were based on 51 control subjects who took the first two

assessments as part of the larger study. #PROB could range from 0 problems to a

maximum value of 8 problems.

Table 2 SRT Word Lists

 

Test 1
Shine
Disagree
Fat
Wealthy
Drunk
Pin
Grass
Moon
Prepare
Prize
Duck
Leaf

Test 2
Bowl
Passion
Dawn
Judgment
Grant
Bee
Plane
County
Choice
Seed
Wool
Meal

 

Test 3
Throw
Lily
Film
Discreet
Loft
Beef
Street
Helmet
Snake
Dug
Pack
Tin

 

 

Test Instrument for Mental Status

The Mini-Mental State Examination (MMSE; Folstein, Folstein, & McHugh,

197 5) is a 30-item form that served as a cognitive impairment (i.e., mental status)

examination. This screening test provides a brief measure of orientation, registration,

attention and calculation, recall, and language. Based on a review of the MMSE by

Tombaugh and McIntyre (1992), the mean coefficient alpha across five studies was .72

(range = .54 to .96) while the mean stability coefiicient across 25 samples was .82 (range

= .80 to .97; time interval 1 day to 2 months). The sensitivity of the MMSE across 27

13

samples that examined demented subjects had a mean value of .79 (range = .20 to 1.00)
while the specificity across 21 samples had a mean value of .83 (range = .46 to 1.00).
Construct validity, based on 14 studies between the MMSE and other cognitive status
measures, ranged from .70 to .90. The MMSE was included for two reasons. First, to
identify any difference in cognitive status that may have been present between those
participants who only agreed to a single testing session and those that took all three.
Second, since cognitive impairment (i.e., a low MMSE) is also associated with poorer
memory, it was used as a screening measure for the Stayers to avoid any confounds. The
23/24 cutoff recommended by Folstein et al. (1975) was used for the present study.
Scores could potentially range fi'om a low of 0 indicating severe impairment to. a maximum
value of 30 indicating an unimpaired mental status.
Procedure

Subjects were assessed on the three depression scales, two objective memory
scales, the 2-item subjective memory problem scale, and the test of cognitive status on
three occasions at approximately three month intervals. Testing was done by clinicians
enrolled in Michigan State University's clinical psychology program. Approximately seven
hours of training were given to each of the clinicians in administering and scoring the tests.
During their training phase, each clinician observed an assessment by another clinician
with at least 1 year experience and was then subsequently observed by the same clinician
during their first testing session. Furthermore, all tests were re-scored by the author to
ensure accuracy. Whenever, differences in total score existed, totals were retabulated

until the same score was produced on two successive occasions.

RESULTS

Data analyses were subdivided into four components. First, the psychometric
properties of the depressive and memory indices were assessed. Second, the "stayers"
were compared to the "decliners. " Third, stayers who participated in the relaxation or
memory workshops (p = 27) were compared to the stayers who did not participate (p =
18) in order to determine the presence of any treatment effects. Fourth, the stayers' levels
of depressive symptoms and amount of memory ability were evaluated for the presence of
any causal relationships.

Psychometric Properties

First, each of the stayer's total scores from the depression and memory measures
were transformed into ; scores and summed to produce a depression factor (DEP) and a
memory factor (lyfl/D. Construct validity was then empirically tested with confirmatory
factor analyses (CLA). An initial CF A with the Hamilton Depression Scale (HAM)
indicated that 4 test items had zero variance and 6 others had negative loadings on a factor
comprising all the HAM test items. It was assumed that the test items with negative
factor loadings were measuring something different from the construct of depression being
assessed by the other test items. Consequently, all ten of these test items were removed so
the HAM total score did not reflect these items in all subsequent analyses. Next, CFA's
were performed to test how each depression and memory test loaded on their respective
factor. The results of these analyses can be found in Table 3.

Next, stability was assessed using Cronbach's (1951) alpha. The resulting stability

coefficients can be found in Table 4. In summary, it appears that there was adequate
l4

15

validity and reliability for the measures used in this analysis. A more detailed description
of these psychometric analyses can be found in Appendix B, as well as, the stayers' raw

scores for the depression and memory measures.

Table 3 DEP and MEM Test-Factor Correlations

 

DEP1* DEP2 DEP3
HAM .71 .79 .83
BDI .79 .72 .78
GDS .75 .96 .71

MEMI * MEM2 MEM3
LM .73 .74 .71
SRT .73 .74 .71

* the number following the DEP and MEM indicates the test period from which the
factors were calculated

 

Table 4 Coefficient Alphas for the DEP and MEM Factors

 

Time 1 Time 2 Time 3
DEP .79 .86 .82
MEM .69 .71 .67

 

Stayers versus the Decliners

Since there was a large number of participants who were involved in the first mood

and memory assessment but declined to take all three assessments, these analyses were

l6

conducted to determine if any differences existed between the stayers and decliners that
could affect the results generalizability. The following variables were examined for
differences: general characteristics (i.e., age, gender, education), severity of depressive
symptoms (i.e., HAM, BDI, GDS, DEP), explicit memory (i.e., LM, SRT, MEM),
subjective memory complaints (i.e., PROB, #PROB), cognitive mental status (i.e.,
MMSE), and if they declined or chose to participate in one of the depression/memory
workshops (i.e., Cfl). Those who took part in the relaxation (p = 11) or memory
workshop (p = 16) were combined due to small samples and since results from the larger
study (N. Abeles, personal communication, April 15, 1992) indicated that the treatment
effect was not significantly different between these two groups. In addition to these
variables, a SR MEM factor, which was based on the summed g-scores for the two
subjective memory complaint measures, was calculated. For this variable, as an
individual's number of subjective memory complaints increases, their ; score will become
progressively more positive.

The means and standard deviations for each of these variables within the two
groups (i.e., "stayers" and "decliners") were computed and are listed in Appendix C.
From these, d statistics were calculated for each variable and then transformed to point
biserial correlations (Appendix C). Stayers were found to be younger, have a higher ratio
of men to women, have higher recall scores on the word list task, have more individuals
with subjective feelings of memory problems and number of problems, be more cognitively
unimpaired, and have a higher proportion of individuals who participated in the
workshops. Another measure of variability between groups, the standard deviation ratio
(2), was also calculated (Hunter, 1993b) and can be found in Appendix C. This measure
compares the standard deviation of the above variables between the stayers and decliners.
Results of this analysis indicated that the mental status of the decliners had greater

variability then was found in the stayers.

