Cognitive and Regulatory Mediation of

Relationships Between Affective States and

Momentary Work Performance

Howard M. Weiss

Purdue University

Neal M. Ashkanasy

University of Queensland

and

Daniel J. Beal

Purdue University

Draft – Not for quotation

Send correspondence to:

Howard M. Weiss

Dept. of Psychological Sciences

Purdue University

West Lafayette, IN 47907

Attentional and Regulatory Mediation of Relationships Between Affective States and Momentary Work Performance

Abstract

Affective Events Theory (Weiss & Cropanzano, 1996) drew a distinction between the immediate performance consequences of transient affective states and the behavioral consequences of more stable attitudes partly and distally influenced by affective experiences. They showed that the organizational literature at emphasized the later at the expense of the former. However, they only briefly discussed the processes by which affective states influence momentary or episodic work performance. In this paper we use the basic literature on the cognitive and behavioral consequences of affective states and the literature on attention, regulation and regulatory resources to describe a general framework for studying affect and work performance.

Attentional and Regulatory Mediation of Relationships Between Affective States and Momentary Work Performance

The study of affect in work settings, and particularly the study of the performance implications of affective states, has a long but particularly disappointing history. We say disappointing for a number of reasons. First, even the smallest amount of reflection will convince anyone that organizations are settings of emotional intensity. If emotions are generated by appraisals of the reaching or impeding of important personal goals or values, then where is this more likely to play out than in work settings? Each day at work, our needs, desires, and identities are challenged and affirmed. So, one would think that the study of emotions at work would be a core topic of organizational research. It is not. Second, performance has been the criterion of interest for organizational researchers. So, one would think that the relationship between emotional states and work performance would receive great attention. It has not.

Third, research on the performance implications of true affective states was there at the beginning of work psychology. Hersey (1932) examined daily moods states (he didn’t call it that but he used what today would be considered a mood checklist) among blue-collar workers. He tracked those mood states with daily performance measures. He showed that on negative mood day’s performance deteriorated quite a bit, but on positive mood days there was little enhancement of performance. Yet, this promising beginning was lost in a paradigm shift that substituted the attitude construct of job satisfaction for affect constructs like moods and emotions (Weiss & Brief, 2001). For almost 60 years the study of affect was pushed aside for the study of job satisfaction and the correlation of job satisfaction and performance. One of the most ubiquitous findings in this arena is the negligible relationship between job satisfaction and performance (Iaffaldano & Muchinsky, 1985). Many, confusing satisfaction with affect, took this as an indication that the performance implications of affect were not worth studying.

The 1980s and 1990s saw a rebirth of interest in affect research (Brief & Weiss, 2002) and with that a rebirth of interest in the performance implications of affective states. Weiss & Cropanzano (1996) developed an overarching framework for studying affect at work that has spawned a good amount of research on the topic (Ashkanasy, Härtel, & Daus, 2002, Fisher, 2000; O’Shea, Ashkanasy, Gallois & Härtel, 2000; Weiss, Nicholas & Daus, 1999; Weiss, Suckow & Cropanzano, 2001). With regard to performance, Weiss and Cropanzano argued that affect-performance relationships required better process models that focused on the consequences of true affective experiences and studied those relationships in an episodic time structure.

We think progress in this area can now be made by paying close attention to advances in the basic literature on the behavioral and cognitive effects of affective states. Organizations may be rather unique social settings, but behavior at work still conforms to the same psychological processes that exist everywhere. In this paper we will tell you of our own attempts to integrate some of these findings, many of which are discussed in papers presented in this symposium, into a general model of episodic work performance. Our discussion will focus particularly on using this model to explore the work performance consequences of affective states.

Affect and Performance: The Traditional Paradigm and its Limitations

That organizational psychology is an applied field generally means that few phenomena of work life are studied without considering their performance implications. Certainly, this is true for the study of affect at work. Unfortunately, for decades it has been impossible to develop any true understanding of the relationship between affective experiences and performance because of the confusion between the constructs of job satisfaction and affect.

Since the 1930s, organizational psychologists have believed that they have been studying work related affect when they studied job satisfaction. Operationally, job satisfaction is assessed as an attitude one has about one’s job. Rigorous and ad hoc attitude scales are developed and used to measure evaluative judgments about jobs or elements of job experience (pay, co-workers, etc.). Theoretically, these operations had been thought to capture “emotions about one’s job” (see Locke, 1976, for example) and so for most organizational researchers the construct of job satisfaction was synonymous with the construct of emotion. When these researchers studied job satisfaction they thought they were studying affect, emotions, etc.

