Content uploaded by Jeffrey Dean Webster
Author content
All content in this area was uploaded by Jeffrey Dean Webster
Content may be subject to copyright.
Journal of Gerontology:
PSYCHOLOGICAL SCIENCES
1993,
Vol. 48, No. 5.P256-P262
Copyright 1993 by The Gerontological
Sociery
of America
Construction and Validation
of the Reminiscence Functions Scale
Jeffrey Dean Webster
Department of Psychology, Vancouver Community College, Langara Campus, British Columbia.
This research introduces the Reminiscence Functions Scale (RFS), a 43-item questionnaire that can be used to assess
reminiscence functions over the lifecourse. Adults (710) ranging in age from 17 to 91 (mean age = 45.76 years)
completed a 54-item Reminiscence Functions Scale-prototype measure, the results of which were submitted to a
principal components analysis. Results indicated the viability of a 43-item,
7-factor
solution with good reliability.
Factors were
labeled:
Boredom Reduction, Death Preparation, Identity/Problem-Solving, Conversation, Intimacy
Maintenance, Bitterness Revival, and Teach/Inform. A separate validity study demonstrated the predictive validity of
the RFS. Directions for future research are discussed.
T
HE process of recapturing salient memories from one's
personal past continues to interest researchers con-
cerned with naturalistic memory dynamics (Webster & Cap-
peliez, 1993). As a vehicle for examining ecologically valid
memory, reminiscence offers researchers a window into not
only the mechanics of memory, but its functions as well.
Unfortunately, although myriad potentially adaptive func-
tions of reminiscence have been postulated in the literature
(e.g., improving self-esteem, coping with depression,
self-
understanding, death preparation, etc.), there currently exists
no reliable and valid questionnaire measure of multiple remi-
niscence functions.
Despite the rapid expansion of published articles concern-
ing some aspect of reminiscence, many reviewers (Haight,
1991;
Kovach, 1990; Merriam, 1980; Molinari & Reichlin,
1984-1985;
Moody, 1988; Romaniuk, 1981; Thornton &
Brotchie, 1987; Webster & Cappeliez, 1993) have indicated
that there remain several important obstacles to future pro-
gress.
These faults concern problems with operational
definitions, sampling characteristics, and methodological
shortcomings. This article addresses the latter concern, spe-
cifically the lack of psychometrically sound instrumentation.
Lack of Valid and Reliable Measures
One of the most common stumbling blocks to integration
of existing evidence is the paucity of psychometrically sound
instruments for assessing various reminiscence dimensions.
Without such assessment tools, researchers have frequently
been forced to assemble ad hoc questions tailored to their own
unique research agendas. In the majority of cases, reliability
and validity data have either not been presented, or are of
questionable quality. In order for meaningful comparisons
between research projects to occur, clinical and experimental
investigators need a readily available, psychometrically
sound instrument that produces comparable scores across
studies.
To my knowledge, there has been only one published
attempt to construct a paper-and-pencil measure of reminis-
cence functions. Romaniuk and Romaniuk (1981) developed
a measure called the Reminiscence Uses Scale (RUS), a 13-
item questionnaire in which subjects responded either yes,
no,
or not sure to the uses of reminiscence listed in the scale.
According to Romaniuk and Romaniuk, all items were intui-
tively generated by the researchers following suggestions
contained in the literature. The final sample in the study was
composed of 91 highly educated older adults (mean age =
78.5,
range = 58-98) who were —80% female and were
living in one of three new retirement communities. A factor
analysis revealed three factors — labeled (1) Self-Regard/
Image Enhancement, (2) Present Problem-Solving, and (3)
Existential/Self-Understanding — accounting for 17.7%,
16.6%,
and 13.2% of the total variance, respectively.
Romaniuk and Romaniuk's (1981) work represents a good
initial attempt to identify reminiscence uses in a homogene-
ous sample of elderly adults. It suffers from several limita-
tions,
however, which render it less than optimal for its stated
purpose. The most serious shortcomings are listed below. In
terms of its sample, the preponderance of females precludes
an analysis on the basis of gender; the ethnic composition is
not given, eliminating a discussion of potential ethnic differ-
ences;
and the homogeneous nature of the sample in terms of
education, health, and age characteristics limits its generali-
zability.
In terms of its psychometric properties, the coefficients of
internal consistency are modest (i.e., .64, .69, and .57 for
factors 1,2, and 3, respectively), the total number of items
constituting the scale are truncated at 13, which limits the
number of factors that could potentially emerge, and some
inclusion criteria for item retention are somewhat problem-
atic.
