3. mEthodology JUST Capital partnered with NORC Results of both surveys were weighted at the University of Chicago to to U.S. Census parameters for gender, MORe On MAxdiFF design and execute the two surveys age, education, and race/ethnicity MaxDi昀昀 (Maximum Di昀昀erence Analysis) derives relative featured in this report. Both surveys to ensure representativeness. All importance scores to just business behavior, which we use as weights in our company Rankings model. The survey were conducted using NORC’s margins of error include “design questions present respondents with a set of items, from AmeriSpeak online panel, a nationally e昀昀ects” to adjust for the e昀昀ects of which they are then asked to identify the most and least representative probability-based weighting. Some questions were important. In our implementation, we presented respondents survey panel in which respondents asked of random subsets of the total with a random selection of three (of seven) Issues at a time, and repeated the question until the respondent had are recruited using traditional sample, which increases the margins answered for all combinations of the seven Issues. As probability methods, and those without of error. More information about respondents go through these choice tasks, their strength internet access complete surveys NORC’s AmeriSpeak panel can be of preference, based on the consistency of their choices, by telephone. The 昀椀rst survey was found at AmeriSpeak.NORC.org. The is developed. Sawtooth software implements a statistical model – Hierarchical Bayes – to summarize the information conducted May 1 to 16, 2018, and relative importance weights for the and calculate how much of the relative importance each Issue consisted of 3,582 respondents (375 Issues and Components were derived receives, summing to 100 percent. These relative importance by telephone and 3,207 online) with a using a MaxDi昀昀 design and Sawtooth measures become the weights we then use in our Rankings. margin of error of +/- 2.2 percentage software. Our full body of survey We repeated the MaxDi昀昀 process on the 3-9 Components within each issue to obtain the relative importance weights for points, and the second was conducted work this year also includes 14 focus the Components within each issue. June 7 to July 1, 2018, and consisted of groups conducted by NORC and a 4,083 respondents (626 by telephone YouGov survey of 1,500 respondents, The MaxDi昀昀 surveys for Issues and Components were conducted separately. The MaxDi昀昀 surveys for Issues and 3,457 online) with a margin of error which was used for internal research consisted of 3,582 respondents, and these estimates have of +/- 2.1 percentage points. purposes and is not included in a margin of error of +/- 2.2 percentage points. The MaxDi昀昀 this report. All di昀昀erences between surveys for Components were conducted separately: numbers reported in this document Environment and Communities each consisted of 1,070 respondents (MOE +/- 4.2%), Company Leadership & are statistically signi昀椀cant at the 95% Shareholders consisted of 1,071 respondents (MOE +/- 4.2%), con昀椀dence level. Customers and Products each consisted of 932 respondents (MOE +/- 4.5%), and Workers and Jobs each consisted of 1,010 respondents (MOE +/- 4.3%). 4 | JUST CAPITAL
JUSTCapital SurveyReport2018 Page 5 Page 7