Emotion Norms in Interpersonal Events1

David R. Heise and Cassandra Calhan
Department of Sociology
Indiana University
Bloomington, IN 47405

This is a pre-publication draft. For the final text see the printed article in:

Social Psychology Quarterly, 58 (1995): 223-240


A graphically-structured adjective check list was used to assess prescriptive and reactive norms of emotion in 128 social events. Both prescriptive and reactive emotion norms exist for all 128 events. Norms center on a specific emotion in about a third of the events and otherwise on a general affective tone. Both prescriptions and reactions occasionally cluster into conflicting norms regarding the expected emotion for an event. Prescriptive norms diverge from reactive norms in almost half of the events, though in only a few events do prescriptions require the opposite of people's emotional reactions. Feeling no emotion occasionally is the norm, and prescribed non-emotionality occurs more often for medical relationships, but not in relationships of other institutions based on rational objectivity. Prescriptive norms are the same for a woman and a man, except a woman is expected to emote more intensely in some interactions. However, reactive norms vary by sex in about a fifth of the events, and on the whole females are disposed to emote with somewhat more displeasure, arousal, and vulnerability than males.

Emotions are supposed to fit the circumstances. For example, a funeral is a situation where grief and sorrow are warranted, and a norm of joylessness applies. Happiness being inappropriate, people regulate themselves so as to feel sorrow, and they censure displays of happiness by others.

Recognizing that there are social norms for emotions departs from the early scientific framing of affective displays as biological phenomena (e.g., Darwin, 1872/1965). People do not emote simply because the animal within responds instinctively to certain predicaments. Rather, people intelligently guide their feelings and their displays of emotions to fit what is expected. At a funeral, sorrow over human loss is invoked at grave-side, yet joy at renewing relationships with the living may be released at the wake. Even deviance from emotion norms can serve as intelligent rhetoric -- for example, a person's irritability at a funeral may be broadcasting ineffable resentment toward the deceased.

Emotion norms, their cultural reproduction, and their adaptation as means to ends are primary concerns of the sociology of emotions that emerged during the last two decades (for summaries, see Thoits, 1989; Kemper, 1993; Heise and O'Brien, 1993; Smith-Lovin, 1994). "The most studied topic in the sociology of emotions is probably the effect of emotion norms upon experience and expression" (Gordon, 1990: 163). The centrality of emotion norms characterizes almost every line of inquiry.

One issue in the sociology of emotions is how emotions, with their obvious physiological concomitants, can be so normative. Kemper (1978) proposed that emotions are not simply socially controlled but are socially engendered, and emotions fit social circumstances so often because particular social circumstances automatically produce specific emotions. In that sense, emotions are social automatisms. Affect-control theory (Heise, 1979; Smith-Lovin, 1990; MacKinnon, 1994) offers a similar argument: emotions automatically emerge from the behavioral confirmation and disconfirmation of socially defined situations.

Interactionists focus on the paradox that emotions simultaneously are social automatisms emerging from particular circumstances and yet sometimes can be deviant. A key to understanding this is that influential people can shape emotion norms to reflect their ideology and to create resources for personal gain. Then others with alternative social understandings may emote differently than norms require. Hochschild (1983) analyzed such a pattern in the airline industry and introduced the concept of emotional labor to describe how flight attendants resolve discrepancies between their own emotional reactions and the prescriptions set by employers. Emotion work -- generating prescribed emotions that over-ride automatic reactions -- has become a key concept for understanding how people respond in social arrangements where prescribed emotion norms conflict with emotional reactions (Thoits, 1990).

Thoits (1985) considered the psychological burdens of emotion reactions that violate prescriptive norms. Recurrent deviant feelings require explanation, and individuals who attribute blame to themselves may conclude that they are deranged. Clark (1990) analyzed how individuals resourcefully concoct unauthentic emotion displays as interaction stratagems. Emotion norms provide one basis for doing this, as when someone displays prescribed guilt for behavioral deviance to forestall labeling or punishment (Heise, 1989; Robinson, Smith-Lovin, and Tsoudis,1994; Smith-Lovin and Tsoudis, 1993).

