A Classification of Trivariate Regression Models

✍ Scribed by Dr. Heinz Holling

• Publisher: John Wiley and Sons
• Year: 1986
• Tongue: English
• Weight: 376 KB
• Volume: 28
• Category: Article
• ISSN: 0323-3847
• DOI: 10.1002/bimj.4710280704

✦ Synopsis

The preeent study begina with a discuseion of a topic frequently mentioned in the relevant literature, namely the divieion of multiple regreasion models with two predictom into five mutually excluding oategoriee. The theoretical basis for this cleesification is criticized and 8 system of three mutually excluding &tegoriea is euggeeted which is free of the criticized inconsistencies. Key word: Regression model Usually, multiple regression models with two predictors and a single criterion are divided into five mutually excluding categories (see, e.g. COHIEN & COHEN, 1975, BOBTZ, 1977, TZELCIOV & STERN, 1978). This scheme appears, however, to be in need of revision since the suppressor concept in the sense of CONGER (1974)) upon which this claesification is based, can be criticized on several levels, as shown in. recent studies (VELICER, 1978, HOLLINCI, 1983).

The starting point of Conger's suppressor concept is the following question : When is sufficient irrelevant variance removed in predictor variables to make the standardized semipartial regression coefficients in multiple regression models (with the aame symptom) numerically greater than the corresponding standardized simple regression coefficients or the predictor--crihrion correlations?