Psychology and Aging 1990, Vol. 5, No. 1,48-57

Copyright 1990 by the American Psychological Association, Inc. 0882-7974/90/100.75

A Self-Rating Scale for Evaluating Memory in Everyday Life Thomas H. Crook III and Glenn J. Larrabee

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Memory Assessment Clinics, Inc. Bethesda, Maryland A memory self-rating scale is described that includes 21 abih'ty-to-remember items, 24 items assessing frequency of occurrence of memory failures, and 4 global rating items assessing overall comparison to others, comparison to the best one's memory has ever been, speed of recall, and concern or worry over memory function. Factor analysis yielded 5 orthogonal Ability to Remember and 5 orthogonal Frequency of Occurrence factors. The factor structure was not affected by age or sex, and level of complaint on the factor scores was not strongly associated with age.

Herrmann (1982) observed that psychologists have been interested in the successes and failures of memory in everyday life since the times of William James and Sigmund Freud. However, only in recent years have attempts been made to develop reliable and valid self-rating scales with adequate normative data for assessing memory in everyday life. The utility of such scales would extend to both research and clinical applications, including the assessment of memory deficits associated with neurological insults or disorders such as head injury, stroke, or Alzheimer's disease. In discussing the importance of self-rating scales for assessing memory in older adults, Gilewski and Zelinski (1986) cited four major factors. First, there is evidence of a reliable correlation between memory complaints and memory performance among healthy elderly persons in the community (Zelinski, Gilewski, & Thompson, 1980). Second, complaint of memory impairment may be important in detecting early signs of a dementing disorder. Note, however, that there may be a closer relation between complaint and performance in the earnest stages of dementia than later in dementing disorders, when significant cognitive deterioration has occurred (van der Cammen et al., 1987). At the later stages of Alzheimer's disease and related dementing disorders, there may be no correspondence between complaint and performance, with the patient totally denying memory deficit (Reisberg, Ferris, de Leon, & Crook, 1982). The third factor cited is that memory report may be useful in differentiating dementia and depression. In this vein, it is noteworthy that whereas there may be minimal or no memory complaints in patients suffering significant memory deficit secondary to dementia, complaint of memory change may be magnified in depressive conditions in the context of average or above-average memory test performance (Popkin, Gallagher, Thompson, & Moore, 1982; Williams, Little, Scates, & Blockman, 1987). Finally, the fourth factor is that memory complaints are important in their own right, because they provide information on how people view their general cognitive functions as they age.

Although there are multiple applications of memory self-rating scales for elderly subjects, the scales now available may be of limited utility. Gilewski and Zelinski (1986) and Herrmann (1982), in their excellent reviews of existing scales, noted that validity has proved to be a challenge in scale development in that correspondence between self-report of memory function and actual performance on memory tasks has frequently been quite low. Herrmann (1982) observed that several previous investigations reporting no association between self-report and performance may be misleading, because few could meet the standard of isomorphic question-to-task agreement. That is, questions on the self-report scales used bore little resemblance to the performance tests used to assess memory. Other limitations of existing scales noted by Herrmann (1982) and by Gilewski and Zelinski (1986) include inadequate normative data, vague and often negative wording of items, and failure to consider the complex, multivariate nature of memory. In addressing this latter issue, authors of both reviews recommended a factor analytic approach in scale development. The Memory Assessment Clinics Self-Rating Scale (MAC-S; Winterling, Crook, Salama, & Gobert, 1986) was developed to address limitations of existing scales such as those identified by Herrmann (1982) and Gilewski and Zelinski (1986). The original MAC-S consisted of 102 items describing specific memory tasks or problems encountered in everyday life and global memory evaluation items. The items were divided into two subscales of equal length: Ability, on which individuals indicated their ability to remember specific types of information (e.g., "the name of a person just introduced to you") and Frequency of Occurrence, on which individuals indicated how often specific memory problems occur (e.g., "forget what you intended to buy at a grocery store or pharmacy"). Responses were recorded on a 5-point Likert scale from very poor (1) to very good (5) on the Ability scale, and from very often (1) to very rarely (5) on the Frequency of Occurrence scale. Winterling et al. (1986) reported a factor analysis of the MAC-S based on the performance of 343 normal subjects. Five memory Ability factors were identified and labeled Everyday Task-Oriented Memory, Remote Memory, Reading Recall, Numeric Recall, and Word Recall. On the Frequency of Occurrence scale, five factors were also identified, and these were la-

