Académique Documents
Professionnel Documents
Culture Documents
碩士論文
Graduate Institute of Psychology
College of Science
National Taiwan University
Master Thesis
台灣健康成年人中文字詞次序學習測驗之
常模研究
A Normative Study on the Word Sequence Learning Test
in Healthy Individuals in Taiwan
陳威嘉
Wei-Chia Chen
指導教授:花茂棽 博士
中華民國 103 年 6 月
June, 2014
摘要
背景:中文字詞次序學習測驗隸屬於語文事件記憶測驗,本測驗主要特徵為
字詞間關係屬任意語意連結的特質,而具有此特質的學習測驗與內側顳葉結構
連性。阿茲海默病的神經病理研究指出,嗅緣皮質是早期出現神經病理現象的腦
區之一,因此該腦區所負責的語文事件記憶能力,也被視為阿茲海默病早期認知
功能改變的項目之一。然而,目前在台灣缺乏偵測該阿茲海默病前期(pre-AD)記
憶缺陷的工具與常模資料。目的:本研究的研究目的是建立台灣健康成年人中文
字詞次序學習測驗的常模資料,同時探討人口學變項對該測驗表現的影響,以及
地理分布等變項進行分組。隨機選取 30 位受試者進行信度的檢驗,另外再隨機
選取 33 位受試者進行校標關聯及建構效度的考驗。並且以回溯性方式收集 40
位具有失憶型輕度認知功能障礙(aMCI)的病人資料,以供進一步驗證本測驗之建
構效度。結果:人口學變項的影響主要出現在年齡與教育程度兩變項上,亦即本
測驗的表現會隨年齡增加而衰退,但隨著教育程度增加而測驗表現提升。本測驗
在校標關聯效度上,本測驗除了再認分數外,與其他語文事件記憶測驗的分數有
著中等程度的相關(.35 至.75)。除此以外,在病人回溯性研究結果顯示,本測驗
i
在偵測 aMCI 病人記憶缺陷的比例,較魏氏記憶量表第三版之邏輯記憶測驗為高。
結論:研究結果顯示年齡及教育變項皆明顯影響中文字詞次序學習測驗的表現,
因此在常模建立上,進行人口學變項的校正有其必要性。此外,研究結果亦顯示
中文字詞次序學習測驗具有適當的心理計量特性。有鑑於此,根據病人回溯性研
究結果,我們可預期該測驗的表現可作為偵測阿茲海默病前期的認知標記。未來
則應持續進行前瞻性病人研究,以了解中文字詞次序學習測驗在診斷阿茲海默病
前期的貢獻。
關鍵詞:中文字詞次序學習測驗、神經心理測驗、事件記憶、常模研究、人口學
變項、信效度、校正分數
ii
A Normative Study on the Word Sequence Learning
ABSTRACT
Background: The Word Sequence Learning Test (WSLT) is one of the verbal
Processing this type of episode memory task is thought to be highly associated with
the perirhinal cortex functioning. Studies have shown that the pathological change of
the perirhinal cortex is one of the main features of the pre-Alzheimer’s disease
arbitrary-associates verbal memory tests, such as the WSLT, would be evident in the
pre-AD. However, currently there is a lack of normative data of the WSLT in Taiwan.
Objective: The purpose of the present study was to establish the normative data of the
WSLT in Taiwanese healthy adults. Issues of the effects of demographic variables and
psychometric properties of the WSLT were also investigated. Methods: The study
included 307 healthy participants stratifying by age (ranging from 16 to 90 years old),
education level (ranging from 0 to 18 years), sex, and different geographical regions
Thirty-three participants were also selected for examining the criterion-related and the
iii
construct validity. Forty patients with amnestic mild cognitive impairment (aMCI)
were also retrospectively collected for verifying the construct validity. Results:
Effects of age and education can be found in the WSLT. The correct, the position, the
learning, and the cued scores decreased as age increased. All sub-scores increased as
education level increased. The correlation coefficient of the retest reliability was
around 0.6 and the alternate-form reliability varied across different sub-scores
(ranging from .14 to .73). Sub-scores of the WSLT correlated moderately (ranging
from .35 to .75) with those of other verbal episodic memory measures, except for the
recognition score. The patient study revealed that the proportion of memory deficits
evident in patients with aMCI measured by the WSLT was higher than by the Logical
Memory subtest of the Wechsler Memory Scale-Third Edition. Conclusion: Age and
education have an influence on the WSLT, and thus the demographic-corrected norm
is needed. In addition, the WSLT has adequate reliability and validity. Based on the
retrospective patient study, it appears that the test performance could be taken as a
cognitive marker for AD. Further prospective study to validate this preliminary result
is necessary.
iv
Table of Contents
Introduction ............................................................................................................... 1
Methods...................................................................................................................... 9
Participants ......................................................................................................... 9
Measurements .................................................................................................. 10
Procedure…….................................................................................................. 12
Statistical Analyses........................................................................................... 13
Results ...................................................................................................................... 15
Demographic Characteristics............................................................................ 15
Effects of Demographic Variables and Intelligence................................... 15
Psychometric Properties................................................................................. 18
Norm Established........................................................................................... 20
Discussion ................................................................................................................ 22
Effects of Demographic Variables and Intelligence on the WSLT......... 22
Psychometric Properties................................................................................ 25
Appropriacy of Norms..................................................................................... 28
Conclusion ............................................................................................................... 30
References................................................................................................................ 32
Appendix.................................................................................................................. 44
A: Informed Consent Form ............................................................................. 44
B: Administration and Scoring......................................................................... 46
Tables....................................................................................................................... 49
Figures..................................................................................................................... 68
v
Introduction
complaints are one of the most common problems (Lezak, 2004). Memory
Alzheimer’s disease (AD) or temporal lobe epilepsy (TLE) (Bell, Lin, Seidenberg, &
Hermann, 2011). A great number of studies have shown that episodic memory is the
earliest and the most prominent cognitive changes in AD, and is vital in early
(Backman, Jones, Berger, Laukka, & Small, 2005; Salmon & Bondi, 2009; Schwindt
& Black, 2009). Furthermore, episodic memory impairment is more likely to predict
the development of probable AD (Elias et al., 2000; Landau et al., 2010). Therefore,
neuropsychological assessment.
Medial temporal lobe (MTL) structures mainly include the hippocampus, the
perirhinal cortex, the entorhinal cortex, and the parahippocampal cortex (Squire &
Zola-Morgan, 1991). For decades, MTL structures have been demonstrated to play a
crucial role in episodic memory based on those patients receiving the surgical removal
1
of pathological tissues or suffering lesions primarily involving hippocampus and/or
amygdala (Frisk & Milner, 1990; Scoville & Milner, 1957; Squire, 2009; Victor &
Agamanolis, 1990; Zola-Morgan, Squire, & Amaral, 1986). Lately, brain imaging
research has reported that MTL activations are strongly linked to both the episodic
encoding (Cabeza & Nyberg, 2000; Martin, 1999) and the episodic retrieval (Nyberg,
McIntosh, Houle, Nilsson, & Tulving, 1996; Schacter, Alpert, Savage, Rauch, &
Albert, 1996). The prefrontal cortex (PFC) is also engaged in episodic memory
(Cabeza & Nyberg, 2000). More specifically, the left-sided PFC activation is
associated with the episodic encoding for verbal stimuli, and the PFC right-sided
activation is associated with the episodic retrieval (Cabeza & Nyberg, 2000).
