Society for Research on Adolescence (SRA) Biennial Meeting
April 12-14, 2018 Minneapolis, Minnesota Presentation Title: "Pattern-Centered Approach to High School Math Motivation" (poster) Authors: Ta-yang Hsieh, Sandra Simpkins Abstract Diversifying the STEM workforce in terms of gender, ethnicity, and class is still a challenge (NSF, 2014; NSF, 2017; Watt et al, 2016). Expectancy-value theory posits that people are more likely to pursue a math-related goals if they think math is valuable and they expect to do well in math (Eccles & Wigfield, 2002). Most previous work is variable-centered, isolating each specific motivational belief while holding others constant. In reality, however, motivational beliefs hardly function in isolation of one another. Only a few recent studies tested the theoretical proposition that students need both high value and ability self-concept (Wang et al, 2016; Simpkins & Davis-Kean, 2005). The current study addresses this gap by utilizing a pattern-centered approach that identifies the most prevalent profiles of motivation across multiple beliefs. The current study asks 1) what do the motivation profiles look like 2) are there demographic differences across profiles 3) how are the profiles associated with math achievement and class effort? The sample (N=15,606; 50% female) comes from the 2012 nationally representative High School Longitudinal Study. The 11th grade participants are diverse in ethnicity (63% White, 11% Black, 17% Hispanic, 9% Asian American) and class (32% were 185% below poverty line). Math motivational beliefs included identity (2 items; α = .88), interest (3 items, .78), self-efficacy (4 items, .89), and utility (3 items, .82). Math outcomes included standardized math achievement scores and effort in math class (4 items, .74). ROPstat’s pattern-centered analysis (Ward’s method) with k-means relocation were used to identify profiles (reached acceptable error sum of squares (59%) and homogeneity coefficient (<.80)). Group differences in demographics and outcomes were tested with ANOVA and adjusted standardized residuals. Six math motivation profiles were identified: 1) overall high, 2) low utility otherwise average, 3) low identity otherwise average, 4) low interest otherwise average, 5) overall below average, and 6) overall low. Profile membership was not evenly distributed across gender, ethnicity, and income level (Table 1). Low math identity (otherwise average motivation) was most demographically disproportionate, showing overrepresentation of females, Hispanic, Black, and low-income students. Such overrepresentation is largely driven by students at the intersection of those three marginalized characteristics. Black and Hispanic students showed smaller under- or overrepresentation in profile memberships than White and Asian American. The profiles were more different in their association with effort than achievement. High/low motivation profiles were associated with highest/lowest math test score and class effort respectively, but profiles of average motivation (i.e., profiles 2-4) ranked differently depending on outcome. Low math identity was more negatively associated with achievement than effort, whereas low interest is more negatively associated with effort than achievement. Pattern-centered approaches capture the complexity of motivation by allowing multiple aspects to interact with each other, while maintaining a parsimonious presentation. Three distinct profiles of ‘average motivation’ were identified, and they in fact were associated with different math outcomes. Profile membership aligns with demographic-based stereotypes about math. This study provides a logical analytical strategy for future works. Comments are closed.
|