We acknowledge the assistance of Anastasia Zadeik in data analysis. Correspondence concerning this article should be addressed to Thomas H. Crook III, Memory Assessment Clinics, Inc., 8311 Wisconsin Avenue, Bethesda, Maryland 20814. 48

49

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

SELF-REPORT MEMORY SCALE

beled Concentration, Everyday Task-Oriented Memory, Word Recall, Spatial Memory, and Facial Recognition. On the basis of these factor analyses, the MAC-S was shortened to 21 Ability items and 24 Frequency of Occurrence items by eliminating (a) items that did not load significantly on any factor and (b) items that loaded on specific factors but appeared redundant. Four global items were also incorporated into the scale. We conducted a study in which the 49 items on the revised MAC-S were administered to 1,106 subjects. Principal components analyses (PCAs) were undertaken to examine the factor structure of the new scale. Because of previous reports of memory complaint factor structure and level of performance differing with age and sex (Gilewski & Zelinski, 1986; Hultsch, Hertzog, & Dixon, 1987), these variables were included for analysis in this study. The effects of the moderator variables education, verbal intelligence, and affective status were also assessed. We were particularly interested in the association between memory report and affective status because three recent investigations have demonstrated significant associations between memory complaint and depression (Larrabee & Levin, 1986; West, Boatwright, & Schleser, 1984; Williams et al., 1987).

Method Subjects Eleven hundred and six volunteers (430 men and 676 women) from the metropolitan Washington, D.C., area were recruited through the print and electronic media. Subjects included the 343 persons from the Wintering et al. (1986) study, plus an additional 763 subjects. The subjects from the Winterling et al. study were included to expand the sample size; consequently, the 45 items omitted on the revised MAC-S were dropped from the MAC-S data collected on these subjects. The mean age of subjects was 56.21 years (SD = 12.93), with a range of 18 to 92 years, and average level of education was 15.89 years (SD = 2.75), with a range of 8 to 25 years. The mean age-corrected scaled score on the Vocabulary subtest of the Wechsler Adult Intelligence Scale (WAIS) was 13.3 (SD = 1.63), and the mean score on the Geriatric Depression Scale (CDS; Yesavage et al., 1983) was 7.06 (SD = 5.46), with most of the sample scoring in the nondepressed range (Yesavage, 1986). No subjects manifested clinical signs of depression, and all participants were healthy, with no history of psychiatric or neurologic disease.

Procedure The revised memory questionnaire contained 21 Ability items and 24 Frequency of Occurrence of memory problem items. All items were rated using a 5-point Likert scale. The following global items were also included in the scale: "In general, as compared to the average individual, how would you describe your memory?" (range = very poor to very good); "How would you describe your memory, on the whole, as compared to the best it has ever been?" (range = much worse to much better); "Compared to the best your memory has ever been, how would you describe the speed with which you now remember things?" (range = much slower to much faster); "How much concern or distress do you feel about your memory at this time?" (range = very serious concern to no concern). The four global items were not administered to the 343 subjects from the Winterling et al. (1986) study. The Ability and Frequency of Occurrence scales were both submitted to PCA with varimax rotation. Pairwise deletion of cases with missing values was employed. The effects of age on factor structure were exam-

ined by dividing the total sample at the median age of 59 into groups of 60 and older (n = 541, mean age = 66.53) or 59 and younger (n = 565, mean age = 46.33). The total sample was also divided into groups of men (n = 430) and women (n = 676). Separate PCAs were conducted on the young versus old and male versus female subjects. Analyses were undertaken to examine the effects of age, education, sex, affective status, and WAIS Vocabulary score on the factor scales. The effects of these variables on the global items were also analyzed.

