See

discussions, stats, and author profiles for this publication at: [Link]

Self-Efficacy and Memory Aging: The Impact of a

Memory Intervention Based on Self-Efficacy

Article in Aging Neuropsychology and Cognition July 2007

DOI: 10.1080/13825580701440510 Source: PubMed

CITATIONS READS

109 503

3 authors, including:

Dana Bagwell Alissa dark-freudeman

Gannon University University of North Carolina at Wilmington
33 PUBLICATIONS 258 CITATIONS 15 PUBLICATIONS 216 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Dana Bagwell on 05 June 2014.

The user has requested enhancement of the downloaded file.

Aging, Neuropsychology, and Cognition, 15: 302329, 2008
[Link]
ISSN: 1382-5585/05 print; 1744-4128 online
DOI: 10.1080/13825580701440510

Self-Efficacy and Memory Aging:

1744-4128
1382-5585/05
NANC
Aging, Neuropsychology, and Cognition
Cognition, Vol. 0, No. 0, May 2007: pp. 00

The Impact of a Memory Intervention

Based on Self-Efficacy
Memory
Robin L. Intervention
West et al. and Self-Efficacy

ROBIN L. WEST, DANA K. BAGWELL* AND ALISSA DARK-FREUDEMAN

University of Florida, Gainesville, FL, USA

ABSTRACT
A multifactorial training program was developed for older adults, to improve self-
efficacy and memory performance. Elements designed to raise self-efficacy were
integrated into class discussions, homework readings, and practice exercises, includ-
ing an emphasis on memory potential at any age, self-set goals, and opportunities for
mastery. Strategy training focused on association, organization, attention, imagery,
and PQRST. Significant improvements for the training group, as compared to a wait-
list control group, were observed for memory self-efficacy, locus of control, name
recall and story recall. Trained participants were using effective strategies more so
than the control group. Final test scores were predicted by self-efficacy, condition
assignment, and baseline ability (with some variation across the three tasks). These
findings suggest that an integrated and comprehensive training program that incorpo-
rates principles of self-efficacy theory has great potential for improving older adults
memory ability.

INTRODUCTION
Self-regulatory beliefs, such as self-efficacy, have the power to influence
performance over and above the impact of ability (Bandura, 1997). Extant
literature on social cognitive theory shows that self-efficacy acts as a regu-
lator of behavior, affecting task-related energetic activation in many
domains of function. Self-efficacy beliefs are important for memory per-
formance (Berry, 1999), school success (Schunk, 2003), organizational
skills (Wood & Bandura, 1989), exercise regimens (Conn, 1998), hearing

*Dana Bagwell is now at The Johns Hopkins University, Bloomberg School of Public Health and Center
on Aging and Health.
Address correspondence to: Robin Lea West, Ph.D., P.O. Box 112250-Psychology, University of
Florida, Gainesville, FL 32611, USA. E-mail: west51@[Link]

2007 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business
 MEMORY INTERVENTION AND SELF-EFFICACY 303

aid use (Smith & West, 2006), problem-solving (Artistico, Cervone, &
Pezzuti, 2003), and medical outcomes (Bandura, 1998). In recent years,
there is also growing recognition of the importance of self-efficacy for reg-
ulating cognitive success for older adults (Cervone, Artistico, & Berry,
2006; Stine-Morrow, Miller, & Hertzog, 2006). Although self-efficacy
beliefs are clearly influential in aging and memory, almost no memory
training programs have targeted self-efficacy beliefs. Among those that
have targeted beliefs in some way, few have been successful, that is, most
investigations do not provide evidence for performance increases co-
occurring with positive changes in beliefs about memory.1 To further
explore the value of a self-efficacy-based memory intervention, this study
investigates a comprehensive memory training program focusing on self-
efficacy.
Why should we target self-efficacy beliefs in memory training?
Memory interventions typically address common everyday memory activi-
ties such as remembering names or numbers. By the time a person is 65
years of age, they will have experienced years of added difficulty on such
tasks due to normal age-related changes in encoding (Craik, Anderson,
Kerr, & Li, 1995), retrieval (Zacks, Hasher, & Li, 2000), and attentional
(McDowd & Shaw, 2000) processes. Such deficits will lead to reduced
confidence for memory, which has been well-documented (Berry, 1999).
There are also strong social stereotypes about age-related losses in mem-
ory (Heckhausen, Dixon, & Baltes, 1989). Stereotypes act as a form of
social persuasion that serves to lower self-efficacy (Bandura, 1997). Thus,
overcoming a loss of confidence may be important for boosting and sus-
taining the impact of a memory intervention. In the context of a compre-
hensive memory intervention, this means that when initial efforts at
learning simple memory skills result in some success, trainees will feel
more self-efficacious. However, when more complex strategies are taught
or trainees attempt to apply these newly learned techniques in the real
world, moderate to high levels of self-efficacy may be needed to encour-
age continued effort (West, Welch, & Yassuda, 2000). Therefore, pro-
grams that raise self-efficacy along with performance are more likely to be
successful.

1
Our review of the literature does not include memory training studies in which there was no control
group (e.g., a general discussion or wait-list control group). Without a control group, it is not possible to
evaluate whether any observed changes in beliefs or performance are due to the elements of the inter-
vention or due to a practice or reactivity effect. For example, in one of the more interesting studies
focusing on subjective beliefs, a cognitive restructuring group showed comparable changes to a tradi-
tional memory training group on an immediate posttest, and then exceeded the training group on some
measures at a 9-week followup session (Caprio-Prevette & Fry, 1996). The meaning of these results is
unclear without a wait-list or practice-only control condition.
304 ROBIN L. WEST ET AL.

Motivational and self-regulatory issues have been a part of memory

intervention review articles for decades (e.g., Treat, Poon, Fozard, & Popkin,
1978). Past research has shown that subjective memory beliefs can be
changed during training (Floyd & Scogin, 1997), although changes in sub-
jective beliefs are more difficult to achieve than changes in performance (cf.
Floyd & Scogin, 1997; Verhaeghen, Marcoen, & Goossens, 1992). It is
surprising that few memory and aging studies have targeted changes in effi-
cacy,2 because there is an extensive literature on the power of increased self-
efficacy to build and sustain behavioral change (Bandura, 1997). While
some interventions have measured subjective memory, such as complaints
or self-ratings, many of these studies have used traditional strategy training
to try to achieve changes in beliefs, without incorporating any self-regulatory
elements in the intervention itself (e.g., Hill, Sheikh, & Yesavage, 1987). In
another case, self-regulatory memory monitoring was the focus of training
but no data on beliefs change were collected (Dunlosky, Kubat-Silman, &
Hertzog, 2003). The studies targeting change in beliefs have all had mixed
success, some showing significant change in beliefs with weak or no signifi-
cant changes in performance (cf. Rapp, Brenes, & Marsh, 2002; Schmidt,
Zwart, Berg, & Deelman, 1999; Schmidt, Berg, & Deelman, 2000), others
improving performance without consistent changes in beliefs (Rebok &
Balcerak, 1989; Hill et al., 1987; Scogin, Storandt, & Lott, 1985; Woolverton,
Scogin, Shackelford, & Black, 2001; Zarit, Cole, & Guider, 1981a; Zarit,
Gallagher, & Kramer, 1981b). Studies emphasizing strategy training plus
memory education to enhance positive beliefs have also shown mixed results
(Troyer, 2001; Turner & Pinkston, 1993).
For example, in one study comparing a traditional strategic approach
with an educational expectancy change intervention intended to change
beliefs (Best, Hamlett, & Davis, 1992), the training group improved perfor-
mance only and the expectancy group showed only altered beliefs. In addi-
tion, the relationship between measures of complaint and performance was
weak. The authors felt that memory training groups should not only provide
strategy training to facilitate memory performance, but should also address
context and expectations about memory (Best et al., 1992, p. 414). Other
investigators have also called for the combination of self-efficacy and strat-
egy training (Cervone et al., 2006; West et al., 2000), but to date, few mem-
ory interventions have taken this advice to heart. The two best examples will
now be considered.