Emotion norms concentrate certain feelings in particular situations, and consequently social happenings take on characteristic ambiances. Collins (1990) linked this aspect of emotion norms to the garnering and distribution of social energies. Solidarity rituals produce moods that people carry away with them, mobilizing other social enterprises by the diffusion of activated states. By the same principle, the normative emotions at religious rituals can diffuse good will to further cooperative efforts.

Virtually every essay on the sociology of emotions presupposes that there are standard ways of emoting in particular circumstances--that norms exist to "specify the type of emotion, the extent of emotion, and the duration of feeling that are appropriate in a situation" (Smith-Lovin, 1994). Yet very little is known about the normative aspects of emotions. Instead, empirical work in the sociology of emotions has focused largely on how a range of emotional experience is shaped by social institutions or on the variety of social conditions that engender a specific emotion (e.g., Cancian and Gordon, 1988; Goffman, 1956; Hochschild, 1983; Scherer, Wallbott, and Summerfield, 1986; Stearns and Stearns, 1986; Simon, Eder, and Evans, 1992; Scheff and Retzinger, 1992).

This study focuses specifically on emotion norms. We address a number of questions about how emotional states are linked with social happenings in hopes of clarifying several issues in the sociology of emotions.

First, do prescriptive emotion norms define how an individual should feel in just relatively few consequential happenings or are people's feelings socially controlled most of the time? Clear emotion norms might regulate feelings only in rites of passage and other situations noted for their emotional tone, leaving a vast arena for untrammeled personal feeling. On the other hand, prescribed emotion norms might accompany nearly every social transaction, being a primary avenue for the social control of minds.

Second, are there reactive norms of emotion as well as prescriptive norms? Conceivably, emotional responses to social situations might so depend on idiosyncratic self processes (Denzin, 1983) that there is no central tendency to what people naturally feel in social situations. On the other hand, the view of emotions as social automatisms suggests that a given happening might evoke largely uniform feelings for most people who participate in it.

Third, what is the relation between prescriptive emotion norms and emotional reactions? People sometimes are disposed to emote differently than norms require, and therefore they must work to bring their feelings in line with expectations. This could happen if no reactive norms exist so very few people happen to emote as they should. Or it could happen because reactive norms do exist but often are different than the prescriptive norms. On the other hand, it may be true that prescriptive emotion norms generally are accompanied by matching reactive norms, making emotion work unnecessary except in special circumstances.

Fourth, are emotional temperaments and styles regulated by the norms of emotion cultures? We approach this issue, raised by Gordon (1990), in terms of gender comparisons. Brody and Hall (1993: 452) reviewed the literature on gender differences in emotionality and found

that in this culture, females are both stereotyped to be and in fact are more intensely expressive of both positive and some negative emotions .... In some contexts, females have been found to be relatively weaker than are males in both expressing and recognizing anger and other outer-directed emotions (e.g., contempt). Males have also been found to report more pride in the self than do women, and fewer of the intropunitive affects (e.g., shame, embarrassment, guilt, and anxiety).
Thus, past research suggests that the sexes have different emotion cultures and different emotional temperaments. We will investigate whether prescriptive norms diverge for males and females; if they do, and female and male emotional reactions align with prescriptions, then cultural norms could be regulating differentiation in female and male emotionality. On the other hand, finding that sex differences in emotional reactions have little relation to prescriptions would imply that differing temperaments arise from gender differentiation itself rather than normative shaping.