(Ronald C. Petersen et al., 2006). In fact, studies (Braak & Braak, 1991; Gomez-Isla
brain region corresponding to the medial portion of the perirhinal cortex (Taylor &
Probst, 2008) and the entorhinal cortex in the pre-AD. Impairment of episodic
memory has been viewed as a neurocognitive marker for AD, and the clinical
tests primarily include the verbal and non-verbal ones. The verbal episodic memory
2
tests can be further classified into prose recall nature, such as the Logical Memory
1997), and the list learning one, such as the California Verbal Learning Test (CVLT)
(Delis, Kaplan, Kramer, & Ober, 2000) or the Rey Auditory Verbal Learning Test
(RAVLT) (Schmidt, 1996). List learning tasks are perhaps the most favorable
measures of episodic memory when used to assess early cognitive changes in MCI
and AD, and to predict the conversion of MCI into AD. Tremont and colleagues (2010)
found that 80% of the MCI participants have defective performance on the delayed
32.2% on the delayed recall of the LM (LM-II). Likewise, Rabin and colleagues
(2009) reported similar findings that the delayed recall or learning score of list
learning tests had a higher sensitivity to the diagnosis of the episodic memory
Recently, Saling (2009) proposed that different episodic memory tasks should
studies (Rausch & Babb, 1993; Saling et al., 1993) have demonstrated that left-sided
mesial temporal foci are correlated with the learning performance on unrelated word
pairs of the Paired Associate Learning subtest of the WMS (WMS-PAL) (Wechsler,
1945) but not immediate recall of the passage. Because there are arbitrary associates
3
among these unrelated words, it makes patients with TLE hard to acquire this kind of
verbal materials, while it makes no difficulty in learning of the prose material, such as
the LM. Arbitrary associates or prosemantic learning can be seen as rote learning with
low semantic connections between word pairs (e.g., the WMS-PAL), or words in the
verbal list learning tests. Consequently, the dissociation of arbitrariness and semantic
components in verbal learning tasks which are respectively correlated with lateral and
depending on the extent of semantic loading (Saling, 2009). Researches (Fernández &
Tendolkar, 2006; Weintrob, Saling, Berkovic, Berlangieri, & Reutens, 2002) further
demonstrated the perirhinal cortex specifically playing an essential role in the learning
of arbitrary associates. Thus, it seems that the impairment of the learning tasks with
The Word Sequence Learning Test (WSLT) (Hua, 1986) belongs to list
learning tests with several learning trials and a delayed recall phase, while the WSLT
is characterized by its abstract word list as the item materials. The semantic relations
among these words are low as well. In contrast to many other verbal episodic memory
tests mixing both arbitrary and semantic components, lesser extents of semantic
structures are involved in the WSLT. Based on the literature (Hermann, Seidenberg,
4
Haltiner, & Wyler, 1992; Saling, 2009), an increasing sensitivity to detect the
perirhinal cortex functioning using the WSLT for its underlying feature of arbitrary
associates learning would thus be expected. Moreover, the WSLT is also characterized
ecological validity, because we always recall what has happened in daily life without
forewarning.
WSLT also requires the contribution of frontal lobes. In the learning phase of the test,
there is a position index score measuring the individual’s sequential learning ability.
and sequential learning tasks, and suggest that sequential learning is linked with the
prefrontal cortex which also involves in WM (for a review, see Janacsek & Nemeth,
Accordingly, the position index score of the test would reflect the left prefrontal
memory with widespread neural networks, mainly involving frontal and temporal
cortices.
5
tasks, particularly the RAVLT and the CVLT, both of the test-retest and the
alternative-form reliabilities of most indices in normal adults are above the marginal
level of correlation coefficient ( > 0.6), with the exception of scores of the Trial 1
correct of the CVLT (Crawford, Stewart, & Moore, 1989; Delis, et al., 2000;
Mitrushina & Satz, 1991; Strauss, Sherman, & Spreen, 2006; Uchiyama et al., 1995).
In addition, the Word List Test of the WMS-III Taiwan version (Hua et al., 2005)
normal age groups (the total recall index: .44 & .69; the long delayed recall index: .35
& .72). With respect to the validity, the available literature shows that most of the
indices of the RAVLT significantly correlates to other verbal episodic memory tasks,
such as the CVLT based on the normal adults and patients with traumatic brain injury
(Crossen & Wiens, 1994; Stallings, Boake, & Sherer, 1995) and the LM based on the
outpatient group (Johnstone, Vieth, Johnson, & Shaw, 2000). However, the Word List
Test correlates to the LM at an ordinary level ranging from .30 to .56 (Hua, et al.,
2005). Moreover, the CVLT-Taiwan version has been proved to have a sound validity
Demographic effects on these typical list learning tasks have also been
investigated. Some studies (Mitrushina, 2005; Schmidt, 1996) reported the age effect
6
years old and a significant decline also appears in groups beyond 59 years old, while
the age effect is lesser apparent among 20- to 59-year-old group. Nonetheless, the
consistent age effect on performances of the RAVLT and the CVLT in which the
number of words recalled decreases as age increases has been reported in most
literature (Bleecker, Bolla-Wilson, Agnew, & Meyers, 1988; Chang, et al., 2010;
Dunlosky & Salthouse, 1996; Geffen, Moar, O'Hanlon, Clark, & et al., 1990;
Mitrushina, Satz, Chervinsky, & D'Elia, 1991; Uchiyama, et al., 1995). The
intelligence effect has also been reported, that is, a positive impact of FSIQ scores on
the total recall performance of the RAVLT and the CVLT (Delis, et al., 2000;
Steinberg, Bieliauskas, Smith, Ivnik, & Malec, 2005), and a correlation between IQ
scores and words recalled in learning trials of the RAVLT (Bolla-Wilson & Bleecker,
1986; Query & Megran, 1983). However, the education effect has demonstrated
level and the testing performance in the RAVLT and the CVLT (Delis, et al., 2000;
Magalhaes & Hamdan, 2010; Miatton, Wolters, Lannoo, & Vingerhoets, 2004;
Norman, Evans, Miller, & Heaton, 2000; Pontón et al., 1996; Van der Elst, Van Boxtel,
Van Breukelen, & Jolles, 2005), but some did not (Bolla-Wilson & Bleecker, 1986;
Mitrushina, et al., 1991; Wiens, McMinn, & Crossen, 1988). In addition, Chang et al.
(2010) and Lee et al. (2002) found that the age and the education have a two-way
7
interaction that the performance on the CVLT and the RAVLT tends to be better in
younger and higher education groups than older and lower education groups. The
gender effect has been inconsistent. RAVLT studies (Bleecker, et al., 1988; Harris,
Ivnik, & Smith, 2002; Lee, et al., 2002) reported that women outperformed men while
others didn’t display the gender difference (Kurylo, Temple, Elliott, & Crawford,
heterogeneity of the MTL structures in terms of memory function, the candidate of the
earliest and optimal neurocognitive marker of AD, and clinical utilities, the primary
aim of the present study was to build the normative data of the WSLT in Taiwanese
healthy adults. Thus, three specific objectives of the present study are as follow. First,
the study was to examine psychometric properties of the WSLT; secondly, the study
was also to explore the effects of demographic variables on the performance of the
test, and thirdly, the appropriacy of the normative data established by the present
8
Methods
Participants
Three hundred and seven volunteers (139 males and 168 females) included in
this study were healthy and community dwelling Taiwanese people. According to data
from the 2012 Population Statistical Yearbook (Ministry of the Interior, 2013),
participants were stratified for age, education level, and different geographical regions
in Taiwan (see Table 1). All participants were independent of their personal care and
activities in daily life and society. Via a semi-structured clinical interview, individuals
who had a current or past history of alcohol or drug abuse, brain injury, stroke,
hypertension or other systemic problems under medical control were not excluded.
abilities. Individuals were excluded when the performance on the MMSE was lower
than 24/30, or 18/30 for individuals with less than 2 years of education (Wang, 2007).
Participants were asked to sign a written informed consent form (see Appendix A)
Taiwan University.
9
The retrospective study was also taken and included 40 patients with amnestic
mild cognitive impairment (aMCI: 27 single domain and 13 multiple domain). The
demographic characteristics of the patient group are presented in Table 2. This group
was used for examining the construct validity. Each patient had been diagnosed by
co-workers (2004). All patients were rated as 0.5 in Clinical Dementia Rating (CDR)
Measurements
The Word Sequence Learning Test (Hua, 1986). A 6-item word list in
which each word belongs to abstract words and has little semantic associations
between each other was read aloud by the examiner. Each participant was then asked
to immediately recall the word list exactly in the same word order as the examiner
presented. The learning phase stopped when the subject recalled correctly in two
consecutive trials, otherwise the subject learned up to 10 trials. Three sub-scores were
calculated in the learning phase including the correct, the position, and the learning
scores. The correct score is the number of words correctly recalled in each trial. The
position score is the number of words correctly recalled in the absolute position of the
10
word list. Subjects got a total of two points in the learning score when recalling the
word list correctly and in right order, while some conditions are counted one point in
the learning score (see Appendix B for detailed scoring rules). After 10-minute delay,
the recall phase which is composed of the free recall, the cued recall and the yes/no
recognition was applied to subjects incidentally. In the free recall trial, subjects were
asked to recall words which had learned before without forewarning. The order of the
word list was not required in the free recall. The cued recall trial should be given
when subjects didn’t recall all of the six words. In the cued recall trial, subjects were
given the first character of words which didn’t answer correctly in the free recall trial.