2
Many studies have targeted subjective beliefs; this statement refers specifically to self-efficacy. Within
social cognitive theory, measures of self-efficacy are not equivalent to other types of self-ratings. A
detailed discussion of important differences between the alternative formats for assessing self-rated
memory is beyond the scope of this paper (see Hertzog & Hultsch, 2000; West, Dennehy-Basile, &
Norris, 1996).
 MEMORY INTERVENTION AND SELF-EFFICACY 305

Lachman, Weaver, Bandura, Elliott, and Lewkowicz (1992) developed

a 3 hour cognitive restructuring treatment and a 6 hour skills training class,
offering both, one, or neither to participants. Their cognitive restructuring
program emphasized several important concepts from social cognitive the-
ory, including adaptive and maladaptive beliefs, and the notion that memory
was controllable. They were able to show significant changes in beliefs, but
no greater memory performance for their trained participants than their con-
trol group. Valentijn et al. (2005) compared self-taught participants, who
used a training manual, to collective training (small group meetings), where
both groups received information on memory, aging, strategies, and self-
efficacy. Their training group showed improvement over controls only on a
few of their measures (Valentijn et al., 2005). In fact, no previous compre-
hensive interventions targeting self-efficacy change have found significant
training-related changes with concomitant improvements in performance on
most target variables. One possible reason is that scholars have not inte-
grated traditional skills training with elements designed to change beliefs.
Instead, the two approaches have been compared (Best et al., 1992) or
offered sequentially (Lachman et al., 1992). In other cases, it appears that the
training program may have had only a few topics related to restructuring of
beliefs (Rapp et al., 2002; Valentijn et al., 2005). It stands to reason that the
strongest strategy training program to simultaneously raise self-efficacy and
performance would be one where self-efficacy principles guide every ele-
ment of the program. That is the approach used here.
The goal of the Everyday Memory Clinic (EMC) training program was
to develop an effective multifactorial approach to training centered on
increasing self-efficacy. Bandura (1997) argues that mastery, verbal persua-
sion, vicarious experience (e.g., modeling), and physiological states are the
key factors determining self-efficacy beliefs. These elements were incorpo-
rated into this strategy training by building task mastery gradually, encour-
aging participants to set their own attainable goals, using a relaxed teaching
format, providing models for trainees and extensive practice for skill learn-
ing, and by reframing age stereotypes and negativity to focus on individual
learning potential, regardless of age. The approach was multifactorial, fol-
lowing previous successful memory interventions (Neely, 2000), including
educational materials about aging, and broad strategy training (including list,
name, story and attentional strategies), while incorporating elements
designed to raise self-efficacy.
We hypothesized that the elements designed to build efficacy, com-
bined with strategy training, would lead to improvements in memory self-
efficacy, as well as improvements in performance on three memory tasks. In
addition to self-efficacy beliefs, metamemory scales focusing on locus of
control, achievement, and anxiety were administered. Because these are
more general assessments of self-reported memory, rather than task-specific
306 ROBIN L. WEST ET AL.

memory self-efficacy scales, we did not expect significant changes on the

metamemory scales. Finally, individuals receiving a multifactorial interven-
tion should demonstrate increased use of the memory strategies they learned
in the program.

METHODS
Participants
Participants over age 50 were recruited for training. The 84 participants
(age range = 5492, M = 69.9, SD = 8.5) were healthy (mean self-rated
health = 3.0 on a scale from 1 excellent health to 10 very poor health),
well-educated (M = 15.3 years, SD = 2.9) volunteers. All participants were
recruited for a 6-week memory training class, using ads in newsletters or
newspapers and letters sent to research lists. Participants recruited at lifelong
learning programs were informed about the research aspects of the program
at the first session (N = 47); other participants knew about the research
project before signing up for training (N = 37). Seven groups, varying in size
from 9 to 18 people, were randomly assigned to control or training condi-
tions. Participants (N = 4) were dropped from the study due to recent strokes,
difficulty completing the interview, or use of anticholinergic medications.
Assessment Procedures and Materials
Memory Test Procedures
All measures were administered in group interviews, each lasting
1.5 to 2 hours. Instructions were presented bimodally, with individuals
reading printed instructions as the experimenter read them aloud. Level 1
and Level 2 tests were administered for each of three immediate recall
tasks shopping lists, names, and stories. To develop the Level 2 tests,
we added more to-be-remembered items or sentences to those items pre-
sented at Level 1. Four equivalent versions of each test were assigned to
groups by counterbalancing. At Level 1, 1 min of study time was permit-
ted, with 5 min retrieval time. Questionnaires were administered between
Level 1 and Level 2 tests. At each interview, participants were permitted
to take notes during study time on the Level 2 tests (they were told you may
take notes while you study, but the notes must be set aside during recall),
and they were allowed up to 5 min study time, with 5 min retrieval time for
names and lists, and 10 min to write the recalled story. Each Level 2 recall
test was followed by a strategy checklist. All to-be-remembered items
were presented in a notebook. Research assistants (one for every three or
four participants) were present to ensure that participants opened (for
encoding) and closed the notebook (for retrieval) at the appropriate time
and followed all instructions.
 MEMORY INTERVENTION AND SELF-EFFICACY 307

Pretest (week 1), posttest-1 (week 5), and posttest-2 (week 9) inter-
views were conducted. Waiting-list control participants were trained after
posttest-1 (N = 17; control at posttest-1 and training at posttest-2) or after
posttest-2 (N = 25; control participants at all three interviews). The main
group of trainees began the program after the pretest. All tests and question-
naires were administered at the pretest and posttest-2. Posttest-1 included
only the memory tests and strategy checklists. Primary analyses for training
impact on memory compared the posttest-1 results for the training group
(N = 42) with both wait-list groups (N = 42). As a replication, we examined
posttest-2, comparing control and trainees on self-efficacy and memory.
Finally, posttest-2 scores were compared to posttest-1 scores for the main
group of trainees, to examine maintenance.
Memory Test Materials
Shopping lists at Level 1 consisted of 15 items, and Level 2 lists con-
sisted of 35 items, with all lists partially categorizable (see West, Welch, &
Thorn, 2001 for list construction details). Participants were asked to write
down the words they recalled and correct recall was the dependent variable.
Level 1 name recall consisted of 12 names and faces and Level 2 con-
sisted of 24 names and faces. Each face was a close-up portrait in color pre-
sented in a 2.53-inch rectangle (e.g., Bell, 2000), and there were four faces
shown on each page of the notebook. During encoding, the persons first
name was printed with each face, using fairly familiar names, such as Rob-
ert, Terry, Oliver, Anne, Emma, and Charlotte. All versions of the faces
maintained a balance of age, gender, and ethnicity. During retrieval, partici-
pants saw the same 12 portraits in a different order and they were asked to
write down the recalled names. The dependent measure was the number
correct.
For the Level 1 story recall task, participants read a one-paragraph
story of eight sentences, typed on one notebook page, in 14-point black font.
The Level 2 stories had 24 sentences, and each one was a continuation of the
Level 1 story for that same version. All stories were taken from a set of 25
structurally equivalent stories (see Dixon, Hultsch, & Hertzog, 1989 for
details). Among the 25 stories, four stories were chosen by a multi-ethnic
group of adults as stories that represented common experiences for any eth-
nic group. After encoding, participants wrote the remembered story as pre-
cisely as possible. Stories were scored using a propositional analysis
(Dixon et al., 1989), with two scorers for each protocol; correlations
between raters ranged from .89 to .98 (average r = .94). The dependent mea-
sure was the percentage of propositions recalled. Final scores represented
the average of the two raters, who never differed by more than 7%.
The purpose for setting up two test levels, with the Level 1 items
embedded in the Level 2 tests, was to ensure that participants would see their
308 ROBIN L. WEST ET AL.

recall scores improve even at the first interview. That is, before we told them
that memory improvement was possible, they would have observed their
own improvement from Level 1 to Level 2 during the pretest interview. In
addition, the extended encoding interval for the Level 2 tests provided a
greater opportunity for participants to apply strategies if they wished,
because they had extra time and support (note-taking) for their strategic
efforts.
Strategy Checklists
Each Level 2 memory test was followed by a corresponding strategy
checklist where participants reported on their memory techniques, including
I concentrated and paid attention, I created a visual image . . ., I cov-
ered the material and tested myself. Self-report strategy measures are a
practical alternative to think-aloud protocols which are cumbersome to use
and complex to score. The three checklists were adapted from the Strategies
for Memory Questionnaire (SMQ; Camp, Markley, & Kramer, 1983). We
created independent variables from checklist responses, to compare the per-
formance of individuals who reported use of a particular strategy with those
who did not, and to examine changes in strategy use across training and control
groups.
Measures for Beliefs
The Memory Self-Efficacy Questionnaire (MSEQ-4) described
shopping list, object location, name, and story recall tasks (West, Thorn, &
Bagwell, 2003). For each task, there were five items ordered from most to
least difficult. Participants were asked to indicate how confident they were
that they could complete each task, on a scale from 0 to 100%. All responses
were averaged to reflect overall memory self-efficacy strength (SES,
range = 0100; = .93). Following standard procedures for assessing effi-
cacy change, an overall score was utilized rather than individual items or
subscales and the MSEQ-4 was administered following the baseline memory
tasks for appropriate comparison with later self-efficacy evaluations (see
West et al., 2001).
Three subscales of the Metamemory in Adulthood (MIA) assessment
(Dixon, Hultsch, & Hertzog, 1988) were utilized: achievement (It is impor-
tant to me to have a good memory; = .82), anxiety (I get upset when I
cannot remember something; = .88), and locus of control (If I were to
work on my memory, I could improve it; = .74). Responses were aver-
aged across all items for each scale (range = 15).
Due to the time required for these lengthy beliefs questionnaires, they
were administered only at the pretest and posttest-2. This prevented a test for
immediate training impact on self-efficacy at posttest-1, but was deemed
necessary to leave time at the posttest-1 session for class discussion.
 MEMORY INTERVENTION AND SELF-EFFICACY 309

Other Measures
Basic demographic information was collected, including age, educa-
tion, gender, and health information. In addition, the digit symbol subtest
(Wechsler, 1981) was administered as a general measure of processing
speed because it is known to impact strategy usage (Verhaeghen & Marcoen,
1994), and a vocabulary scale was utilized as a general measure of verbal
ability (Shipley, 1940).