Fifth, what kinds of relationships require that people should display no emotion at all? The question arises from the prominence of "affective neutrality" as a pattern variable distinguishing primary and secondary relationships in Parsonian sociology (1951). Theoretically, people might be obligated to suppress emotions in work and professional encounters, presenting a neutral demeanor and maintaining conditions for enhanced rationality. On the other hand, Hochschild's (1983) research indicates that, to the contrary, workers can be required to emote as part of their jobs. Additionally, affect-control theory proposes that information about institutional roles is borne in affect and much institutional activity is generated via affective processes (Heise, 1979), and that implies that diverse emotions will arise in both primary and secondary relationships.

Scope of the Study

We investigated prescriptive emotion norms by asking college students how a participant should feel in a particular event. For example, one item in the questionnaires was "A funeral director shushes a woman. The woman should feel". Reactive norms were investigated by rephrasing the same items in more personal and immediate terms -- e.g., "Imagine a funeral director is shushing you. How do you feel at that moment?" Our instrument for recording emotion allowed respondents to choose quickly among 25 different emotional states. The 25 options constitute about one-fourth of the English emotion lexicon and are representative of the rest.

A modal response to the "should feel" question defines a prescriptive emotion norm in this study. The assumption is that plurality in specifying how a person should feel correlates with social awareness of that expectation and with individuals' tendencies to regulate themselves accordingly, and that plurality also correlates with the likelihood of contrary emotions being socially censured. We treat the modal response to the "how do you feel" question as a reactive norm, inasmuch as a reactive norm intrinsically is a statistical description of convergent responses. The assumption in this case is that emotions felt while imagining an event correlate with emotions that would be experienced in the lived event {2}.

The 128 happenings that we studied are stratified within different social institutions (e.g., law, medicine, religion, etc.). They involve a systematic range of interactants -- from good, powerful, and active people such as physicians to bad, weak, and inactive people such as addicts. Behaviors, too, were systematically selected to represent supportive actions such as helping and congratulating, as well as detrimental actions such as seducing and insulting. Variability of interactant identities and behaviors was assured by consulting dictionaries with quantified measures of connotations that were assembled originally for research in affect control theory (Heise, 1978; Smith-Lovin and Heise, 1988).



We developed a recording instrument graphically configured as a spiral to let respondents select from 25 specific emotional states like shame, anger, or depression while avoiding the tedium and frustrations of using classic mood-adjective check lists (Zuckerman and Lubin, 1965). The spiral configuration was derived from research on the structure of emotions, though we forego discussing the details here.

The instruction page from the study of emotional reactions is shown in Figure 1. The centerpiece is the emotion spiral. The bottom graphic shows the underlying structure of the spiral: the top-bottom dimension represents Pleasure, with pleasant feelings at the top. The left-right dimension represents Arousal, with feelings of activation on the right. The overlapping sectors in the spiral represent Dominance, with impotent, vulnerable feelings on the inside. The check boxes in the center of the spiral allow respondents to record emotional intensity and to select "no emotion" as an answer.

Figure 1. Sample instruction page for an emotion-spiral questionnaire.

You will choose emotions you would feel in different situations. Circle the one emotion on the diagram that is closest to the appropriate emotion. Then check a box to indicate how intense you think the emotion would be. If you think you would feel no emotion, then circle nothing and check the "No emotion" box. If you do not understand the situation, then draw an X over the diagram. Here is an example.



Note. Listed in order from the inner-most cell to the outermost, the emotion words are: embarrassed, scared, ashamed, nervous, flustered, impatient, excited, thrilled, overjoyed, pleased, happy, proud, calm, at ease, contented, unhappy, joyless, depressed, annoyed, disgusted, bitter, furious, outraged, mad.

Our questionnaires presented descriptions of social events and asked how one of the interactants in each situation should or does feel. An emotion spiral appeared with each stimulus, and four stimulus-spiral combinations appeared on each page. On the average, respondents in our studies took about 20 seconds to read each one-line vignette, circle an appropriate emotional response, and check a level of emotion intensity.


For the study of prescriptive norms, each event description presented Ego (named as "a man" or "a woman") and Alter (named by a social identity), with either Ego or Alter performing a social behavior on the other. The respondents were asked to specify the emotion that Ego should feel.