17

Workshop versus Controls
Initially, control and workshop groups were analyzed separately for the presence

of any changes in their DEP and MEM factors between time 1 and 2, time 2 and 3, and
time 1 and 3 (Appendix D). These changes between test periods were then expressed as
treatment correlations that had been corrected for attenuation (Hunter, 1994). Further
information on correcting for attenuation can be found in Appendix J. Analysis of the
control condition indicated that there was a high probability that depressive symptoms
increased in severity between test period 2 and 3 (Odds ratio = 4.38). Calculations also
indicated a high probability that memory recall increased between time 2 and 3 (Odds ratio
= 7.87) and between time 1 and 3 (Odds ratio = 12.48). Analysis of the workshop
condition indicated a high probability that depressive symptoms became less severe
between time 1 and 2 (Odds ratio = 34.58). Results also indicated that it was highly
probable that memory recall increased between time 1 and 2 (Odds ratio = 15.69) and time
1 and 3 (Odds ratio = 13.24). These analyses also indicated that there was no significant
interaction for either the control or workshop condition (p>.05). In other words, once
measurement error was controlled for, each of the participants within a condition (i.e.,
workshop or control) changed approximately the same amount in terms of depressive
symptoms and memory recall between the test periods listed above. Ninety five percent
confidence intervals with worst case, best case, and odds ratio that provided the basis for
the above results are also tabulated in Appendix D for both the control and workshop
conditions.

Next, treatment correlations for the control and workshop conditions were
compared during each of the time periods. Gain scores (i.e., change scores) were
calculated based on the differences in treatment correlations between the control and
workshop conditions (Appendix D). Results of this analysis indicated that it is highly
probable that there was a difference in the DEP treatment effect between the control and

workshop conditions when time 1 was compared to time 2 (Odds ratio = 3.67). In this

18

case, it was due to the workshop group evincing a greater decrease in self-reported
depressive symptoms during this time frame than was found in the control group.
Causality

First, a stability analysis (Heise, 1969) was conducted on the DEP and MEM
factors over the three time periods to assess their temporal stability independent of
measurement error. A value of 1.00 would indicated perfect stability across time. As can
be seen in Table 5, the values for the DEP and MEM factors were very stable. In other
words, the participants' self-reported depression and explicit memory ability changed very
little across time. These changes were so minimal that any causal relationship that may

have existed between the two factors could not be determined.

Table 5 Stability Analysis of the DEP and MEM Factors

 

DEP Factor MEM Factor

Stability between Time 1 and 2 .95 .99
Stability between Time 2 and 3 .96 .97
Stability between Time 1 and 3 .92 .96

 

Table 6 displays the correlations between the DEP and MEM factors. The time
periods are designated by the number following the DEP and MEM factor. For example,
DEP] indicates the DEP factor at time 1. In each case, there is a negative correlation
between the DEP and MEM factors. In addition, these negative correlations are slightly
larger across time (Egg = -.38) then within the same time period (ran = -.35). These
correlations were corrected for attenuation. A path diagram of the DEP and MEM factors

can be found in Appendix E.

19

Table 6 Correlations Between DEP and MEM Factors

 

DEP 1
DEP2
DEP3
MEM l
MEM2
MEM3

DEP l

1.00
1.00
-.38
-.44
-.46

DEP2

1.00
-.46
-.35
-.38

DEP3

-.39
-.33
-.32

MEM] MEM2 MEM3

1.00 -
1.00 1.00 -

 

DISCUSSION

The analyses for this study were subdivided into four components. First, the
psychometric properties of the depressive and memory indices were determined. Second,
the 45 "stayers" were compared to the 163 "decliners." Third, the 27 stayers who
participated in the relaxation or memory workshops were compared to the 18 stayers who
did not participate in order to assess the presence of any treatment effects. Fourth,
depressive and memory indices among the stayers were examined for the existence of any
causal relationships. This same format will be used to discuss the findings of these
analyses.

Psychometric Properties

Results indicated that there was reasonable evidence of construct validity and
reliability of the selected measures of depression and memory after test items were
removed from the Hamilton Depression Rating Scale (HAM) due to a lack of variance and
negative factor loadings. One possibility for the problematic test items encountered with
the HAM may be due to its use on a population for which it was not designed for. Even
though the HAM has been extensively used with other populations (Hedlund & Vieweg,
1979), it was originally designed for assessing the severity of depression in individuals
already diagnosed depressed. That population is quite different from the present sample
which reported relatively few depressive symptoms. Another possibility is that the
clinicians did not give enough time for the HAM interview in order to bring out responses
to each of the test items (Hamilton, 1960). However, each of the clinicians in this study

had at least one year experience in supervised geropsychiatric assessments so this
20

21

explanation seems unlikely. One factor which apparently contributed was the 23-item
version of the HAM used in this study. This version contains the original 17 test items
plus an additional 6 test items. Within this study, three of these additional test items had
zero variance while another 3 had negative factor loadings. It maybe that the 23 -item
version was psychometrically unsound from its inception since the psychometric properties
repOrted in the literature for the HAM appear to only be based on the original 17-item
test.
Stayers versus Decliners

Although the stayers and decliners had a high probability of being different on
several variables, it is doubtful whether these differences are clinically meaningful. For
example, even though both groups differed on mean recall scores (i.e., Selective
Reminding Test) and mental status scores (i.e., Mini-Mental State Examination), both
group's mean scores are in the healthy range based on normative data (Masur et al., 1989;
Tombauch & McIntyre, 1992). In addition, the difference between the average number of
subjective memory problems being perceived between the two groups was less than half a
problem. This also does not suggest a meaningful difference in perceived memory 1
problems.
Workshop versus Controls

The first surprising finding was that, after correcting for measurement error, all of
the individuals in the workshop changed approximately equal amounts in memory ability
across the three test periods. Although others have reported high levels of individual
variability in memory training treatment effects (Schaffer & Poon, 1982; Yesavage et al.,
1990), no mention was made in these studies of correcting for the effects of measurement
error. The unfortunate consequence of this practice is that misconceptions, such as high
individual variability being typical among memory training programs, results. When
measurement error is corrected for, the overall efficacy of these programs appears to be

much better.

