An Adjusted ABS List (with 'plus' and 'minus' tendencies)

32 Pages Posted: 25 May 2011

See all articles by Andreas G. F. Hoepner

Andreas G. F. Hoepner

Smurfit Graduate Business School, University College Dublin; European Commission - DG FISMA

Date Written: May 13, 2011

Abstract

This Table displays estimated 'Adjusted ABS scores', which aim to be less exposed to biases, rounding margins and randomness than the four point ABS rating. The 'Adjusted ABS scores' are estimated on the basis of those criteria, which the ABS list claims to underly its rating. The estimation process comprises three steps: First, the four point ABS ratings are regressed on seven criteria underlying the ABS assessment to capture all influences on the ABS rating beyond these criteria (e.g. biases, rounding margins or randomness) in the error term of the regression. This regression has an adjusted Rsquared of 0.806. These seven underlying criteria (their regression coefficients and respective p-values) are: natural logarithm of Cites in RAE 2008 (0.0618, 0.0000), Mean GPA in RAE 2008 (0.3222, 0.0000), Impact Factor [Yes? No?] (0.1192, 0.0370), Average Impact Factor Quartile (0.1321, 0.0000), World Elite Count (0.0109, 0.3780), Average Previous Journal Rating [ABS 2009, Aston 2008, Cranfield 2009, Durham 2006, Imperial 2004, Kent 2007, Warwick, 2003] (0.6378, 0.0000) and Previous Journal Rating [Yes? No?] (-0.8274, 0.0000). Second, the error term of the regression and all criteria with insignificant regression coefficients (i.e. only World Elite Count) are ignored from the further estimation process to avoid exposing the 'Adjusted ABS scores' to unnecessary biases, rounding margins or randomness. Third, 'Adjusted ABS scores' displayed in column 13 [the third from the right] are estimated as the sum of the products of the criteria's regression coefficients and a journal's value in the respective criterion [see columns 5-12] plus the significant intercept of the regression (intercept -0.2772, p-value: 0.0648). The estimations presented in this Table is based on an Ordinary Least Squares regression, but estimations based on Poisson or Negative Binomal (Quasi) Maximum Likelihood regression resulted in 'Adjusted ABS scores' which correlate greater than 98% with the estimations presented in this Table. The remaining columns in this Table display the following information: Columns 1-4 display a journal's International Standard Serial Number (ISSN), its field, title and ABS 2010 rating. The second last column presents a tendency to the ABS 2010 assessment based on the 'Adjusted ABS score'. If an 'Adjusted ABS score' is more than 0.25 larger or smaller than the ABS 2010 assessment, the tendency ' ' or '-' is indicated. This tendency aims provide a meaningful and easily interpretable results which is based on both the virtues of the four point ABS 2010 rating and the accuracy and reduced biases, rounding margins and randomness of the 'Adjusted ABS scores'. The last column tranfers the ABS 2010 with tendency assessment in an intuitive, potential 10 point (ABS) rating.

Suggested Citation

Hoepner, Andreas G. F., An Adjusted ABS List (with 'plus' and 'minus' tendencies) (May 13, 2011). Available at SSRN: https://ssrn.com/abstract=1840347 or http://dx.doi.org/10.2139/ssrn.1840347

Andreas G. F. Hoepner (Contact Author)

Smurfit Graduate Business School, University College Dublin ( email )

Blackrock, Co. Dublin
Ireland

European Commission - DG FISMA ( email )

2 Rue de Spa
Brussels, 1000
Belgium

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
36
Abstract Views
402
PlumX Metrics