For example, several factor loadings are quite conserva-
tive (i.e., below .49), and items were assigned to factors on
the basis, "that the factors make intuitive or theoretical
sense" (p. 480). The latter is cause for some consternation,
because item 5 in the scale (i.e., to entertain) loaded .48 on
factor 3 and .47 on factor 1, yet was included, without
adequate explanation, in the latter factor. Finally, no validity
checks using external, psychometrically established mea-
sures in a predictive or concurrent fashion were conducted.
Recently, Wong and Watt (1991) discussed how various
types of reminiscence may be differentially related to suc-
P256
REMINISCENCE FUNCTIONS SCALE
P257
cessful aging. Through content coding of verbatim reminis-
cence protocols, they derived six types of reminiscence that
they labeled: integrative, instrumental, transmissive, escap-
ist, obsessive, and narrative. Their results suggest that higher
levels of integrative and instrumental, and lower levels of
obsessive, reminiscence types are related to successful aging.
Wong and Watt's classification system is a welcome shift
toward methodological refinement and an increased focus on
the multidimensional nature of reminiscence procedures. Un-
fortunately, there are two disadvantages to their approach.
First, it is relatively time-consuming and labor-intensive,
because protocols must be elicited from subjects in a face-to-
face format, and raters must be trained to code the resulting
qualitative data.
Second, their work introduces a crucial confound between
a taxonomic division, or type, and a specific memorial use,
or function. In the operational definitions of their types,
which they explicitly state are "mutually exclusive catego-
ries"
(p. 276), they include a host of functions that overlap
taxonomic divisions. For instance, the function of increasing
self-esteem is a characteristic of three types (i.e., integra-
tive,
transmissive, and escapist) and life satisfaction en-
hancement is a core component of both integrative and
instrumental types.
To illustrate the potential danger of operational overlap
discussed above, consider a specific coding example pro-
vided by the authors. One subject reflected on how the marital
breakup of
his
or her parents caused resentment as a teenager,
but that now the subject had achieved reconciliation with
them due to the subject's more mature outlook. This passage
is coded as integrative because it "conveys the idea of
resolving past conflicts" (p. 276). This seems reasonable
enough, but it does not demonstrate exclusivity, because
another function of such an excerpt might be to pass on a
moral lesson (e.g., do not judge others too harshly without
full understanding). If the latter is the case, then the above
scenario is clearly an example of transmissive reminiscence,
because moral instruction and personal wisdom are funda-
mental constituents of this type. Despite this concern, Wong
and Watt (1991) provide interrater reliability data (i.e., kappa
= .88 for two judges) that would seem to attenuate this
criticism.
Finally, Wong and Watt's (1991) taxonomy is derived
from elderly adults all older than 65 (mean age = 77.8
years),
once again potentially limiting the generalizability of
the results to the aged.
In summary, there has been only one previous attempt at
constructing a paper-and-pencil measure of reminiscence
functions. Perhaps due to the limitations identified above, it
has been rarely used in published studies of reminiscence.
The purpose of
the
present article is to develop a psychomet-
rically sound instrument that improves on earlier work.
METHOD
Development of
the
Reminiscence Functions Scale-Prototype
(RFS-p)
Pilot study I: item generation. — As part of earlier work
(Webster, 1989) and classroom demonstrations, participants
were asked to write down two different reasons why they
reminisced, and two different reasons why other people might
reminisce. From this larger pool of respondents, 40 subjects
(19 males and 21 females) were selected. The respondents
ranged in age from 18 to 76 years (mean age = 42.42, SD =
18.84),
with approximately equal numbers falling in each age
category (i.e., Young: N = 12, range = 18-29; Middle-age:
N = 14, range = 30-49; Old: N = 14, range = 50-76).
The sample had a mean education level of 12.55 years (range
= 8-17, SD = 1.78) and a mean self-perceived health rating
of 5.27 (range = 3-7, SD =
1.17),
where scores could range
from 1 = "very poor" to 7 = "excellent." In terms of
marital status, 15%, 30%,
37.5%,
and 17.5% of the sample
were separated, single, married, and divorced, respectively.
The sample was predominantly Caucasian (77.5%), followed
by Chinese (10.0%), Native Indian (5.0%), East Indian
(2.5%),
and Other (2.5%). Between them, these subjects
generated a total of 115 statements to the stems: "I reminisce
because. . . ," and "Others reminisce because. ..." The
verbatim statements were then edited for spelling and to make
them grammatically flow from the stem, "I reminisce: . . . ."
