Are we overestimating IQ reliability? Inconsistent intraindividual comparability of Screening IQ and Full-Scale IQ for three different test batteries.

Grieder, S (Silvia)¹, Odermatt, S (Salome)¹, Bünger, A (Anette)¹, Schweizer, F (Florine)¹, Grob, A (Alexander)^1
1Department of Psychology, University of Basel, Missionsstrasse 62, 4055, Basel

 

Submission type

Poster only

Scheduled

Hallway, 22-07-2016, 16:00 - 17:00

Keywords

Intelligence, screening, psychometrics, validity, reliability, individual level

Summary

Group- and individual-level comparability and validity of Screening IQ versus Full-Scale (FS)IQ were investigated for three intelligence test batteries: The Intelligence and Development Scales–2 (IDS-2; N = 1,946, age 5;00 to 20;11 years),  the Stanford-Binet Intelligence Scales–Fifth Edition, German Adaptation (SB5; N = 2,172, age 4;00 to 80+ years), and the Reynolds Intellectual Assessment Scales, German Adaptation (RIAS; N = 2,145, age 3;00 to 99;00 years). Internal consistencies (.93-.99) and intercorrelations (.80-.91) for all IQs were high. On an individual level, 95% CIs overlapped for between 57.6% (SB5) and 98.4% (RIAS) of participants. Between 0.1% (IDS-2: Profile IQ vs. FSIQ) and 13% (SB5) of participants had an IQ difference of one SD or more. Regression analyses with individual characteristics (including ADHD, language group, cooperation in the test situation) revealed that only age and FSIQ were significant predictors for some of the differences. Finally, regression analyses showed incremental validity for the FSIQ over and above the ScrIQ in predicting school grades, especially for mathematics. Determinants of a reliable and valid measurement of intelligence on an individual level are discussed. Results challenge the use of confidence intervals based on internal consistencies for individual diagnostics and have implications for test construction and practice.

 

 

Auteurs