Numerous reviews and meta-analyses have consistently shown that job satisfaction shows negligible relations with job performance, heroic efforts at saving the construct through statistical corrections notwithstanding (Judge, Thoreson, Bono, & Patton, 2001). Consequently, organizational psychologists were hard pressed to acknowledge that affect had much to do with performance. Thorough process models need not be entertained to explain relationships that do not exist. Instead, it was believed that performance could best be explained by models of workers deciding among courses of action with different payoffs.

As research on emotions heated up across the fields of psychology, organizational psychologists began to realize that the satisfaction paradigm was too narrow (see Weiss & Brief, 2001, for a history of research on affect in organizations.) Emotional experiences, the fear of losing one’s job, the anger of being passed over for promotion, the elation from being noticed by higher management, are experiences not captured by the attitudinal framework of job satisfaction. Neither are the day-to-day rhythms of moods. Moods and emotions offered richer phenomena to study and, moreover, the basic research on these phenomena described consequences of clear relevance to organizational performance.

Through the decade of the 1990s, organizational research on emotions blossomed and numerous studies of affect and performance were conducted. By and large these studies described affect in terms closer to its real meaning and followed the directions pointed out by the basic research. So, for example, organizational studies have shown that positive mood states enhance worker creative problem solving (Estrada, Isen, & Young, 1997; Madjar, Oldham, & Pratt, 2002), encourage prosocial behavior (helping) among individual workers and work teams (George, 1990; 1991) and facilitate cooperation in negotiations (Forgas, 1998).

Limitations

It is fair to say that the more recent research on affect and performance has been substantially more sophisticated than prior research on satisfaction and performance. Much of this new research has been well grounded in basic theory and findings, particularly those having to do with the behavioral and judgment consequences of mood states. This is a good thing. It is also fair to say, however, that this new research remains overly narrow in its perspectives and that the narrowness owes much to the domination of applied thinking and to the vestiges of the satisfaction paradigm.

One illustration of this narrowness is the overwhelming emphasis on the effects of mood states at the expense of studying the consequences of discrete emotions. Certainly, moods vary through the course of a day and these variations have performance implications. However, so too do discrete emotions. People respond to the events of the workday with various emotional reactions. Happiness, guilt, pride, fear, and anger comprise an important part of affective life and organizational researchers have been slower to study these states than they have mood states. Why is this? The applied nature of the field encourages us to look for easy and direct connections to performance. Organizational research is still too often driven by particular criteria. Researchers list important performance criteria, search the basic literature for affect predictors and then demonstrate the relationship in organizational settings. In the affect area, so far, the mood effects have been easier to connect to criteria (mood biases judgments, mood affects helping, etc.) than have discrete emotions. The consequences of emotional states have generally not been studied in so direct a fashion. We will have more to say about this later.

Another limitation comes from the satisfaction tradition itself. That tradition examined the effects of between-person differences in attitudes on between-person differences in performance. The classic question is whether more satisfied workers are more productive workers. Much of the newer affect literature follows this between-persons paradigm. So, for example, researchers ask whether people who report typically being in positive states are more likely to do X or Y. Even when purportedly assessing “affect states” researchers use between-person assessments that strain the concept of state. George (1991), for example, looked at “mood states” by asking people to report their general affect levels over a week’s time. Apart from whether people can make such judgments accurately, such between-person assessment of aggregated affect states ignore one of the most important aspects of affective experience, their momentary and transient nature (see Robinson & Clore, 2002, for a discussion of episodic versus semantic memory for emotional experiences).

The between-person model is severely limiting for the study of affect-performance relationships. Yet, while troublesome, the dominance of this paradigm in organizational research is not surprising. As we have noted elsewhere (Beal & Weiss, 2002), the prevailing model of organizational research looks at the associations among various characteristics of organizationally relevant entities. This approach assumes that organizational phenomena are best modeled as relationships among features or characteristics that vary across entities. In most cases the entities are people but in other cases the entities are teams or even organizations as a whole. So, for example, organizational researchers study whether individual properties such as intelligence or conscientiousness predict performance, with performance taken as a relatively stable feature of the person (i.e. What kind of performer is he?). One consequence of this prevailing approach is the almost exclusive use of between-persons designs to study organizational processes, comparing people (or teams or organizations) with different levels of the characteristic of interest. Given the fundamental acceptance of this approach it is easy to see why satisfaction was, for a very long time, the accepted way to study affect. With satisfaction, affect could be studied as a property of individual workers and the association of that property with properties of tasks or jobs could be examined.