Subsequently, all 115 statements were readministered to a
new group of subjects.
Pilot study 2: item reduction/selection. — A new group of
119 subjects selected from an introductory psychology class
at Vancouver Community College, a demographically di-
verse educational institution, were asked to rate each of the
115 statements, generated in Pilot Study 1, on a
6-point
scale
where
1
= "never reminisce" (for the stated reason) to 6 =
"very frequently reminisce" (for the stated reason). The
sample included 44 males and 72 females ranging in age from
18 to 46 years (mean age = 22.62 years, SD = 4.84) with a
mean education level of 13.05 years (range = 8-17, SD =
1.26).
In terms of ethnic composition, 41.18%,
39.5%,
1.68%,
0.84%, 5.04%, and 11.76% were Caucasian, Chi-
nese,
Japanese, Black, East Indian, and Other, respectively.
Two means of culling redundant items were used in this
phase of instrument construction. First, an informal principal
components factor analysis was performed on the 115 state-
ments. The number of factors was set a priori at 10 in order to
both ensure that as wide a net as possible was cast at this early
stage of instrument development, while at the same time
limiting the number of potential factors to a theoretically and
pragmatically workable size. Items were then ranked within
each factor in terms of their loading.
Second, a research assistant was instructed to hand-sort
each statement into between 5-10 groupings, subsequently
producing eight relatively distinct clusters. Comparing the
overlap between the two methods resulted in a rationally
derived compromise of nine groupings of six items each. If a
factor had more than six items, the top six with the highest
loadings from the informal factor analysis were included. If
a
factor had less than six items, all the original items were
included and additional items were written that seemed to tap
the same underlying dimension. At this point, the groupings
were given the tentative labels of: Boredom Reduction, Death
Preparation, Identity Consolidation, Problem-Solving, Con-
versation, Intimacy Maintenance, Obsessive/Pathological,
Self-Esteem Enhancement, and Teach/Inform.
P258
WEBSTER
In summary, a two-stage process of item generation and
reduction/selection yielded a prototype instrument with the
following strengths: the items were generated by a diverse
sample spanning age, sex, education, ethnic, and other
demographic dimensions; it covered an expansive range of
potential functions; items reflected input from subjects them-
selves as well as input directed by theoretical considerations
indicated in the literature; and it had high face validity. The
next step involved administration of the finalized prototype
to a new sample.
Measures
The Reminiscence Functions Scale prototype (RFS-p) is a
54-item questionnaire in which subjects are asked to respond
on a 6-point scale how often they reminisce with a particular
function in mind. Subjects read the following introduction:
At different points throughout their
lives,
most adults think
about their
past.
Recalling earlier
times
can happen spontane-
ously or deliberately, privately or with other
people,
and may
involve remembering both happy and sad episodes. The
process of recalling memories from our personal past
is
called
reminiscence, an activity engaged in by adults of all ages.
This questionnaire concerns the why, or functions, of
reminiscence. That is, what purpose does reminiscence fulfil,
or, what goal does retrieving certain memories help you
accomplish?
The 54 items were randomly ordered in the questionnaire
and presented as completions to the stem: "When I reminisce
it is: . . .". For example, an item following the stem might
read: "to pass the time during idle or restless hours." Re-
sponses ranged from 1 = "never" (reminisce for the stated
purpose) to 6 = "very frequently" (reminisce for the stated
purpose). Subjects recorded their answers on an attached
answer sheet by filling in the circle containing the number that
best represented their choice (i.e., 1-6). The questionnaire
typically took between 15-25 min to complete.
Procedure
Subjects were recruited through a variety of means, in-
cluding hand delivery of questionnaires to retirement hous-
ing complexes and community centers; volunteers from
introductory psychology classes at Vancouver Community
College and St. Thomas University; and having students
solicit volunteers from the ranks of family members, peers,
and neighbors. The bulk of respondents were gained from
the latter category, where students received nominal class
credit for soliciting volunteers from the community. All
subjects volunteered, and it was stressed to students that if
potential subjects they were attempting to solicit appeared
reluctant in any way, even after it had been emphasized that
responses were anonymous and that there was no obligation
to complete any or all of the questionnaire, that students
would thank the person for their time and select an alternate
person. Information concerning the voluntary and anony-
mous nature of the questionnaire was explicitly detailed in a
cover letter that contained two telephone numbers that per-
sons could contact about concerns or comments. Only ques-
tionnaires with all 54 RFS-p questions answered completely
were used. In this manner, 710 usable questionnaires were
obtained.