The recall and the cued recall scores are words correctly recalled. The recognition
trial should be given when subjects still didn’t correctly get the 6-word cued recall. In
the recognition trial, subjects identified six target words from 24 foils, also abstract
words. The recognition score derives from an equation that hits plus 24 and substrate
false positives. Detailed instructions and scoring rules were listed in Appendix B.
the present study. Therefore, the estimated full-scaled intelligence quotient (FSIQ)
Taiwan version (Chen & Chen, 2002). The short-form was advised to include the
Similarity, the Matrix Reasoning, the Arithmetic, and the Digit-Symbol Substitution
11
subtests (Chen, Hua, Zhu, & Chen, 2008). The WSLT is one kind of list learning tasks
which generally considered as a form of verbal episodic memory. Thus, the Logical
Memory subtest and the Word List Test of the WMS-III Taiwan version (Hua, et al.,
addition, the WSLT is characterized by its abstract word list and very low semantic
relations among words. The LM and the Word List Test could be assigned to high to
moderate semantic relations. Hence, these two tests were also contained in the
Procedure
Each participant was introduced the aim and the following procedure of tests,
firstly. An informed consent form was then asked to signed, and subjects filled in a
Afterwards, the MMSE was administered for the screening of cognitive abilities, and
assessments including the WAIS-III short-form (Taiwan version), the LM subtest and
the Word List Test of the WMS-III (Taiwan version) were selectively administered to
some participants in a standardized procedure. The whole process of the testing took
20 minutes to one and half hours, the time varying with the participants’ performances
and the number of tests administered. Thirty participants were randomly selected for
12
examining the test-retest and the alternate-form reliability. Thirty-three participants
were also randomly selected for examining the criterion-related and the construct
validity. In the retest group, subjects receive the second administration of the WSLT at
least two months later (ranging from 57 to 132 days). In the alternate-form group, two
other forms with different word lists of the WSLT were administered in a
Statistical Analyses
SPSS version 22.0 was used in all statistical analyses. Descriptive statistics
were calculated for raw sub-scores of the WSLT and demographic characteristics of
all samples. The Kolmogorov-Smirnov test was used to examine whether or not the
and curve estimates were used to determine the most fitted model for the data of these
these different demographic groups differed on various sub-scores of the WSLT, and
13
procedures for conducting multiple comparisons in the post hoc test were proposed by
Siegel & Castellan (1988). The post hoc test was employed to specify the pairwise
data, demographic effects on sub-scores of the WSLT were adjusted, and the
regression coefficients were used to derive corrected equations and then generate
adjusted sub-scores and their percentile ranks. With regard to the psychometric
properties of the WSLT, Pearson correlation and Spearman’s rho were employed to
calculate the test-retest reliability and the alternate-form reliability. Wilcoxon signed
rank test was conducted to examine the practice effect. Likewise, similar correlation
procedures were also employed to investigate the relationships between the WSLT
and other verbal episodic memory tasks to explore the criterion-related and construct
validity. Age effects on the WSLT performances were analyzed as well to examine
construct validity. In addition, the patient group and healthy controls were compared
by Mann-Whitney U test, and the percentage of scores below 5th percentile in two
memory tests (the WSLT and the LM) was also analyzed to provide further evidence
of construct validity.
14
Results
Demographic Characteristics
No age difference was noted in gender (t = 0.81, p > .05), while male participants had
higher education level than their female counterparts in this study (t = -3.42, p < .05).
groups are presented in Table 4. Generally, mean sub-scores are lower as age
noted between male and female participants by the Mann-Whitney test (all sub-scores
p > .05).
The relationship between two demographic variables (age and education) and
sub-scores was examined by scatter plots and curve estimates. It presented that all
sub-scores had a quadratic relationship to age and education, with the exception of a
linear relationship between the recall score and age, and between the learning score
and education. The linear and/or quadratic terms of age and education were centered
for reducing multicollinearity, and then entered into multiple regression analysis to
determine which variable significantly predicted the sub-scores (for the result, see
Table 5). However, there was no gender effect after the above statistical analyses were
15
used. Stepwise regression analysis is presented in Table 6. Education accounted for
the most variances in the correct, the position and the learning scores (31.4% to
35.9% of variance), followed by age (3% to 6.7% of variance) and education2 (0.8%
to 4.8%). In the delayed recall, the cued and the recognition scores, education2 could
account for the most variances (9.8% to 14.2% of variance), followed by age2 (2.4%
quadratic term had more conspicuous effects on explaining variances in the WSLT
Table 7. The result showed that a significant between-group difference of age groups
was evident in the correct, the position, the learning, and the cued recall scores (all
sub-scores p < .005), while the recognition score approached the significant level (p
= .053). In the pairwise comparisons of different age groups, the post hoc test (see
Table 8) showed that the oldest group ( > 74 years old) had a significant decline in
these four sub-scores compared to groups 16-24 and 25-34, with the exception of
non-significant decline in the correct score compared to the 25-34 group and in the
cued score compared to the 16-24 group. The 65-74 group also showed a significant
discrepancy with groups 16-24 and 25-34 in the four sub-scores, with the exception of
non-significant discrepancy with the 16-24 group in the cued score. When the 65-74
16
group and the above-74 group emerged, the new group ( > 64 years old) performed
significantly worse than groups 25-34 and 45-54 in the four sub-scores, and also had a
decline in the correct, the position, and the learning scores compared to the 16-24
group and/or the 35-44 group. Thus, there seems an obvious decline in the WSLT
sub-scores for those above the 64 year-old subjects. Because there was no age effect
on the recall score, we further controlled learning trials by dividing subjects into three
groups: 2-4 trials group, 5-7 trials group, and 8-10 trials group. The ANOVA result
different education groups in all sub-scores. The post hoc test further showed that the
above-12th grade group had a better performance in the correct, the position, and the
learning scores compared to other three groups. In addition, the above-12th grade and
the 10-12th grade group also performed significantly better than the 0-6th grade group
sub-scores and the estimated FSIQ. The position, the learning, and the recall scores
were positively correlated with the estimated FSIQ (ranging from .26 to .30, p < .05).
However, the correct, the cued and the recognition scores had a non-significant
correlation with the estimated IQ (ranging from .09 to .22, p > .05).
17
Psychometric Properties
the position, the learning and the cued scores had above marginal test-retest reliability
(ranging from .61 to .64). The recall and the recognition scores had an approaching
marginal test-retest reliability (.54 and .51 respectively). In addition, the practice
effect was found in the position, the learning, the recall, and the cued scores with at
least 2 months of interval time (mean = 87.93 days). With regard to the alternate-form
reliability (see Tables 12 & 13), the coefficient fell among .26 to .61 for form 1 versus
form 2, .14 to .73 for form 1 versus form 3, and .37 to .65 for form 2 versus form 3. In
addition, the performances between form 1 versus form 2, form 1 versus form 3, and
form 2 versus form 3 showed no significant differences except for form 2 versus form
with other measures of verbal episodic learning and memory including the LM and
the Word List Test of the WMS-III (see Table 14). The correct, the position, the
learning, and the recall scores correlated moderately well with corresponding
sub-scores in two criteria (ranging from .42 to .75). The cued score correlated
significantly with the LM-II (r = .35), while no significant correlation was found with
the Word List Test-II (r = .15). Moreover, the recognition score didn’t significantly
18
correlated with corresponding recognition score in two criteria (-.01 and .29
respectively).
of the extent of semantic relations was differently assigned to the WSLT, the Word
List Test, and the LM. We considered the LM to be abundant of semantic structures in
its learning stimuli, and followed by the Word List Test and the least extent of
semantic relations in these three tests, the construct validity was verified by
tasks. The result showed that the WSLT correlated much higher with the Word List
Test in all sub-scores except for the recognition score (see Table 14).
