A Conceptual Overview of the Regression Discrepancy Model for Evaluating Severe Discrepancy Between IQ and Achievement Scores

1990 ◽  
Vol 23 (7) ◽  
pp. 406-412 ◽  
Author(s):  
Larry D. Evans
1992 ◽  
Vol 15 (3) ◽  
pp. 167-174 ◽  
Author(s):  
Larry D. Evans

Increased Type I error resulting from multiple IQ-achievement comparisons may inflate learning disability identification rates for discrepancy models. The magnitude of such inflation was investigated with 87 referred students given the WISC-R and Woodcock-Johnson (R) Tests of Achievement. Five IQ-achievement comparisons were made for each student. Correction for multiple comparisons significantly decreased the number of students and achievement areas found to demonstrate significant IQ-achievement differences. However, less dramatic decreases were found for students and achievement areas determined to show discrepancies. It was concluded that the magnitude of inflation is a function of number of comparisons, degree of correction for multiple comparisons, and discrepancy model used.


1996 ◽  
Vol 19 (4) ◽  
pp. 242-249
Author(s):  
Larry D. Evans

Regression discrepancy-model equations fail to account for the common practice of obtaining more than one achievement score for a discrepancy area (e.g., reading comprehension). This article presents the equations required to calculate composite achievement scores based upon two achievement scores. Equations to calculate the composite's reliability and correlation with IQ are also given. Calculating composite scores offers a method to combine achievement scores from the same discrepancy area to produce a single standard score that is often more comprehensive and reliable than either individual score, and more suitable for regression equations. Composite scores may not always be necessary in discrepancy assessment, but can help overcome problems with interpretation of regression results.


2008 ◽  
Vol 16 (1) ◽  
pp. 1-22 ◽  
Author(s):  
Zaiton Ali ◽  
Stanley McGreal ◽  
Alastair Adair ◽  
James Webb

SAGE Open ◽  
2021 ◽  
Vol 11 (2) ◽  
pp. 215824402110041
Author(s):  
Mohammad Salehi ◽  
Samaneh Gholampour

Cheating is an academically dishonest behavior about which there has been a thrust of research. However, it has not been extensively researched in an Iranian context. Therefore, the current study was conducted with 310 Iranian students. A cheating questionnaire was devised and administered to the participants. Certain demographic variables were investigated. Both descriptive and inferential statistics were employed to analyze the obtained data. The results of the descriptive statistics revealed that cheating was common among participants, and most students did not harbor any negative attitude toward cheating or at least were neutral about it. The most common method of cheating was “letting others look at their papers while taking exams.” The most common reason for cheating was “not being ready for the exam.” As for inferential statistics, one-way analysis of variance, an independent t-test, and correlational analyses were used to test the effect and relationship of demographic variables on and between the cheating behaviors of the participants. It was found that none of the two demographic variables of gender and year level had any effect on students’ cheating behaviors. Furthermore, achievement scores and age were not significantly correlated with cheating behavior scores.


Sign in / Sign up

Export Citation Format

Share Document