Spanish and English Neuropsychological Assessment Scales Guiding Principles

Spanish and English Neuropsychological Assessment Scales - Guiding Principles and Evolution Friday Harbor Psychometrics Workshop 2005

A Brief Detour • Th(eta) estimation - a simple minded approach • Information Friday Harbor Psychometrics Workshop 2005

P(yij=1|£)=1/(1+exp(-aj(£i-bj)) Friday Harbor Psychometrics Workshop 2005

Basic Data Structure Subject Item 1 Item 2 Item 3 Item 4 S 1 X 12 X 13 X 14 S 2 X 21 X 22 X 23 X 24 S 3 X 31 X 32 X 33 X 34 S 4 X 41 X 42 X 43 X 44 Friday Harbor Psychometrics Workshop 2005

Sample size considerations • Large samples required for stable estimation § 150 -200 for 1 PL § 400 -500 for 2 PL § 600 -1000 for 3 PL Friday Harbor Psychometrics Workshop 2005

Information Curves Friday Harbor Psychometrics Workshop 2005

Back to Our Regular Program Friday Harbor Psychometrics Workshop 2005

Purpose 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 Friday Harbor Psychometrics Workshop 2005

Psychometric 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 Friday Harbor Psychometrics Workshop 2005

Guiding Principles Initial Scale Development • Scales targeted to assess neuropsychologically relevant cognitive domains § Domains that are affected by dementia and other agerelated 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 Friday Harbor Psychometrics Workshop 2005

Original SENAS Scales and Abilities Measured Friday Harbor Psychometrics Workshop 2005

Psychometric 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 Friday Harbor Psychometrics Workshop 2005

Friday Harbor Psychometrics Workshop 2005

Friday Harbor Psychometrics Workshop 2005

English and Spanish Item Characteristic Curves for “Stone/Piedra” Item Friday Harbor Psychometrics Workshop 2005

English and Spanish Item Characteristic Curves for “Lamb/Cordero” Item Friday Harbor Psychometrics Workshop 2005

Model for Item Selection Friday Harbor Psychometrics Workshop 2005

Test Information for 3 MS and SENAS Object Naming Friday Harbor Psychometrics Workshop 2005

Initial 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 Friday Harbor Psychometrics Workshop 2005

Phase 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 Friday Harbor Psychometrics Workshop 2005

Scales by Information Values • High - 10 -12 § Object Naming § Picture Association § Spatial Localization § Verbal Attention § Visual Attention § Verbal Comprehension § Verbal Memory 1 Friday Harbor Psychometrics Workshop 2005 • Moderate - 8 -10 § Verbal Conceptual § Pattern Recognition § Verbal Memory 2 • Low - 4 -8 § Verbal Expression § Non-Verbal Conceptual

Phase 3 Data Collection • New goals § Measures of executive function • Fluency • Working memory § African American recruitment and evaluation § Additional Hispanics and Caucasian recruitment and validation Friday Harbor Psychometrics Workshop 2005

Phase 3 Data Collection • Community recruitment - health care systems and community outreach § 200 African Americans § 300 Hispanics § 200 Caucasians • Full spectrum of cognitive ability Friday Harbor Psychometrics Workshop 2005

Clinical Diagnosis by Ethnicity Friday Harbor Psychometrics Workshop 2005

Supplementary Data • Demographic and linguistic variables • Reading measures • Clinical variables § Diagnosis, global cognition, global function • Volumetric MRI § Brain matter, white matter hyperintensity, hippocampus Friday Harbor Psychometrics Workshop 2005

Previous Work • Scale construction and psychometric characteristics § Mungas et al. , Neuropsychology, 2000 § Mungas et al. , Psychological Assessment, 2004 Friday Harbor Psychometrics Workshop 2005

Previous 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 Friday Harbor Psychometrics Workshop 2005

Mean Object Naming Ability by Group DIF Adjusted and Unadjusted Friday Harbor Psychometrics Workshop 2005

Mean White-Hispanic Difference in Object Naming Ability Friday Harbor Psychometrics Workshop 2005

Previous 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 Friday Harbor Psychometrics Workshop 2005

Language Utilization English 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 Friday Harbor Psychometrics Workshop 2005

Ethnicity Effect Sizes for Raw Scores and Demographically Adjusted Scores Friday Harbor Psychometrics Workshop 2005

Previous 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 Friday Harbor Psychometrics Workshop 2005

Cognition by Clinical Diagnosis for Hispanics and Caucasians Friday Harbor Psychometrics Workshop 2005

New Data • Adjustment for reading and education § 114 African Americans § 145 Hispanics § 77 Caucasians Friday Harbor Psychometrics Workshop 2005

Ethnicity Effect Sizes for Raw Scores and Demographically Adjusted Scores Friday Harbor Psychometrics Workshop 2005

New Data • Adjustment for reading and education • Age and education effects on relationship with MRI Friday Harbor Psychometrics Workshop 2005

MRI Effect Sizes and Age and Education Adjustment - Full Sample (n=135) Friday Harbor Psychometrics Workshop 2005

MRI Effect Sizes and Age and Education Adjustment - Whites (n=68) Friday Harbor Psychometrics Workshop 2005

MRI Effect Sizes and Age and Education Adjustment - Minorities (n=67) Friday Harbor Psychometrics Workshop 2005

Questions • 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 Friday Harbor Psychometrics Workshop 2005

Questions • Deconstructing ethnic differences in distributions of test scores § Premorbid ability or expected test score • Broader external validation § Clinical diagnosis (including African Americans) § MRI Friday Harbor Psychometrics Workshop 2005