Title: Spanish and English Neuropsychological Assessment Scales Guiding Principles and Evolution
1Spanish and English Neuropsychological Assessment
Scales - Guiding Principles and Evolution
2A Brief Detour
- Th(eta) estimation - a simple minded approach
- Information
3P(yij1)1/(1exp(-aj(i-bj))
4Basic Data Structure
5Sample size considerations
- Large samples required for stable estimation
- 150-200 for 1PL
- 400-500 for 2PL
- 600-1000 for 3PL
6Information Curves
7Back to Our Regular Program
8Purpose of SENAS Project
- Create matched English and Spanish language
neuropsychological tests for ages 60 - New scales based upon neuropsychological model of
cognitive functioning - Scales psychometrically matched
- Within English and Spanish language versions
- Between English and Spanish language versions
- Distribution of item difficulty appropriate to
elderly population
9Psychometric Considerations
- Concept of item and test bias central to project
- A test is biased if its validity for measuring a
particular attribute is different in different
groups - Bias is present if individuals from different
groups having the same ability have a different
expected score - Empirical question
- Definition of ability
- Capacity to successfully respond to test items
- Net result of all genetic and environmental
influences - Measured by scales composed of homogenous items
10Guiding PrinciplesInitial Scale Development
- Scales targeted to assess neuropsychologically
relevant cognitive domains - Domains that are affected by dementia and other
age-related causes of cognitive impairment - Verbal and non-verbal measures
- Non-timed
- New scales, not translations of existing scales
- Examiner administered
- Item generation
- 90-100 items generated per scale to comprise
initial item pool - Broad range of difficulty
11Original SENAS Scales and Abilities Measured
12Psychometric Methods
- Based upon Item Response Theory
- Modern approach to psychometric test development
- Item selection
- Identify items without language DIF
- Based upon empirical data
- Use these items as foundation for scale
construction - Add English and Spanish items to achieve target
test information curve
13(No Transcript)
14(No Transcript)
15English and Spanish Item Characteristic Curves
for Stone/Piedra Item
16English and Spanish Item Characteristic Curves
for Lamb/Cordero Item
17Model for Item Selection
18Test Information for 3MS and SENAS Object Naming
19Initial Subject Sample
- Community dwelling older persons (60)
- English or Spanish speakers
- 208 English
- 200 Spanish
- Recruited using representative sampling methods
and targeted outreach - Hispanics and non-Hispanics (mainly Caucasians)
- Used for preliminary item calibration and
selection
20Phase 2 Data Collection
- English Administration
- SALSA Hispanics
- Random Sample (All scales) - 141
- Screened (Subset of scales) - 156
- Woodland Community Sample - 153
- Spanish Administration
- Random Sample (All scales) - 264
- Screened (Subset of scales) - 287
- Used for item calibration and scale construction
21Scales by Information Values
- High - 10-12
- Object Naming
- Picture Association
- Spatial Localization
- Verbal Attention
- Visual Attention
- Verbal Comprehension
- Verbal Memory 1
- Moderate - 8-10
- Verbal Conceptual
- Pattern Recognition
- Verbal Memory 2
- Low - 4-8
- Verbal Expression
- Non-Verbal Conceptual
22Phase 3 Data Collection
- New goals
- Measures of executive function
- Fluency
- Working memory
- African American recruitment and evaluation
- Additional Hispanics and Caucasian recruitment
and validation
23Phase 3 Data Collection
- Community recruitment - health care systems and
community outreach - 200 African Americans
- 300 Hispanics
- 200 Caucasians
- Full spectrum of cognitive ability
24Clinical Diagnosis by Ethnicity
25Supplementary Data
- Demographic and linguistic variables
- Reading measures
- Clinical variables
- Diagnosis, global cognition, global function
- Volumetric MRI
- Brain matter, white matter hyperintensity,
hippocampus
26Previous Work
- Scale construction and psychometric
characteristics - Mungas et al., Neuropsychology, 2000
- Mungas et al., Psychological Assessment, 2004
27Previous Work
- Scale construction and psychometric
characteristics - Mungas et al., Neuropsychology, 2000
- Mungas et al., Psychological Assessment, 2004
- Impact of DIF (Object Naming)
- Mungas et al., Psychological Assessment, 2004
28Mean Object Naming Ability by GroupDIF Adjusted
and Unadjusted
29Mean White-Hispanic Difference in Object Naming
Ability
30Previous Work
- Scale construction and psychometric
characteristics - Mungas et al., Neuropsychology, 2000
- Mungas et al., Psychological Assessment, 2004
- Impact of DIF (Object Naming)
- Mungas et al., Psychological Assessment, 2004
- Demographic effects - Hispanics and Caucasians
- Mungas et al., Neuropsychology, 2005
31Language UtilizationEnglish and Spanish
- How well do you speak English
- 0 - Not at all
- 1 - A little
- 2 - Well
- 3 - Very well
- How well do you speak Spanish
- 0 - Not at all
- 1 - A little
- 2 - Well
- 3 - Very well
32Ethnicity Effect Sizes for Raw Scores and
Demographically Adjusted Scores
33Previous Work
- External validation
- Global cognition/function, Hispanics and
Caucasians - Mungas et al., Neuropsychology, 2000
- Mungas et al., Neuropsychology, 2005
- Clinical diagnosis, Hispanics and Caucasians
- Mungas et al., JINS, 2005
34Cognition by Clinical Diagnosis for Hispanics and
Caucasians
35New Data
- Adjustment for reading and education
- 114 African Americans
- 145 Hispanics
- 77 Caucasians
36Ethnicity Effect Sizes for Raw Scores and
Demographically Adjusted Scores
37New Data
- Adjustment for reading and education
- Age and education effects on relationship with MRI
38MRI Effect Sizes and Age and Education
Adjustment - Full Sample (n135)
39MRI Effect Sizes and Age and Education
Adjustment - Whites (n68)
40MRI Effect Sizes and Age and Education
Adjustment - Minorities (n67)
41Questions
- Measurement characteristics in more specific
ethnic/linguistic groups - Especially different English speakers
- Hispanics, African Americans, Caucasians
- DIF other than language
- Psychometric properties of executive measures
42Questions
- Deconstructing ethnic differences in
distributions of test scores - Premorbid ability or expected test score
- Broader external validation
- Clinical diagnosis (including African Americans)
- MRI