We generated stimulus events by systematically combining Evaluation-Potency-Activity (EPA) configurations for behavior and for Alter identity. (Evaluation, Potency, and Activity are universal dimensions of affective meaning -- Osgood, May, and Miron, 1975). Dichotomizing each EPA dimension as positive versus negative yields eight configurations (good-potent-active, good-potent-inactive, etc.) of behavior or identity, and 64 behavior-identity combinations. EPA ratings of identities and behaviors were drawn from an available dictionary (Heise and Lewis, 1988){3}. In order to increase diversity of our events, we also systematically varied the institutional affiliation of Alter, assigning an identity associated with academic, business, justice, medicine, religion, laity, family, or intimacy relations.

We used an 8x8 Graeco-Latin square to construct stimuli, with patterns of behavior defining rows, patterns of alter identity defining columns, and institutions as the diagonal factor. The Graeco-Latin square generated 64 scenarios in which Ego was the object of Alter's action. To illustrate, we needed to create an event in a business setting with a good, potent, active (+++) actor (such as "flight attendant") engaging in a good, potent, active (+++) behavior (such as "assists") toward a man. Thus, one of our scenarios is: "A flight attendant assists a man. The man should feel:". We used the same design to generate another 64 scenarios in which Ego acted on Alter. For example, we required a lay relationship in which a woman engages in a good, powerful, active (+++) behavior (such as "rescuing") on a +++ object person (such as "hero"), so one of our stimuli is "A woman rescues a hero. The woman should feel:". The 64 stimuli with Ego as object plus the 64 stimuli with Ego as actor constitute the basic 128 events studied in this project.

In order to examine gender differences in prescriptive norms, we created additional sets of the same events with "man" and "woman" interchanged, for a grand total of 256 stimuli. These stimuli were distributed equally across four different forms, such that no identity-behavior combination appeared more than once in a form, and events in each form featured men and women equally, and object-egos and actor-egos equally.

The study of reactive norms used the same 128 events. In this case, though, the respondent became Ego in the stimulus. For example, the stimuli mentioned above were rephrased as "Imagine a flight attendant is assisting you. How do you feel at that moment?" and "Imagine you're rescuing a hero. How do you feel at that moment?" (A slightly different phrasing was used with nine instances of insinuated behaviors -- e.g., "You realize your landlord is evading you. How do you feel at that moment?") Gender comparison is based on the sex of the respondents. The 128 stimuli were distributed into two questionnaires, following the same principles as in the prescriptive-norm study.

Respondents were encouraged to skip stimuli that they did not understand, and events involving "grind," "teetotaler," "quack," "goody," "neurotic," and "killjoy" were skipped by unusual numbers of respondents (though more than 75 percent responded in every case). We included such off-beat characters in stimuli in order to present a range of characters in different social institutions and to sample diverse perspectives within American culture. Some rare but familiar events were included for the same reason. For example, "A maniac worships a woman/man" does not refer to an everyday happening, but the phenomenon is a familiar part of American culture due to massive media coverage of incidents like movie actress Jodie Foster being worshiped by John W. Hinckley, Jr., who sought her attention via the attempted assassination of President Ronald Reagan in 1981.

Respondent Measurements

The questionnaires elicited respondent's sex, sexual orientation, age, race, citizenship, and parental income{4}.

The questionnaire included three items related to gender ideologies: "Women should take care of running their homes and leave running the country up to men" (agree, disagree, not sure); "Do you approve or disapprove of a married woman earning money in business or industry if she has a husband capable of supporting her?" (approve, disapprove, don't know); "Most men are better suited emotionally for politics than are most women" (agree, disagree, not sure). The questionnaire also collected information about the respondent's commitment to being "health conscious," "financially practical," "religious," "law abiding," "family-minded," "educated," "normal," and "loving, caring." Commitment questions used the frame: "How important to you is acting as a ____ person? (I'm not concerned with this; I give this some effort; I give this a lot of effort.")