Subjects
Subjects consisted of a diverse convenience sample, in-
cluding 289 males and
421
females ranging in age from
17
to
91 years (mean age = 45.76, SD = 21.69). In terms of
education, the sample ranged from those with no formal
schooling whatsoever, to those with at least some postgradu-
ate work (mean education = 12.57, SD = 2.58).
Self-
perceived health scores ranged from 1 = "very poor" to 7
= "excellent" (mean health = 5.09, SD =
1.27).
Table 1
provides a breakdown for the above demographic variables
for the sample as a whole and decomposed by age decade.
RESULTS
Exploratory Principal Components Analysis
The RFS-p was submitted to a principal components
analysis with varimax rotation using Systat for the PC
(version 5.1; Wilkinson, 1990). The criteria adopted for
item and factor retention were that: (a) eigenvalues must
have a minimum value of 1, (b) minimal item loading on a
factor was set at 5s .50, and (c) at least 4 or more items
meeting the S
2
.50 criterion would be the minimum to
constitute a factor (in order to facilitate reliability checks).
Using these criteria and rotating the loadings to facilitate
interpretation produced seven clearly distinct factors, la-
beled as follows: (1) Boredom Reduction, (2) Death Prepa-
ration, (3) Identity/Problem-Solving, (4) Conversation, (5)
Intimacy Maintenance, (6) Bitterness Revival, and (7)
Teach/Inform. A total of 43 of the original 54 items were
retained in the analysis. Table 2 lists the factor loadings on
each factor for all 43 RFS questions.
The seven factors that emerged closely mirror those pro-
duced by the informal analysis during the RFS-p develop-
ment, with two major exceptions being (1) the elimination of
a self-esteem factor and (2) the collapsing together of the
Identity Consolidation and Problem-Solving groupings.
With respect to exception (1), three items
(i.
e., Q18: remem-
bering previous achievements makes me feel proud of my-
self;
Q34: reexperience feelings of pride due to past accom-
plishments; and
Q44:
to build up my self-esteem by recalling
earlier life successes) had respective factor loadings of .76,
Table 1. Demographic Characteristics for Total
Sample and By Decade
Sample
Demographics
Gender
Male
Female
Age
Mean
SD
Education
Mean
SD
Health
Mean
SD
10s
31
88
18.3
.6
12.5
.6
5.3
1.1
20s
32
73
22.5
2.6
13.1
1.0
5.3
1.2
30s
39
47
34.2
2.7
13.5
1.9
5.2
1.1
ge (by decade)
40s
38
39
44.4
2.6
13.4
2.4
4.9
1.3
50s
41
39
53,
2.
13
2.
5.
.8
.7
.4
.5
.2
.9
60s
44
58
64.6
2.8
12.0
3.4
5.1
1.3
70s
44
58
73.1
2.6
11.3
3.7
4.8
1.6
80s
3
20
18
83.2
2.8
10.3
2.9
4.9
1.7
Total
289
421
45.7
21.6
12.6
2.6
5.1
1.3
a
One 91-year-old male was included in the 80s decade.
REMINISCENCE FUNCTIONS SCALE
P259
.72,
and
.61,
qualifying them on two of the inclusion criteria
outlined earlier. However, their intercorrelations were low
(range = .40-.60), and reliability estimates for only three
items would be suspect. They were, therefore, eliminated
from further analysis, although they were clearly suggestive
of a self-esteem/pride factor that warrants further investiga-
Table
2.
Reminiscence Functions Scale Factor Loadings
Factor Loadings
tion. Abbreviated RFS item examples for all seven RFS
factors are given in Table 3. The complete RFS can be
obtained by writing to the author.
Reliability
As a means of assessing factor reliability, internal consis-
tency scores were computed using coefficient alpha for all
items.
Alpha levels and factor intercorrelations for the RFS
are reported in Table 4.