temporal and frontal lobe (Salthouse, Atkinson, & Berish, 2003; Schretlen et al., 2000;
St. Jacques, Rubin, & Cabeza, 2012) which were the main brain regions involved in
the WSLT. Age effect was then provided to further evaluate the construct validity of
the WSLT. The results of the ANOVA test and the regression analyses revealed
obvious age effects on each sub-scores of the WSLT (see Tables 5, 6, and 7). The post
19
hoc comparison also showed that a significant decline in the performance was found
A retrospective patient study was also employed for examining the construct
validity. Patients with aMCI performed worse than healthy controls in all sub-scores
of the WSLT (see Table 15). We further investigated performances on two memory
tests among these patients with aMCI. The result found that 77.5% (31/40) of patients
had a defective picture in the WSLT compared to only 20% (8/40) in the LM (see
Figure 4-1), and the proportion differed significantly ( (1) = 21.16, p < .001). In
patients presented memory deficits in the WSLT compared to only 11.11% (3/27) in
the LM (see Figure 4-2), and the proportion discrepancy reached a statistically
Norms Established
Regression coefficients were used for ruling out demographic effects on the
WSLT performances, and corrected equation of each sub-score was then provided as
follow:
Adjusted score = raw sub-score – bedu × (edu – 13.84) – bedu2 × (edu – 13.84)2
20
The bedu, bedu2, bage, and bage2 refer to the unstandardized coefficients of
education, the square of education, age, and the square of age respectively. The
numbers 13.84 and 42.16 derived from the average age (by year) and the average
education level (by year) of all samples. Corrected equations of all sub-scores are
presented in Table 9. All raw sub-scores of our samples were adjusted by these
corrected equations and were ranked by percentile. The referential table for percentile
rank of adjusted sub-scores is presented in Table 10. To identify the defective level of
these adjusted sub-scores, Lezak and colleagues (2004) advised that the fifth
21
Discussion
The present study was aimed to identify effects of age, education level, gender,
and intelligence on the WSLT, and further to provide equations for correcting the
Effect of age. The age-related decline in the correct, the position, and the
learning scores is reported in the results of the ANOVA test and the multiple
regression analyses. These are consistent with previous findings that words recalled
on the learning phase show greater age effects (Dunlosky & Salthouse, 1996; Incalzi,
Capparella, Gemma, Marra, & Carbonin, 1995; Mitrushina, et al., 1991; Uchiyama, et
al., 1995). However, a linear relationship between scores on the learning phase and
was found. The present results also exhibited that the recognition score had an
found that recognition shows less age effects compared to other memory indices
(Bleecker, et al., 1988; Mitrushina, et al., 1991). It is worth noting that the recall score
shows no age effect in the ANOVA test and regression analyses (also see Figure 2-2)
22
which contrast with previous studies (Incalzi, et al., 1995; Magalhaes & Hamdan,
2010). It is possibly due to the learning material of the WSLT comprises abstract
words and the unrelated word list that distinctly differ from other list learning tests. In
addition, another reason could be the stop rule of the consecutive two-trial corrects in
the learning phase causing less exposure to the learning material. Most young subjects
get higher learning scores will have fewer trials to learn, and get poorer score in the
recall. Accordingly, we further investigated the age effect in the recall score with a
control of learning trials. However, the result revealed that even under similar
learning trials, no age effect was found. Thus, the reason for a lack in the age effect on
the recall score is still unclear, and further investigations are required.
and education accounts for more variances than age. This is not common in most
previous studies, but still many studies have reported education effects on the RAVLT
and the CVLT (Messinis, Tsakona, Malefaki, & Papathanasopoulos, 2007; Miatton, et
al., 2004; Norman, et al., 2000; Van der Elst, et al., 2005). The post hoc test shows
that the highest-education group performed significantly well than other three groups
in the learning phase. The reason for this phenomenon might be resulted from the
words used in the WSLT. These words belong to abstract and function words. Subjects
with higher education level might be proficient in abstract verbal ability and further in
23
favor of learning these abstract words. Moreover, in the post hoc test, subjects with
recall phase. Education also has a quadratic model in the recall, the cued and the
Mandarin and Taiwanese. In our samples, some low-education subjects are Taiwanese
language users, and this causes great difficulty in learning these words. Nevertheless,
an apparent education effect on the WSLT should be explored strictly in the future.
intelligence, our data showed no gender difference in all sub-scores of the WSLT.
However, the position, the learning, and the recall scores significantly correlate with
the estimated FSIQ. Comparing to other studies, correlations between the WSLT and
the intelligence are milder. It might be partially associated with different measures of
the intelligence (Bolla-Wilson & Bleecker, 1986). Another reason might be due to
found that the AVLT correlated better at a moderate level of the FSIQ than higher and
lower IQ.
24
Psychometric Properties
Reliability. In the retest reliability, most of the scores fall above marginal
level (ranged from .61 to .64). The recall and the recognition scores fall near to
marginal level (.54 and .51 respectively). The correlation coefficients are all
significant and are similar to the previous researches (Mitrushina & Satz, 1991;
Uchiyama, et al., 1995). In addition, many other studies found practice effects on the
RAVLT after at least 1 month retest. Our results also showed the same practice effect
on the position, the learning, the recall, and the cued scores. Hence, in the clinical
practice, it might be noticed that the practice effect still exists at least 2 months retest
in normal subjects. As for the alternate forms in the WSLT, results show nearly no
appeared in sub-scores of the recall phase. Thus, it should be careful to use corrected
equations in forms II and III of the WSLT which might cause more biases in
explaining the adjusted scores. The possible explanation for low level of retest and
alternate-form reliabilities in the recall score is the violation of the incidental recall.
Subjects who received the WSLT more than twice might realize a following recall
phase and violated the incidental recall. In addition, due to the rules about the
selective administration in the cued and the recognition, the truncated sample in these
25
two sub-scores might also influence the reliability. However, the WSLT generally
significantly with the corresponding indices in the LM and the Word List Test. The
result is consistent with prior findings (Crossen & Wiens, 1994; Johnstone, et al.,
2000; Stallings, et al., 1995; Strauss, et al., 2006). However, the recognition score had
between the AVLT and the CVLT has also been reported by Crossen & Wiens (1994).
The reason for the non-significant correlation in the recognition score might be due to
low difficulty in the recognition of the Word List Test. Foils in the Word List Test are
not involved words of the interference list. Thus, subjects with higher recognition
score in the Word List Test might perform variously in the WSLT. In addition, the
truncated problem in the recognition subtest might also lower the criterion-related
validity.
memory measures, we expect a closer correlation between the WSLT and the Word
List Test than to the LM. The results (see Table 14) showed that all sub-scores of the
WSLT except for the recognition had higher correlation coefficient with the Word List
26
Test. This finding verifies the semantic-syntactic continuum in the verbal episodic
memory measures, and confirms that the WSLT has the least semantic structures in its
word list compared the other tests. That is, the WSLT could be considered as a higher
arbitrary-associates loaded test which specifically associates with the perirhinal cortex
functioning.
Construct validity: aging effect. The WSLT is associated with the temporal
lobe which dominates learning and memory, and is associated with the frontal lobe
which is responsible of the sequential learning. Likewise, these brain regions are
found to correlate with aging process, and age-related decline in the episodic learning
and memory is expected (Salthouse, et al., 2003; St. Jacques, et al., 2012; Stranahan
& Mattson, 2010). Thus, it is expected that the performance of the WSLT should be
worse in the aging group. The results presented in Table 8 and the Figures 2-1 & 2-2
agree with our expectations, and could be a supporting evidence of validity in the
WSLT.
collecting patients’ data retrospectively for further examining the clinical validity of
the WSLT. The result shows that all sub-scores differ between the patient group and
healthy controls (see Table 15). It suggests that the performance on the WSLT can
27
compared performances on the WSLT and the LM, the WSLT reveals a higher
aMCI-single domain (see Figures 4-1 & 4-2). This finding is in accordance with the
previous studies (Rabin, et al., 2009; Tremont, et al., 2010). According to the criteria
of MCI proposed by Petersen (2004), patients with subjective complaints should also
result reveals that the WSLT can reflect more memory deficits of patients with
subjective memory complaints. It may suggest that the WSLT is a more sensitive tool
Appropriacy of Norms
present study included 307 participants stratified by age, education, gender and
year-old group, the above-75 year-old group, the 0-6 education group, and the 7-9
education group, our samples matched the distribution of population. Therefore the
representativeness of the sampling is adequate. Second, the norms collected from May
2011 to April 2014 have good recency. Third, the normative data was corrected by
demographic variable, and the samples are relevant to the clinical patients for the
28
same cultural background and compatible demographic characteristics. Thus, the
29
Conclusion
The present study provides a normative data of the WSLT. In addition, the
normative data. The psychometric properties are also examined, and the result reveals
In the clinical use, it has always been lacking a verbal episodic memory task
with good psychometric properties suitable for Taiwanese people. Thus, to establish
the normative data of the WSLT could be helpful in the clinical practice of Taiwan for
according to results of the construct validity in the present study, the WSLT is proven
to have more arbitrary associates than other verbal memory measures. The
task-specificity is confirmed in the present study. The patient study also shows that
the WSLT has a higher sensitivity in detecting memory deficits in aMCI compared to
the LM. Hence, we infer that the WSLT has high sensitivity in detecting the perirhinal
cortex functioning, and play a crucial role in the early detection of memory deficits in
Alzheimer’s disease.
more participants in the highest education level group ( > 13 years) and less
participants in the highest age range ( > 74 years old) were included in the present
30
study. With a trend of aging society, the normative data might lose its
representativeness in the old-old subjects ( > 74 years old). Future studies could
recruit the old-old subjects specifically for exploring the performance on the WSLT.
Second, the interpretation of the patient study should be taken with care for its
clinical significance of memory deficits which are supported by the WSLT, and even
the alternative-form reliability shows low level of correlation coefficient in the recall,
the cued, and the recognition scores, and the corrected equations might be unsuitable
for forms II and III. Thus, extra normative data for forms II and III is required in
future studies.
