Academy of Management, Research Methods Division

Research Methods Forum, Vol. 4 (Summer 1999).

Lost Time: Reflections on Temporal Issues In Organizational Research

DONALD D. BERGH
Department of Managment and Organization
Smeal College of Business Administration
The Pennsylvania State University
ddb2@psu.edu

Time has always been an important part of organizations and organizational behavior. For example, some of the earliest investigations into and applications of organizational research were on "time and motion", as managers and researchers hunted for ways to identify and improve efficiency within organizations. In addition, issues such as "change over time" have long been of interest to managers and researchers and studies incorporating time effects are becoming increasingly popular in organizational research. For example, a study that I did with Gordon Holbein (Bergh & Holbein, 1997), revealed that 36 percent of articles published in the Strategic Management Journal from 1980 to 1993 incorporated a time dimension. We also found that attention to time had increased over the latter years of the study period (1990-1993) to where over half of all studies were longitudinal. And, as we look forward to the new millennium, we find ourselves wondering what the organizations of the 21st century will look like and what pressing issues researchers will be contemplating. Time, therefore, is a central part of our lives, our organizations and our research.

Despite this important role, my research has shown that time is not well conceptualized, operationalized or analyzed within organizational research. In fact, one can even say that our field is "behind the times" in developing an understanding of the effects of time. In an effort to make up for "lost" time, I would like to summarize my surveys on the application of time effects in organizational research and also provide some recommendations as to how we as organizational researchers can improve our use of time.

Making Time a Requisite Feature of Organizational Theorizing

First, most theories in organizational research do not contain an element or dimension of time. Although there are some notable exceptions - such as population ecologies' emphasis on selection over time, evolutionary theories, and punctuated equilibrium models - many popular theories (agency theory, the growing interest in the Resource-based View in strategic management for instance, to name but a few) do not provide any explicit predictive or explanative element that directly incorporates or considers time. As a consequence, important questions are left unasked and unaddressed.

Fortunately, however, some researchers have tried to address this issue. In most examples, researchers take existing theories and try to apply them for the purposes of explaining dynamic relationships. One interesting advance by Gersick (1991) helps us see how and where time effects can work their way into our conceptual models. She suggests that by challenging the traditional assumptions of change, we can identify new ways of thinking about time. Her model indicates that time can be addressed in the context of periods, stability and speed of change, raising new conceptual vistas that are driven by the definition of time and its role in theoretical explanations. Other notable efforts in this regard are offered by Ginsberg (1988) and Porter (1991). In addition, other researchers have attempted to offer new depictions of the role of time in organizations and theories. For example, the Strategic Management Journal recently (1996) published studies featuring different evolutionary approaches, views that present some new and promising theoretical paths.

So, taken together, what we see is that some seeds have been planted. But to date, the overall theoretical stalk has not gained enough strength to produce the conceptual flowers that time effects can offer, ones that can move the organizational literatures forward.

I believe that one fruitful way to bring time into more of a theoretical blossom is to incorporate the suggestions made by Kelly and McGrath (1988). They suggest that theorists should specify the temporal shape of the causal lag between their independent and dependent variables as well as identify the scale of time in the relationships. More specifically, Kelly and McGrath observe that time dimensions allow us to consider ordering of treatments and observations, consider duration of intervals between events, reaction times, rate and pace of effects, forms of longitudinal processes (series versus parallel development) and the idea that concepts may exhibit causal interdependence over time. What grabs me about this work is the idea that, unless we specify otherwise, we implicitly assume that longitudinal relationships are linear and constant. Although that assumption may hold for some subjects and relationships in the physical sciences, it doesn't seem to make much sense in organizational research. Quite to the contrary, one can easily believe that longitudinal relationships in organizational research are non-linear and would exhibit decay, delayed effects, cyclicalities and so on. By considering these issues, one gains a new perspective for viewing questions and theories in organizational research.

Matching Hypothesized and Operationalized Effects of Time

Second, there are several ways to conceptualize and operationalize time effects. I tried to capture these different alternatives in two earlier papers (Bergh, 1993a, 1993b). Based on a survey of articles published in two leading journals, I found that time effects have generally been conceptualized and operationalized in one of five ways: (1) lagged relationships, whereby time is used for sequentially depicting associations; (2) moderation, where relationships hold over or depend on time; (3) pooling, which occurs when variables gathered over a period are aggregated together; (4) as a dependent variable, and finally (5) as a reference point or to partition relationships.

These results indicate that time effects have been measured quite differently in organizational research. Apparently, time effects can take many different forms. The strange thing, though, is that very rarely do researchers incorporate these various approaches to viewing time in their theoretical models or discussions. That tendency is unfortunate because it limits theory development. Let me illustrate. In a review of the diversification strategy and performance literature (Bergh, 1995), I found that researchers assume that longitudinal relationships between diversification and performance are constant over time. That premise is hard to believe, as I know from personal experience that if I go out and buy something expensive, then my own financial statements show it shortly thereafter. The same logic applies to companies; if they go and buy something, then their performance will reflect that outflow of resources. However, of course, managers are not rewarded for losing money, so it is expected that eventually, the acquisitive company's performance will raise again. When I looked at the longitudinal relationships between diversification and performance, I found it to resemble mine: Initially performance falls but after a short period, it rises again. This nonlinear relationship, because it incorporates time into the mix, extends the previous understanding of the diversification strategy and performance relationship. To me, that result suggests that researchers should think carefully about the role that time plays in the theoretical relationship.