As can be seen, the internal consistency of the factors is
good, ranging from .79 for Conversation to .89 for Identity/
RFS Items
Factor
1
Q20
Q47
Q14
Q26
Q4
Q24
Factor
2
Q41
Q43
Q37
Q48
Q2
Q12
Factor
3
Q40
Q49
Q39
Q45
Q23
Q15
Q52
Q33
Q5
Qll
Q13
Q31
Factor
4
Q28
Q7
Q36
Q9
Factor
5
Q6
Q32
Q17
Q51
Factor
6
Q53
Q50
Q16
Q22
Q19
Factor
7
Q29
Ql
Q35
Q25
Q38
1
.83
.80
.73
.73
.66
.61
.16
.09
.10
.27
.02
-.00
.11
.10
.06
.02
-.02
.27
.10
.03
.12
.09
.05
.12
19
.06
.09
.18
.06
.04
.05
.17
.05
.18
.21
.18
.20
.23
-.05
-.00
-.00
-.00
-.08
2
.05
.16
.09
.12
.08
.04
.76
.76
.74
.65
.59
.58
.14
.18
.18
.14
.01
.04
.03
.17
.09
-.13
.20
.20
10
.04
.16
-.00
.17
.16
.16
.26
.04
.16
.12
.09
-.10
-.01
.25
.28
.24
-.02
.38
3
.05
.04
.09
.10
.18
.26
.17
.07
.12
.26
.12
.11
.73
.73
.70
.64
.64
.64
.63
.59
.56
.53
.52
.50
13
.19
.17
.12
.20
.04
.01
.02
.14
.11
.11
.14
.25
.19
.06
-.08
.10
.37
.04
4
.07
.14
.09
.07
.10
.10
.06
.09
.04
.09
.10
.07
-.06
.17
.15
.06
.22
.09
.10
.02
.10
.11
.01
.13
.71
.68
.67
.66
.61
.10
.05
.11
.04
-.01
.06
.02
-.00
.12
.13
.02
.22
.21
.08
5
.05
.03
.07
.01
.13
.07
.11
.15
.13
.06
.12
.13
-.01
.08
.10
-.09
.12
.11
.01
.13
-.06
-.09
.04
.06
04
.11
.07
-.00
.13
.84
.83
.74
.59
.07
.04
.20
.01
-.02
.12
.17
.14
.07
.25
6
.19
.08
.16
.20
.11
.20
.07
.00
.06
.14
.00
.06
.15
-.01
.07
.13
.06
.15
.27
.11
-.02
.19
.12
-.00
.07
-.01
.07
.06
-.01
-.00
.04
.12
.21
.82
.80
.68
.60
.59
.01
-.03
-.04
.01
-.12
7
-.01
-.04
-.12
.11
-.03
-.01
.07
.24
.19
.03
.21
.27
-.04
.09
.07
.05
.09
-.08
.01
.18
-.03
.02
.10
.20
17
.
i /
.21
-.01
.26
.03
.16
.23
.14
.06
-.10
-.06
-.07
.14
.02
.78
.72
.63
.59
.58
Problem-Solving (mean alpha = .84, SD = .04). The
Table:
Factor
1:
Q20
Q47
Factor
2:
Q41
Q43
Factor
3:
Q40
Q49
Factor
4:
Q42
Q28
Factor
5:
Q6
Q32
Factor
6:
Q53
Q50
Factor
7:
Q29
Ql
3.
Sample Items From
All 7
Factors
of
the Reminiscence
Functions Scale
Boredom Reduction
to reduce boredom (.83)
a
for something
to do (.80)
Death Preparation
because
I
feel less fearful
of
death after
I
finish reminiscing
(.77)
because
it
helps
me
see that
I've
lived
a
full life and
can
therefore accept death more calmly
(.76)
Identity/Problem-Solving
to
try to
understand myself better (.73)
to see how
my
strengths
can
help me solve
a
current problem
(.73)
Conversation
to create ease
of
conversation
(.71)
to create
a
common bond between
old and
new friends
(.68)
Intimacy Maintenance
to keep alive
the
memory
of
a dead loved one
(.84)
to remember someone who has passed away
(.83)
Bitterness Revival
to keep memories
of
old hurts fresh
in
my mind
(.82)
to rekindle bitter memories
(.80)
Teach/Inform
in order to teach younger persons about cultural values
(.78)
to teach younger family members what life was like when
I
was young
and
living
in a
different time
(.72)
a
Values
in
parentheses
are
factor loadings.
Factor
1
2
3
4
5
6
7
Alpha
Note.
nnni
|
P
,
. uuu i
le
*p
=
Table
4.