The gender ideology responses were coded 1 for the first option (agree or approve), 2 for the third option, and 3 for the second option. A component analysis showed that responses to the gender ideology items were unifactorial; additionally we found that the responses correlated with religious commitment. We formed a single scale by doubling the range of the ideology items, then adding religious commitment and the "money" item, and subtracting the "homes" and the "politics" items. These scores, dichotomized at the median, identify respondents with traditional gender ideology. Forty percent of our "traditional" respondents accepted one or more statements of traditional gender ideology, and 30% more were unsure about one or more of the positions. Thirty percent were "traditional" just by virtue of their religiosity.

Data Collection

The prescriptive-norm questionnaires were distributed in three sociology classes at a large Midwestern university. 154 people were approached and 152 volunteered to fill out a questionnaire; 149 provided adequate data for all phases of the study, yielding about 37 respondents for each of the four forms. The reactive-norm questionnaires were distributed in two other sociology classes, and 132 of 133 students approached agreed to fill in the questionnaire; 125 provided usable data for all analyses, or about 62 respondents for each of the two forms. The questionnaire administrator orally interpreted the instruction sheet, and then participants spent about 25 minutes of class time completing the questionnaires.

Results for Prescriptive Norms

This section reports results for stimuli that presented an event involving a man or a woman and asked how the man or woman should feel.


We conducted a multivariate analysis of variance of frequency of usage of all 24 emotion words, as a function of sex and the questionnaire form that the respondent answered. Respondents' sex had no significant effect on how often they used different emotion words.

The events were presented twice -- once with a female as the focal character, and once with a male as the focal character -- to male and female respondents. Consequently, we can examine whether males and females prescribe different emotions for people in social situations and also whether emotion prescriptions depend on the sex of the focal character.

For each event separately, we cross-tabulated sex of respondent and sex of focal character with categories of emotional response. Tabulations using all 24 emotion options plus "no emotion" were beyond our resources because populating a 2x2x25 table would take a respondent sample much larger than ours. Instead, the tabulations employed the eight sectors on the emotion spiral plus "no emotion" for a 2x2x9 table that retains part of the qualitative richness of the data. However, even the 2x2x9 tables were underpopulated, so we exercised the following cautions. Statistical significance was tallied only when the probability associated with the likelihood ratio Chi-Square and the probability associated with Pearson Chi-Square both were less than .10 and the probability associated with one statistic was less than .05. Instances are considered in detail only when both statistics are associated with probabilities less than .04.

Data for ten events met the first criterion of significance, and five met the second criterion of significance. This is what would be expected by chance if there are no significant interactions in the data. That is, in 128 analyses we should expect to get five significant tables using the .04 cut-off, and we should expect to get between six and 13 tables significant at a level between .05 and .10. Moreover, the median probability over all tables is .44 (based on Pearson Chi-Square), which is close to the .50 that would be expected if interactions are absent in the tables. These analyses indicate that emotion prescriptions are not conditioned by the gender of the observer or by the gender of the observed character.

Secularization and the feminist movement may have eliminated disparate emotion prescriptions for males and females among many educated people. However, gender effects might continue among those who still maintain remnants of traditional gender ideology. Analyses restricted to the 73 "traditional" respondents halve our sample size, so we ignored respondent sex, examining only whether more traditional respondents had distinctive emotion expectations for men as opposed to women, employing a 2x9 table for each event. We found 13 instances of gender-emotion interaction by our first criterion of significance, and five events met our second criterion of significance. Again the numbers of significant results could be due to chance. Additionally, the differences that did arise hardly conform to stereotypes of how traditional men and women should emote (e.g., A vigilante sentences a man/woman: a man's emotion was expected to be in the Embarrassed sector; a woman's emotion in the Furious sector). Thus, our results indicate that there is no significant relation between emotion prescriptions and the sex of the emoting character, even for respondents with more traditional gender ideology.

