About the ROR Indicator

The ROR Indicator begins with the development of robust measurement models of the reputation in individual industries. An industry's model reflects the specific and unique expectations stakeholders have for that industry. The measurement model provides a highly reliable index of the industry's reputation and that of its companies.

Our analytic models analyze the statistical relationship between the Reputation Index and key outcome variables which are derived either from the survey data or from other publicly available data sources. The predictions shown in the ROR Indicator are based on regression models analyzed from observed data.

As with any regression model, the predicted outcomes shown are subject to the caveat of ceteris paribus (assuming all other variables except those under immediate consideration are held constant) and are subject to multiple sources of known and unknown error, including sampling error for survey data, error as a function of self-reported outcomes and all other sources of survey error that cannot be precisely measured or estimated. Furthermore, all measures of correlation (including regression analysis) do not imply causation.

Here are more details on the models for each dimension of the Return on Reputation Indicator:

Consumer Behavior (Spending): Linear regression model: y (Dependent variable) = Annual consumer spending (self-reported dollar spent at individual retailers); x1 (Primary independent variable) = Reputation Index (of individual companies). Constant = -$4367.58; B (Reputation Index) = $133.05; Standard Error of B (Reputation Index) = $25.14; p = .000.

Consumer Behavior (Loyalty): Linear regression model: y (Dependent variable) = Loyalty (self-reported proportion of the time shopped at each store compared to other similar stores, scale from 0-100%); x1 (Primary independent variable) = Reputation Index (of individual companies). Constant = 5.688; B (Reputation Index) = 0.481; Standard Error of B (Reputation Index) = 0.020; p = .000.

Consumer Behavior (Promoters): Binary logistic regression model: y (Dependent variable) = Promoter (0='Definitely Would Not', 'Probably Would Not', or 'May or May Not'; 1 = 'Probably Would' or 'Definitely Would' recommend store to friends or acquaintances); x1 (Primary independent variable) = Reputation Index (across all companies). Constant = -2.936; B (Reputation Index) = .059; Exp B (Δx of 1) (Reputation Index) = 1.061; Standard Error of B (Reputation Index) = .002; p = .000.

Community Activism (Retail Advocates): Binary logistic regression model: y (Dependent variable) = Advocate (0 = Not involved in any of five activities in the past 12 months; 1 = Involved in any of five activities in the past 12 months); x1 (Primary independent variable) = Reputation Index (across all companies). Constant = -4.32; B (Reputation Index) = .0218; Exp B (Δx of 1) (Reputation Index) = 1.022; Standard Error of B (Reputation Index) = .002; p =.000. Odds ratio applied to current number of Retail Advocates estimated based on 18.4% reported in survey out of total Community Activist population (representing 10% of total U.S. population 18 and over from most recent Census data).

Community Activism (Retail Critics): Binary logistic regression model: y (Dependent variable) = Critic (0 = Not involved in any of four activities in the past 12 months; 1 = Involved in any of four activities in the past 12 months); x1 (Primary independent variable) = Reputation Index (across all companies). Constant = -4.41; B (Reputation Index) = .0140; Exp B (Δx of 1) (Reputation Index) = 1.014; Standard Error of B (Reputation Index) = .003; p =.000. Odds ratio applied to current number of Retail Critics estimated based on 10.8% reported in survey out of total Community Activist population (representing 10% of total U.S. population 18 and over from most recent Census data).

Policy Environment: Binary logistic regression model: y (Dependent variable) = Support for average policy proposal (0 = Not support industry position on an index of support across 16 different policy proposals; 1 = Support industry position on an index of support across 16 different policy proposals); x1 (Primary independent variable) = Reputation Index (across all companies); Control for party identification. Constant = -1.174; B (Reputation Index) = .015; Exp B (Δx of 1) (Reputation Index) = 1.015; Standard Error of B (Reputation Index) = .005; p = .001.

Litigation Environment: Binary logistic regression model: y (Dependent variable) = Defenders (0 = rating of 0-5 on 10-point scale for giving company the benefit of the doubt on litigation; 1 = 6-10 on 10-point scale for giving company the benefit of the doubt on litigation); x1 (Primary independent variable) = Reputation Index (of individual companies). Constant = -3.68; B (Reputation Index) = .051; Exp B (Δx of 1) (Reputation Index) = 1.052; Standard Error of B (Reputation Index) = .002; p = .000. Odds ratio applied to current number of Defenders estimated based on 50.6% reported in survey out of total U.S. general public 18 and over from most recent Census data.

Employee Engagement: Binary logistic regression model: y (Dependent variable) = Loyal Employees (0 = rating of 0-7 on 10-point scale of likelihood of staying with company for the next 12 months; 1 = 8-10 on 10-point scale of likelihood of staying with company for the next 12 months); x1 (Primary independent variable) = Reputation Index (of individual companies). Constant = -.895; B (Reputation Index) = .0178; Exp B (Δx of 1) (Reputation Index) = 1.018; Standard Error of B (Reputation Index) = .002; p = .000. Odds ratio applied to current number of Loyal Employees estimated based on 65.4% reported in survey out of total retail employee population from U.S. Department of Labor, Bureau of Labor Statistics data (42,815,656).

Financial Value: Adapted Market Value of Equity Model (linear regression model): y (Dependent variable) = Market capitalization of company at end of 2009; Independent variables: xassets = Total assets at end of 2009, xliab = Total liabilities at end of 2009, xearn = Abnormal earnings at end of 2009 (adjusted for cost of capital), xRI = Reputation Index (of individual companies). Constant = $2.618; B (Reputation Index) = $5.216; Standard Error of B (Reputation Index) = $2.846; p=.064.

Back to the Return Indicator