Dissin' the Deadwood or Coddling the Incompetents? Patterns and Issues in Employee Discipline and Dismissal in the States

Abstract

Relying on several extensive databases, this study finds state employee dismissal rates to be low with many managers also complaining about discipline and dismissal processes as a management impediment. Although interstate variation was evident, extent of civil service coverage, the presence of collective bargaining, or past efforts to simplify dismissal processes had limited explanatory power, however. The implications of the findings for civil service reform and the adoption of at-will employment arrangements are discussed.

Keywords:

• state employee dismissal
• impact of civil service and collective bargaining on dismissal
• desirability of at-will employment

Notes

5. Stahl, 1983.

10. Assuring comparability in the status of state employees was a major methodological challenge. Data were sought on turnover for full-time, permanent, classified employees. For most states these are the data used in the analysis. A few states could not report turnover solely for such employees. In Missouri, Ohio, and Virginia, turnover was reported for classified and unclassified employees combined. Readers interested in more on the study's methods should contact the senior author of this article, or see Elling, R.C., Ibid., and Elling, R.C. Dissin' the Deadwood? Patterns in the Dismissal of Civil Servants in American State Bureaucracies. The Annual Meeting of the Southwestern Political Science Association, San Antonio, TX, April 1–3, 1999.

12. Ibid, p.171.

15. In the case of both surveys, questionnaires were sent to all but the smallest or least significant state administrative agencies. In larger agencies a number of division and bureau chiefs were contacted in addition to the agency head. If an agency had field operations, the heads of some of its field offices were also contacted. A total of 847 managers responded to the 1982 survey, for a response rate of 40%. Individual state responses varied from a low of 35% for Vermont to a high of 50% for Arizona. A total of 683 managers responded to the 2000 Comparative State Management Survey, for a response rate of 37%. Individual state response rates varied from a low of 25% In New York to 48% in Vermont. Studies that depend on the willingness of state managers to complete surveys have experienced declining response rates in recent decades (see Bowling, C., Chung-lae, C., Wright D.S. Exploring and Explaining the Preferences of State Agency Heads for Governmental Expansion—A Typology of Administrator Growth Postures, 1964–1998: Understanding Minimizing as Well as Maximizing Bureaucrats. Presented at the Annual Meeting of the Midwest Political Science Association, Chicago: April 27–30, 2000. Chung-lae, C., Wright, D.S. The Devolution Revolution in the 1990s: Changing Patterns and Perceptions of State Administrators. Paper presented at the annual meeting of the American Political Science Association, San Francisco: August 29–September 2, 2001). Given this fact, response rates for the 1982 and 2000 Comparative State Management surveys are respectable. For more on the characteristics of the samples in these two studies see Elling. R.C., Ibid.; and Elling, R.C., Thompson, L.T., Monet, V. Ibid.

16. The response options were “not a problem,” a “minor problem,” a “serious problem,” or a “very serious” problem.

17. In both 1982 and 2000 the state means in the severity of employee discipline/dismissal as a problem differed significantly (One-Way ANOVA, F-test, significant at .01 level). Yet the absolute differences in the proportion of managers who considered this to be a problem were often quite small. So what is going on here? Significant mean state differences can result if only one or two states have scores that differ dramatically from those of the remaining states. Moreover, the F-statistic does not tell us where the differences lie. In each of the surveys one or two states were often outliers with respect to the severity of the various management problems. This is true with respect to interstate means in the severity of discipline/dismissal. A multiple comparison of means test (Scheffe Test) is one way to get at this. In the case of both the 1982 and 2000 data, no significant differences (.05 level) in the mean severity of employee discipline/dismissal existed between any given pair of states.

18. A rule of thumb is that a coefficient of variation greater than 0.30 indicates considerable variability. In the case of civil service coverage across the responding units the C.V. was greater than 0.40 in four states: Arizona, Indiana, South Dakota, and Texas. Note that the average extent of civil service coverage for a particular state used here is for the sample of responding units. This proportion may differ from that which exists for all administrative units in a given state. The important point is to see whether—for the units in question in a particular state—whether variation in extent of civil service coverage matters with respect to dismissals.

19. These states were Arizona, California, Idaho, Illinois, Iowa, New Hampshire, Nevada, Ohio, Oklahoma, and Wisconsin.

20. One reviewer of our manuscript objected to the use of dismissal rates for probationary employees as an indicator of what might happen if the just-cause and due process aspects of traditional civil service dismissal processes were eliminated. This reviewer argued that, “probation is part of the selection process and is not a disciplinary action. Passing probation is like passing an examination.” This may be so, but the fact remains that—procedurally at least—probationary employees can, like at-will employees, be fired more readily. The reviewer admits this when she/he observes that probationary dismissal ‘has none of the substance of dismissal once one has “permanent status” and none of the standards of just cause or due process apply.” This is our point exactly. Yet why is a probationary employee dismissed? Is it not because of doubts that the employee will perform at adequate levels if granted permanent status? Hence, while we would grant that a dismissal of a probationary employee is not identical to the dismissal of an employee in the classified service, statistics on dismissal rates for the former do provide insight into the likely consequences of eliminating the “just cause” standard for dismissal.

21. For a more extended discussion of these data see Elling, Ibid. . . . A major shortcoming of the data was that while these ten states reported the number of dismissals of probationary and non-probationary employees separately, none provided information on the total number of employees in each category in a given year. Hence we lack the denominator for the dismissal rate calculation. This is a very important point. While discussions with state personnel officials suggested that a proportion of five to ten percent probationary employees was about right, these proportions likely varied from state to state in the same year, or across individual states in different years. For example, a state with high turnover would—assuming it filled those vacancies—have a higher proportion of probationary employees than would one with lower turnover, all other things equal. Hence, these findings should be interpreted with caution.

23. Elling, 1992.

24. Florida, Texas, and Michigan were at the procedurally more rigid end of the scale, with scores of 2, 3, and 4, respectively. North Carolina, Iowa, and Virginia were at the more flexible end of the spectrum with scores of 15, 16, and 18, respectively.

25. The Pearson's r between procedural flexibility and dismissal rates for individual years is .067 in 1993, .115 in 1994, and -.220 in 1995. Only the last correlation has a probability value that approaches conventional standards of statistical significance (p = .183). But the sign of the relationship is not in the hypothesized direction.

26. Steiber, Ibid.

27. Barrett & Greene, Ibid.

28. For the twenty states for which dismissal data are available, the average percentage of state employees for the 1994-96 period that were unionized ranged from 8% in Virginia and 12% in Texas, to 61% in Minnesota and 73% in Florida.

29. Collective bargaining was significantly and negatively correlated with the flexibility of a state's termination processes (Pearson's r = -.47, p = .04). So, while greater unionization did not reduce dismissal rates, it was associated with a state having more complex processes for dismissal.

30. Managers were not asked about this potential problem in the 1982 version of the survey.

36. Elling, Ibid.

38. Elling, Ibid.

43. Ban, Ibid.

45. Stahl, Ibid.

47. Ban, Ibid., p. 24.

48. Ibid.

49. Ibid., p. 13.

50. Ibid., p. 271.

52. In 2000 Florida ranked 23^rd in average monthly earnings of full-time employees while Georgia ranked 36^th [U.S. Bureau of the Census, State Government and Employment Payroll Data. 2002. Available at www. census.gov/govs/www/apesst.html]. (accessed January 26, 2006). In the 20-state turnover study, Florida had the highest turnover rate—averaging almost 15% across the 1993–95 period. Its “quit rate” was in the 11% to 13% range for these three years. Only Missouri had a higher quit rate.