Training Procedures and Materials

Training Materials
Every training session was scripted with a particular format and con-
tent, and didactic instruction on particular strategies, as noted in Table 1.
The instructor was observed and/or videotaped at intervals, to establish qual-
ity control throughout the program. All readings were written at the seventh-
grade reading level.
Class materials included a videotape (West, 1988), lists to practice
association and organization, short stories to practice the PQRST method
(West, 1985), and color photographs for attentional training. Memory exer-
cise packets for class practice and homework were developed for serial
recall, free recall and recognition of lists, names, and stories; object location
recall, and face recognition. In the packets, exercise difficulty was varied by
adding more items or setting time limits on encoding and retrieval. Trainees
were encouraged to work at their own pace, beginning with the easier exer-
cises, and to repeat the exercises as often as they wished. Table 1 indicates
how assigned exercises were distributed across the training program.

Training Procedures
The typical format for each 2 hour class involved a short overview of
previous readings, followed by small-group discussion of homework ques-
tions, overall group discussion, didactic instruction on strategies, and then
individual practice using the newly trained techniques. Training focused on
everyday memory tasks. The instructor facilitated discussion, answered
questions, and noted individual progress. All participants were sent home
with readings, homework questions, and practice exercise booklets designed
to help them learn about memory and aging and to master the trained strate-
gies (West, 2003).
The entire program was structured to enhance self-efficacy. Easier
strategies, simpler homework readings and exercises, and less demanding
memory skills were presented first, followed by more difficult ones, to
encourage early mastery. The training program required at least 3 hours per
week for those who were fully compliant with homework instructions. At
310 ROBIN L. WEST ET AL.

TABLE 1. Training program

Weekly format
Week 1
Assessment
Didactic strategy lesson: none
Homework: What is Memory? How Memory Works (13 pages); 4 homework questions;
1 story recall exercise
Week 2
Didactic strategy lesson: organization, association
Class practice: 9 shopping list recall exercises, 1 picture association exercise, 1 word association exercise
Homework: Age and Memory What Happens to Memory as You Grow Older (13 pages);
5 homework questions; 9 shopping list recall exercises+the 9 above
Week 3
Didactic strategy lesson: image-name-match method
Class practice: 14 names of people introduced on videotape
Homework: How to Remember Names (6 pages); 3 homework questions, videotape on
imagery techniques; 9 name and face memory exercises
Week 4
Didactic strategy lesson: PQRST, attention (active observation)
Class practice: 2 story recall exercises, 4 photograph recall exercises
Homework: Memory Improvement Techniques (30 pages); 3 homework questions; 6 object
location recall exercises, 6 story recall exercises
Week 5
Posttest-1
Didactic strategy lesson: None
Homework: Controlling Your Memory (4 pages); 3 homework questions; 20 min self-
selected memory challenge
Week 6
Presentation of posttest-1 change data
Discussion of DASH reading
Homework: The Memory DASH (DASH stands for Decide, Attend, Strategize, Hold)
(4 pages); list of 60 memory activities to challenge oneself at home
Week 7 and Week 8
No meeting
Week 9
Posttest-2
Self-efficacy elements in the training program
Enactive mastery
Easier strategies first
Each skill practiced in class and at home
Easier practice exercises first
Self-set goals for practice and achievement
In class success remembering names of class members
In class success with PQRST for a story
In class success with active observation
In class success discussing homework
Vicarious experience
Emphasis on learning from each other in class
Trainer gives examples, models each strategy
Whole group practices strategy together
Small-group discussion of homework questions
Personal stories provided by group members

(Continued)
 MEMORY INTERVENTION AND SELF-EFFICACY 311

TABLE 1. (Continued)

Verbal persuasion
Emphasize what you already know
Readings emphasis potential at any age
Videotape shows how to use imagery
Group provides social support
Research presented on strategy effectiveness and learning potential of seniors
Positive feedback from trainer in session
Positive feedback from trainer on posttest scores
Trainer reframes all negative comments to emphasize potential
Physiologic and affective states (reducing anxiety)
Multiple sessions to provide enough time for learning
Readings reviewed each week to help trainees understand
Emphasis on process/learning, not on 100% score
Class discussion shows that others have similar problems
Focus on potential in readings and class sessions
Trainer emphasizes that everyone has memory failures
Self-set goals

the same time, trainees were encouraged to work at their own pace, to iden-
tify their own ways to challenge their memory, and not to be concerned
about achieving a high score. In class, the instructor emphasized the poten-
tial for change, learning from peers, planning in order to remember, the
importance of applying effort, the benefits of practice, and the value of set-
ting realistic learning goals: Were all in here to learn more about memory
and how it affects us . . . Weve all had good [experiences] where we had
memory success and bad ones where we forgot. Some of us have more skill
than others, but in this room, we all need to work together . . . This class is
not about getting 100%. Its about each one of you learning and improving . . .
To facilitate peer modeling, other participants contributed answers for ques-
tions, as well as the instructor. Classmates reported their own memory fail-
ures and ways to overcome them, and the class proceeded at a relaxed pace,
both of which served to reduce anxiety. In the last group session, after post-
test-1, participants were explicitly informed about their improved scores.
Table 1 details the topics covered in each class, and the specific features of
the class that were designed to enhance self-efficacy.

RESULTS
Overview
Primary analyses for training effects used an interview (pretest and post-
test-1) by condition (training and control) analysis, with the three types of
312 ROBIN L. WEST ET AL.

memory tasks (list, name, and story recall) at Level 1 and Level 2 as the depen-
dent measures. In a replication, pretest and posttest-2 scores were compared
(note that this is a conservative test for the training effect because it includes
individuals who finished strategy training a month earlier). Training effects
were indicated by significant interactions of interview and condition. Subse-
quent analyses examined maintenance of training from posttest-1 to posttest-2
for those who were trained before posttest-1. All significant differences were
evaluated using 95% confidence intervals and effect sizes reported by SPSS.
Group Comparisons Pretest
In preliminary analyses, control and training conditions were compared
on basic demographic variables, and baseline ability measures. There were
no significant group differences for age, self-rated health, digit symbol, or
vocabulary scores, but there were significant differences between conditions
for education, F(1, 82) = 15.42, p < .001, 2 = .16. Post hoc tests indicated
that one of the groups assigned to the control condition (education M = 17.0
years, SD = 3.3) was significantly more educated than one of the groups
assigned to training (education M = 12.9 years, SD = 2.6). Therefore, educa-
tion was utilized as a covariate in the analyses.

Attrition
For posttest-1, 82% of the original sample returned, and for posttest-2,
77% returned. Those who missed posttest-1 had an average age similar to
that of the full sample (M = 70.9, SD = 8.7), but they were significantly older
than those who completed posttest-1 (M = 65.3, SD = 6.0), F(1, 82) = 5.7,
p < .05, 2 = .07. In addition, those who failed to attend posttest-2 showed a
significantly higher pretest score (M = 10.6, SD = 2.2), for Level 1 list recall
than those who attended (M = 9.4, SD = 2.0) posttest-2, F(1, 82) = 4.7, p < .05,
2 = .05. Individuals who signed up for research showed less attrition by
posttest-2 (18.9%) than individuals recruited in lifelong learning classes
(27.7%), 2 (1, 84) = 7.0, p < .01. However, most of the dropouts from the
lifelong learning classes left the program immediately after the pretest (N = 8).
After that, there were no differences in attrition as a function of recruitment,
and attendance did not vary between the two recruitment types, F(1, 84) <
1.0, p > .25. No other variables showed significant differences as a function
of attrition. These data suggest that factors other than ability influenced attri-
tion in this particular sample.

Memory Tasks
Training Gains Posttest-1
Multivariate analyses for name recall, story recall, and list recall were
examined in a repeated measures analysis using two levels for each memory
 MEMORY INTERVENTION AND SELF-EFFICACY 313

