BACKGROUND: The Internal-External frame of reference (IE) model suggests that as self-concept in one domain goes up (e.g., English) self-concept in other domains (e.g., mathematics) should go down (ipsative self-concept hypothesis). AIMS: To our knowledge this assumption has not been tested. Testing this effect also provides a context for illustrating different approaches to the study of growth with longitudinal data. SAMPLE: We use cohort sequential data from 2,781 of Year 7 to Year 11 Australian high school students followed across a total of 10 time waves 6 months apart. METHOD: Three different approaches to testing the ipsative self-concept hypothesis were used: Autoregressive cross-lagged models, latent growth curve models, and autoregressive latent trajectory models (ALT); using achievement as a time varying covariate. RESULTS: Cross-lagged and growth curve models provided little evidence of ipsative relationships between English and math self-concept. However, ALT models suggested that a rise above trend in one self-concept domain resulted in a decline from trend in self-concept in another domain. CONCLUSION: Implications for self-concept theory, interventions, and statistical methods for the study of growth are discussed.
BACKGROUND: The Internal-External frame of reference (IE) model suggests that as self-concept in one domain goes up (e.g., English) self-concept in other domains (e.g., mathematics) should go down (ipsative self-concept hypothesis). AIMS: To our knowledge this assumption has not been tested. Testing this effect also provides a context for illustrating different approaches to the study of growth with longitudinal data. SAMPLE: We use cohort sequential data from 2,781 of Year 7 to Year 11 Australian high school students followed across a total of 10 time waves 6 months apart. METHOD: Three different approaches to testing the ipsative self-concept hypothesis were used: Autoregressive cross-lagged models, latent growth curve models, and autoregressive latent trajectory models (ALT); using achievement as a time varying covariate. RESULTS: Cross-lagged and growth curve models provided little evidence of ipsative relationships between English and math self-concept. However, ALT models suggested that a rise above trend in one self-concept domain resulted in a decline from trend in self-concept in another domain. CONCLUSION: Implications for self-concept theory, interventions, and statistical methods for the study of growth are discussed.