A similar observation can be extended to the relationship between conceptual and empirical models. My own surveys of the literature have revealed that many researchers have assembled longitudinal data bases for testing non-longitudinal theoretical relationships. This occurs when researchers predict relationships with no reference to time, or they implicitly assume that the theorized relationships are constant over the study period. They gather longitudinal data, usually aggregate it into a pool, and then analyze the data in a cross-sectional manner. This practice fails to take advantage of the benefits associated with longitudinal data. I once called this "killing time", as here is a situation whereby researchers have the opportunity to test for patterns and shifts in the relationships over time, findings that would most likely represent conceptual advancements in the literature, and yet they examine the data as if it were collected for one period only (Bergh, 1993a; Bergh, 1995). If organizational researchers spend the effort collecting longitudinal data, then, in my opinion, they should consider how time effects factor into their theoretical relationships.

Using Time-Sensitive Data Analyses

Third, it is important to note how researchers have analyzed their longitudinal data. For the most part, they have failed to recognize and incorporate the critical statistical assumptions into their analyses. My reviews indicate that most analyses of longitudinal data -- over 95% -- did not recognize or account for the critical assumptions associated with data analysis. This creates the likelihood that empirical results are inflated with errors and that the findings may be incorrect. I have found in replications that findings of statistical significance often become non-significant when the analytical assumptions are recognized. I'm not going to "name names", but the changes in findings have often been quite dramatic, leading to entirely different conclusions than those suggested in the studies (see Bergh, 1995, and Bergh & Holbein, 1997, for demonstrations).

The good news about this problem is that it is quite simple to fix. Researchers can use diagnostic tests to determine whether the temporal assumptions are met. These tests are reported in most major statistical programs. If the test results indicate that the assumptions are not satisfied - the most likely scenario because the subjects of our research generally have characteristics that conflict with longitudinal analytical requirements - then a number of alternatives can be used to account for the problems. Two of the more popular approaches are to (1) use an analytical technique that is not as sensitive to error-term assumptions (such as generalized least squares models, multivariate approaches) or (2) to modify the model parameters to account for the distortions created by the error inflations. Believe it or not, neither of these are hard to do, as I know from experience. My paper on repeated measures analysis (Bergh, 1995) and one with G. Holbein (1997) provides some suggestions about how to assess for the presence of analytical problems and what to do if the critical assumptions underlying longitudinal analyses are not met.

Another analytically-oriented problem concerns the lack of structural components in the analytical model. In almost all cases, researchers fail to recognize that relationships do not hold over time and that a variable is needed to account for the year or period of the measures. For an example, let's return to my paper with Holbein. We find that 96% of the 203 longitudinal studies operationalized time like this: "researchers either pooled, averaged, tested percentage changes, or performed cross-sectional analyses for each measurement period" (1997: 560). These practices, which appear to represent the modus operandi in organizational research, simply fail to recognize that relationships can and do change over time. Again, fortunately, researchers can remedy this particular problem by testing the effects of year or study period, or include a variable that accounts for year within the analytical model.

Conclusion

This final point brings us back to my main exhortion: we cannot assume that time is a constant in our conceptual and empirical models. And, as I reflect on the state of time effects in organizational research, I am hopeful with the recent calls by the Academy of Management for continued conceptual and empirical research on and about time. I am optimistic that efforts to include time effects will yield fascinating insights into organizational phenomena and add new dimensions to our theoretical explanations. It is my final hope that my ramblings will promp others to reconsider how they approach time and its effects in their own research.

References

Bergh, D.D. 1993a. Watch the time carefully: The use and misuse of time effects in management research. Journal of Management, 19: 683-705.

Bergh, D.D. 1993b. Don't 'waste' your time! The effects of time series errors in management research: The case of ownership concentration and research and development spending. Journal of Management, 19: 897-914.

Bergh, D.D. 1995. Problems with repeated measures analysis: Demonstration with a study of the diversification and performance relationship. Academy of Management Journal, 38: 1692-1708.

Bergh, D.D. & Holbein, G.F. 1997. Assessment and redirection of longitudinal analysis: Demonstration with a study of the diversification and divestiture relationship. Strategic Management Journal, 18: 557-571.

Ginsberg, A. 1988. Measuring and modeling changes in strategy: Theoretical and empirical directions. Strategic Management Journal, 9: 559-575.

Kelly, J.R., & McGrath, J.E. 1988. On time and method. SAGE: Newbury Park, CA.

Porter, M.E. 1991. Towards a dynamic theory of strategy. Strategic Management Journal, Winter Special Issue, 12: 75-94.

[back to top]

• Copyright © [Research Methods Division, Academy of Management]. All rights reserved.
• This page maintained by RM Web Associate, David A. Harrison, Department of Management, University of Texas, Arlington.
• Last updated on 11/14/1999.