task across the pretest and posttest-1 interviews (with education as a covariate).
This analysis resulted in a main effect for education, F(3, 61) = 8.6, p < .001,
2 = .30, and an interaction of interview and condition, F(3, 61) = 7.0, p < .001,
2 = .26, due to higher scores on posttest-1 than pretest for those who were
trained. Multivariate interactions of level and education, F(3, 61) = 5.1, p <
.005, 2 = .20, and level and condition, F(3, 61) = 4.7, p < .005, 2 = .19,
were also significant due to greater pretest to posttest-1 gains on Level 2
than Level 1 for the trained group and greater gains from Level 1 to Level 2
for the more educated.
Univariate repeated measures ANCOVAs were used to further explore
these multivariate findings. These separate followup analyses for name
recall, story recall, and list recall showed significant training effects (interac-
tions of interview and condition) for story recall and name recall, but not list
recall (see Table 1). Name recall showed the expected training interaction,
with greater posttest-1 scores for the trained than the control group, F(1, 63)
= 5.6, p < .05, 2 = .08, and no significant group differences at pretest (see
Figure 1). Similarly for story recall, there were no significant condition dif-
ferences at pretest, but higher posttest-1 scores for the trained than the control
group, resulting in a significant interaction of interview and condition,
F(1, 63) = 16.3, p < .001, 2 = .20 (see Figure 1). The interaction of training
condition and level, F(1, 63) = 13.6, p < .001, 2 = .18, for story recall was
superseded by the three-way interaction of level, interview, and condition,
F(1, 63) = 4.1, p < .05, 2 = .06. Post hoc comparisons showed that at pre-
test, the trained participants remembered significantly less of the story than
the control group at Level 2 (although their performance was comparable at
Level 1), and at posttest-1, trained participants remembered significantly
more of the story than the control group at both Level 1 and Level 2 (see
Table 2). There were significant effects of education on list recall, F(1, 63) = 8.7,
p < .005, 2 = .12, name recall, F(1, 63) = 7.3, p < .01, 2 = .10, and story
recall, F(1, 63) = 25.9, p < .001, 2 = .29. Also, the more highly educated
participants gained more from Level 1 to Level 2, as shown by an
EducationLevel interaction for list recall, F(1, 63) = 11.7, p < .001, 2 = .16,
and name recall, F(1, 63) = 9.6, p < .005, 2 = .13.
Training Gains Posttest-2
The replication analyses examined posttest-2 gains relative to pretest
using a similar multivariate (name recall, story recall and list recall) repeated
measures (two task levels and two interviews) design. There was a main
effect of education on performance, as before, F(3, 58) = 4.4, p < .01,
2 = .19, and an interaction of education and level, F(3, 58) = 2.9, p < .05,
2 = .13. Significant training effects (an interaction of condition and inter-
view) were observed F(3, 58) = 6.8, p < .001, 2 = .26, as well as higher
scores for Level 2 than Level 1, F(3, 58) = 3.6, p < .05, 2 = .16, and a
314 ROBIN L. WEST ET AL.

FIGURE 1. Performance on three tasks across three interviews, as a function of training condition.

70
60
50

Stor y Recall
Level 2
40 control
30 trained
20
10
0
pretest posttest follow-up
Interview

20
18
16
Name Recall
Level 2

14 control
12 trained
10
8
6
pretest posttest follow-up
Interview

24
22
20
List Recall
Level 2

18 control
16 trained
14
12
10
pretest posttest follow-up
Interview

significant three-way interaction of level, interview, and condition, F(3, 58)

= 3.1, p < .05, 2 = .14.
Univariate repeated measures analyses comparing posttest-2 and the
pretest were used to follow-up these multivariate findings. These data repli-
cated the posttest-1 results. Story recall showed a significant training effect,
that is, an interaction of condition and interview, due to significantly higher
scores by the trained group at posttest-2, but not at pretest, F(1, 60) = 17.7,
p < .001, 2 = .23. The percentage scores for stories were significantly lower
for the Level 2 story than the Level 1 story, F(1, 60) = 5.3, p < .05, 2 = .08,
 MEMORY INTERVENTION AND SELF-EFFICACY 315

TABLE 2. Means and standard errors as a function of training

condition and interview session

Pretest Posttest-1 Posttest-2

Recall scores
List recall
Level 1
Control 9.7 (.37) 9.5 (.38) 9.5 (.40)
Training 9.4 (.43) 9.8 (.43) 10.2 (.29)
Level 2
Control 19.1 (.91) 19.0 (.76) 19.3 (1.1)
Training 20.0 (1.0) 20.5 (.87) 20.3 (.76)
Name recall
Level 1
Control 4.8 (.46) 5.2 (.48) 5.5 (.59)
Training 4.8 (.53) 6.1 (.55) 5.8 (.42)
Level 2
Control 10.7 (1.1) 11.7 (1.1) 10.6 (1.3)
Training 10.8 (1.2) 14.3 (1.2) 13.5 (.91)
Story recall
Level 1
Control 50.8 (2.1) 54.7 (1.9) 52.9 (3.3)
Training 53.5 (2.4) 63.8 (2.1) 62.3 (2.4)
Level 2
Control 43.2 (2.1) 47.4 (1.6) 38.1 (2.8)
Training 35.3 (2.4) 53.3 (1.8) 50.1 (2.0)
Number of strategies
List strategies
Control 3.8 (.22) 4.0 (.26) 4.1 (.31)
Training 3.4 (.25) 4.2 (.29) 4.2 (.22)
Name strategies
Control 4.3 (.30) 4.6 (.39) 4.4 (.45)
Training 3.8 (.33) 4.7 (.43) 5.4 (.31)
Story strategies
Control 5.0 (.31) 4.3 (.33) 5.2 (.44)
Training 4.1 (.34) 5.3 (.36) 5.8 (.31)
Beliefs
MIA-Achievement
Control 4.1 (.09) 3.9 (.10)
Training 4.0 (.06) 4.0 (.06)
MIA-Anxiety
Control 3.4 (.17) 3.4 (.15)
Training 3.4 (.11) 3.2 (.10)
MIA-Locus
Control 3.8 (.11) 3.8 (.09)
Training 3.8 (.07) 4.1 (.06)
Self-efficacy
Control 45.6 (3.8) 43.5 (3.5)
Training 47.3 (2.5) 53.5 (2.3)
316 ROBIN L. WEST ET AL.

FIGURE 2. Change in beliefs as a function of training condition (standardized scores).

Change in Beliefs (SDs)

4
3
2
1 control
0
training
1
2
3
4
Achieve Anxiety Locus Self-Eff.

due to its length and difficulty, and this effect was larger at posttest-2, F(1,
60) = 5.1, p < .05, 2 = .08, and interacted with education, F(1, 60) = 4.6, p <
.05, 2 = .07. Both name and story recall showed a significant three-way
interaction between level, interview, and condition (see Table 2). At Level 2,
the posttest-2 name recall scores for the trained group exceeded all other
scores (pretest for both groups, posttest-2 for the control group), F(1, 60) = 4.9,
p < .05, 2 = .08 (see Figure 1). Training gains were significant at Level 1
and Level 2 for story recall, F(1, 60) = 4.5, p < .05, 2 = .07, whereas the
control group showed modest gains at Level 1 and modest declines at Level
2 from pretest to posttest-2 (see Table 1). List recall results revealed that
higher gains from Level 1 to Level 2 occurred for the more highly educated
adults, F(1, 60) = 5.3, p < .05, 2 = .08. Name recall showed a similar inter-
action with education, F(1, 60) = 7.7, p < .01, 2 = .11.
Maintenance
Comparisons of posttest-1 and posttest-2 for those trained before post-
test-1 yielded important information about the maintenance of training
effects over 1 month, in comparison to testing-only gains that might have
occurred for the control group (individuals who obtained training between
posttest-1 and posttest-2 were excluded from this analysis; group means are
provided in the text where appropriate). In reporting this multivariate analy-
sis of the three memory tasks across two levels and two interviews, we have
focused only on data related to maintenance across interviews, rather than
repeat all findings noted above. The analyses showed significant effects for
condition, F(3, 42) = 5.4, p < .005, 2 = .28, with higher scores for the train-
ees, and a significant interaction of InterviewCondition, F(3, 42) = 4.8, p <
.01, 2 = .26, InterviewLevel, F(3, 42) = 4.1, p < .01, 2 = .23, and
InterviewLevelCondition, F(3, 42) = 2.8, p = .05, 2 = .17. The multi-
variate analysis also showed that those with higher education had higher
scores, F(3, 42) = 3.2, p < .05, 2 = .18, and there was a significant interac-
tion of InterviewLevelEducation, F(3, 42) = 4.4, p < .01, 2 = .24.
 MEMORY INTERVENTION AND SELF-EFFICACY 317

Follow-up univariate analyses for each memory task showed that there
was a greater improvement from posttest-1 to posttest-2 for Level 2 (18.5 to
19.4) lists than for Level 1 (9.3 to 9.7) lists, F(1, 44) = 8.2, p = .01, 2 = .16,
and that these gains were larger for those with more education (significant
InterviewLevelEducation interaction), F(1, 44) = 7.8, p < .01, 2 = .15.
The three-way interaction of InterviewLevelEducation was also signifi-
cant for names, F(1, 44) = 4.0, p = .05, 2 = .08, due to a decline in Level 2
scores from posttest-1 to posttest-2 for those with less education. For stories,
there was a significant decline between posttest-1 and posttest-2 for the con-
trol group (48.7 to 40.0) while the trained group (55.3 to 57.7) maintained
higher scores, F(1, 44) = 12.6, p < .001, 2 = .22, and there was a significant
interaction of InterviewLevelCondition, F(1, 44) = 6.8, p = .01, 2 = .14,
due to the fact that the trainees improved significantly on Level 1 scores
from posttest-1 to posttest-2 (31.5 to 34.1), while the control group declined
significantly (70.6 to 53.7) on Level 2 scores. These data showed that the
trainees retained higher scores than controls and maintained their posttest
gains during the maintenance period, and for stories, trainees even improved
their scores while the control group declined.