22

The second finding was that the workshop condition only had a minimal treatment
effect. Based on the number of studies cited in the introduction that observed significant
treatment effects, this came as a surprise. However, Schafi‘er and Poon (1982) contend
that this is not atypical and instead hypothesize that their may be many unpublished
memory training studies with relatively neutral or negative results. Looking back, we
might have achieved greater changes in memory ability had we included memory tests
more relevant to the memory strategies that were taught in the workshop. For example,
one strategy. that the elderly were taught was to group like-words (e.g., all meats or all
vegetables) in order to increase their level of recall. If a shopping list task had been used
instead of the Selective Reminding Test, it is possible that greater recall scores would have
resulted. Also, although the elderly were assessed for prose recall using Logical Memory,
no strategies for improving prose recall were given during the workshop. However, the
workshops were purposely structured this way in order to assess the potential for transfer
of training to other different but related memory tasks. Based on this study, proven
memory training strategies do not appear to transfer to related memory tasks as were used
in our memory assessments.

Another factor that may have negatively impacted the level of recall in the
workshop condition was the presence of any participant anxiety. Anxiety has also been
associated with decreased memory performance (Bhagia & Pal, 1986). However, since an
anxiety measure was not part of the test battery and this relationship does not imply
causality, this possibility remains only conjecture. Also, participants may have reverted to
their former memory strategies to increase their "comfort zone" or, since they were not
expressly asked to use their "new" memory strategies, they may not have chosen to use
them.

Causality
A stability analysis revealed that our relatively well-educated elderly sample had a

very stable level of self-reported depressive symptoms and explicit memory ability across

23

time. Because of these minimal changes, causality could not be determined. However, a
negative relationship was found between depressive symptoms and explicit memory
ability. In other words, as the level of depressive symptoms increased, total recall scores
of the two measures of explicit memory decreased. The negative correlations encountered
were similar to those observed by Feehan et al. (1991) and King et al. (1991).
Unexpectedly, this relationship was found to be equally strong across time periods. One
could hypothesize that an individual's present level of depressive symptoms influences
future memory ability and that one's present memory ability influences future levels of
depressive symptoms. However, due to the minimal variation across time, these
hypotheses of causality cannot be determined from this sample. It was hoped that the
results of this study could have been used to design future memory training programs
and/or memory remediation interventions more effectively. Ifdepression does decrease
memory recall, the client could first be treated for depression before entering a memory
training or rehabilitation program or, alternatively, the program could have a depression
component within its design. However, it remains the task of future studies to determine
if such a relationship exists.
Suggestions for Future Research

Ideally, examination of the Johnson and Magaro's (1987) hypothesis should be
carried out concurrently on depressed and non-depressed psychiatric patients, patients
with high levels of depressive symptoms but not meeting formal diagnositc criteria for
depression, and those who are affectively healthy and do not meet the criteria for a
psychiatric diagnosis. With these four groups, a study should be able to tease out the
effects of depression, psychiatric hospitalization, and depressive symptoms on memory
recall. As for the memory measures, ideally visual, as well as, verbal explicit memory
should be assessed in order to more fully delineate any effects of depressive symptoms on
explicit memory. These measures could be even further refined by having verbal explicit

memory broken down into aural, visual, and mixed input. Also, adding some "everyday"

24

explicit memory measures like a shopping list task or a newspaper article may reveal
something not observed in the more "removed-from-the-real-world" laboratory measures.
As for the problems of limited change in memory and depression levels across time, having
assessments over additional time periods may solve this problem. Using a time series
analysis in which the participants are assessed for 5 or even 10 test periods may be able to
tease out any causal relationships that do exist.

Work could also be focused on the effects of depressive symptoms on the other
types of memory suffering age-related declines (e.g., episodic memory, semantic memory,
prospective memory). Does Johnson and Magaro's (1987) hypothesis extend to these
forms of memory also? Does this hypothesis extend to other forms of memory that do not
appear to be susceptible to the effects of age (e.g., implicit memory)? Is there an
interaction effect, where those forms of memory exhibiting age-related declines are
differentially susceptible to the effects of depressive symptoms? Also, do depressive
symptoms effect memory in the same temporal plane or do they have their greatest impact
on some future time frame? In addition, does memory ability affect depressive symptoms?
Furthermore, do all depressive symptoms exert these effects or are just certain symptoms
associated with this occurrence? In summary, there are many questions that still need to

be addressed in this area of mood and memory.

APPENDIX A

APPENDIX A

Mummies of the Meta-Analysis

Medline and Psychlit were searched for relevant articles published from 1979 to
1993. Relevancy was based on the following criteria: (a) the sample needed to include
elderly subjects (i.e., 55 years of age or older), (b) subjects had to be assessed for
depressive symptomatology, (c) the article had to be published in English, and (d) explicit
memory had to be assessed by either a word list or prose task. This analysis was limited
to studies with elderly individuals for purposes of comparison with this study and due to
the possibility that the relationship between memory impairment and depressive
symptomatology may be moderated by age (Lichtenberg, Manning, & Turkheimer, 1992).
In addition, studies were limited to those using word list and prose recall tasks since they
are explicit memory measures that are analogous to many everyday situations (e.g.,
shopping list, to-do list, newspaper and magazine articles) that the elderly encounter.
Relevant studies cited in the articles obtained through Medline and Psychlit were also
acquired. The resulting studies, with the depressive and memory measures used in each,
are found in Table 7.

Statistical analysis of these studies was done with the d and 1 statistics. The d
statistic was used since it is analogous to the "1" statistic but is independent of sample size.
For this analysis, _c_l was based on the difference between the means of the two groups
divided by the within-group standard deviation. The within-group standard deviation was
used since it has approximately half the sampling error found in the control-group standard

deviation (Hunter & Schmidt, 1990). The point biserial correlation (g) was used as
25

26

another measure of treatment effect and was calculated by transforming the d Statistic
(Hunter & Schmidt, 1990). This correlation describes the degree of relationship that
exists between the group (i.e., depressed or control) and the level of explicit memory
ability (i.e., total recall score). In addition, since the point-biserial correlation is affected
by sample size, the r_s for studies having unequal sized groups were corrected for
attenuation (Hunter & Schmidt, 1990; p. 274). The larger correlation that results from
this correction is what would have occurred had the samples been of equal size.