Factor Intercorrelation Matrix
and Internal Consistency Data
for the Reminiscence Functions Scale
rsctor
12
3 4 5 6 7
—
.32 .36 .35 .24 .48 .05*
—
.39 .34 ,40 .23 .52
—
.44 .23 .40 .28
—
.29 .20 .41
—
.22 .42
—
.02*
—
Internal Consistency
.88
.85 .89 .79 .84 .82 .82
Except where indicated,
all
correlations are significant
at
the
p <
not significant.
P260
WEBSTER
intercorrelations indicate that all but two factors on the RFS
are significantly correlated with one another, ranging from
small to moderate in magnitude (M correlation =
.31,
SD =
.13,
range = .02-.52). On average, the variance accounted
for between factors is < 10%, indicating an expected associ-
ation between factors that is nevertheless small enough to
validate the divergent nature of the subscales.
Predictive Validity
Two separate means of assessing the predictive validity of
the RFS were conducted. The first method examined the
relationships between reminiscence functions and personal-
ity traits. The second investigated developmental differences
in reminiscence functions over the life span.
Personality. — A new and separate sample of 123 intro-
ductory psychology students at Vancouver Community Col-
lege served as subjects. Participants included 45 males and
78 females ranging in age from 18 to 45 (mean age = 24.6,
SD = 6.9). Although there is a dearth of empirical studies
examining the relationship between personality characteris-
tics and reminiscence dimensions (Fry, 1992; Webster &
Cappiliez, 1993), what limited evidence does exist suggests
there may be a relatively stable relationship between various
dimensions of personality and reminiscence. For instance,
both Fry (1992) and Webster (in press) found that the
personality domain of openness was associated with simple
reminiscence frequency.
For the validity study, Costa and McCrae's (1985) NEO-
FFI (Form S) personality measure was used as the dependent
variable in a series of regression analyses using the RFS
factors as predictor variables. The NEO-FFI was chosen
because it is an extensively used instrument of high reliabil-
ity and validity, normed on a sample spanning the entire
adult age spectrum, and because it lends itself to specific
hypotheses vis-a-vis reminiscing.
The NEO-FFI measures five personality domains, three of
which have been extensively researched and used in the RFS
validity study. The Neuroticism (N) domain comprises the
six facets of anxiety, hostility, depression, self-conscious-
ness,
impulsiveness, and vulnerability. Persons scoring high
on this domain are more likely to manifest high levels of
anxiety and negative emotions such as anger and embarrass-
ment. The Extraversion (E) domain comprises the six facets
of warmth, gregariousness, assertiveness, activity, excite-
ment-seeking, and positive emotions. Persons scoring high
on this domain are more likely to manifest high levels of
sociability, assertiveness, and stimulation-seeking. The
Openness (O) domain comprises the six facets of fantasy,
aesthetics, feelings, actions, ideas, and values. Persons
scoring high on this domain are more likely to manifest high
levels of introspection, exploration of novelty, and intellec-
tual curiosity.
Given the constituents of the three personality domains,
the following a priori hypotheses were tested: namely, that
(a) Factor 6 would be positively correlated with Neuroti-
cism, (b) Factor 4 would be positively correlated with
Extraversion, and (c) Factor
3
would be positively correlated
with Openness.
The results support these predictions, providing prelimi-
nary evidence of the predictive validity of the RFS. Speci-
fically, Factor 6 correlated .42 with Neuroticism
{p
<
.01,
n
= 120, one-tailed), Factor 4 correlated .33 with Extraver-
sion (p < .0\,n = 123, one-tailed), and Factor 3 correlated
.18 with Openness (p < .05, n = 120, one-tailed).
Developmental. — Age differences on the RFS, full
details of which can be found in Webster's current work
(available on request), also provide evidence of
its
predictive
validity. Because some authors (Kiernat, 1983; Webster &
Young, 1988) explicitly call for a life-span approach to
reminiscence research, it would be instructive to examine
possible age differences in the RFS. Two specific examples
will suffice here. First, there is an age difference in Factor 2
(Death Preparation). Consistent with Butler's (1963) notion
of the life review, one of the functions of reminiscence in
advanced old age is to prepare to accept one's impending
biological demise and, therefore, it is expected that elderly
adults, in general, being closer to death than their younger
counterparts, would use reminiscence for this reason more
frequently. We can see from Figure 1 that the results are
clear and strongly support this prediction. Specifically, there
is a monotonic increase in the use of reminiscence in the
service of death preparation with age (F = 18.24, df = 7,
693,
p < .0001). Second, there is also a significant relation-
ship (F = 26.19, df = 7, 693, p < .0001) between age and
Factor 7 (Teach/Inform). We can see from Figure 2 that
older adults use reminiscence in this fashion with greater
frequency than younger adults. This finding is consistent
with the greater accumulation of experience as a corollary of
advanced chronological age. In other words, older adults'
extensive navigation through multiple life events has pro-
20
15
c
o
2
(0
a
2 10
a.
