DIF in the Spanish Version of the Verbal Selective Reminding Test Using Samples From Hispanics in the United States, Mexicans, and Spaniards.

Manuel Morales-Ortiz,Fabiola Peña-Cardenas

doi:10.3389/fpsyg.2019.02687

Abstract

AimBefore a test can be used in the target population, it is necessary to demonstrate that there is measurement equivalence. One way to do this is by studying differential item functioning (DIF).ObjectiveIn this study, we used the Mantel–Haenszel procedure and logistic regression to analyze DIF in the Spanish version of the Verbal Selective Reminding Test between Spaniards, Mexicans, and Hispanics in the United States.MethodThree balanced samples, matched by age, education, and sex, were studied: Spaniards, Mexicans, and Hispanics [616 healthy participants from Spain (n = 211), Mexico (n = 201), and the United States (n = 205)]. A six-trial version of the Spanish version of the Verbal Selective Reminding Test was administered and scored according to standard procedures.ResultsStatistical analysis showed some DIF between the samples of Spaniards, Mexicans, and Hispanics. A bootstrap validation of results confirmed that the item pollo showed DIF: the Mexicans and Hispanics outperformed the Spaniards, holding a total score constant. The item oído also showed DIF and was remembered in greater measure by the Spaniards.ConclusionThe Spanish version of the Verbal Selective Reminding Test can be used with Hispanic populations.

Highlights

In recent years the Hispanic/Latino population has been increasing, both in absolute terms and as a percentage of the total population of the United States
Using ETS criteria based on effect size, two items appeared with a high differential item functioning (DIF) score: dado and pollo
Gómez-Benito et al (2009) have pointed out that the combination of the criteria of significance and effect size reduces the rate of false positives in DIF studies

Summary

Introduction

In recent years the Hispanic/Latino population has been increasing, both in absolute terms and as a percentage of the total population of the United States. The most recent census data indicate that the figure is already over 59 million, which is more than 18% of the United States population (United States Census Bureau, 2017). This percentage is even higher if we include those who are unregistered, a population that has been estimated at 11.2 million (Puente et al, 2015). This strategy is not considered suitable, even when it includes back translation (Salinas et al, 2016), since several studies have shown the effect of culture and ethnicity on the results of neuropsychological tests, with participants of non-English-speaking origin, and those of Spanish-speaking origin in particular, performing worst (Acevedo et al, 2000; Boone et al, 2007; Razani et al, 2007; Saez et al, 2014; Milman et al, 2018; Rosselli et al, 2019)

Methods

Results

Conclusion