The overall results of statistics d and g were expressed as a weighted average and
weighted standard deviation. These measures were used to take into account the variation
in sample size across the studies. This average sample correlation (pm), which can be
considered an unbiased estimate of the population correlation (Hunter & Schmidt, 1990),

appears congruent with the relationship hypothesized by Johnson and Magaro (1987).

27

Table 7 Depression and Memory Measures

 

 

Depression Measures Memory Measures
Gibson (1982)

Psychiatrist rating lO-item word list
Zarit (1982) Zung Depression Scale 15-item shopping list task

Brief Symptom Inventory 15-item word list

O'Hara et al. (1986)
RDC diagnosis 20-item word list
Schedule for Affective Disorders and Schizophrenia (SADS-L) diagnosis
Center for Epidemiological Studies Depression Scale (CES-D)

Hart, Kwentus, Taylor, & Hamer (1987)
DSM-III diagnosis Selective Reminding Test
Hamilton Depression Scale

Hart, Kwentus, Taylor, & Harkins (1987)
DSM-III diagnosis Logical Memory Form I
Hamilton Depression Scale

Poitrenaud et al. (1989)

DSM-III diagnosis 20—item shopping list task
Montgomery and Asberg Depression Rating Scale
Geriatric Depression Scale

Feehan et al. (1991)
DSM-III diagnosis Logical Memory Form I
Geriatric Depression Scale

King et al. (1991)
DSM-III-R diagnosis lO-item word list
Hamilton Depression Scale

Lichtenberg et al. (1992)
Geriatric Depression Scale Logical Memory Form I

APPENDIX B

APPENDIX B

Psychometric Properties
First, psychometric analyses of the three depressive measures (i.e., HAM, BDI,

GDS) and the two explicit memory measures (i.e., LM, SRT) were undertaken. In
addition to these variables, a DEP factor and MEM factor were also analyzed. The DEP
factor was based on the 45 participants' three depression scores. Each score was
transformed into a z-score and these were aggregated to yield a DEP factor.
Consequently, as the respondents endorsed fewer depressive symptoms across the three
measures, their score on the DEP factor would become more negative or progressively
more positive as more symptoms were acknowledged. The MEM factor was based on the
summed z—scores of the individual's two explicit memory measures. As a result, lower
recall scores would produce more negative 2 scores for the MEM factor while greater
recall scores would result in more positive 2 scores. Z-scores were used as the basis of
these factors since the total score ranges varied across each test measure.

Initially, the construct validity of each of the depressive measures (i.e., HAM, BDI,
GDS) were empirically tested using confirmatory factor analysis. Confirrnatory factor
analyses (CF A; Hunter & Cohen, 1969) were performed on each of the depressive
measures based on all participants who were assessed across each of the three time
periods. For each CFA, all test items making up that depressive instrument were grouped
under one factor. Test contamination was then defined as any test item which had a
negative loading on its corresponding factor. Test items with negative factor loadings

were assumed to be measuring something different from the construct of depression being
28

29

assessed by the other test items and were removed to increase the internal consistency of
the measure. In each analysis, the correlations between the test items and their
corresponding factor were corrected for attenuation (Hunter & Cohen, 1969). Also,
CFA's were only performed on the depression measures since the scores for the two

memory measures were not based on multiple test items (Table 8).

Table 8 Item-Factor Correlations for Depressive Measures

 

HAM BDI GDS

 

Time 1 (p = 208)

Range .17 - .71 .26 - .63 .27 - .64

M .44 .45 .46
Time 2 (p = 120)

Range .20 - .72 .20 - .71 .17 - .76

M .48 .45 .45
Time 3 (p = 45)

Range .06 - .87 .12 - .63 .06 - .68

M .41 .38 .37

 

Analysis of the CFA'S for the BDI and GDS indicated that loadings between test
items and their corresponding factors were all positive across each of the three time
periods. The HAM, however, had 4 test items eliminated due to zero variance, and six
removed due to negative test item-factor loadings. In all subsequent analyses, the HAM
total score did not reflect these "contaminated" test items. Across each of the three

depressive measures, the mean loading was lowest in the third test period indicating a

lower level of internal consistency.

30

Next, stability was determined for the three depressive measures using Cronbach's
(1951) alpha. The alphas for each of the depressive measures are found in Table 9. For
each depression scale, the time period containing the highest mean test item-factor loading
resulted in the highest coefficient alpha and the time period with the lowest mean test

item-factor loading resulted in the lowest coefiicient alpha.

Table 9 Coefficient Alphas for Depressive Measures

 

 

HAM BDI GDS
Time 1 .75 .84 .89
Time 2 .77 .84 .88
Time 3 .70 .78 .83

 

Following this, construct validity of the DEP and MEM factor were empirically
tested by determining the loadings of the HAM, BDI, and GDS with the DEP factor and
the SRT and LM with the MEM factor. As Table 10 indicates, the factor loadings with
the DEP factor ranged from .71 to .96. The HAM loaded highest with the DEP factor at
time 3, the BDI loaded highest at time 1, and the GDS at time 2. The loadings of the LM
and SRT with the MEM factor were consistently in the low .7 range across the three time
periods with both memory tests having their highest loading with the MEM factor during
the second test period. The stability of the stayers' DEP and MEM factors were then
determined across the three time periods. The resulting coefficient alphas for the DEP and
MEM factors are reported in Table 11. Here, the alphas are found to be highest in the
second period while the lowest alpha varied with the measure. For the DEP factor, time 1

was associated with the lowest alpha, while for the MEM factor, time 3 had the lowest

31

alpha. As can be seen, reliability was moderately higher for the DEP factor than for the
MEM factor. Also, for those readers that prefer the raw scores of the depression and

memory measures, these are included in Table 12.

