Poster #7 - Longitudinal Relations Between Schoolwork-Related Task Values and Performance Among Three Cohorts of Children and Adolescents

• In Event: 3-001 - Poster Session 2 during Breakfast

Abstract

The expectancy-value model of achievement-related behavior posits that when one holds high task values they are more likely to engage in a goal and show better performance. Many empirical studies have found supporting evidence for the model, including longitudinal studies. Some research findings also suggest that task values and performances have reciprocal effects. The aim of the present study was to examine the relations between schoolwork-related task values and performance from a developmental perspective using three different cohorts of Korean children and adolescent samples collected during the same timeframe. Data came from the Korean Children and Youth Panel Survey (KCYPS) conducted by the National Youth Policy Institute from 2010 to 2016. Data included three cohorts, where each cohort were in their first (C1; N = 2342, 48% girls), fourth (C2; N = 2378, 47% girls), and seventh grade (C3; N = 2351, 50% girls) at the first year of study. For the present study, 5th, 6th, and 7th wave of data were used for C1 (Mage = 11.98, SDage = 0.18) and C2 and Mage = 14.95, SDage = 0.23), and 3rd, 5th, and 6th wave of data were used for C3 (Mage = 14.89, SDage = 0.34). The reason for selecting different waves was because of how the data were collected. Task values were measured with five items (e.g., “School work is very important to me.”) using a 4-point scale. Academic performance was measured through asking how well one performed in language arts, math, and English using a 5-point scale. Using three waves of data for each cohort, cross-lagged structural equation models were fitted using AMOS 21.0, and four competing models were compared (i.e., bidirectional relations model, task values predicting performance, performance predicting task values, and no relations model). Results revealed that the bidirectional model fitted best among C2 and C3, while performance predicting task values model fitted best for the youngest cohort (C1). Time, age, and cohort should be considered interpreting the results, as each cohort were in different age with slightly different intervals between measurements while data collection occurred during similar historical time. Advantages of using open source panel data is that scholars get access to large-scale samples of longitudinal data from different cohorts, which help us disentangle the effect of time, age, and cohort, one of the core interests of developmental scientists. Limitations of panel data include less control over measures and data collection time points.

Author

• Bora Lee, Korea University
  Presenting Author