Following is a correction to Figure 5 (a–c) of Bueno and Dunning’s publication in World Politics (2017). After correcting minor coding errors in their quality of numbers analysis, the authors found their results largely unchanged. These corrections are reflected in the revised figure and in the revised analyses that contain measures of the quality of candidates– numerical identification in the updated supplementary material. The replication files have also been audited and revised.
Differences in the Quality of Candidates’ Numeric Codes for White/Nonwhite and Elected/Nonelected Politiciansa
aThe figure shows the weak relationship between race and the “good number” measure among candidates for federal deputy (a), state deputy (b), and city councilor (c). Good number is the sum of two components: the number of adjacent digits and the number of adjacent repeated integers in a candidate’s numeric code. We present differences of means using three race measures—dichotomous ibge, Afro-descent, and black or white. To accommodate race classifications with nonunique modes, we present results with both the whitest and blackest mode. The final row shows the strong relationship between good numbers and electoral victory. The dashed vertical line is drawn at the point of zero difference. Horizontal lines are bootstrapped 95 percent confidence intervals.
The authors regret these errors.
