A logistic regression analysis of the economic factors affecting the survival of sister newspapers in the United States

In an attempt to provide the diversity of voices necessary in an informed democracy, most sister newspapers —morning and evening newspapers published in the same city by the same firm—are fiercely competitive, yet are not protected by any regulations designed to encourage or sustain this diversity. To the degree their existence is governed by the laws of economics, an understanding of the reasons for their decline in number requires an analysis of the economic factors which appear to have an effect on their survival. To this end, a qualitative response model—with survival as the dichotomous dependent variable—was analyzed using the logistic regression (Logist) procedure available in the SAS statistical package for mainframe computers.

Based on a review of the literature, factors appearing to have an effect on a newspaper's survival were operationally defined as independent variables and included in the models to be tested. Data were collected from public sources, because most financial information about newspaper operations is proprietary and, therefore, generally unavailable.

A backward elimination of independent variables included in the models to be tested by the logistic regression procedure revealed that none, separately or in combination, were statistically significant at the .05 level. None of the variable sets tested had a predictive ability greater than 10 percent. The most valuable variable examined was the percentage of service workers in an area's economy and the current ratio of morning/evening circulation, yet they were still far from being statistically significant predictors of a newspaper's survival.

The answer to the question why no variables tested were found to contribute to—much less be able to predict —the decline of sister newspapers, based on the clarity of the trend, must lie in the variables examined and their operational definitions. The existing literature tends to oversimplify both the problems and the contributing factors, economic and sociological. Better independent variables, based on more intricate combinations of data, must be developed to reveal all hidden factors if the problems of sister newspapers are ever to be fully understood.