Motivation for Mathematics: The Development and Initial Validation of an Abbreviated Instrument

This study outlines the development and initial validation of an abbreviated instrument intended to measure motivation for mathematics of university students in developmental algebra courses. I look across many of the predominant theories on motivation with the aim of representing several of these theories as latent constructs in a single instrument that is short enough to be administered in a reasonable amount of time, but inclusive enough that it could incorporate subscales representing multiple distinct latent factors. This study answers a call by researchers expressing a need to investigate relationships between disparate theories on motivation and is a response to recent studies that have used several subscales from many published instruments in whole or in part as lengthy combined instruments to measure motivation across theories. The practice of utilizing many separate instruments to measure across theoretical frameworks may be unwieldy leading to validity concerns based on response processes, and the practice of taking individual items from separate instruments may potentially be incomplete leading to validity concerns based on the internal structure of the instrument and underrepresentation of the intended construct. To answer these concerns and develop a tool for future research, I conducted a three phase study. Phase one of this study asked experts in motivation to comment on and pick the best items from a pool of 122 items sourced from several popular previously published instruments that contained factors associated with self-determination, self-efficacy, achievement goals, and expectancy-value. The commentary by experts gave insight into item alignment with theory, and all items with at least 40% endorsement by experts proceeded to phase two. In phase two, cognitive interviews of students and instructors provided insight into the cognitive processes employed in responding to the 53 items endorsed in phase one. Two researchers coded these qualitative interview data with a grounded theory approach and quantified the data using intra-respondent matrices. Effect sizes of each code provided evidence of content validity of preferred items, and concerns over social dynamics, misrepresentation of factors associated with poor wording, and the use of words like “very much” that forced students to quantify their cognitive processes provided evidence against non-preferred items. During phase three I administered an instrument containing the surviving 34 items from phase two to 186 participants from twelve developmental algebra courses. Concerns over the broadness of the domain of mathematics led to the removal of self-efficacy and task-value items, and concerns over the abbreviated nature of the instrument led to the removal of items associated with extrinsic motivation. Concerns over the multilevel nature of achievement structured items led to their removal. Thus an exploratory and confirmatory factor analyses of the remaining 16 items representing intrinsic motivation, mastery orientations, performance orientations, and expectancy led to a four factor model that discriminated along theoretical lines and was a good fit for the data. A regression of achievement on the four latent factors from this model revealed expectancy to be the only significant predictor of achievement. With gender included as a moderating variable, performance and expectancy were both significant indicators of achievement for females, but expectancy was the only significant indicator for males. The latent factors from the instrument developed for this study had strong bivariate correlations to subscales from previously published instruments that represented similar constructs. Several sources provided evidence of content validity. Qualitative data provided evidence in the form of commentary from experts and cognitive interview data from students and instructors. A structural equation model provided evidence of validity based on relationships to other variables. For this model the dependent variable achievement was regressed upon the latent motivation variables with gender included as a moderating variable. Exploratory and confirmatory factor analyses provided validity evidence based on the internal structure. Validity based on consequences and response processes was controlled by using an anonymous process where participation was blind to instructors and researchers, and the administration of an abbreviated measure in a familiar paper and pencil face-to-face format reduced construct irrelevant variance. This process produced a four factor 16 item Motivation for Mathematics Abbreviated Instrument measuring intrinsic motivation, mastery orientation, performance orientation, and expectancy while accumulating validity evidence for three out of five sources of validity. The result of this inquiry was a psychometric instrument that may be used by researchers, practitioners, and grant writers who desire a tool to measure motivation for mathematics across several of the predominant theories on motivation