Table 10 Stayers Test-Factor Correlations

 

DEPl DEP2 DEP3
HAM .71 .79 .83
BDI .79 .72 .78
GDS .75 .96 .71

MEM] MEM2 MEM3
LM .73 .74 .71
SRT .73 .74 .71

 

Table 11 Stayers Coefficient Alphas for the DEP and MEM Factors

 

Time 1 Time 2 Time 3
DEP .79 .86 .82
MEM .69 .71 .67

 

32

Table 12 Stayers Raw Score Means and SDS

 

TIME 1 TIME 2 TIME 3
M SD M S_D .M. SD
HAM 3.6 4.1 3.1 3.3 2.5 3.0
BDI 6.8 5.9 6.1 5.3 6.4 5.1
GDS 7.1 5.1 5.4 4.8 6.2 4.8
LM 6.2 2 4 6 7 2 S 7 O 2 6

SRT 38.4 12.9 38.7 11.7 39.0 11.7

 

APPENDIX C

APPENDIX C

Stayers versus Decliners

The means and standard deviations for the stayers and decliners were initially
determined and are listed in Table 13. Secondly, point biserial treatment correlations were
calculated. The point biserial correlations represent the degree of relationship between the
group (i.e., stayers and decliners) and each of the above variables (e. g., age, HAM).F or
each point biserial correlation, a l-sided, 95% confidence interval with worst case, best
case, and odds ratio was calculated (Hunter, 1993a). A one-sided confidence interval was
used since it was hypothesized that decliners would be more depressed and have a poorer
memory than stayers. The 95% confidence interval gives the interval that 95 times out of
100 should contain the treatment correlation. It is bounded by the worst case, which
represents the lowest possible correlation that would be expected in 95 times out of 100,
and the best case, which represents the highest possible correlation that would be expected
in 95 times out of 100. The true treatment correlation should lie somewhere in the middle
of this interval. The odds ratio gives the probability that the correlation is positive. The
greater the odds ratio, the greater the chance that the two groups are different on that
variable. Variables with odds ratios that were equal to or greater than 3 :1 were
considered to have a high probability of being different between groups. A value of 3 :1
was chosen since it results in a 75% chance that there is a positive correlation in the
population. These results are presented in Table 14.

Finally, the standard deviation ration was calculated. The standard deviation ratio

( M ) is formed by the larger standard deviation divided by the smaller standard deviation.
33

34

It provides another index of differences in variation that may have existed between the two
groups on the variables assessed. For each of the standard deviation ratios, 3 2-sided 95%
confidence interval with worst and best case were calculated (Hunter, 1993c). This gives
us the lowest (i.e., worst case) standard deviation ratio and the highest (i.e., best case)
standard deviation ratio that would be expected in 95 times out of 100. The true standard
deviation ratio is expected to occur somewhere in the middle of these two values.
Intervals not containing 1.00 are considered to have a high probability of being different.
As can be seen in Table 15, only the variable MMSE filled this criteria. It appears that the
standard deviation is twice as big in the "decliners" as it is in the "stayers" (Decliners

MMSE S_D: 2.9; Stayers MMSE S_D; 1.6).

35

Table 13 Comparison of Stayers and Decliners Means and SDs

 

 

STAYERS DECLINERS
p = 45 p = 163
M 5.2 M S_D
AGE 68.0 09.8 70.5 09.4
GENDER 00.3 00.5 00.2 00.4
EDUCATION 13.3 02.9 13.2 02.8
HAM 03.6 04.1 03.5 03.7
BDI 06.8 05.9 07.3 06.0
GDS 07.1 05.1 07.0 05.8
DEP -00.1 02.5 00.0 02.7
LM 06.2 02.4 05.8 02.9
SRT 38.4 12.8 34.9 12.8
MEM 00.3 01.6 -00.1 01.8
PROB 00.8 00.4 00.8 00.4
# PROB 02.5 02.0 02.2 01.8
SR MEM 00.2 01.8 -00.1 01.8
MMSE 28.0 01.6 27.3 02.9

C/W 00.600 00.5 00.4 00.5

 

36
Table 14 Stayers and Decliners Correlation Coefficients

 

 

[ REL. WORST BEST ODDS
CASE CASE RATIO
AGE 0.11 1.00 0.00 0.22 16.62
GENDER 0.05 1.00 -0.06 0.17 03.55
EDUCATION 0.02 1.00 -0. 10 0.13 01.51
HAM 0.01 0.75 -0.13 0.14 01.12
BDI 0.04 0.84 -0.09 0.16 02.27
GDS 0.01 0.89 -0.11 0.13 01.15
DEP 0.01 0.87 -0.11 0.13 01.23
LM 0.05 0.79 —0.07 0.19 03.64
SRT 0.11 0.81 0.00 0.25 18.29
MEM 0.09 0.76 -0.02 0.24 10.17
PROB 0.06 0.77 -0.06 0.20 04.18
# PROB 0.06 0.66 -0.07 0.21 04.20
SR MEM 0.07 0.80 -0.05 0.20 04.88
MMSE 0.11 0.82 0.00 0.25 17.21

C/W 0.20 1.00 0.10 0.31 925.2

 

37
Table 15 Stayers and Decliners Standard Deviation Ratios

 

 

STANDARD WORST BEST
DEVIATION CASE CASE
RATIO ( y)
AGE 1.04 0.79 1.29
GENDER 1.03 0.33 1.34
EDUCATION 1.02 0.73 1.26
HAM 0.90 0.69 1.11
BDI 1.03 0.73 1.23
GDS 1.14 0.37 1.42
DEP 1.09 0.33 1.35
LM 1.19 0.91 1.43
SRT 1.00 0.76 1.24
MEM 1.13 0.86 1.40
PROB 1.11 0.34 1.37
# PROB 1.13 0.86 1.40
SR MEM 1.02 0.73 1.27
MMSE 1.86 1.41 2.30

C/W 1.02 0.78 1.27

 

APPENDIX D

APPENDIX D

Workshop versus Controls

The means and standard deviations for the control and workshop conditions were
initially computed and are listed in Table 16. Next, treatment correlations that had been
corrrected for attenuation were determined for each condition and time period
comparisons (Table 17). Here attenuation refers to the reduced size in the treatment
correlation due to the effects of random error of measurement. Consequently, correction
for attenuation restores, or increases the size of the correlation to what it would have been
had perfect measurement been used (Hunter, 1990). These treatment correlations can be
considered a measure of the size of the average treatment effect that occurred between the
different time frames. Following this, the treatment correlations were determined for a
95% confidence interval for both the control (Table 18) and workshop (Table 19)
condition.