31
References
Backman, L., Jones, S., Berger, A.-K., Laukka, E. J., & Small, B. J. (2005). Cognitive
Bell, B., Lin, J. J., Seidenberg, M., & Hermann, B. (2011). The neurobiology of
154-164.
Bleecker, M. L., Bolla-Wilson, K., Agnew, J., & Meyers, D. A. (1988). Age-related
403-411.
Bolla-Wilson, K., & Bleecker, M. L. (1986). Influence of verbal intelligence, sex, age,
Cabeza, R., & Nyberg, L. (2000). Neural bases of learning and memory: functional
Chang, C. C., Kramer, J. H., Lin, K. N., Chang, W. N., Wang, Y.-L., Huang, C.-W., et
al. (2010). Validating the Chinese version of the Verbal Learning Test for
32
screening Alzheimer’s disease. Journal of the International
Chen, H. Y., & Chen, Y. H. (2002). Wechsler Adualt Intelligence Scale-third edition
Science Corporation.
Chen, H. Y., Hua, M. S., Zhu, J., & Chen, Y. H. (2008). Selection of Factor-Based
Crawford, J., Stewart, L., & Moore, J. (1989). Demonstration of savings on the AVLT
16(2), 190-194.
Delis, D. C., Kaplan, E., Kramer, J. H., & Ober, B. A. (2000). California Verbal
Corporation.
33
multitrial free recall. Aging, Neuropsychology, and Cognition, 3(1), 2-14.
Elias, M. F., Beiser, A., Wolf, P. A., Au, R., White, R. F., & D'Agostino, R. B. (2000).
Fernández, G., & Tendolkar, I. (2006). The rhinal cortex: ‘gatekeeper’ of the
Frisk, V., & Milner, B. (1990). The role of the left hippocampal region in the
Geffen, G., Moar, K., O'Hanlon, A., Clark, C., & et al. (1990). Performance measures
of 16- to 86-year-old males and females on the Auditory Verbal Learning Test.
Gomez-Isla, T., Price, J. L., McKeel, D. W., Jr., Morris, J. C., Growdon, J. H., &
4491-4500.
Harris, M. E., Ivnik, R. J., & Smith, G. E. (2002). Mayo's Older Americans Normative
Hermann, B. P., Seidenberg, M., Haltiner, A., & Wyler, A. R. (1992). Adequacy of
34
Language Function and Verbal Memory Performance in Unilateral Temporal
Hua, M. S., Chang, B. S., Lin, K. N., Yand, J. M., Lu, S. R., & Chen, H. Y. (2005).
Incalzi, R. A., Capparella, O., Gemma, A., Marra, C., & Carbonin, P. U. (1995).
Janacsek, K., & Nemeth, D. (2013). Implicit sequence learning and working memory:
Johnstone, B., Vieth, A. Z., Johnson, J. C., & Shaw, J. A. (2000). Recall as a function
Kapur, S., Tulving, E., Cabeza, R., McIntosh, A. R., Houle, S., & Craik, F. I. (1996).
Kurylo, M., Temple, R. O., Elliott, T. R., & Crawford, D. (2001). Rey Auditory Verbal
35
Learning Test (AVLT) Performance in Individuals With Recent-Onset Spinal
Landau, S., Harvey, D., Madison, C., Reiman, E., Foster, N., Aisen, P., et al. (2010).
Lee, T. M. C., Yuen, K. S. L., & Chan, C. C. H. (2002). Normative Data for
Press.
Magalhaes, S. S., & Hamdan, A. C. (2010). The Rey Auditory Verbal Learning Test:
Normative data for the Brazilian population and analysis of the influence of
Martin, A. (1999). Automatic activation of the medial temporal lobe during encoding:
Mehrens, W. A., & Lehman, I. J. (1987). Using standardized tests in education (4th
Messinis, L., Tsakona, I., Malefaki, S., & Papathanasopoulos, P. (2007). Normative
36
data and discriminant validity of Rey's Verbal Learning Test for the Greek
Miatton, M., Wolters, M., Lannoo, E., & Vingerhoets, G. (2004). Updated and
Ministry of the Interior, T., ROC. (2013). 2012 population statistical yearbook. from
http://sowf.moi.gov.tw/stat/year/list.htm
47(6), 790-801.
Mitrushina, M., Satz, P., Chervinsky, A., & D'Elia, L. (1991). Performance of four age
Norman, M. A., Evans, J. D., Miller, W. S., & Heaton, R. K. (2000). Demographically
Nyberg, L., McIntosh, A., Houle, S., Nilsson, L., & Tulving, E. (1996). Activation of
37
medial temporal structures during episodic memory retrieval. Nature,
380(6576), 715-717.
Petersen, R. C., Parisi, J. E., Dickson, D. W., Johnson, K. A., Knopman, D. S., Boeve,
Pontón, M. O., Satz, P., Herrera, L., Ortiz, F., Urrutia, C. P., Young, R., et al. (1996).
Query, W. T., & Megran, J. (1983). Age-related norms for AVLT in a male patient
Rabin, L. A., Pare, N., Saykin, A. J., Brown, M. J., Wishart, H. A., Flashman, L. A., et
al. (2009). Differential memory test sensitivity for diagnosing amnestic mild
Rausch, R., & Babb, T. L. (1993). HIppocampal neuron loss and memory scores
before and after temporal lobe surgery for epilepsy. Archives of Neurology,
38
50(8), 812-817.
Saling, M. M., Berkovic, S. F., O'Shea, M. F., Kalnins, R. M., Darby, D. G., & Bladin,
Savage, R. M., & Gouvier, W. D. (1992). Rey auditory-verbal learning test: The
effects of age and gender, and norms for delayed recall and story recognition
Schacter, D. L., Alpert, N. M., Savage, C. R., Rauch, S. L., & Albert, M. S. (1996).
39
Schmidt, M. (1996). Rey Auditory and Verbal Learning Test. A handbook. Los
Schretlen, D., Pearlson, G. D., Anthony, J. C., Aylward, E. H., Augustine, A. M.,
6(1), 52-61.
45(1), 181-190.
Scoville, W. B., & Milner, B. (1957). Loss of recent memory after bilateral
11-21.
Siegel, S., & Castellan, N. J. (1988). Nonparametric statistics for the behavioral
Squire, L. R., & Zola-Morgan, S. (1991). The Medial Temporal Lobe Memory System.
40
St. Jacques, P. L., Rubin, D. C., & Cabeza, R. (2012). Age-related effects on the
33(7), 1298-1310.
Stallings, G., Boake, C., & Sherer, M. (1995). Comparison of the california verbal
learning test and the rey auditory verbal learning test in head-injured patients.
Steinberg, B. A., Bieliauskas, L. A., Smith, G. E., Ivnik, R. J., & Malec, J. F. (2005).
Mayo's Older Americans Normative Studies: Age- and IQ-Adjusted Norms for
the Auditory Verbal Learning Test and the Visual Spatial Learning Test. The
Taylor, K. I., & Probst, A. (2008). Anatomic localization of the transentorhinal region
Tremont, G., Miele, A., Smith, M. M., & Westervelt, H. J. (2010). Comparison of
41
verbal memory impairment rates in mild cognitive impairment. Journal of
Uchiyama, C. L., D'Elia, L. F., Dellinger, A. M., Becker, J. T., Selnes, O. A., Wesch, J.
E., et al. (1995). Alternate forms of the Auditory-Verbal Learning Test: Issues
Van der Elst, W. I. M., Van Boxtel, M. P. J., Van Breukelen, G. J. P., & Jolles, J.
(2005). Rey's verbal learning test: Normative data for 1855 healthy
participants aged 24-81 years and the influence of age, sex, education, and
11(03), 290-302.
Victor, M., & Agamanolis, D. (1990). Amnesia due to Lesions Confined to the
Wechsler, D. (1997). Wechsler Memory Scale. Third edition manual. San Antonio:
42
The Psychological Corporation.
Weintrob, D. L., Saling, M. M., Berkovic, S. F., Berlangieri, S. U., & Reutens, D. C.
Learning Test: Development of norms for healthy young adults. The Clinical
Zola-Morgan, S., Squire, L. R., & Amaral, D. G. (1986). Human amnesia and the
6(10), 2950-2967.