Results Ability Factors Through inspecting the scree plot and applying the KaiserGuttman criterion, a five-factor solution was deemed appropriate for the performance of the 1,106 subjects on the 21 Ability scale items. This solution accounted for 59.7% of the variance. The eigenvalues for this five-factor solution were 6.52, 1.75, 1.61, 1.52, and 1.13. Varimax rotation was applied. The five Ability factors include Remote Personal Memory, Numeric Recall, Everyday Task-Oriented Memory, Word Recall/Semantic Memory (with a secondary representation of Reading Recall), and Spatial and Topographic Memory. The specific items loading on each factor are shown in Appendix A. Numeric Recall also had a moderately strong loading (.53) on an item measuring recall of the name of a person who has just been introduced. The appearance of this item suggests that the Numeric Recall factor probably involves everyday attentional processes. Three items did not demonstrate strong loadings on the five Ability factors: "Facts that must be recalled very quickly as in a game or television show," "The date and day of the week," and "What you read in the newspaper one day ago." These items were retained for the analyses of the associations among Ability factor scores, global rating items, and total Ability scale score (see Appendix B for the associations). In general, the Ability factor structure of the 1,106 subjects is the same as that reported by Winterling et al. (1986). Both studies yielded Everyday Task-Oriented Memory, Remote Memory, and Numeric Recall factors. In the Winterling et al. study, Word Recall and Reading Recall emerged as separate factors, whereas they combined in the present study as a Word Recall/Semantic Memory factor. Also, a Spatial and Topographic Memory factor emerged in the present PCA, which was not found in the Winterling et al. analysis.

Frequency of Occurrence Factors PCA of the Frequency of Occurrence scale items was also conducted on the total sample of 1,106 subjects. Inspection of the scree plot was consistent with the Kaiser-Guttman criterion, suggesting a five-factor solution, which accounted for 54% of the variance. Eigenvalues for the five factors were 7.97, 1.51, 1.41,1.20, and 1.05. Varimax rotation was conducted. The five Frequency of Occurrence factors include Word and Fact Recall or Semantic Memory, Attention/Concentration, Everyday Task-Oriented Memory, General Forgetfulness, and Facial Recognition. Specific items loading on each factor are shown in Appendix A. Both the Everyday Task-Oriented Memory factor and General Forgetfulness factor appear to be measuring aspects of gen-

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

50

THOMAS H. CROOK III AND GLENN J. LARRABEE

eral forgetfulness in everyday memory. In the former case, however, forgetfulness appears to involve action-oriented everyday memory tasks requiring storing, seeking, or retrieving important memoranda, whereas in the latter case, the tasks are not as action-oriented. Four Frequency of Occurrence items did not demonstrate strong loadings on the five factors: "Forget that you told someone something and tell that person the same thing again," "Forget an entire event, such as attending a party or having a visitor," "Call someone you recently met by the wrong name," and "Pass the point where you intended to exit while driving a car or taking public transportation." These items were retained for the analyses of the associations among Frequency of Occurrence factor scores, global rating items, and the total Frequency of Occurrence scale score (see Appendix C for the associations). In general, as in the case of the Ability factors, the Frequency of Occurrence factor structure of the combined data set of 1,106 is highly similar to that obtained in the original analyses of Winterling et al. (1986). The only variation is the failure to obtain a Spatial/Topographic Memory factor and the appearance instead of a General Forgetfulness factor.

Age and Sex Effects on Ability and Frequency of Occurrence Factor Structure The same five Ability and five Frequency of Occurrence factors were obtained in the total sample, the 60 and older sample, and the 59 and younger sample. There were slight variations in the order in which the factors were extracted. The same five Ability and five Frequency factors were obtained in the total sample, the male-only sample, and the female-only sample. Again, slight variations were observed in the order in which the factors were extracted.

Effects of Age, Sex, Education, and Vocabulary on Level of Performance The combined data set of 1,106 was used to generate orthogonal factor scores. As in the original PCAs, there was pairwise deletion of cases with missing data. The minimum number of pairwise cases was 1,046. The correlations of Ability factor scores with demographic variables are displayed in Table 1. The correlations of Frequency of Occurrence factor scores with demographic variables are shown in Table 2. The minimum number of pairwise cases was 1,049. Table 3 displays the correlations of the four global rating items with the demographic variables. As can be seen in Tables 1,2, and 3, age was significantly correlated in a negative direction with Ability total score, Numeric Recall, and Spatial/Topographic Memory; Frequency of Occurrence total score, Word and Fact Recall/Semantic Memory, and Everyday Task-Oriented Memory; the global rating for the item asking subjects to compare their memory with the best it has ever been; and the global rating for the item gauging speed of recall. Although significant, these relations were small, accounting for, at maximum, 4.8% of the variance. Tables 1,2, and 3 also show minimal associations of MAC-S scores with sex, WAIS Vocabulary, and educational level. Although 7 of 16 correlations of sex with MAC-S scores were sig-