Memory Beliefs
Training Gains Posttest-2
Analyses for training effects for beliefs evaluated the pretest and post-
test-2 measures in a multivariate analysis with dependent variables of self-
efficacy, MIA anxiety, MIA achievement, and MIA locus of control, using
education as a covariate. As with performance, education showed a signifi-
cant main effect, F(4, 56) = 2.5, p < .05, 2 = .15. There was also evidence
for a training effect in the interaction of condition and interview, F(4, 56) =
2.6, p < .05, 2 = .16.
Univariate follow-up analyses of these multivariate findings showed sig-
nificant training effects for self-efficacy, but not for achievement and anxiety,
as expected. Surprisingly, locus of control also showed significant training
effects (see Figure 2). As shown in Table 1, there were modest reductions in
beliefs for the control group from pretest to posttest-2, and significant increases
for the trained group, in self-efficacy, F(1, 59) = 3.8, p < .05, 2 = .06, and locus
of control, F(1, 59) = 8.5, p < .005, 2 = .13. Education had a significant effect
only for anxiety, F(1, 59) = 4.2, p < .05, 2 = .07, with higher memory-related
anxiety on the MIA reported by the less well-educated participants.

Memory Strategies
People recorded their memory strategies on a checklist after each Level 2
memory task. Initially, we examined each individual strategy to determine
whether those who used the strategy performed better than those who did not
318 ROBIN L. WEST ET AL.

use the strategy. We also evaluated combinations of strategies, to fully

understand the impact that strategies may have had on performance.
Strategy Usage: List Recall
Organization was consistently related to improved list recall. There
was a significant main effect, with higher scores for those who used organi-
zation at the pretest, F(1, 81) = 9.6, p < .005, 2 = .11 (organize-yes M =
20.3, SE = .54; organize-no M = 15.4, SE = 1.6), at posttest-1, F(1, 64) = 7.4,
p < .01, 2 = .10 (organize-yes M = 14.9, SE = .37; organize-no M = 11.3,
SE = 1.3), and at posttest-2, F(1, 61) = 4.0, p < .05, 2 = .06 (organize-yes M =
20.5, SE = .62; organize-no M = 16.6, SE = 1.8). Associating items that
occurred together in the store also resulted in higher posttest-2 scores, F(1,
61) = 5.5, p < .05, 2 = .08 (associate-yes M = 22.3, SE = 1.1; associate-
no M = 19.2, SE = .69). Interestingly, rote repetition was associated with poorer
test results at posttest-2, F(1, 61) = 8.2, p < .01, 2 = .12 (rote-yes M = 17.1,
SE = 1.2; rote-no M = 21.0, SE = .65), but was not significant at posttest-1
(p > .25), and approached significance at pretest, p < .08 (rote-yes M = 17.1,
SE = 1.5; rote-no M = 20.1, SE = .56). We inferred that those who were using
rote memorization during the extended study time for Level 2 performed more
poorly because they were not utilizing more effective encoding methods.
All participants used more list strategies on each successive interview,
reporting more strategy usage at posttest-1 than pretest, F(1, 63) = 5.2, p < .05,
2 = .08, and more at posttest-2 than pretest, F(1, 60) = 9.9, p < .005, 2 =
.14, with no change in the number of list strategies reported (maintenance)
between posttest-1 and posttest-2 for those trained before posttest-1 (see
Table 2). We also evaluated the frequency with which control group mem-
bers and trainees employed the specific strategies, and we looked at the rela-
tionship between pretest and posttest strategy usage. These analyses
revealed a significant difference between the training and control groups at
posttest-1 for strategy usage, 2(2, 69) = 6.2, p < .05, due to the fact that 50%
of the trainees reported both association and organization as common strate-
gies and only 3% reported using neither method. In contrast, only 23% of the
control group used both association and organization and 13% reported nei-
ther strategy. We observed a shift in strategy usage from pretest to posttest-1
for the training group, 2(4, 30) = 8.7, p = .05; the training group showed a
significant shift from 23% not using organization at the pretest to only 7%
not using organization at posttest-1, 2(1, 30) = 7.0, p < .01. A similar shift
occurred for association, with the training group shifting to greater use of
association at posttest-1, 2(1, 30) = 5.1, p < .05.
Strategy Usage: Name Recall
Almost no one reported using the full image-name-match-method,
although many of its component strategies were used successfully to aid
 MEMORY INTERVENTION AND SELF-EFFICACY 319

memory. Attention to names led to significantly higher performance at the

pretest, F(1, 80) = 7.9, p < .01, 2 = .09 (attend-yes M = 12.5, SE = .84; attend-
no M = 8.9, SE = .98), and posttest-1, F(1, 64) = 6.4, p < .01, 2 = .09
(attend-yes M = 14.3, SE = .90; attend-no M = 10.6, SE = 1.2), but a trend in
that direction at posttest-2 was not significant (p > .10). Forming a familiar
name association did not result in significantly higher scores at the pretest
(p > .20), but was significant at posttest-1 (association-yes M = 14.6, SE =
.96; association-no M = 11.1, SE = .94), F(1, 64) = 6.4, p < .01, 2 = .09, and
posttest-2, F(1, 62) = 5.3, p < .05, 2 = .08 (association-yes M = 14.0, SE =
.93; association-no M = 10.8, SE = 1.04). Use of a familiar face association
provided a performance advantage at the pretest interview, F(1, 80) = 9.0,
p < .005, 2 = .10 (familiar face-yes M = 13.3, SE = .99; familiar face-no M =
9.4, SE = .83), as well as posttest-1, where those who used face associations
performed better (M = 15.5, SE = 1.16) than those who did not (M = 11.5,
SE = .90), F(1, 64) = 7.5, p < .01, 2 = .11. There was no advantage for
familiar face association at posttest-2. Finally, self-testers performed better at
the pretest, F(1, 80) = 4.1, p < .05, 2 = .05 (self-test-yes M = 13.8, SE = 1.6;
self-test-no M = 10.4, SE = .72), and posttest-2, F(1, 62) = 4.4, p < .05, 2 =
.07 (self-test-yes M = 14.7, SE = 1.2; self-test-no M = 11.6, SE = .86), but not
at posttest-1 (p > .50).
All participants showed an increase in name strategies across interview
sessions, reporting more strategies at posttest-1 than pretest, F(1, 63) =
6.9, p < .01, 2 = .10, and more at posttest-2 than pretest, F(1, 60) = 16.1, p
< .001, 2 = .21, with no change in the number of name strategies reported
(maintenance) between posttest-1 and posttest-2 for those trained before
posttest-1 (see Table 2). Those who combined name and face strategies did
well at posttest-2 (M = 15.0, SE = 1.4), in comparison to those who did not
use both strategies (M = 11.8, SE = .82), F(1, 62) = 6.4, p < .01, 2 = .09.
There was a trend for the trainees to use combined name and face strategies
more than controls at posttest-1 (57% trained vs. 40% control) and posttest-2
(76% trained vs. 62% control), and for the trained participants to use name
associations more at posttest-2 than control participants (30 vs. 18%), but these
differences were not significant in chi-square analyses. From pretest to posttest-
1, the control group had significantly more people who stayed with their previ-
ous strategies rather than shift, 2(1, 39) = 5.7, p < .05. In addition, trainees
shifted significantly toward use of face strategies from pretest to posttest-2 (37%
of the pretest non-users utilized this technique at posttest), 2(1, 43) = 4.0, p <
.05, but there was no significant strategy shift for controls.
Strategy Usage: Story Recall
Based on comparisons of those who used and did not use the tech-
niques (below), the PQRST (Preview, Question, Read, Summarize, Test)
was an effective strategy. Individuals were taught to scan the story (Preview)
320 ROBIN L. WEST ET AL.

for the main idea, then ask Questions about who-what-when-where-why (the
Ws strategy), Read the story, summarize the main points and key details and
then Test themselves. Participants who noted the main idea remembered a
higher percentage of the story at the pretest, F(1, 81) = 7.1, p < .01, 2 = .08
(main-yes M = 40.9, SE = 1.48; main-no M = 32.1, SE = 2.94), and at posttest-
2, F(1, 62) = 5.8, p < .05, 2 = .09 (main-yes M = 48.1, SE = 1.82; main-no
M = 37.3, SE = 4.09), but not at posttest-1 (p > .20). Those who used the Ws
strategy scored higher at the pretest (M = 43.1, SE = 2.06) than those who
did not endorse that strategy (M = 36.2, SE = 1.74), F(1, 81) = 6.6, p < .01,
2 = .08. Those who specifically noted names and places in the story also
showed a pretest advantage (M = 41.4, SE = 1.43), over those who did not
(M = 29.9, SE = 2.85), F(1, 81) = 12.8, p < .001, 2 = .14, and an advantage
at posttest-2, F(1, 62) = 4.2, p < .05, 2 = .06 (name-place-yes M = 47.4, SE
= 1.76; name-place-no M = 36.4, SE = 5.09), but not at posttest-1 (p > .20).
All participants showed an increase in story strategies across interview
sessions, reporting more strategies at posttest-2 than pretest, F(1, 60) = 11.5,
p < .001, 2 = .16, with no change in the number of story strategies reported
(maintenance) between posttest-1 and posttest-2 for those trained before post-
test-1 (see Table 2). Significantly more strategies were reported by trainees
than controls across the two posttests, F(1, 63) = 11.4, p < .001, 2 = .11. Chi-
square analyses were utilized to determine whether particular story strategies
were used more or less by trainees and control group members. One-third of
the training group shifted to use a Ws strategy on posttest-1 after not using it at
pretest, compared to only a few shifting in the control group, 2(1, 30) = 4.2,
p < .05. At posttest-2, nearly half of the trained participants reported the Ws
approach, 2(1, 43) = 5.4, p < 05, with relatively few in the control group using
this strategy. None of the control group used the PQRST. Four trainees at post-
test-1 and nine trainees at posttest-2 used the full PQRST method.
The strategy results indicate that all participants were spontaneously
trying more strategies on each successive assessment, which has been
reported previously (Camp, 1998). At the same time, it is clear that the more
effective strategies were being utilized more often by trainees, as evidenced
by the impact of the particular strategies they reported, their shifts to more
effective strategies after training, and their significantly higher scores,
especially on the more difficult Level 2 tests.3