Finally, the treatment correlations were compared between the control and
workshop conditions (Table 20). From these, one-sided, 95% confidence intervals with
worst case, best case, and odds ratios were calculated (Hunter, 1993a). One sided
confidence intervals were chosen since it was hypothesized that the workshop condition
should improve more than the control condition. Here, the 95% confidence interval
reflects that region which captures the actual population treatment correlation 95% of the
time. It is bounded by the worst and best case. The worst case is the lowest possible
treatment correlation that should result in 95 times out of 100. The best case is the

highest possible treatment correlation that should occur in 95 times out of 100. The odds
38

39

ratio, which indicates the chances of the treatment correlation being positive, was

considered to have a high probability of being different if it was 3 :1 or greater.

40
Table 16 Means and SDS for Control and Workshop Conditions.

 

A. Control Condition

TIME 1 TIME 2 TIME 3
DEP Factor
M 0.185 -0.334 -0.096
S_D 2.391 2.160 2.264
MEM Factor
M -0.060 -0.030 0.352
S_D 1.596 1.741 1.502

B. Woglgshop Condition

DEP Factor

M 0.521 -0.118 -0.238
g 3.018 2.897 2.488
MEM Factor

M -0.302 0.048 0.079
SQ 1.918 1.776 1.888

 

41
Table 17 Treatment Correlations for Control and Workshop Conditions.

 

TIME1&2 TIME2&3 TIME1&3

CONTROL

DEP -0.13 0.06 -0.07
MEM 0.01 0.14 0.16
WORKSHOP

DEP 0.12 -0.02 -0.15
MEM ‘ 0.11 0.01 0.12

 

Table 18 Control Treatment Correlations

 

 

WORST CASE BEST CASE ODDS RATIO

DEP Factor

TIME 1& 2 -0.26 0.01 0.07

TIME 2 & 3 -0.05 0.17 4.38

TIME 1 & 3 -0.21 0.08 0.28
MEM Factor

TIME1&2 -0.15 0.17 1.19

TIME 2 & 3 -0.05 0.33 7.87

TINIE 1 & 3 -0.02 0.34 12.48

 

42
Table 19 Workshop Treatment Correlations

 

 

 

 

WORST CASE BEST CASE ODDS RATIO
DEP Factor
TIME 1& 2 0.02 0.22 34.58
TINIE 2 & 3 -0. 12 0.07 0.05
TIME 1 & 3 -0.25 -0.05 0.01
MEM Factor
TIME 1 & 2 -0.01 0.23 15.69
TIME2&3 -0.11 0.13 1.25
TIME 1 & 3 -0.01 0.25 13.24
Table 20 Comparisons of Treatment Correlations
WORST CASE BEST CASE ODDS RATIO
DEP
TIME 1& 2 -0.26 0.75 3.67
TIME 2 & 3 -0.59 0.43 0.65
TIME 1 & 3 -0.59 0.42 0.65
MEM
TIL/IE 1 & 2 -0.41 0.61 1.69
TIME 2 & 3 -0.64 0.38 0.51
TIME 1 & 3 -0.54 0.46 0.82

 

APPENDIX E

APPENDIX E

_Ca_usal_itx

The path diagram was based on correlations that had been corrected for
attenuation. As can be seen in Table 21, due to the minimal change, the stability
coefficients were 1.00 between the depressive measures and between the memory

measures across the three time periods.

Table 21 Path Diagram

 

 

 

 

 

 

MEM] ”'0 sueuz ‘4'" 91413143

 

 

43

LIST OF REFERENCES

LIST OF REFERENCES

American Psychiatric Association. (1980). Diagnostic and statistical manual of mental
disorders (3rd ed). Washington, DC: American Psychiatric Association.

American Psychiatric Association. (1987). Diagpostic and statistical manual of mental
disorders. (Revised 3rd ed.). Washington, DC: American Psychiatric Association.

Anschutz, L., Camp, 0, Markley, R., & Kramer, J. (1985). Maintenance and

generalization of mnemonics for grocery shopping by older adults. Experimental Aging
Research, 1_1, 157-160.

Beck, A., Steer, R., & Garbin, M. (1988). Psychometric properties of the Beck

Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review. 8,
77-100.

Beck, A., Ward, C., Mendelson, M., Mock, J ., & Erbaugh, J. (1961). An inventory for
measuring depression. Archives of GeneraiPsychiatg, 4, 53-63.

Bhagia, N., & Pal, H. (1986). The effect of anxiety on memorization. Perspectives in
Psychological Researches. 2, 60-61.

Blackford, R., & LaRue, A. (1989). Criteria for diagnosing age-associated memory
impairment: Proposed improvements from the field. Developmenta_l
Neuropsychology, 5, 295-306.

Bowden, S., & Bell, R. (1992). Relative usefirlness of the WMS and WMS-R: A comment

on D'Elia et al. (1989). Jouthl of Clinical_and Experimental Neuropsychology, M,
340-346.

Buschke, H. (1973). Selective reminding for analysis of memory and learning. Journal of
Verbal Learning_and Verbflehavior. M, 543-550.

Buschke, H., & Fuld, P. (1974). Evaluating storage, retention, and retrieval in disordered
memory and learning. Neurology, 2_4, 1019-1025.

44

45

Cattell, R. (1966). The scree test for number of factors. Multivariate Behavioral Research,
1, 245.

Cronbach, L. (1951). Coefficient alpha and the internal structure of tests. Psychometrika,
1_6_, 297-334.

Craik, F. (1991). Memory functions in normal aging. In T. Yanagihara & R. Petersen
(Eds), Memtfl disorders reseaych and clinical practice (pp. 347-367). New York:
Marcel Dekker.

Crook, T. (1989). Diagnosis and treatment of normal and pathological memory
impairment in later life. Seminars in Neurology, 2, 20-30.

 

Crook, T., Bartus, R., Penis, S., Whitehouse, R, Cohen, G., & Gershon, S. (1986). Age
associated memory impairment: Proposed diagnostic criteria and measures of clinical
change-Report of a National Institute of Mental Health Work Group. Developmental

Neuropsychology, _2_, 261-276.

Derogatis, I., & Spencer, P. (1982). The Brief Symptom Inventory (BS1): I.
Administratiop, scoring, and procedures manual. Baltimore, MD: Johns Hopkins

University Press.

Erickson, R., & Scott, M. (1977). Clinical memory testing: A review. Psychological
Bulletin, 8_4, 1130-1149.

Feehan, M., Knight, R., & Partridge, F. (1991). Cognitive complaint and test performance
in elderly patients suffering depression or dementia. International Journal of Geria_t_ri_c
Psychiatgy, a, 287-293.