nificant, the largest amount of variance accounted for by sex was 5.6% (the correlation of sex with the Remote Personal Memory score). Positive correlations reflect better reported performance by women. Scores on the WAIS Vocabulary test correlated significantly, generally in a positive direction, with 5 of the 16 MAC-S variables in Tables 1,2, and 3. The only correlations accounting for more than 6% of the variance were the correlations of WAIS Vocabulary with the Word Recall/Semantic Memory Ability factor (13% of test score variance) and the Attention/Concentration Frequency factor (6.6% of test score variance). The WAIS Vocabulary and Word Recall/Semantic Memory Ability factor correlation was expected and supports the concurrent validity of this particular rating scale dimension. The relative absence of other meaningful correlations between the MAC-S and WAIS Vocabulary suggests that the MAC-S is not substantially affected by differences in subjects' vocabulary level, within the performance range investigated in this study. Education correlated significantly, in a positive direction, with 4 of the 16 MAC-S variables in Tables 1,2, and 3. Education level accounted for 5.7% of the variance in the Word Recall/Semantic Memory Ability factor, 0.96% of the variance in Frequency of Occurrence total score, and 4.2% of the variance in the Attention/Concentration Frequency of Occurrence factor. Relations among MAC-S factor scores, total scores, and global items are shown in Appendix B for the Ability scale and in Appendix C for the Frequency of Occurrence scale. All but one of the correlations shown were significant. (Note: the Ability factor scores correlated at zero, as did the Frequency of Occurrence factor scores, because these were based on an orthogonal rotation.) However, the relations were not so strong as to render the various scores redundant and unnecessary. In no cases did the total scores or global scores account for more than 25% of the variance in the factor scores.

Relation of the GDS to MAC-S Performance The GDS correlated significantly, in a negative direction, with 15 of 16 MAC-S variables, shown in Tables 1 through 3. Only seven of these correlations accounted for greater than 6% of MAC-S test score variance: Ability total (9.8%); Frequency of Occurrence total (12.7%); Attention/Concentration Frequency of Occurrence factor (6.7%); and global items dealing with overall comparison to the average individual (8.9%), comparison to the best one's memory has ever been (7.3%), speed of recall (6.5%), and concern or distress about one's memory (17%). These data suggest that memory ratings based on total scores and on global rating items are more susceptible to influence by affective variation than those based on factor scores. The one correlation between the GDS and MAC-S factor score accounting for more than 6% of the variance was the correlation between the GDS and the Attention/Concentration Frequency of Occurrence factor. As indicated earlier, none of the subjects manifested clinical signs of depression, and most scored below the GDS cutoff of 11 suggested by Yesavage et al. (1983). To explore the relation between self-reported affective status and self-reported memory function further, we contrasted the MAC-S scores of those sub-

51

SELF-REPORT MEMORY SCALE

Table 1 Correlation of Self-Report Ability Factor Scores and Ability Scale Total With Demographic Variables Factor

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Variable Age Sex Geriatric Depression Scale WAIS Vocabulary Education

Ability scale total

Remote Personal Memory

-.100»* .090*

.011 .237**

-.313** .063 .062

-.157** -.114** -.043

Numeric Recall

Everyday TaskOriented Memory

Word Recall/ Semantic Memory

Spatial Topographic Memory

-.077* .038

-.039 .124**

.003 -.144**

-.140** -.118**

-.116** -.052 -.047

-.188** -.058 -.031

-.145** .361** .238**

-.092* .037 .052

Note. A positive correlation indicates better ability; positive correlations with sex reflect better performance for women. WAIS = Wechsler Adult Intelligence Scale. Minimum pairwise number of cases = 1,046. *p

A self-rating scale for evaluating memory in everyday life.

A memory self-rating scale is described that includes 21 ability-to-remember items, 24 items assessing frequency of occurrence of memory failures, and...
1MB Sizes 0 Downloads 0 Views