3
Detailed tables showing strategy usage for all interviews are available on request. On many strategies, there
were also significant interactions between education and level, such that those with higher education per-
formed better on the Level 2 test relative to those with less education. These results were not included here
because they do not add anything to the primary test score results reported earlier. Also note that it was not
possible, for many individual strategies, to examine the impact of strategy usage in the same analyses with
condition effects, due to the differential distribution of strategy usage across conditions, e.g., there were a
number of people who utilized a PQRST strategy at posttest-2, but they were all in the training condition.
 MEMORY INTERVENTION AND SELF-EFFICACY 321

Prediction of Performance
Regression analyses were conducted to predict the level of perfor-
mance achieved on the final memory tests. For each task, the Level 1 pretest
score was considered an assessment of baseline ability. We began by pre-
dicting baseline scores separately for name recall, list recall, and story recall,
to identify demographic and ability factors (age, education, health rating,
vocabulary, digit symbol score, and gender) that had a significant impact on
pretest scores for each specific task. Then, those demographic and ability
factors that had a significant impact at baseline were entered along with
beliefs (self-efficacy and locus of control) and condition (as a dummy vari-
able) to predict scores on the last assessment (posttest-2 Level 2) for each of
the three memory tasks. Finally, we examined the impact of interactions,
using centered variables, for Self-efficacyCondition, ConditionBaseline
Ability, and Self-efficacyBaseline Ability (interactions with the MIA locus
of control scale were not examined because it was not a significant predictor
for any of the tests). It is useful to note that when independent variables are
entered together in multiple regression, each one is tested with the variance
removed for all others. It is therefore important to ensure that the indepen-
dent variables are not so highly correlated as to interfere with an independent
test of their predictive effects. Multicollinearity diagnostics were employed
to ensure this independence, and all tolerance levels were acceptable.
For list recall, scores at baseline (Level 1 pretest) were predicted sig-
nificantly, F(6, 67) = 4.4, p < .001, R2 = .28, by age ( = .30, p < .05) and
gender ( = .21, p < .05). Final trial scores were predicted significantly by
age ( = .37, p < .001), baseline ability ( = .26, p < .01), condition ( =
.20, p < .05), and self-efficacy ( = .35, p < .001), with over 50% of the vari-
ance accounted for in predicting final list recall scores, F(6, 56) = 13.5, p <
.001, R2 = .59. Gender and locus of control were not significant. When
examining interaction effects, condition showed a significant interaction
with self-efficacy ( = .29, p < .005), due to the fact that participants who
retained lower self-efficacy even after training showed the lowest scores and
those with high self-efficacy performed best overall. Baseline ability also
interacted with condition ( = .32, p < .005), as those with higher baseline
skills in the training group had higher posttest-2 scores than those with lower
baseline skills, and baseline ability accounted for a larger percentage of final
trial performance in the control group, as compared to the trained group.
This analysis suggests that, even though there was not a main effect for con-
dition in the MANCOVAs reported for list recall above, condition did have a
significant effect on final list recall scores, in interaction with other factors,
including self-efficacy.
For name recall, scores on the baseline pretest interview were predicted
significantly, F(6, 67) = 4.2, p < .001, R2 = .27, by digit symbol ( = .27, p < .05)
322 ROBIN L. WEST ET AL.

and vocabulary ( = .31, p < .01). Final trial scores were predicted signifi-
cantly by vocabulary ( = .29, p < .01), baseline ability ( = .45, p < .001),
and digit symbol ( = .32, p < .005), with over 50% of the variance
accounted for, F(6, 52) = 12.8, p < .001, R2 = .60. Self-efficacy, locus of
control, and condition had no direct impact on the final trial score, however,
the interaction of condition and baseline ability was significant ( = .43, p <
.001). This interaction occurred because the impact of baseline ability on
performance, significant in both groups, was larger in the control group,
accounting for more than twice as much variance as in the trained group.
These results confirm that when baseline ability is considered, training
condition has a significant impact on final name recall scores. Surprisingly,
self-efficacy did not have a significant effect on name recall when the basic
ability measures were included in the analysis.
For stories, the prediction of baseline test scores by demographic and
basic ability measures was significant, F(6, 67) = 4.5, p < .001, R2 = .29,
with age as a significant predictor ( = .32, p < .01). No other demo-
graphic or ability variables were significant. Final trial scores were pre-
dicted significantly by condition ( = .30, p < .01), baseline score ( = .40,
p < .001), age ( = .21, p = .05), and self-efficacy ( = .30, p < .005), F(5,
58) = 12.5, p < .001, R2 = .52. Locus of control had no significant effect (p >
.25). The interaction of self-efficacy and condition was a significant predic-
tor ( = .38, p < .005), due to higher scores for those who had high self-effi-
cacy after training and lower scores for control participants and trainees
with lower self-efficacy at posttest-2. The interaction of condition and base-
line ability ( = .23, p < .05) reflected the fact that baseline scores
accounted for twice as much variance in final test scores in the control
group, as compared to the trained group, although the impact of baseline
scores was significant in both groups. These results confirm the primary
analyses, and further support the notion that self-efficacy is a key vari-
able in memory improvement for stories.

DISCUSSION
The results of this training program verify that a self-efficacy based
approach to memory training for older adults can be successful, showing
improvements in memory self-efficacy and performance on most of our
everyday memory tasks, as well as increases in locus of control over mem-
ory. In regression analyses, the final levels of performance achieved after
training were predicted by self-efficacy, training condition, or interactions
between training condition and other factors, with baseline ability acting as
a significant predictor as well. On story recall, for example, final scores
were predicted by training condition, self-efficacy, baseline ability, age,
and interactions of condition with self-efficacy and ability. To our
 MEMORY INTERVENTION AND SELF-EFFICACY 323

knowledge, this is the first investigation of memory training showing per-

formance increases on most dependent measures, with concomitant
improvements in self-efficacy that exceeded the gains made by a waiting-
list control group.
Significant gains for the training group, over and above the control
group, were observed on Level 1 and Level 2 name and story recall, with
only modest changes on shopping list recall (see Figure 1). Gains by trainees
were maintained for 1 month after the class was completed, with additional
gains for story recall. The gains observed on test scores were mirrored, in
part, by changes in strategy usage, as reported in other studies (Dunlosky
et al., 2003; Saczynski, Rebok, Whitfield, & Plude, 2007; Troyer, 2001;
Verhaeghen & Marcoen, 1996; Woolverton et al., 2001). In regression anal-
yses, all final trial scores were predicted significantly by training condition
in interaction with baseline ability, among other factors. It is not clear why
shopping list recall showed very little improvement over time, even though
strategy usage increased. It may be that participants were able to use organi-
zation to selectively recall most of the categorizable items (about half of
each list), but found it very challenging to encode the miscellaneous items
on the randomly ordered shopping lists, as needed to reach higher levels of
recall.
In past training research, investigators have often operated on the
assumption that strategy change predicted score gains, without examining
strategies (West et al., 2000). It is not surprising that investigators have
shied away from assessing strategies directly there are few memory tests
that lend themselves to direct measures of strategy usage. Here, evidence
from the SMQ checklists validated previous work showing the increased
use of list strategies after training (Saczynski et al., 2007) and the benefits
of self-testing (Dunlosky et al., 2003). New evidence pointed to the
increased use of story and name strategies, and greater story strategy usage,
in particular, for trainees. More important for future studies are the data
showing that trainees mainly chose not to use the more complex, effortful
strategies, as has been noted by other investigators (e.g., Verhaeghen &
Marcoen, 1996). For example, few reported using the image-name-match-
method or the complete PQRST, preferring to use some elements from
these techniques (e.g., face association or the Ws method) as a partial strat-
egy rather than employing all of the steps (note that lack of endorsement of
the complex trained strategies also verifies that respondents were not sim-
ply checking off strategies due to demand characteristics). The fact that
higher scores on many of the tests were obtained by those who used partial
strategies suggests that the checklist serves as a useful method for exploring
strategic processing. It would be interesting to find out why participants
chose not to use the complex strategies, and how they decided which strate-
gic elements to apply. It would also be useful, in future training studies, to
324 ROBIN L. WEST ET AL.