CO
a
Age (by decade)
Figure I. Age differences on the RFS Death Preparation subscale.
REMINISCENCE FUNCTIONS SCALE
P261
20
15
10
Age (by decade)
Figure 2. Age differences on the RFS Teach/Inform subscale.
vided a wealth of knowledge on which they may draw. The
relatively sharp break from youth to middle adulthood may
be a consequence of parenthood more than age per se.
DISCUSSION
The RFS is a valid, reliable, and relatively comprehensive
measure of many fundamental uses of reminiscence in adult-
hood. It builds on, strengthens, and elaborates earlier work
by Romaniuk and Romaniuk (1981), and is consistent with
many findings of Wong and Watt's (1991) recent taxonomy.
There exist clear parallels between the present findings
and Romaniuk and Romaniuk's (1981) earlier RUS. Both
measures derive a Problem-Solving component and an Iden-
tity or Existential/Self-Understanding factor. In Romaniuk
and Romaniuk's scale, these emerged as separate factors,
whereas they were combined into a single factor in the RFS.
This combination in the RFS is somewhat problematic and
needs further refinement for the following reasons.
First, Identity and Problem-Solving items were assumed
to constitute separate subscales during the second pilot study
and were expected to remain distinct. Second, an inspection
of how the items clustered within the single scale revealed
that they are not randomly intermixed, but rather tended to
group together with problem-solving items generally group-
ing together first, followed by the cluster of identity items.
Third, other factor rotation patterns (i.e., equimax) produce
distinct factors for Identity and Problem-Solving. Finally,
the Romaniuks' work demonstrates that these two functions
may indeed constitute separate subscales.
These concerns suggest that future refinements of the RFS
may attempt to tease apart the Identity/Problem-Solving fac-
tor into separate subscales, allowing for finer grained analysis
of certain hypotheses. As only one example, problem-solving
has been postulated to be higher in middle-age than in old age
(Lieberman & Falk, 1971). Such a finding may be masked in
the RFS if there is an opposite effect for Identity, whereby a
higher Identity score among elderly adults would cancel out
the lower problem-solving score.
Further, consistent with the Romaniuks' RUS, a
self-
esteem factor emerged in the RFS, but, as noted above, did
not meet all the inclusion criteria established for factor
retention. In future refinement of the RFS, additional items
reflecting self-esteem components might be included in an
attempt to meet the inclusion criteria established previously.
Alternatively, the inclusion criteria themselves could be less
stringent, but this would seem to defeat the major aim of this
research, which is to develop an assessment instrument
whose psychometric properties are as high as possible.
Parallels also exist between the RFS and Wong and Watt's
(1991) typology. The RFS has several advantages over the
latter typology, including ease of administration and scor-
ing, greater comprehensiveness in terms of identified fac-
tors,
and clearer differentiation between factors. Neverthe-
less,
findings of the present study are clearly consistent with
many facets of Wong and Watt (1991). Specifically, strong
similarities emerge between (1) Factor 6 (Bitterness Re-
vival) of the RFS and the Obsessive type, (2) Factor 7
(Teach) and the Transmissive type, and (3) The Integrative
and Instrumental types which, in combination, match
closely with Factor 3 (Identity/Problem-Solving). Less
strongly related are Factor 1 (Boredom Reduction) and the
Escapist type. The latter involves a clear deprecation of the
present relative to the past, whereas the former probably
involves more pedestrian recollections of a positive nature.
Finally, inasmuch as one of the main functions of Narrative
reminiscence is to "recount past anecdotes that may be of
interest to the listener" (p. 274), it shares certain characteris-
tics with Factor 4 (Conversation). Factor 2 (Death Prepara-
tion) and Factor 5 (Intimacy Maintenance) find no counter-
parts in Wong and Watt's (1991) typology, although one
suspects that there should be a correlation between Factor 2
of the RFS and the Integrative type. The fact that two
separate methods should result in such similar findings
regarding reminiscence functions bodes well for the conver-
gent validity of the RFS.
