Abstract
BACKGROUND: Spelling assessments can provide a valuable marker of cognitive change in Alzheimer's disease and related dementias (ADRD) and play an important role in ADRD research. However, most commercial assessments are not well-suited to the needs of researchers or participants; they are expensive and often require face-to-face administration by a trained examiner. To help overcome these barriers and foster progress in ADRD research, the National Institute on Aging (NIA)-funded Mobile Toolbox (MTB) offers a library of cognitive measures that can be self-administered remotely on a participant's own smartphone, including a brand-new Spelling test. OBJECTIVE: The goal of this paper is to describe the design, piloting, calibration, and validation of the MTB Spelling test. DESIGN: We describe a pilot study, calibration study, and three validation studies, all of which use a cross-sectional design. SETTING: The pilot study, calibration study, and validation studies 2 and 3 were conducted remotely, while validation study 1 was conducted in the lab. PARTICIPANTS: Participants for all of the studies were recruited from the general population by a thirdparty market research firm and the samples were stratified by age, gender, race, ethnicity, and education to represent the U.S. POPULATION: The pilot sample included 1,950 participants and the calibration study included 1335 participants over the age of 8. Validation study 1 included 92 participants ages 20 to 84, validation study 2 included 1021 participants ages 18 to 90, and validation study 3 included 168 participants ages 28 to 87. MEASUREMENTS: Participants in each of the studies completed the MTB Spelling test. Participants in validation studies 1 and 2 completed measures from the NIH Toolbox including Oral Reading Recognition as a measure of convergent validity, and Visual Reasoning and the Rey Auditory Verbal Learning as measures of divergent validity. As an additional measure of convergent validity, participants in study 1 also completed the Spelling subtest from the Wechsler Individual Achievement Test, 4th Edition. RESULTS: The MTB Spelling test demonstrated evidence of internal consistency (r=.79 to .83) convergent validity (r=.56 to .81, p<.01), discriminant validity (r = .23 to .36, p <.01), test-retest reliability ( ICC=.63 ), and correlations with normal cognitive aging (r = -.06 to -.04, p >.01). CONCLUSION: Findings suggest the MTB Spelling test is a reliable and valid measure of English spelling abilities in general population samples, and has potential in ADRD research.