Fitzgibbon, M., Cella, D., & Sweeney, J. (1988). Redundancy in measures of depression.
Journal of Clinical Psychology, 451, 372-374.

Folstein, M., Folstein, S., & McHugh, P. (1975). "Mini-Mental State" A practical method
for grading the cognitive state of patients for the clinician. J ouml of Psychiatnl
Research, 12, 189-198.

Gibson, A. (1981). A further analysis of memory loss in dementia and depression in the
elderly. Britis_h Jourml of Clinical Psychology, 29, 179-185.

Hamilton, M. (1960). A rating scale for depression. Journal of Neurolcgy, Neurosurgery,
and Psychiatgg, 2_3_, 56-62.

Hannay, H., & Levin, H. (1985). Selective Reminding Test: An examination of the

equivalence of four forms. JournaLof Clinical and Experimental Neuropgrchology, 1,
251-263.

46

Harper, R., Kotik-Harper, D., & Kirby, H. (1990). Psychometric assessment of depression
in an elderly general medical population. Journal of Nervous and Mental Disea_sa, 13,
113-119.

Hart, R., Kwentus, J ., Taylor, J., & Hamer, R. (1987). Selective reminding procedure in
depression and dementia. Psychology and Aging, 2, 111-115.

Hart, R., Kwentus, J., Taylor, J ., & Harkins, S. (1987). Rate of forgetting in dementia and
depression. Journal of Consultingand Clinical Pavehology, M, 101-105.

Heath, J ., Grant, W., Kamps, C., & Margolin, E. (1991). Outpatient geriatric assessment:
Associations between referral sources and assessment findings. Journal of the Americap
Geriatrics Socieg 39 267-272.

’—,

Hedlund, J., & Vieweg, B. (1979). The Hamilton Rating Scale for Depression: A
comprehensive review. Journal of Operational Psychiatg, 19, 149-162.

Heise, D. (1969). Separating reliability and stability in test-retest correlation. American
Sociological Review, 35, 93-101.

Hill, R., Storandt, M., & Simeone, C. (1990). The effects of memory skills training and
incentives on free recall in older learners. Journal of Gerontology, 15, 227-232.

Hulicka, 1., & Grossman, J. (1967). Age-group comparisons for the use of mediators in
paired-associate learning. Journal of Gerontology, _2_2, 46-51.

Hunter, J. (1990). Attethtion due to random error of measurement. Unpublished
manuscript.

Hunter, J. (1993a). CORRECT: A progzam in Basica to correct correlations for various
artifacts (1st revision, Sep. 1993). Unpublished manuscript.

 

Hunter, J. (1993b). SE: A progrgam to compute standard errors. Unpublished manuscript.

Hunter, J. (1993c). CONFINT: A proggam in Basica to compute confidence intervals and
statistical power analysis (2nd revision, Sep. 1993). Unpublished manuscript.

Hunter, J. (1994). WITHINBAS- The manual- A prog1_'am to compute the analysis for
a "within subjects" des_igr_1. Unpublished manuscript.

Hunter, J., & Cohen, S. (1969). PACKAGE: A system of computer routines for the
analyses of correlational data. Education and Psychological Measurement. 2_9, 697-700.

Hunter, J. & Schmidt, F. (1990). Methods of meta-analysis. Newbury Park: Sage
Publication.

47

Johnson, M.,-& Magaro, P. (1987). Effects of mood and severity on memory processes in
depression and mania. Psychological Bulletin, Mi, 28-40.

Johnston, L., & Gueldner, S. (1989). Remember when? Using mnemonics to boost
memory in the elderly. Journmf Gerontological Nursing 15 22-26.

,—,

Kahn, R., Zarit, S., Hilbert, N., & Niederehe, G. (1975). Memory complaint and
impairment in the aged. Archives of General Psychiatg, 32, 1569-1573.

Kendall, P., Hollon, S., Beck, A., Hammen, C., & Ingram, R. (1987). Issues and
recommendations regarding use of the Beck Depression Inventory. Cognitive Therapy
and Research 11 289-299.

,.—,

King, D., Caine, E., Conwell, Y., & Cox, C. (1991). The neuropsychology of depression
in the elderly: A comparative study of normal aging and Alzheimer's disease. 113
Journal of Neurgrsychiatgy and Clinical Neurosgiencea, 3, 163-168.

Koenig, H., Meador, K., Cohen, H., & Blazer, D. (1988). Self-rated depression scales and
screening for major depression in the older hospitalized patient with medical illness.

Journal of the American Geriatrics Sociagy, 36, 699-706.

Lichtenberg, P., Manning, 0, & Turkheimer, E. (1992). Memory dysfunction in depressed
spousal caregivers. Clinical Gerontologist, M, 77-80.

Lichtenberg, P., Steiner, D., Marcopulos, B., & Tabscott, J. (1992). Comparison of the
Hamilton Depression Rating Scale and the Geriatric Depression Scale: Detection of
depression in dementia patients. Ps cholo 'cal Re orts 70 515-521.

,—’

Lowenthal, M., Berkman, P., Brisset, G., Buehler, J ., Pierce, R., Robinson, 3., & Trier,
M. (1967). Aging and mental disorder in San Francisco. San Francisco: Jossey-Bass.

Lyons, J ., Strain, J ., Hammer, J ., Ackerrnan, A., & Fulop, G. (1989). Reliability, validity,
and temporal stability of the Geriatric Depression Scale in hospitalized elderly.
International ioumal of Psychiatg in Medicine, L9, 203-209.

Maier, W. (1990). The Hamilton Depression Scale and its alternatives: A comparison of
their reliability and validity. In P. Bech & A. Coppen (Eds), The Hamilton Scales
(pp. 64-71). New York: Springer-Verlag.

Masur, D., Fuld, P., Blau, A., Crystal, H., & Aronson, M. (1990). Predicting development
of dementia in the elderly with the Selective Reminding Test. Journal of Clinical and

Experimental Neuropsychology, M, 529-538.

48
Masur, D., Fuld, P., Blau, A., Thal, L., Levin, H., & Aronson, M. (1989). Distinguishing
normal and demented elderly with the Selective Reminding Test. J our@ of Clinical and
Experimental Neuropsychology, M, 615-630.