43
Appendix
參與研究同意書
國立臺灣大學心理學研究所
您好:
首先,感謝您參與本次的研究。本研究名稱為「台灣正常人中文字詞次序學
習測驗之常模研究」。此研究之主要目的在於建立台灣地區中文字詞次序學習測
驗之常模,並探討台灣地區人口學變項與中文字詞次序學習測驗表現之相關性。
同時評估中文字詞次序學習測驗在台灣正常人使用上之信度與效度等,期能使臨
床實務工作者在衡鑑與追蹤病人認知功能狀況時能做更有效的評估。
本研究過程為透過問卷填寫以及神經心理測驗評估之方式完成,對個人並無
侵入性危險。主要分成以下幾個部分:
1. 施測前:主試者簡介研究目的,研究參與者閱讀並填寫同意書、基本資料、
與疾病史等相關問卷。
2. 正式施測:由研究者對參與者進行神經心理功能測驗評估。
基於我們對於您個人權益的尊重,本研究對您有以下的承諾:
※您有權隨時停止作答,並且不會因此而受到懲罰。
※您有權要回您的資料,並且撤銷被納入分析。
※您有權在本研究結束後知道研究結果。
※如有任何疑問,您可隨時經由參與者聯中所附之聯絡方式與研究人員進行
聯絡。
本研究之結果,僅供學術使用,將不做其他用途;本研究的資料沒有對錯或
好壞的判斷,您的所有資料將被匿名保密,請您放心填答,謝謝您的合作!
如您同意參與本研究,請於簽上您的姓名與聯絡方式,謝謝您。
本人已詳細閱讀本同意書,並同意參與研究。
參與者簽名:__________ 聯絡方式:___________日期:____年____月____日
再次感謝您。
實驗/主試者:陳威嘉
國立臺灣大學心理學研究所/臨床心理學組/碩士班研究生
聯絡地址:國立台灣大學心理學系南館 118 室
聯絡電話/手機:02-23695438/0963117197
指導老師:花茂棽
44
國立臺灣大學心理學系教授
聯絡地址:國立台灣大學心理學系南館 118 室
聯絡電話/手機:02-23695438
研究者簽名:__________ 日期:_____年____月____日
45
Appendix B: Administration and Scoring
施測:
指導語:
一、學習:等一下我會唸出一些字詞讓你學習,你就照你聽到的把它背誦出
來,由於這些字詞較長一點,可能你第一次無法完全記得,但是
我會重複念這些字詞讓你學。
二、回想:(10 分鐘後)之前我有請你記一些詞,你還記得有哪些嗎?你回憶
看看。
三、線索回憶:(如果受試者無法回想)現在我給你剛才那些字詞的第一個字,
你把完整的字詞說出來。
四、再認:(如果受試者也無法完成線索回憶)現在我念一些字詞,你告訴我
是不是剛才曾經聽過的。
計分:
一、學習:每一個字詞 1 分。橫列計正確分數,每次 trial 有 6 個,共 10 次
trials,滿分 60 分。縱列計位置分數,有 6 個詞,每個詞有 10
次 trials,滿分 60 分。學習分數是以受試者的答案符合位置且項
目皆正確來計算,每題 2 分,共 10 題,滿分 20 分。倘若受試者
的答案是符合位置、項目正確連續二次,即停止施測,以下 trials
皆以 2 分計之。在如下四種情形可給予每一 trial1 分:
1. 漏了一個詞,但其他項目、位置全對。
2. 多了一個詞,但其他項目、位置全對。
3. 錯了一個詞,但其他項目、位置全對。
4. 全部的詞皆對,只是鄰近的兩個詞位置顛倒。
二、回想:每個詞 1 分,共 6 個詞,滿分 6 分。
三、線索回憶:每個詞 1 分,共 6 個詞,滿分 6 分。
四、再認:計算受試者 O (Hit ≤ 6)中對了幾個,減去 N 中錯的個數,再加上 24。
滿分 30 分。
46
中文字詞次序學習測驗
國立台灣大學心理學系 花茂棽教授
姓名: 施測編號:
性別: 出生日期:
學歷: 施測日期:
圈出施測之序列
選擇題本 刺激詞
Ⅰ 相當 原來 不但 時候 服務 從此
Ⅱ 已經 還是 雖然 什麼 其中 覺得
Ⅲ 至少 以後 原則 任何 完全 當時
受試者之反應--逐字記錄在空表上,受測者能連續正確地背誦出兩次時,就停止
繼續施測,同時對於其他的嘗試給予分數。
正確 學習
分數 分數
1. _______ _______ _______ _______ _______ _______ ___/6 ____
位置分數 總學習分數
47
Recall (10 minutes later) Cueing
Response Correct score Response Correct score
1._______ _______ 1._______不 _______
2._______ _______ 2._______相 _______
3._______ _______ 3._______時 _______
4._______ _______ 4._______原 _______
5._______ _______ 5._______從 _______
6._______ _______ 6._______服 _______
Total _______ Total _______
Recognition
I
1.原來______O 11.雖然______N 21.時候______ O
2.同時______N 12.相當______O 22.當時______ N
3.覺得______N 13.應該______N 23.表示______ N
4.不但______O 14.完全______N 24.已經______ N
5.至少______N 15.從此______O 25.任何______ N
6.原則______N 16.當然______N 26.可以______ N
7.開始______N 17.其中______N 27.以後______ N
8.如何______N 18.特別______N 28.服務______ O
9.繼續______N 19.還是______N 29.什麼______ N
10.可能______N 20.結果______N 30.決定______ N
(1) HIT=______/6
(2) FALSE ALARM=______/24
(3) TOTAL=______/30 【(1) + 24 – (2)】
48
Tables
Table 1
Demographic Distribution of Samples
Age
16-24 25-34 35-44 45-54 55-64 65-74 ≥75 Total Exp. N
Area
Northern 32 31 23 15 23 7 4 135 135
Western 8 24 10 14 6 2 2 66 74
Southern 9 20 11 21 16 13 5 95 84
Eastern - 3 3 2 2 1 - 11 7
Education
0-6 - - - 1 10 11 6 28 46
7-9 1 - - 5 5 3 1 15 40
10-12 10 10 5 21 16 4 3 69 96
≥13 38 68 42 25 16 5 1 195 118
Gender
Female 23 39 28 33 24 17 4 168 151
Male 26 39 19 19 23 6 7 139 149
Total 49 78 47 52 47 23 11 307 300
Exp. N 48 56 56 56 44 22 18 300
Note. Exp. N: expected number of sampling.
49
Table 2
Demographic characteristics of Samples for the Construct Validity
N Agea Educationa % of female
NC 40 70.10 (7.20) 13.18 (2.79) 50.00
aMCI 40 (27/13)b 70.13 (8.40) 13.10 (3.49) 50.00
Note. NC: normal control group.
a
mean (standard deviation). b number of single domain and multiple domain
50
Table 3
Descriptive Statistics of Demographic Characteristics
N Age Edu
a b
Total 307 42.16 (16.95) [16-90] 13.84 (3.68) [0-18]
Age
16-24 49 21.63 (1.79) 14.18 (1.78)
25-34 78 28.24 (2.76) 15.91 (1.77)
35-44 47 38.79 (2.50) 16.09 (1.95)
45-54 52 49.87 (2.72) 13.77 (3.00)
55-64 47 59.62 (2.66) 11.40 (3.76)
65-74 23 68.74 (2.58) 9.52 (4.81)
≥75 11 80.09 (5.43) 7.91 (5.41)
Education
0-6 28 66.93 (7.80) 5.25 (1.94)
7-9 15 56.73 (14.69) 8.87 (0.52)
10-12 69 46.77 (16.90) 12.00 (0.00)
≥13 195 35.85 (13.29) 16.11 (1.38)
Gender
Female 168 (54.72%) 42.87 (16.51) 13.22 (4.12)
Male 139 (45.28%) 41.30 (17.49) 14.60 (2.91)
Note. a mean (standard deviation). b [minimum-maximum].