train only the simpler elements of the complex strategies; perhaps more
trainees would benefit from training and use their newly learned strategies
if that were the case (Camp, 1998).
Changes in self-efficacy and locus of control were clearly observed in
the training condition and not the control group. Other investigators have
noted changes in control beliefs after training (Rapp et al., 2002; Turner &
Pinkston, 1993) or have seen improvements on single-item efficacy scales
that covary with performance (Hill et al., 1987; Rebok & Balcerak, 1989),
but no one has reported significant score improvements on most memory
measures that covaried with self-efficacy change (cf. Best et al., 1992;
Rebok & Balcerak, 1989; McDougall, 2000; Zarit et al., 1981a, 1981b).
Why was this program successful? Because the length of this training pro-
gram was comparable to other comprehensive training programs, we cannot
conclude that the results occurred because of a greater time commitment for
this program. Other than self-efficacy, the dependent measures in this study,
for the most part, were comparable to past intervention research. We believe
that the key variable explaining program impact is the way in which self-
efficacy was incorporated into our intervention. Unlike past investigations
(Lachman et al., 1992; Valentijn et al., 2005), this intervention program inte-
grated self-efficacy into every aspect of the program. Each element of the
training program (readings, class discussions, weekly homework, practice
exercises) was designed to enhance self-efficacy by providing easier exer-
cises or strategies first to foster mastery (West et al., 2000), by encouraging
self-paced work and self-set goals (Camp, 1998), and by focusing on ones
potential for learning through effort and goal setting rather than dwelling on
test score results (Dweck & Leggett, 1988; Elliott & Lachman, 1989).
Embedding efficacy change elements in every aspect of the training makes it
possible for participants to maximize their gains from memory training, just
as trainees in other domains have benefited from self-efficacy-based treat-
ments (Bandura, 1997).
The multivariate effect sizes show that the changes in beliefs were less
powerful than the performance changes, consistent with past metaanalyses
(cf. Floyd & Scogin, 1997; Verhaeghen et al., 1992). Although individual
investigations do not always mirror that pattern of weaker change for beliefs
(cf. Lachman et al., 1992; Rapp et al., 2002; Schmidt et al., 1999; Scogin
et al., 1985; Troyer, 2001), it does appear, as Floyd and Scogin (1997, p. 159)
report, change in subjective assessment of memory is difficult to achieve
and is best accomplished by addressing both the improvement of skills
(mnemonic training) and the development of more adaptive attitudes toward
memory performance (expectancy modification) (Floyd & Scogin, 1997,
p. 160). Even though we obtained significant improvements in both locus of
control beliefs and self-efficacy, locus of control did not predict final trial
scores. Our regression analyses revealed that responses on the MSEQ-4
 MEMORY INTERVENTION AND SELF-EFFICACY 325

predicted performance for list recall, even when baseline ability factors and
training condition effects were controlled for. For stories, scores were high-
est for trainees who had higher levels of self-efficacy. This supports the
position that changes in self-efficacy beliefs can have a significant impact on
performance outcomes (Bandura, 1997).
One issue to consider in evaluating this research was the education
level of the sample. On the one hand, a highly educated sample would be
expected to learn strategies without training, as spontaneous strategy usage
commonly occurs (Camp, 1998). In fact, there was some evidence that con-
trol participants tried more strategies on each subsequent interview, even
without training. To the extent that spontaneous improvements occurred for
controls, it would mitigate against significant training effects. On the other
hand, this well-educated population could have been more receptive to a
self-efficacy based approach than a less educated or lower functioning sam-
ple. It is possible that a less educated sample might not have been as eager to
do the extensive reading, practice, and discussion incorporated into this
training program.
This investigation uncovered evidence for significant changes in self-
efficacy and performance, and for prediction of most final test scores by
self-efficacy and ability interacting with condition. It would still be informa-
tive to establish that a training program that emphasizes self-efficacy is more
valuable than a traditional training program. It would also be interesting, in
more fine-tuned comparisons, to identify which elements for enhancing self-
efficacy are key to the observed effects, or to determine how much benefit
would be gained by practice only in an environment emphasizing positive
change and increasing confidence. With the homework required for this pro-
gram, participants became accustomed to memory effort at home, which
should encourage the transfer of newly learned techniques to everyday mem-
ory situations. Participants subjectively reported this type of transfer on
course evaluation forms, but future studies could address this type of transfer
more systematically.
Some prior evidence has suggested that training is more effective with
younger participants (e.g., Verhaeghen et al., 1992). Here, participants were
widely distributed in terms of age, with 25% younger than age 63 and 25%
over age 77. We made no special attempt to recruit particular ages, thus this
distribution suggests that there is a demand for memory training across a
wide age range. Age predicted final scores for list memory, and for story
memory, the task that showed the strongest training effects, but age was not
a significant predictor for name recall, a task that older adults often complain
about (Best et al., 1992; Leirer, Morrow, Sheikh, & Pariante, 1990). By
including a diverse age sample, we hoped to reinforce several aspects of the
self-efficacy approach: (1) participants saw a wide range of ability in the
group, emphasizing that being older does not control performance; (2) with
326 ROBIN L. WEST ET AL.

many skill levels represented, individuals who were performing better

served as models for the more impaired; and (3) individuals of all ability lev-
els were able to observe some gains, supporting the view that learning was
more important than scoring 100%.
This is the first memory training investigation focusing on self-efficacy
in training that has resulted in significant changes on most performance mea-
sures, along with significant memory self-efficacy change. We believe that
the primary difference between this approach and that of past investigators is
the extent to which self-efficacy issues were integrated into the training pro-
gram. Participants did not just talk about self-efficacy or myths about aging
during one or two of their classes. Instead, an emphasis on mastery, verbal
persuasion, reductions in anxiety, and modeling of skills occurred through-
out the training. Looking across studies, it appears that trainees have to be
convinced that they should believe in themselves and their ability to gain
from practice and effort. When this happens, trainees gain new memory
strategies, higher test scores, stronger beliefs in their ability to control mem-
ory, and higher self-efficacy. In future investigations of this approach to
training approach, we hope to be able to demonstrate that self-efficacy-based
training programs will yield long-term benefits for the daily memory ability
of seniors.

ACKNOWLEDGEMENTS
This research was supported by a grant from the Retirement Research Foundation. We are
grateful for their support. Special thanks to the Lifelong Learning Program at the Villages and
the Shepherd Center for permitting us to recruit participants in their educational programs.
Original manuscript received June 30, 2006
Revised manuscript accepted May 8, 2007
First published online July 25, 2007

REFERENCES
Artistico, D., Cervone, D., & Pezzuti, L. (2003). Perceived self-efficacy and everyday prob-
lem solving among young and older adults. Psychology and Aging, 18, 6879.
Ball, K., Berch, D. B., Helmers, K. F., Jobe, J. B., Leveck, M. D., Marsiske, M., Morris, J. N.,
Rebok, G. W., Smith, D. M., Tennstedt, S. L., Unverzagt, F. W., & Willis, S. L. (2002).
Effects of cognitive training interventions with older adults: A randomized control trial.
Journal of the American Medical Association, 288, 22712281.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W. H. Freeman.
Bandura, A. (1998). Health promotion from the perspective of social cognitive theory. Psy-
chology and Health, 13, 623649.
Bell, A. (2000). Memory pack. London: Carlton Books.
Berry, J. M. (1999). Memory self-efficacy in its social cognitive context. In T. M. Hess &
F. Blanchard-Fields (Eds.), Social cognition and aging (pp. 7096). San Diego, CA:
Academic Press.
 MEMORY INTERVENTION AND SELF-EFFICACY 327

Best, D. L., Hamlett, K. W., & Davis, S. W. (1992). Memory complaint and memory perfor-
mance in the elderly: The effects of memory-skills training and expectancy change.
Applied Cognitive Psychology, 6, 405416.
Camp, C. J. (1998). Memory interventions for normal and pathological older adults. In R. Schulz,
G. Maddox, & M. P. Lawton (Vol. Eds.), Annual review of gerontology and geriatrics:
Vol 18. Focus on interventions research with older adults (pp. 155189). New York:
Springer.
Camp, C. J., Markley, R. P., & Kramer, J. J. (1983). Spontaneous use of mnemonics by
elderly individuals. Educational Gerontology, 9, 5771.
Caprio-Prevette, M. D., & Fry, P. S. (1996). Memory enhancement program for community-
based older adults: Development and evaluation. Experimental Aging Research, 22, 281304.
Cervone, D., Artistico, D., & Berry, J. M. (2006). Self-efficacy and adult development. In
C. H. Hoare (Ed.), Handbook of adult development and learning (pp. 169195). London:
Oxford University Press.
Conn, V. S. (1998). Older adults and exercise: Path analysis of self-efficacy related con-
structs. Nursing Research, 47, 180189.
Craik, F. I. M., Anderson, N. D., Kerr, S. A., & Li, K. Z. H. (1995). Memory changes in nor-
mal ageing. In A. D. Baddeley, B. A. Wilson, & F. N. Watts (Eds.), Handbook of memory
disorders (pp. 211241). Chichester, UK: John Wiley.
Dixon, R. A., Hultsch, D. F., & Hertzog, C. (1988). The Metamemory in Adulthood (MIA)
Questionnaire. Psychopharmacology Bulletin, 24, 671688.
Dixon, R. A., Hultsch, D. F., & Hertzog, C. (1989). A manual of twenty-five three-tiered
structurally equivalent texts for use in aging research. Collaborative Research Group on
Cognitive Aging Technical Report No. 2. University of Victoria and Georgia Institute of
Technology.
Dunlosky, J., & Hertzog, C. (2001). Measuring strategy production during associative learn-
ing: The relative utility of concurrent versus retrospective reports. Memory & Cognition,
29, 247253.
Dunlosky, J., Kubat-Silman, A. K., & Hertzog, C. (2003). Training monitoring skills
improves older adults self-paced associative learning. Psychology and Aging, 18,
340345.
Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to motivation and person-
ality. Psychological Review, 95, 256273.
Elliott, E., & Lachman, M. E. (1989). Enhancing memory by modifying control beliefs, attri-
butions, and performance goals in the elderly. In P. S. Fry (Ed.), Psychological perspec-
tives of helplessness and control in the elderly (pp. 339367). Amsterdam: Elsevier-North
Holland.
Floyd, M., & Scogin, F. (1997). Effects of memory training on the subjective memory func-
tioning and mental health of older adults: A meta-analysis. Psychology and Aging, 12,
150161.
Heckhausen, J., Dixon, R. A., & Baltes, P. B. (1989). Gains and losses in development
throughout adulthood as perceived by different adult age groups. Developmental Psychol-
ogy, 25, 109121.
Hertzog, C., & Hultsch, D. F. (2000). Metacognition in adulthood and old age. In F. I. M.
Craik & T. A. Salthouse (Eds.), The handbook of aging and cognition (pp. 417466).
Mahwah, NJ: Erlbaum.
Hill, R. D., Sheikh, J. I., & Yesavage, J. (1987). The effect of mnemonic training on per-
ceived recall confidence in the elderly. Experimental Aging Research, 13, 185188.
Lachman, M. E., Weaver, S. L., Bandura, M., Elliott, E., & Lewkowicz, C. J. (1992). Improv-
ing memory and control beliefs through cognitive restructuring and self-generated strate-
gies. Journal of Gerontology, Psychological Sciences, 47, 293299.
328 ROBIN L. WEST ET AL.

