Abstract
In this study, when the assumptions of linearity and homogeneity of regression slopes of conventional ANCOVA are not met, a new approach named as SEYHAN has been suggested to use conventional ANCOVA instead of robust or nonlinear ANCOVA. The proposed SEYHAN’s approach involves transformation of continuous covariate into categorical structure when the relationship between covariate and dependent variable is nonlinear and the regression slopes are not homogenous. A simulated data set was used to explain SEYHAN’s approach. In this approach, we performed conventional ANCOVA in each subgroup which is constituted according to knot values and analysis of variance with two-factor model after MARS method was used for categorization of covariate. The first model is a simpler model than the second model that includes interaction term. Since the model with interaction effect has more subjects, the power of test also increases and the existing significant difference is revealed better. We can say that linearity and homogeneity of regression slopes are not problem for data analysis by conventional linear ANCOVA model by helping this approach. It can be used fast and efficiently for the presence of one or more covariates.
Similar content being viewed by others
References
Huitema BE (2011) The analysis of covariance and alternatives: statistical methods for experiments, quasi experiments, and single-case studies, 2nd edn. Wiley, New York
Keppel G, Wickens TD (2004) Design and analysis: a researcher’s handbook, 4th edn. Prentice Hall, New Jersey
Harwell M (2003) Summarizing Monte Carlo results in methodological research: the single-factor, fixed-effects ANCOVA case. J Educ Behav Stat 28:45–70
Lazar AA, Zerbe GO (2011) Solutions for determining the significance region using the Johnson-Neyman type procedure in generalized linear (Mixed) models. J Educ Behav Stat 36(6):699–719
Miller JW, Stromeyer WR, Schwieterman MA (2013) Extensions of the Johnson-Neyman technique to linear models with curvilinear effects: derivations and analytical tools. Multivar Behav Res 48(2):267–300
O’Connor BP (1998) SIMPLE: all-in-one programs for exploring interactions in moderated multiple regression. Educ Psychol Measur 58:836–840
Wilcox RR (1997) ANCOVA based on comparing a robust measure of location at empirically determined design points. Br J Math Stat Psychol 50:93–103
Wilcox RR (2016) ANCOVA: a global test based on a robust measure of location or quantiles when there is curvature. J Modern Appl Stat Methods 15(1):12–31
Mayer A, Nagengast B, Fletcher J, Steyer R (2014) Analyzing average and conditional effects with multigroup multilevel structural equation models. Front Psychol 5:304
D’Alonzo KT (2004) The Johnson-Neyman procedure as an alternative to ANCOVA. West J Nurs Res 26(7):804–812
Rast P, Rush J, Piccinin A, Hofer SM (2014) The identification of regions of significance in the effect of multimorbidity on depressive symptoms using longitudinal data: an application of the Johnson-Neyman technique. Gerontology 60:274–281
Moses T, Klockars A (2005) Testing for aptitude-treatment interactions in analysis of covariance and randomized block designs under assumption violations. J Modern Appl Stat Methods 4(2):460–472
Wu YB (1984) The effects of heterogeneous regression slopes on the robustness of two test statistics in the analysis of covariance. Educ Psychol Measur 44:647–663
Friedman JH (1991) Multivariate adaptive regression splines. Ann Stat 19(1):1–67
Barron AR, Xiangyu X (1991) Discussion on Friedman’s multivariate adaptive regression. Ann Stat 19(1):67–82
Harwell MR, Serlin RC (1988) An empirical study of a proposed test of nonparametric analysis of covariance. Psychol Bull 104(2):268–281
Soyoung K (2010) Alternatives to analysis of covariance for heterogeneous regression slopes in educational research. Korean J Teacher Educ 26(1):73–91
Rutherford A (1992) Alternatives to traditional analysis of covariance. Br J Math Stat Psychol 45:197–223
Kim S (2007) Type I error rate and statistical power of ANCOVA, Potthoff’s modified Johnson–Neyman method, and Wilcox Trimmed mean method in randomized two-group experiment design. Unpublished doctoral dissertation. University of Georgia, Georgia
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Ankarali, H., Cangur, S. & Ankarali, S. A New SEYHAN’s Approach in Case of Heterogeneity of Regression Slopes in ANCOVA. Interdiscip Sci Comput Life Sci 10, 282–290 (2018). https://doi.org/10.1007/s12539-016-0189-0
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12539-016-0189-0