51
Table 4
Descriptive Statistics of Raw Sub-scores by Age, Education, and Gender
Correct Position Learning Recall Cued Recognition
M Md M Md M Md M Md M Md M Md
(SD) (R) (SD) (R) (SD) (R) (SD) (R) (SD) (R) (SD) (R)
Total 54.08 56 45.79 50 12.75 15 3.48 4 4.87 5 28.64 30
(6.72) (22-60) (12.64) (6-60) (6.09) (0-20) (1.47) (0-6) (1.30) (0-6) (1.98) (14-30)
Age
16-24 56.78 58 51.27 55 15.31 17 3.86 4 4.96 5 29.04 30
(4.52) (33-60) (9.41) (18-60) (5.10) (0-20) (1.23) (2-6) (1.02) (2-6) (1.32) (26-30)
25-34 56.49 58 51.38 54 15.47 17 3.68 4 5.22 5.5 29.14 30
(4.37) (32-60) (9.64) (13-60) (4.67) (0-20) (1.27) (1-6) (0.98) (2-6) (1.29) (25-30)
35-44 55.89 57 48.77 51 14.19 15 3.34 3 4.77 5 28.57 29
(3.70) (43-60) (9.08) (22-60) (4.45) (4-20) (1.40) (1-6) (1.22) (2-6) (1.74) (24-30)
45-54 54.58 57 45.85 49 12.75 14 3.42 4 5.08 6 28.88 30
(5.74) (25-60) (10.35) (15-58) (5.22) (0-19) (1.51) (0-6) (1.23) (1-6) (1.69) (23-30)
55-64 51.70 53 40.02 42 9.43 9 3.66 4 4.96 5 28.55 29
(6.28) (33-59) (12.39) (12-57) (6.04) (0-19) (1.55) (1-6) (1.29) (1-6) (1.84) (23-30)
65-74 45.13 48 31.00 28 5.57 3 2.91 3 3.96 5 27.09 28
(8.63) (28-58) (12.49) (13-56) (5.73) (0-18) (1.62) (0-5) (1.72) (0-6) (3.73) (14-30)
≥75 43.82 46 24.27 22 5.09 2 1.73 1 3.09 3 26.18 26
(11.74) (22-58) (13.19) (6-47) (6.27) (0-17) (1.90) (0-6) (1.97) (0-6) (2.96) (20-30)
52
Table 4 (continued)
53
Table 5
Summary of Multiple Regression Model
Education Education2 Age Age2
t b t b t b t b
Correct 4.87*** 0.556 -4.39*** -0.070 -3.53*** -0.074 -1.22 -
Position 5.28*** 1.149 -1.93 - -5.00*** -0.199 -1.63 -
Learning 7.11*** 0.648 - - -5.23*** -0.103 -0.82 -
Recall 1.58 - -2.84** -0.012 -0.87 - - -
Cued 0.21 - -4.13*** -0.015 -0.37 - -2.83** -0.001
Recognition -0.27 - -3.87*** -0.022 -1.87 - -2.20* -0.001
Note. b: unstandardized coefficients. t: t value.
*p < .05. **p < .01. ***p < .005.
54
Table 6
Explained Variance in Stepwise Multiple Regression Analyses
Education Education2 Age Age2 Total
2 2 2 2
ΔR ΔR ΔR ΔR R2
Correct 35.90% 4.80% 3.00% - 43.70%
Position 34.40% 0.80% 6.70% - 41.90%
Learning 31.40% - 6.50% - 37.90%
Recall 1.20% 9.80% - - 11.00%
Cued - 14.20% - 2.90% 17.10%
Recognition - 13.20% - 2.40% 15.60%
Note. ΔR2: R Square Change.
55
Table 7
Comparisons between Different Age and Education Groups by Kruskal-Wallis
One-Way Analysis of Variance by Ranks
a b
Age group 6 Education group 3
56
Table 8
Pairwise Comparisons between Different Age and Education Groups by Post Hoc Test
with Bonferroni Corrections
Comparisons between different groups
a age groupsa
age groups education groupsb
(add group F’)
(alpha = .0012) (alpha = .0042)
(alpha = .0017)
a, b, c < d;
Correct A, B > F; A > G A, B, D > F’
a<c
A, B, C, D > G; A, B, C, D, E > F’ a, b, c < d;
Position
A, B > E, F; A > C, D A, B > E; A > C, D a<c
A, B, C, D > F’; a, b, c < d;
Learning A, B > E, F, G
A, B > E; A > C, D a<c
Recall - - a, b < c, d
Cued B > F, G B, D > F’ a < c, d
Recognition - - a < c, d
Note. A: 16-24 year-old group. B: 25-34 year-old group. C: 35-44 year-old group. D: 45-54
year-old group. E: 55-64 year-old group. F: 65-74 year-old group. G: > 74 year-old group. F’ : >
64 year-old group. a: 0-6 years. b: 7-9 years. c: 10-12 years. d: > 12 years.
a
Partial out the education effect. b Partial out the age effect.
57
Table 9
Corrected Equations of Raw Sub-scores
Adjusted score Equation
Adj_Correct = Correct – 0.556 × (edu – 13.84) + 0.07 × (edu – 13.84)2 +
0.074 × (age – 42.16)
Adj_Position = Position – 1.149 × (edu – 13.84) + 0.199 × (age – 42.16)
Adj_Learning = Learning – 0.648 × (edu – 13.84) + 0.103 × (age – 42.16)
Adj_Recall = Recall + 0.012 × (edu – 13.84)2
Adj_Cued = Cued + 0.015 × (edu – 13.84)2 + 0.001 × (age – 42.16)2
Adj_Recognition = Recognition + 0.022 × (edu – 13.84)2 + 0.001 × (age –
42.16)2
58
Table 10
Referential Table for Percentile Rank of Adjusted WSLT Sub-scores
Percentile Adj_Correct Adj_Position Adj_Learning Adj_Recall Adj_Cued Adj_Recog.
1 32.70 15.70 -2.36 0.29 2.08 23.50
2 39.11 18.31 -0.08 1.00 2.29 24.97
3 41.97 22.25 2.75 1.04 2.49 25.31
4 43.08 25.47 3.22 1.04 2.85 25.42
5 44.71 26.77 3.94 1.06 3.07 25.82
6 45.74 28.10 4.15 1.13 3.11 26.19
7 47.48 28.87 4.48 1.21 3.17 26.38
8 48.62 29.90 4.98 1.32 3.27 26.44
9 48.97 30.57 5.55 1.74 3.30 26.49
10 49.37 31.32 5.74 2.00 3.40 26.73
11 49.86 32.32 6.18 2.02 3.47 27.09
12 49.99 32.79 6.40 2.04 4.00 27.13
13 50.34 33.54 6.69 2.04 4.08 27.17
14 50.51 34.07 6.85 2.04 4.12 27.23
15 50.82 34.70 7.38 2.06 4.19 27.34
16 51.29 35.56 7.83 2.06 4.25 27.40
17 51.45 36.09 7.95 2.06 4.27 27.40
18 51.71 36.52 8.04 2.06 4.29 27.44
19 52.01 36.75 8.36 2.06 4.32 27.51
20 52.23 37.58 8.52 2.06 4.36 27.57
21 52.70 37.97 8.67 2.21 4.37 27.87
22 52.79 38.93 8.88 2.21 4.42 28.11
23 52.85 39.49 9.15 2.21 4.46 28.14
24 52.98 40.27 9.47 2.23 4.49 28.22
25 53.19 40.67 9.70 2.29 4.50 28.24
26 53.45 40.87 9.91 2.41 4.55 28.28
27 53.60 41.61 10.01 2.74 4.64 28.34
28 53.70 41.92 10.19 3.00 5.02 28.39
29 53.78 42.36 10.34 3.00 5.07 28.45
30 53.86 42.67 10.56 3.02 5.07 28.57
31 53.94 42.93 10.79 3.04 5.09 28.79
32 54.13 43.23 10.91 3.04 5.11 29.08
33 54.52 43.43 11.26 3.04 5.12 29.10
34 54.60 43.57 11.70 3.04 5.13 29.11
(continued)
59
Table 10
Referential Table for Percentile Rank of Adjusted WSLT Sub-scores
Percentile Adj_Correct Adj_Position Adj_Learning Adj_Recall Adj_Cued Adj_Recog.
35 54.68 43.77 11.82 3.05 5.14 29.14
36 54.78 44.09 11.85 3.06 5.15 29.17
37 55.00 44.20 11.89 3.06 5.20 29.20
38 55.19 44.40 11.94 3.06 5.23 29.33
39 55.33 44.81 12.08 3.06 5.26 29.37
40 55.44 45.29 12.26 3.06 5.28 29.40
41 55.52 45.56 12.45 3.21 5.31 29.42
42 55.60 45.91 12.55 3.21 5.35 29.45
43 55.71 46.09 12.62 3.21 5.36 29.48
44 55.75 46.43 12.67 3.40 5.38 29.52
45 55.86 46.72 12.78 3.74 5.40 29.58
46 55.89 47.04 12.89 3.74 5.42 29.66
47 55.93 47.11 13.23 4.00 5.44 29.89
48 55.96 47.20 13.46 4.00 5.46 30.00
49 56.00 47.34 13.58 4.01 5.49 30.05
50 56.06 47.62 13.63 4.02 5.50 30.10
51 56.13 48.19 13.77 4.02 5.52 30.11
52 56.25 48.35 13.88 4.04 5.54 30.11
53 56.32 48.47 14.00 4.04 5.61 30.12
54 56.51 48.71 14.23 4.04 5.76 30.13
55 56.66 48.85 14.27 4.04 5.99 30.14
56 56.70 49.09 14.63 4.04 6.04 30.15
57 56.75 49.20 14.68 4.04 6.08 30.17
58 56.78 49.46 14.80 4.05 6.08 30.19
59 56.85 49.48 14.90 4.06 6.09 30.20
60 56.86 49.75 15.01 4.06 6.10 30.22
61 56.88 50.22 15.04 4.06 6.11 30.23
62 56.88 50.56 15.05 4.06 6.11 30.25
63 56.93 50.79 15.07 4.06 6.13 30.27
64 56.99 50.95 15.12 4.06 6.15 30.28
65 57.02 51.20 15.14 4.06 6.17 30.29
66 57.08 51.36 15.21 4.06 6.19 30.30
67 57.09 51.49 15.25 4.06 6.21 30.30
68 57.15 51.50 15.30 4.18 6.23 30.33
(continued)
60
Table 10
Referential Table for Percentile Rank of Adjusted WSLT Sub-scores
Percentile Adj_Correct Adj_Position Adj_Learning Adj_Recall Adj_Cued Adj_Recog.
