Using their combined research strengths in health and organization communication, Kim and Miller, both, as of this writing, established professors in communication studies, report on their study of feedback message strategies conducted within hospital settings using experienced, full-time, female nurse managers as participants. The study appears in Management Communication Quarterly, a journal that invites articles relevant to the interests of both management and organizational communication professionals. Miller, a former editor of this journal, currently serves on the Editorial Board of this well-respected journal. This paper reviews the study goal and findings, strengths and weaknesses of the article, and concludes with suggestions for future researchers.

In this investigation, Kim and Miller test the hypothesis that supervisors formulate a variety of feedback message strategies as a function of (a) attributions regarding poor subordinate performance and (b) their own feedback goals. Their research results indicate that (a) feedback messages contain substantial counseling and information value in addition to a correction message and that (b) the feedback goal of the supervisor is a driving force in determining the type of feedback strategy used. The attributions or causes of the poor performance do not appear to be a strong factor in the managers' selection of the feedback strategy, a finding that is different from that of other researchers including Green and Mitchell (1979). In this study, experienced supervisors carefully analyze the poor performance situation while formulating feedback messages and use their own goal orientation to drive feedback strategy selection. Further, work-oriented and altruism feedback messages are the predominant types of strategies used with subordinates by the nurse managers. Sanction strategies that are reward, punishment, and esteem-oriented are used less. Because of the very small proportion of esteem category messages generated, this category was dropped from the testing used in the study.

In addition to these findings, Kim and Miller add two important contributions to the study of feedback in supervisor-subordinate communication interactions. They provide a taxonomy for feedback message strategies consisting of fifteen categories and a procedure using free-ranging written responses that increases external validity and is an improvement over preformulated checklist methods. Patterns detected using the taxonomy to code the written responses from nurse managers in response to the scenarios posed by the researchers suggest the following:

• Nurse managers using sanction punishment-oriented feedback tend to describe poor performance specifically (more than those using nonsanction work-oriented feedback) and appear to grant a low level of autonomy to the subordinate;
• Frequent questions as to why the incident occurred and requests for suggestions as to how to improve the performance were found in nonsanction altruism-oriented feedback, granting a moderately high level of authority to the subordinate; and
• The sanction esteem-oriented and reward-oriented feedback messages consisted of positive statements with no negative evaluation statements. These statements refer to positive events from the past and positive events in the future without referencing the present poor performance.

Considered together, these findings provide further information on the determinants used by supervisors as they generate messages that will influence the performance of subordinates.

The key strengths of the study are also key weaknesses. The taxonomy is a key pioneering contribution, but because it has not been used before, there is nothing with which to compare it. The free-ranging written responses solicited from the participants are arguably an improvement over the preformulated checklist method of data collection, but no evidence is given in the article to substantiate their claim of improved external validity and, some have suggested that they somewhat abandon the free form responses when the coding criteria are applied Kim and Miller assume that the written responses will reveal the assumptions and thoughts of the nurse supervisors. The flaw they most criticize, that subjects tend to rate socially undesirable messages low in likelihood of use when messages are preformulated by the researcher, is believable, but Kim and Miller present no evidence to show that distortion is eliminated in the more free-ranging data collection approach. It is conceivable that both approaches fall short of depicting a real-life situation where both the supervisor and the subordinate are interacting in the give and take of a performance feedback scenario.

Another strength/weakness is that the authors intentionally place rigid boundaries around the study, limiting it to the comparison of only two attributions and two goals. Although a literature-based justification is given for selection of these specific variables, these variables may not be the most relevant attributions and goals to investigate and certainly are not the only ones to investigate. Another researcher may find attributions such as mood or task difficulty more appropriate for investigation. Futhermore, without a clear statement in this study that the nurse managers were given precise definitions to follow, the reader is left with an apprehensive notion that the study participants were free to form their own definitions of the attribution elements "ability" and "effort" and that these subjects may be thinking of and using other variables when making their responses based on prior training and experiences in providing performance feedback. On the other hand, if precise definitions were provided to the nurse managers at the outset of the investigation, then the bounded limits of the definitions of the attributions under study do make it possible to determine how the creation of supervisory feedback messages is altered as a function of each variable separately and then by the joint effect of the two variables. The article just does not fully explain the instructions given to participants.

A key weakness not successfully addressed in this study is the potential for gender bias. All participants in this study were female and the researchers give theoretical rather than empirical evidence to justify their belief that it is unlikely that the selection of female managers had a substantive impact on the outcome of the research. I question this finding. I find it difficult to believe that supervisor-subordinate relationships and appraisal approaches do not vary by gender. The male/female difference along with the supervisor-subordinate difference could influence the selection of feedback strategies used in poor performance situations, but this study does not test for or acknowledge that potential and the authors have not convinced me that this factor has not influenced the outcome of their research. The fact that findings in this study conflict with findings in previous studies may suggest that factors are in place that deserve further exploration and explanation.

Coding procedures are explained in this study and raise questions about the appropriateness of the system used which required placement of each message into a single feedback category when multiple strategies were used. The researchers suggest that there is room for further study into the sequencing of multiple feedback strategies and that supervisors may intentionally form a sequence of messages.

An alternative design for this study is to incorporate the Q-methodology (Brown, 1986). This method assesses in a highly structured and systematic way people's understanding of an issue from their own unique standpoint. The Q-method can integrate qualitative and quantitative information. It allows participants to construct the relative importance of elements that make up the scenario presented and allows the researcher to assess similarities and differences and uncover patterns. This method does require careful exploratory pre-testing and refinement to be effective, but it provides a way to capture how meaning is organized and patterned. It has the added features of being enjoyable for respondents to complete, quick to complete, and does not require intense concentration. The Q-methodology could be used to determine if findings are the same or different for people in the nurse manager group and could be used to extend the findings to other occupational groups. Using this technique, those with strong motivational goals, for example, may uniformly show a tendency for altruistic-oriented feedback messaging. Using the Q-method, profiles of occupational groups that use similar shared meaning could emerge and perhaps further explain the findings of Kim and Miller.

Using a scoring scale of 1 to 5 with 1 being poorly conducted research and reporting and 5 being well conducted and reported, I score this article at 3.5 on the scale. With the exception of a few tables that show numbers that differ only slightly from the numbers reported in the text, the information presented is clear and easy to follow, making it possible for the reader to readily grasp the points presented. The purpose of the research is clear, the scholarship used as a basis for the researcher's formulation of the research question is clear, and the method used is creative. The format used in reporting is appropriate and the topic is of interest to those in management as well as organizational communication. The presentation of the complete taxonomy in the article invites future testing and use as a foundation for further explorations. These distinguishing features are both strengths and weaknesses. However, the minimizing of the potential gender bias in this study is a definite flaw that future researchers can overcome by intentionally being inclusive of the male population in future studies.

This page is created and maintained by Sue Soy ssoy@ischool.utexas.edu
Last Updated 01/04/2002
© Copyright 2002 Susan K. Soy
Please feel free to copy and distribute freely for academic purposes with this notice and attribution.
All other rights reserved.