Norris, J ., Gallagher, D., Wilson, A., & Winograd, C. (1987). Assessment of depression in
geriatric medical outpatients: The validity of two screening measures. Journal of the
American Geriatric Socieg 35 989-995.

,—,

O'Brien, K., & Glaudin, V. (1988). Factorial structure and factor reliability of the
Hamilton Rating Scale for Depression. Acta Pachiatrica Scandinavica, M, 113-120.

O'Hara, M., Hinrichs, J ., Wallace,R., Lemke, J ., & Kohout, F. (1986). Memory complaint
and memory performance in the depressed elderly. Psychology and Aging, 1, 208-214.

Parmalee, P., Lawton, M., & Katz, 1. (1989). Psychometric properties of the Geriatric
Depression Scale among the institutionalized aged. Journal of Consulting and Clinical
Psychology, 1, 331-338.

Poitrenaud, J., Moy, F., Girousse, A., Wolmark, Y., & Piette, F. (1989). Psychometric
procedures for analysis of memory losses in the elderly. Archives of Gerontology and
Geriatrics, (Suppl. 1), 173- l 83.

Prigatano, G. (1978). Wechsler Memory Scale: A selective review of the literature.
Journal of Clinical Psychology, 34, 816-832.

Rebok, G., & Balcerak, L. (1989). Memory self-efficacy and performance differences in
young and old adults: The effect of mnemonic training. Developmentalfi Psychology, 2,
714-721.

Rehm, L., & O'Hara, M. (1985). Item characteristics of the Hamilton Rating Scale for
Depression. Journal of Psychiatric Research 19 31-41.

9—3

Robertson-Tchabo, E., Hausman, C., & Arenberg, D. (1976). A classical mnemonic for
older learners: A trip that works! Educational Gerontology, 1, 215-226.

Rose, T., & Yesavage, J. (1983). Differential effects of a list-learning mnemonic in three
age groups. Gerontology, _2_9, 293-298.

Ruff, R., Quayhagen, M., & Light, R. (1988). Selective Reminding Test: A normative
study of verbal learning in adults. Jourial of Clinical and Experimental
Neuropsychology, 11 539-550.

_’

Russel, E. (1975). A multiple scoring method for the assessment of complex memory
functions. Journal of Consulting and Clinical Psychology, M, 800-809.

49

Salamero, M., & Marcos, T. (1992). Factor study of the Geriatric Depression Scale.
Acta Psychiatrica Scandinavica, 86, 283-286.

Schaffer, G., & Poon, L. (1982). Individual variability in memory training with the elderly.
Educational Gerontology, 8, 217-229.

Schmitt, E, Murphy, M., & Sanders, R. (1981). Training older adult free recall rehearsal
strategies. Journal of Gerontolcgy, 3_6_, 329-337.

 

Scogin, F., Storandt, M., & Lott, L. (1985). Memory-skills training, memory complaints,
and depression in older adults. Journal of Gerontolpgy, M, 562-568.

Shelton, M., & Parsons, O. (1987). Alcoholics' self-assessment of their
neuropsychological firnctioning in everyday life. Journai of Clinical Psychology, Q,
395- 403.

Smith, M., Glass, G., & Miller. (1980). Benefits of psychotherapy. Baltimore, MD: John
Hopkins University Press.

Smith, G., Ivnik, R., Petersen, R., Malec, J ., Kokmen, E., & Tangalos, E. (1991). Age-
associated memory impairment diagnoses: Problems of reliability and concerns for
terminology. Psychology and Aging, 9, 551-558.

Spitzer. R., Endicott, J ., & Robbins, E. (1978). Research Diagnostic Criteria: Rationale
and reliability. Archives of General Psychim, 3_6_, 773-782.

Storandt, M., Botwinick, J ., Danziger, W., Berg, L., & Hughes, C. (1984). Psychometric
differentiation of mild senile dementia of the Alzheimer type. Archives of Neurology,
4_1_, 497-499.

Stukenberg, K., Dura, J., & Kiecolt-Glaser, J. (1990). Depression screening scale
validation in an elderly, community-dwelling population. Jourpal of Consulting and
Clinical Psychology, 2, 134-138.

Tombaugh, T., & McIntyre, N. (1992). The Mini-Mental State Examination: A
comprehensive review. Journal of the AmericMGemtrics Society. 40, 922-935.

Tun, P., Perlmutter, L., Russo, P., & Nathan, D. (1987). Memory self-assessment and
performance in aged diabetics and non-diabetics. Ex erimental A 'n Research, _1_3,
151-157.

Wechsler, D., & Stone, C. (1945). Wechsler memoiy Scale. New York: Psychological
Corporation.

50

Weiss, 1., Nagel, C., & Aronson, M. (1986). Applicability of depression scales to the old
old person. JouMal of the American GeriMTic Society. 3_4, 215-218.

West, R., Boatwright, L., & Schleser, R. (1984). The link between memory performance,
self-assessment, and affective status. Experimental Aging Research 10 197-200.

3—’

Williams, J ., Little, M., Scates, S., & Blockman, N. (1987). Memory complaints and
abilities among depressed older adults. Journal of Consulting and Clinical Psychology,
M, 595-598.

Yesavage, J., Brink, T., Rose, T., Lum, 0., Huang, V., Adey, M., & Leirer, V. (1983).
Development and validation of a Geriatric Depression Screening Scale: A preliminary
report. Journal of Psychiatric Research, M, 37-49.

Yesavage, J., & Rose, T. (1983). Concentration and mnemonic training in elderly subjects
with memory complaints: A study of combined therapy and order effects. Psychiatgi
Research, 2, 157-167.

Yesavage, J., Rose, T., & Spiegel, D. (1982). Relaxation training and memory
improvement in elderly normals: Correlation of anxiety ratings and recall improvement.

Experimental Aging Research, 8, 195-198.

Yesavage, J., Sheikh, J ., Friedman, L., & Tanke, E. (1990). Learning mnemonics: Roles of
aging and subtle cognitive impairment. Psychology and Aging, 8, 133-137.

Zarit, S. (1982). Affective correlates of self-reports about memory of older people.
International Jourpai of Behavioral Geriatrics, 1, 25-35.

 

Zarit, S., Cole, K, & Guider, R. (1981). Memory training strategies and subjective
complaints of memory in the aged. The Gerontologist, M, 158-164.

   

"Illlllllllllllllllllllf