Development and Validation of the Computerised Adaptive Beat Alignment Test (CA-BAT)

Abstract

Beat perception is increasingly being recognised as a fundamental musical ability. A number of psychometric instruments have been developed to assess this ability, but these tests do not take advantage of modern psychometric techniques, and rarely receive systematic validation. The present research addresses this gap in the literature by developing and validating a new test, the Computerised Adaptive Beat Alignment Test (CA-BAT), a variant of the Beat Alignment Test (BAT) that leverages recent advances in psychometric theory, including item response theory, adaptive testing, and automatic item generation. The test is constructed and validated in four empirical studies. The results support the reliability and validity of the CA-BAT for laboratory testing, but suggest that the test is not well-suited to online testing, owing to its reliance on fine perceptual discrimination.

The mistuning perception test: A new measurement instrument

“…We estimated IRT parameters using explanatory item response modeling (de Boeck & Wilson, 2004), allowing us to use a cognitive–statistical model of task performance to improve the efficiency of parameter estimation. The item response model took the form of a generalized linear mixed-effect model with a logit link function (also known as mixed-effect logistic regression ) and modified asymptotes (Harrison & Müllensiefen, 2018). This model reproduces a four-parameter logistic IRT model in which the guessing, discrimination, and inattention parameters are constrained not to vary within the item bank (Magis & Raîche, 2012).…”

“…Beat perception ability was assessed using the computerized adaptive beat alignment test (CA-BAT) of Harrison and Müllensiefen (2018). This test uses a 2-AFC task in which each item comprises two versions of a musical excerpt, both superimposed with a metronome-like series of tones termed the beep track .…”

“…The test is intended to assess the participant’s ability to infer the beat in a musical piece. We used the full 25-item test with identical psychometric parameters and adaptive procedure to the original study (Harrison & Müllensiefen, 2018).…”

“…Finally, no previous mistuning tests have used the rigorous methodology of item response theory (IRT). IRT is the state-of-the-art approach to test construction, and has been recently used for assessing melodic discrimination ability (Harrison, Collins, & Müllensiefen, 2017; Harrison, Musil, & Müllensiefen, 2016), beat perception ability (Harrison & Müllensiefen, 2018), and ear training abilities in musically trained listeners (Wolf & Kopiez, 2018). IRT can also form the basis of adaptive testing procedures, which tailor their difficulty to the individual’s ability level through on-the-fly item selection.…”

“…We then used these materials to construct an IRT-based adaptive test. Our procedure followed the automatic item generation approach used in recent IRT-based adaptive musical tests (Harrison et al, 2017; Harrison & Müllensiefen, 2018). This approach can be split into three phases: item construction , test calibration , and test validation .…”