1. PM systems are prone to becoming irrational such as at Southern Bank and/or political such as in South City Irrational systems are disordered Order in the system, if any, is not correspondent with administrative claims as to how the system is intended to operate Political systems are ordered Spoils system is insulated by administrative processes that institutionalize systemic bias

2. South City (city & county of ½ million population) Ranks 16th among the largest southeastern metropolitan areas (urban: approx 250,000) Approximately 4000 civil service employees rated (56% AA, 43% Caucasian) Defendant in 1970’s race discrimination and 1990’s reverse-discrimination cases County/municipal employees maintain highest average salary in southeast US in purchase-power parity

3. Federal intervention in administering civil service system for the past decade RFP for complete overhaul of systems We were invited to audit the PM System PM System Annual, top-down single-source system Employees rated on “Applicable” task statements 1-5 point graphic rating scale (“below” to “exceeds”) Scan sheet-based records of appraisal data Appraisal interview First and second-level Appeal Process Merit rating system

4. 3966 employee records Organized by ratee ID (raters identifiable by ID code) 10 or fewer task dimension ratings Entry, 1 st and 2 nd level Review dates Ratee Agreement and Signature data

5. Racial differences are significant across all ten rated task statements Central tendency in that 60%+ of ratings are at scale midpoint Skewness and Kurtosis statistics are strikingly different between races; more + among Caucasians Positively skewed and leptokurtic ratings centered at scale midpoint Negative range avoidance AA C

6. 11% of employees received one rating below the scale midpoint on any of the 10 task statements 169 of 3966 employees received 2 or more ratings below the scale midpoint (4.28%) Caucasians received 67% more ratings below the scale midpoint though comprising only 46% of the ratee pool (p<.01 and then some) Merit raise is flat 5% added to cost of living increase Ineligible for merit raise (4%) Eligible for merit raise (96%)

7. Atypical ratings distribution at the high end of the rating scale is strongly suggestive of: 1.Absence of incentive for performance exceeding expectations 2.Deliberate ratings strategy to preserve entitlement Absence of a graded merit system produces no incentive for performance that “exceeds expectations”

8. The incidence of a Caucasian employee receiving ratings below the scale midpoint is 165% higher than for African- American employees The incidence of Caucasian employees receiving ratings above the scale midpoint is <80% the rate for African- American employees

9. 95.7% of employees are merit-eligible for the 5% increase to their base salary Caucasian employees tended to be less eligible for merit pay relative to African-American employees; this difference is statistically significant although shy of the disparity forbidden by application of the 4/5ths rule.

10. Supervisory Review Latency The records have numerous examples of such illogical instances as shown by the minimum values (expressed in days) in Table 17 These data include illogical negative time estimates (n=153, 4%) The differences in mean latencies between ethnic groups for both latency measures (employee and supervisory review) are statistically significant (p<.01); Caucasian employees experienced greater latencies.

11. Latency of both reviews was significantly longer for Caucasian employees Race differences existed regardless of department or an imposed minimum of task ratings requirement Only race x position (rater/ratee) cell deviating from the null was when African-American employees were rated by African- American supervisors Alternative merit-system configurations were designed Imposed more stringent performance demands Each alternative imposing greater performance demands results in producing greater levels of disparate impact Mean rating requirements have the most significant impact in the absence of changing other system parameters

12. Race relations History of adversarial relationship with employer Merit qualifications, Review, & Federal Oversight Exaggerated Negative range avoidance Institutionalized entitlement program Feeble HRM Documentation System Ownership of the data closet What’s in your closet?