Unfortunately, the associated properties paradigm is inappropriate for studying affect-performance relationships for three reasons. First, affect itself is not well conceptualized as a property of individuals (Ekman & Davidson, 1994; Larsen, 1987). Affective experiences, whether they are moods or discrete emotions, are transient states. Happiness, sadness, pride, and anger are not characteristics of people. They are experiential states that vary meaningfully over time within individuals. As a consequence, studying them as properties is inherently limiting.

Of course, a counterargument is that there is stability to these states when they are aggregated over time. That is, while it is true that individuals vary over time in their own levels of affect, there are also individual differences in characteristic levels of affective states that can be understood and studied as properties. Although there is meaningful between-person variance in affect, this between-person variance generally is lower than the within-person variance on the same constructs (Fleeson, 2001; Larsen, 1987). Further, and this point will be emphasized later, treating affect as disposition tends to obscure affect process in organizational settings (Weiss, Nicholas, & Daus, 1999).

Although the transient nature of affective states is easy to accept, the meaningfulness of within-person variance in performance is too often dismissed. Organizational researchers have understood for quite a long time that performance varies within-persons (e.g., Kane & Lawler, 1979). Many organizational researchers, however, have generally considered this within-person variability to be error in the context of the associated properties paradigm. However, recent work has both documented how much variance in performance is ignored by dismissing the within-person component and has started to model this component of the total variance in performance. So, for example, Fisher and Noble (2000) measured momentary performance five times a day over a two-week period. They found that 77% of the total (i.e., the sum of the within- and between-persons) variance in performance was within-persons variance. Other researchers conducting variance components analyses of performance have often found nearly 50% or more of the variance due to within-person fluctuations (e.g., Deadrick, Bennett, & Russell, 1997; Miner, 2001). Further, this within-person variance is not simply error, as it is systematically related to other variables measured in the same time frame.

Although it is clear that the percent of total variance accounted for by within- and between-persons factors will vary across people, jobs, and measurement procedures, the lesson will still be the same. Performance varies across time, that variance is systematic not error, and between-persons analyses cannot enlighten the processes that account for these changes.

Within-Person Analyses of Affect and Work Performance

Affective states show meaningful changes over time. Performance shows meaningful changes over time. Emotion research tells us that the consequences of emotions are likely to have performance implications. Taken together, the conclusion is obvious. Meaningful examinations of the relationships between affect and work performance must model the momentary relationship between changes in affective states and changes in episodic performance. In the remaining sections of our paper we will lay out some thoughts about the momentary processes that describe these relationships.

Two Categories of Affect-Performance Processes

Weiss & Cropanzano (1996) made the distinction between affect driven work behaviors and judgment driven work behaviors. Affect driven behaviors are those behaviors, decisions and judgments that have (relatively) immediate consequences of being in particular affective states. Affective states are the proximal causes of these behaviors, and, as such, they are temporally coincident with those states and consequently are time bound. So, for example, people may display more creative problem solving or more helping behavior when in a positive mood. When the mood goes away so do the consequences for creativity and helping. Judgment driven behaviors are those behaviors, decisions, or judgments that are driven by more enduring attitudes about the job or organization. Evaluative judgments, or attitudes, are the proximal causes of these behaviors, although affective experiences may influence those attitudes as well. Generally, these behaviors are determined by more considered, decision making processes. So, for example, job satisfaction and organizational commitment predict turnover. Presumably they do so by entering into a decision process in which attitude toward the job is one factor being considered. Immediate affective states may play a smaller proximal role and some distal role, but they do not drive the behavior in the same immediate and time bound manner as is seen with affect driven behaviors.

Weiss & Cropanzano (1996) made this distinction to call attention to the differences between satisfaction as attitude and true affect and to show why satisfaction/attitude historically has shown weak associations with job performance. True affective experiences do, in fact, have important performance implications but they are relatively immediate, transient, and variable within persons over time. Weiss & Cropanzano further argued that productive analyses of affect and performance would require that assessment of true affective experiences (moods and emotions) replace measures of job satisfaction, that such research be done using within-persons analyses, and that better process models based upon the burgeoning literature on the consequences of affective states be developed.