Abstract
This study examines influences on the degree to which news organizations attend to everyday readers through online feedback, or conversely via advertising and marketing data. Factors at the community level, the market level, and the organizational level are examined as predictors of editors’ exposure and attention to audiences. Results of a national survey indicate that most editors tend to pay attention to both online reader feedback and marketing data. Factors at the organizational level, such as organizational size and public ownership, were more influential than community-level factors in explaining level of attention. Community structural pluralism was not a strong predictor of attention to feedback via marketing data or online. These findings suggest a business motive in editors’ communication and reinforce previous research pointing to the importance of organizational and peer-level factors in shaping journalists’ response to audiences.
Notes
1We do not assume that all organizations conduct formal market research. However, all of these news organizations have an interest in publishing advertising from their communities, and so it is assumed that sales staffs, large or small, gather information that helps target these efforts, even if only informally.
2Papers were originally selected for another study related to mimicry of large papers by smaller papers; this study required representation from particular circulation ranges. However, these ranges are appropriate for the present study, which requires variation in community complexity, a measure that corresponds closely to newspaper size.
3Similar measures have been used by Hindman, Ernst, and Richardson (Citation2001); Demers (Citation1998); and Donohue, Olien, and Tichenor (Citation1989).
4The content analysis was conducted February to May 2008. Each home page was coded once, as these features do not change from day to day. Three coders searched specific terms using the browser's “Find” field to locate html text and scanned home pages for text in image formats. Coders determined if the found term's context was relevant to the feature sought. An initial reliability test on site home pages not included in the sample yielded Scott's pi coefficients less than .80 for three of the five measures. Coders were retrained twice more, yielding a Scott's pi of higher than .80 for all measures.
5This coefficient is just below the generally accepted .70 level, suggesting all items may not be measuring the same latent construct. However, a factor analysis was conducted, and the results did not suggest multidimensional data, and so the summed measure was left intact.
Note. N = 181. Values are Pearson product moment correlation coefficients. DMA = designated market area.
*p < .05. **p < .01.
Note. N = 181. Standardized regression coefficients (partial correlation coefficients in parentheses).
a N = 178.
*p < .05. **p < .01.
6Organizational Size and Structural Pluralism had unacceptable skewness scores, above 1.0. Two cases had outliers on more than two variables and so were dropped from the analysis. Eight cases had outliers on only one variable—Organizational Size—and these were changed so that outlying values equaled the closest value not determined to be an outlier by the box plot. As a result, the new mean for size was 48,626.44 and the new standard deviation was 46,660.69. Durbin-Watson statistics for all regression models fell between 1.9 and 2.1. According to Ott, uncorrelated residuals are evidenced by a Durbin-Watson statistic between 1.5 and 2.5. There was no evidence of multicollinearity among independent variables, as the condition index offered by SPSS reached no higher than 6 (a common rule of thumb is that a condition index of 15 or higher suggests possible collinearity problems).