Leirer, V. O., Morrow, D. G., Sheikh, J. I., & Pariante, G. M. (1990). Memory skills elders
want to improve. Experimental Aging Research, 16, 155158.
Lineweaver, T. T., & Hertzog, C. (1998). Adults efficacy and control beliefs regarding mem-
ory and aging: Separating general from personal beliefs. Aging, Neuropsychology, and
Cognition, 5, 264296.
McDougall, G. (1995). Memory self-efficacy and strategy use in successful elders. Educa-
tional Gerontology, 21, 357373.
McDougall, G. J. (2000). Memory improvement in assisted living elders. Issues in Mental
Health Nursing, 21, 217233.
McDowd, J. M., & Shaw, R. J. (2000). Attention and aging: A functionalist perspective. In F.
I. M. Craik & T. A. Salthouse (Eds.), The handbook of aging and cognition (pp. 221292).
Mahwah, NJ: Erlbaum.
Neely, A. S. (2000). Multifactorial memory training in normal aging: In search of memory
improvement beyond the ordinary. In R. D. Hill, L. Backman, & A. S. Neely (Eds.), Cog-
nitive rehabilitation in old age (pp. 6380). Oxford: Oxford University Press.
Rapp, S., Brenes, G., & Marsh, A. P. (2002). Memory enhancement training for older adults
with mild cognitive impairment: A preliminary study. Aging and Mental Health, 6, 511.
Rebok, G. W., & Balcerak, L. J. (1989). Memory self-efficacy and performance differences in
young and old adults: The effect of mnemonic training. Developmental Psychology, 25,
714721.
Saczynski, J. S., Willis, S. L., & Schaie, K. W. (2002). Strategy use in reasoning training with
older adults. Aging, Neuropsychology, and Cognition, 9, 4860.
Saczynski, J. S., Rebok, G. W., Whitfield, K. E., & Plude, D. L. (2007). Spontaneous production
and use of mnemonic strategies in older adults. Experimental Aging Research, 33, 273294.
Schunk, D. H. (2003). Self-efficacy for reading and writing: Influence of modeling, goal set-
ting and self-evaluation. Reading & Writing Quarterly, 19, 159172.
Schmidt, I. W., Zwart, J. F., Berg, I. J., & Deelman, B. G. (1999). Evaluation of an interven-
tion directed at the modification of memory beliefs in older adults. Educational Gerontol-
ogy, 25, 365385.
Schmidt, I. W., Berg, I. J., & Deelman, B. G. (2000). Memory training for remembering texts
in older adults. Clinical Gerontologist, 21, 6790.
Scogin, P., Storandt, M., & Lott, L. (1985). Memory-skills training, memory complaints, and
depression in older adults. Journal of Gerontology, 40, 562568.
Shipley, W. C. (1940). A self-administering scale for measuring intellectual impairment and
deterioration. Journal of Psychology, 9, 371377.
Smith, S. L., & West, R. L. (2006). Hearing aid self-efficacy of new and experienced hearing
aid users. Seminars in Hearing, 27, 325329.
Stine-Morrow, E. A. L., Miller, L. M. S., & Hertzog, C. (2006). Aging and Self-Regulation in
Language Understanding. Psychological Bulletin, 132, 582606.
Treat, N. C., Poon, L. W., Fozard, J. L., & Popkin, S. J. (1978). Toward applying cognitive
skill training to memory problems. Experimental Aging Research, 4, 305320.
Troyer, A. K. (2001). Improving memory knowledge, satisfaction, and functioning via an
education and intervention program for older adults. Aging, Neuropsychology and Cogni-
tion, 8, 256268.
Turner, M. L., & Pinkston, R. S. (1993). Effects of a memory and aging workshop on nega-
tive beliefs of memory loss in the elderly. Educational Gerontology, 19, 359373.
Valentijn, S. A. M., van Hooren, S. A. H., Bosma, H., Touw, D. M., Jolles, J., van Boxtel,
M. P. J., & Ponds, R. W. H. M. (2005). The effects of two types of memory training
on subjective and objective memory performance in healthy individuals aged 55
years and older: A randomized controlled trial. Patient Education and Counseling,
57, 106114.
 MEMORY INTERVENTION AND SELF-EFFICACY 329

Verhaeghen, P., & Marcoen, A. (1994). Production deficiency hypothesis revisited: Adult age
differences in strategy use as a function of processing resources. Aging and Cognition, 1,
323338.
Verhaeghen, P., & Marcoen, A. (1996). On the mechanisms of plasticity in young and older
adults after instruction in the method of loci: Evidence for an amplification model. Psy-
chology and Aging, 11, 164178.
Verhaeghen, P., Marcoen, A., & Goossens, L. (1992). Improving memory performance in
the aged through mnemonic training: A meta-analytic study. Psychology and Aging, 7,
242251.
Verhaeghen, P., Van Ranst, N., & Marcoen, A. (1993). Memory training in the community:
Evaluations by participants and effects on metamemory. Educational Gerontology, 19,
523532.
Wechsler, D. (1981). Wechsler Adult Intelligence Scale-Revised (WAIS-R) Manual. San
Antonio, TX: Psychological Corporation.
West, R. L. (1985). Memory fitness over forty. Gainesville, FL: Triad.
West, R. L. (1988). Memory training with imagery Bethesda, MD: Memory Assessment Clin-
ics, Inc. [45-minute videotape].
West, R. L. (2003). Everyday memory clinic workbook. Gainesville, FL: Everyday Memory
Clinic.
West, R. L., Dennehy-Basile, D., & Norris, M. P. (1996). Memory self-evaluation: The
effects of age and experience. Aging, Neuropsychology, and Cognition, 3, 6783.
West, R. L., Welch, D. C., & Yassuda, M. S. (2000). Innovative approaches to memory train-
ing for older adults. In R. D. Hill, L. Backman, & A. S. Neely (Eds.), Cognitive rehabilita-
tion in old age (pp. 81105). Oxford: Oxford University Press.
West, R. L., Welch, D. C., & Thorn, R. M. (2001). Effects of goal-setting and feedback on
memory performance and beliefs among older and younger adults. Psychology and Aging,
16, 240250.
West, R. L., Thorn, R. M., & Bagwell, D. K. (2003). Memory performance and beliefs as a
function of goal setting and aging. Psychology and Aging, 18, 111125.
Willis, S. L., & Schaie, K. W. (1994). Cognitive training in the normal elderly. In F. Forette,
U. Christen, & F. Boller (Eds.), Cerebral plasticity and cognitive stimulation
(pp. 91113). Paris: Fondation Nationale de Gerontologie.
Wood, R. E., & Bandura, A. (1989). Impact of conceptions of ability on self-regulatory mech-
anisms and complex decision making. Journal of Personality and Social Psychology, 56,
407415.
Woolverton, M., Scogin, F., Shackelford, J., & Black, S. (2001). Problem-targeted memory
training for older adults. Aging, Neuropsychology, and Cognition, 8, 241255.
Zacks, R. T., Hasher, L., & Li, K. Z. H. (2000). Human memory. In F. I. M. Craik & T. A.
Salthouse (Eds.), The handbook of aging and cognition (pp. 293358). Mahwah, NJ:
Erlbaum.
Zarit, S. H., Cole, K. D., & Guider, R. L. (1981a). Memory training strategies and subjective
complaints of memory in the aged. The Gerontologist, 21, 158164.
Zarit, S. H., Gallagher, D., & Kramer, N. (1981b). Memory training in the community aged:
Effects on depression, memory complaint, and memory performance. Educational Geron-
tology, 6, 1127.
View publication stats