Many of the resulting factors of the RFS have at one point
been nominated as potentially important uses of reminis-
cence in elderly adults (e.g., Factor 2: Butler, 1963; Factor
3:
Lieberman & Tobin, 1983; Factor 4: Boden & Bielby,
1983).
The RFS allows researchers to assess multiple func-
tions simultaneously and allows for direct comparisons
across studies.
Future research directions can include contrasting remi-
niscence functions between groups of interest (e.g., on the
basis of sex, age, culture, religion, marital status, living
arrangements, health, personality, etc.). As a new instru-
ment, the RFS will find its initial place as a descriptive,
research-based tool. Eventually, however, it may find em-
ployment as a diagnostic measure as well. For instance, is
there an RFS "profile" consistent with clinical depression,
or perhaps predictive of the early stages of dementing ill-
nesses such as Alzheimer's disease? It is hoped that these
P262
WEBSTER
and other important questions can be fruitfully explored, in
part, with the RFS.
ACKNOWLEDGMENT
Address correspondence and reprint requests to Jeffrey Dean Webster,
Department of Psychology, Vancouver Community College, Langara Cam-
pus,
100 West 49th Avenue, Vancouver, British Columbia, Canada V5Y
2Z6.
REFERENCES
Boden, D., & Bielby, D. (1983). The past as resource: A conversational
analysis of elderly talk. Human Development, 26, 308-319.
Butler, R. (1963). The life review: An interpretation of reminiscence in the
aged. Psychiatry, 26, 65-76.
Costa, P. T., Jr., & McCrae, R. R. (1985). TheNEO Personality Inventory
(Manual). Odessa, FL: Psychological Assessment Resources.
Fry, P. S. (1991). Individual differences in reminiscence among older
adults: Predictors of frequency and pleasantness ratings of reminiscence
activity. International Journal of Aging and Human Development, 33,
311-326.
Haight, B. K. (1991). Reminiscing: The state of the art as a basis for
practice. International Journal of Aging and Human Development, 33,
1-32.
Kiernat, J. M. (1983). Retrospection as a life span concept. Physical and
Occupational Therapy in Geriatrics, 3, 35-48.
Kovach, C. R. (1990). Promise and problems in reminiscence research.
Journal of Gerontological Nursing,
16,
10-14.
Lieberman, M. A., & Falk, J. (1971). The remembered past as a source of
data for research on the life cycle. Human Development, 14,
132-141.
Lieberman, M. A., & Tobin, S. S. (1983). The experience of old age:
Stress, coping, and survival. New York: Basic Books.
Merriam, S. B. (1980). The concept and function of reminiscence: A
review of the research. The Gerontologist, 20, 604-608.
Molinari, V., & Reichlin, R. (1984-85). Life review reminiscence in the
elderly: A review of the literature. International Journal of Aging and
Human Development, 20, 81-92.
Moody, H. R. (1988). Twenty-five years of the life review: Where did we
come from? Where are we going? Journal of Gerontological Social
Work,
12, 7-21.
Romaniuk, M. (1981). Reminiscence and the second half of life. Experi-
mental Aging Research, 7, 315-336.
Romaniuk, M., & Romaniuk, J. (1981). Looking back: An analysis of
reminiscence functions and triggers. Experimental Aging Research, 7,
477-489.
Thornton, S., &Brotchie, J. (1987). Reminiscence: A critical review of the
empirical literature. British Journal of Clinical Psychology, 26, 93-
111.
Webster, J. D. (1989, October). Individual differences in reminiscence
behaviour: A lifespan perspective. In G. T. Reker (Chair), Reminis-
cence and aging: Partnership in theory and practice. (Invited sympo-
sium address, 18th annual Scientific and Educational Meeting of the
Canadian Association on Gerontology, Ottawa, Ontario.)
Webster, J. D. (in press). Predictors of reminiscence: A lifespan perspec-
tive.
Canadian Journal on Aging.
Webster, J. D., & Cappeliez, P. (1993). Reminiscence and autobiographi-
cal memory: Complementary contexts for cognitive aging research.
Developmental Review, 13,
54—91.
Webster, J. D., & Young, R. A. (1988). Process variables of the life
review: Counselling implications. International Journal of Aging and
Human Development, 26, 315-323.
Wilkinson, L. (1990).
SYSTAT:
The system for statistics. Evanston, IL:
SYSTAT, Inc.
Wong, P. T. P., & Watt, L. M. (1991). What types of reminiscence are
associated with successful aging? Psychology and Aging, 6, 272-279.
Received November 11, 1992
Accepted March 2, 1993