Author:
Abstract
Keywords
Abstract
The aim of this study is to investigate the effect of measurement errors on the estimation of simple linear regression and compare the performance of the classical least squares (CLS) regression method and orthogonal regression according to the standard deviation of the residuals. For this aim, analyses were performed on three data sets of different size. The first set of data was consisting 10 data for hardness and durability of substances. The second set of data was collected from 157 eighth-grade students’ non-routine problem-solving test scores and mathematics scores on Examination for Transition to Basic Secondary Education. The third set of data was collected from 956 eighth-grade students’ PISA Mathematics Literacy Test scores and Level Determination Examination scores. As a result of comparisons on three data sets, ıt was seen that orthogonal regression gave the smaller standard deviation of residuals than CLS regression method. Depending on these findings, it was determined that orthogonal regression has better performance than CLS regression method in the estimating the linear relationship between two variables. The results of this study can be guiding for making important contributions about measurement error modeling to researchers working in social sciences.