I do not find Mintzberg's roles useful in evaluating what CEOs do. Roles are difficult to isolate and measure. In any one interaction, the CEO may be performing several roles simultaneously. How does the observer measure in a valid and reliable way the roles the CEO is performing when, for example, the CEO is watching a subordinate "manage"? Roles may be important for understanding what managers do, but they have not yet been transformed into effective units of measure.
I find Carroll and Gillen's functional approach to be more susceptible to valid and reliable measurement. Their managerial functions include staffing, planning, investigating, coordinating, evaluating, and supervising (Carroll and Gillen 1987). This approach is useful in categorizing managerial work, but Carroll and Gillen do not differentiate what is or is not such work; nor do they focus upon different episodes of work or projects, such as governance or community relations, which managers participate in regardless of function. I believe that episodes of work can be important in that respect-what episodes does the manager choose to manage? What results does he or she achieve? At what cost? Related to what expectations?
I find Mintzberg's time analysis of managerial work useful, not in terms of assessing managerial effectiveness, but in assessing the skills and experience required for specific managerial positions. In connection with the skills and experience required of managers, my research validates Mintzberg's propositions, as abbreviated and shortened by Kurke and Aldrich. (Of Mintzberg's sample of CEOs, one of five was a hospital administrator; of Kurke and Aldrich's sample, one of four was.) The propositions (Kurke and Aldrich 1983) are as follows:
- Managers perform a great quantity of work with little time for breaks, and they must work beyond customary working hours.
- Managers' jobs are characterized by brevity, variety, and fragmentation.
- Managers clearly favor verbal over written contacts (that is, telephone calls and scheduled or nonscheduled meetings over desk work and tours).
- Scheduled meetings consume more of managers' time than any other activity.
- Managers spend little time in open-ended touring.
- Managers are boundary spanners, linking their organizations with the outside in a variety of ways. External contacts consume about half of the manager's verbal contact time.
- Subordinates generally consume about one-third to one-half of managers' total contact time.
- Managers spend relatively little time with superiors.
Consider Ted Graver's Monday. This does not include any time Graver may have spent on work before or after coming to the office, nor did I ask Graver what he was thinking about during work. I did not ask what telephone calls he made or received or what he read during the hours before or after work. I count Graver as participating in 24 different activities and episodes of work during the Monday observed. These ranged from a brief telephone call informing him that the vice-president for human resources had received a threat from a disgruntled worker, to two scheduled meetings of over seven hours regarding a possible merger.
Most of the brief activities lasted five minutes or less. Most of Grover's time, however, was spent on the scheduled meetings, not on desk work or tours (I kept no record of desk work in my log). Of the 24 brief contacts, 20 were with subordinates, four with outsiders, and none with board members. Most contacts were with subordinates, but most time was spent on external matters.
Of Grover's 24 brief interactions, all but 4 involved principally the sharing of information with internal managers. Regarding three of the other contacts, Grover requested information or was asked to supply information; in the last contact, he requested that donation proposals be channeled through the formal organizational structure.
These data are typical of all the days and all the CEOs I observed. To me, they support the model of the CEO in the large health care organization as "communicator, persuader, and shaper of organizational values and decisions" (Peters 1979, 170) rather than as problem solver and decision maker. The task of the CEO, according to Peters, is not "to impose an abstract order on an inherently disorderly process, but to become adept at the sorts of intervention by which he can nudge the organization in the desired direction and to some degree control its course" (page 172). This view runs counter to my own bias toward formulating objectives and measuring CEO performance on the basis of contribution to those objectives. The four organizations studied had never been subjected to any serious price competition that resulted in considerable numbers of patients' leaving one health care organization for another. I believe that, as such competition increases, more and more large health care organizations will specify objectives in order to retain and obtain market share.
Recommendations to Managers
Much of the general management literature focuses on functions of management, such as planning, coordinating, and controlling (Gulick 1937), or on the roles managers assume, such as negotiator, figurehead, or entrepreneur (Mintzberg 1973). As a result of following the four CEOs and parallel reading about episodes of illness as units of analysis in medical care (Strauss et al. 1985, 8-39; Hornbrook et al. 1985), I have become convinced that more attention should be paid, in evaluating CEOs and other managers, to analyzing their actual work rather than their roles and functions. This is particularly true insofar as managerial work relates to organizational objectives. Managerial activities can be categorized by subject. Episodes of work are clusters of activities that happen on the same subjects that are performed or participated by the manager. Although this study was not designed to analyze activities or episodes of CEO work, I found, as the data were analyzed, that they emerged as a focus worthy of managerial consideration.
Episodes of work can be analyzed in terms of subject matter (who is involved in what kinds of interactions); whether the action requires CEO response, and if so, what kind of response; the CEO's choice of which episodes to manage and which to delegate or ignore; how much time the CEO chooses to devote to managing or participating in different episodes; and results per managed episode versus costs in CEO time and other organizational resources.
Episodes of work can be used to analyze how the CEO is spending his or her time. This type of analysis can sharpen the CEO's focus on his or her behavior in relation to organizational objectives. It can also result in changes in or specification of organizational objectives when the CEO's time is being used effectively but not on current objectives.
Griffith points out that many of the activities a CEO performs are the normal "noise" of a working week. He suggests that an activity has a duration of two to four hours and that an episode has a duration of 1 to 24 months. "Things less than 20 minutes are noise, overhead, or something" (Griffith 1987b).