69 57.29 51.58 15.44 4.21 6.24 30.33
70 57.37 51.69 15.51 4.21 6.26 30.33
71 57.49 51.71 15.61 4.21 6.27 30.36
72 57.52 51.89 15.72 4.21 6.27 30.36
73 57.63 51.90 15.75 4.28 6.27 30.38
74 57.67 52.16 15.84 4.74 6.27 30.38
75 57.73 52.34 15.97 4.74 6.28 30.39
76 57.81 52.46 16.00 5.00 6.30 30.40
77 57.87 52.50 16.12 5.01 6.30 30.40
78 57.93 52.59 16.16 5.04 6.30 30.41
79 58.00 52.70 16.20 5.04 6.30 30.42
80 58.00 52.82 16.26 5.04 6.32 30.43
81 58.09 52.89 16.39 5.04 6.33 30.43
82 58.28 53.29 16.55 5.06 6.33 30.44
83 58.43 53.50 16.62 5.06 6.35 30.45
84 58.54 53.65 16.74 5.06 6.36 30.47
85 58.62 53.89 16.76 5.06 6.40 30.47
86 58.68 54.07 16.83 5.06 6.41 30.47
87 58.73 54.28 16.93 5.06 6.43 30.48
88 58.80 54.50 16.98 5.06 6.44 30.50
89 58.84 54.56 17.14 5.06 6.44 30.51
90 58.97 54.89 17.32 5.06 6.45 30.53
91 59.12 55.59 17.69 5.21 6.48 30.55
92 59.32 56.08 17.99 5.21 6.48 30.56
93 59.60 57.01 18.25 5.21 6.52 30.60
94 59.93 57.09 18.41 5.40 6.54 30.61
95 60.27 58.26 18.97 6.00 6.57 30.64
96 60.57 58.62 19.26 6.04 6.64 30.71
97 60.84 58.77 19.71 6.04 6.88 30.83
98 61.06 59.28 19.84 6.06 7.04 31.02
99 61.65 59.66 20.09 6.06 7.41 31.18
Note. The cut-off point is the 5th percentile.
61
Table 11
Test-Retest Reliability of the WSLT Sub-scores
Test and retest performances
Participants
Variable practice
(n = 30 ) rb 1st 2nd effect
45.67 56.63 57.20
Age a Correct .64** (3.02) (2.92) Z = 1.243
(19-90)
13.77 49.33 51.90
Education Position .61** (8.81)
(6-18) (8.51) Z = 2.029*
14.00 15.83
Female (%) 13 (43%) Learning .63** Z = 2.555*
(4.86) (4.29)
Interval time 87.93 3.87 4.40
Recall .54** Z = 2.286*
(day) (57-132) (1.20) (1.38)
5.17 5.63
Cued .62** Z = 2.841**
(0.99) (0.62)
29.00 29.23
Recognition .51** Z = 0.902
(1.55) (1.43)
Note. a mean (range). b Pearson correlation. Z: Z test by Wilcoxon signed rank test.
**p < .01.
62
Table 12
Alternate-Form Reliability of the WSLT Sub-scores
Participants Coefficientsb
Variable (n = 30 ) Form I - II Form I – III Form II - III
a
41.00
Age Correct .61 .73 .51
(19-74)
14.27
Education Position .62 .61 .65
(6-18)
16
Female (%) Learning .64 .68 .62
(53%)
63
Table 13
Comparisons of the performances in the WSLT between different forms
Form Ia Form IIa Form IIIa I vs IIb I vs IIIb II vs IIIb
Correct 55.40 56.27 54.87 Z = -1.07 Z = -0.72 Z = -1.23
Position 48.70 47.50 49.83 Z = -0.70 Z = -0.36 Z = -1.01
Learning 14.00 13.60 14.17 Z = -0.89 Z = -0.28 Z = -0.55
Recall 3.77 3.83 3.30 Z = -0.06 Z = -1.52 Z = -2.24*
Cued 5.23 5.50 5.37 Z = -1.71 Z = -0.69 Z = -0.92
Recognition 29.10 28.50 28.13 Z = -0.60 Z = -1.72 Z = -1.14
Note. a mean scores. b Z test by Wilcoxon signed rank test
* p < .05
64
Table 14
Correlation Coefficients between the WSLT Sub-scores and Criterion
Test (N) Correcta Positiona Learninga Recallb Cuedb Recog.b
LM1c (33) .46** .60** .66** - - -
LM2c (33) - - - .42* .15 -
LM_recog.c (33) - - - - - .29
WL1c (33) .57** .69** .75** - - -
WL_shortc (33) .43* .58** .64** .30 .39* -
WL2c (33) - - - .49** .35* -
WL_recog.c (33) - - - - - -.01
FSIQd (66) .22 .27* .26* .30* .13 .09
Note. LM: Logical Memory subtest of the WMS-III. WL: Word List Test of the WMS-III.
WL_short: short-delayed recall score of the Word List Test. FSIQ: Full scale IQ of the WAIS-III.
a
Pearson correlation. b Spearman’s rho. c raw score. d correlation between FSIQ and age-corrected
sub-scores
*p < .05. **p < .01.
65
Table 15
Comparison between aMCI (N = 40) and Healthy Controls
Correct Position Learning Recall Cued Recog.
a
Z score -2.81** -3.46*** -3.33*** -5.21*** -4.65*** -3.65***
M (SD) of 55.20 47.80 14.28 3.67 5.55 29.18
HC (5.42) (11.31) (5.68) (1.70) (1.50) (2.02)
M (SD) of 50.40 40.44 9.73 1.38 3.71 26.25
P’t (8.41) (10.56) (5.59) (1.42) (1.67) (4.94)
Note. M: mean adjusted sub-scores. SD: standard deviation. a Mann-Whitney U test
**p < .01. ***p < .005
66
Table 16
Comparisons between Different Age Groups in the recall score by Kruskal-Wallis
One-Way Analysis of Variance by Ranks
a
Age group 6
Recall
2-4 trials group 9.55
5-7 trials group 4.59
8-10 trials group 7.81
Note. a Partial out the education effect.
67
Figures
60
Education-corrected Sub-scores
of the WSLT (Median) 50
40
correct
30
position
20 learning
recognition
10
0
16-24 25-34 35-44 45-54 55-64 65-74 75+
Age Group
Figure 1-1. Medians of the WLST sub-scores in different age groups (7 groups)
7
Education-corrected Sub-scores
6
of the WSLT (Median)
3 recall
cued
2
0
16-24 25-34 35-44 45-54 55-64 65-74 75+
Age Group
Figure 1-2. Medians of the WLST sub-scores in different age groups (7 groups)
68
60
Education-corrected Sub-scores
50
0
16-24 25-34 35-44 45-54 55-64 65+
Age Group
Figure 2-1. Medians of the WLST sub-scores in different age groups (6 groups)
7
Education-corrected Sub-scores
6
of the WSLT (Median)
3 recall
cued
2
0
16-24 25-34 35-44 45-54 55-64 65+
Age Group
Figure 2-2. Medians of the WLST sub-scores in different age groups (6 groups)
69
60
Age-corrected Sub-scores
50
0
0~6 7~9 10~12 13+
Education Group
7
Age-corrected Sub-scores
6
of the WSLT (Median)
3 recall
cued
2
0
0~6 7~9 10~12 13+
Education Group
70
100
77.5%
80
60
Percent
40
20%
20
0
WSLT LM
Memory test
Figure 4-1. Percentage of scores below 5th percentile among patients with aMCI
100
80 74.07%
60
Percent
40
20 11.11%
0
WSLT LM
Memory test
Figure 4-2. Percentage of scores below 5th percentile among patients with aMCI-single domain
71