We also constructed quantitative measures of Pleasure, Dominance, and Arousal to analyze gender variations across events.The Pleasure variable was +1 for any response in the Calm, Pleased, or Excited sectors, -1 for any response in the other sectors, and 0 for "no emotion" choices. The Dominance variable was 1 for the Embarrassed sector through 8 for the Furious sector; no-emotion was coded as 4.5. The Arousal variable was +1 for responses in the Excited, Nervous, and Furious sectors, 1 for the Calm and Unhappy sectors, and 0 for other sectors along with no-emotion. Analyses of variance were computed for these three dependent variables, using the mean scores on 128 events as observations, in a factorial design that varied whether the sub-sample of respondents for computing means was male or female, whether the focal character in the event was male or female, and whether the character was the actor or object of action. No sex factor was significant in a main effect or an interaction.

Finally, we coded each event according to whether respondents chose the extrapunitive emotions in the Annoyed and Furious sectors or not. Each event was coded four times: once for female respondents assessing the feelings of a woman, again for females assessing the feelings of a man; once for male respondents assessing the feelings of a woman, again for male respondents assessing the feelings of a man. If more than 33% of the responses were in the Annoyed or Furious sectors, the case was coded as characterized by "male" emotions; otherwise the case was coded as characterized by "non-male" emotions. Chi-square for the cross-tabulation of respondent sex by character's sex by male emotional response was not significant. We repeated the whole procedure, this time coding cases in terms of whether or not respondents selected "female" emotions explicitly mentioned by Brody and Hall (1993) -- embarrassment, shame (ashamed), fear, nervousness, happiness, and sadness (unhappy). The chi-square for the cross-tabulation of respondent sex by character's sex by female emotional response was not significant either. Thus "male" emotions were prescribed for a woman as often as they were prescribed for a man; and "female" emotions were prescribed for a man no less often than they were prescribed for a woman.

Perhaps gender effects arise in the prescribed intensities of emotions, rather than in the quality of the emotions (Brody and Hall, 1993). For each event, we cross-tabulated sex of respondent and sex of character with four levels of intensity ("no emotion" being the lowest), ignoring quality of emotion for these analyses. Applying the same probability criteria as above, we got the following results. Sixteen tables were significant by the first criterion. Data for eight events were significant by the second criterion. The number of significant results is beyond that expected by chance, using either criterion. This suggests that there are some gender differences in the prescribed intensities of emotions. On the other hand, such effects are far from omnipresent. Gender effects in prescribed intensities of emotion did not arise for 87% of the events considered. Examination of the significant items in the intensity analysis suggested that encounters with disvalued others are supposed to be more emotional for women. We checked this with a quantitative analysis and found that both male and female respondents did prescribe unusually high emotional intensity for females involved with disvalued others.

Prescriptive Emotion Norms

With evidence that gender is irrelevant to qualitative choices of emotions, we pooled data from male and female respondents and for male and female characters in order to tabulate the number of times each specific emotion was selected for each of the 128 events. Then we computed Chi-Squares to test whether the 24 emotion terms plus the no-emotion category were equally likely to be chosen as the way a woman or man should feel for each event. The null hypothesis, that each emotion word was chosen by about 1/25 of the respondents, is rejected at the p.04 level of significance for every one of the 128 events (the probabilities are less than .01 for all but two of the events). Thus pervasive prescriptive norms of emotion exist in the sense that people agree at above-chance levels about what emotions should be felt in response to any event.

Highlights of the data are summarized in Table 1. The stimulus event is displayed in the left column of Table 1, with a blank indicating that the event appeared twice in our prescriptive-norm questionnaires, once with "woman" as the focal character, once with "man" as the focal character.

Table 1. Normative Emotions in 128 Events, and Mean Intensity (1.0-3.0) of the Emotions.

Note: *CAP* = majority norm; CAP = modal norm; plain text = diffuse norm; italics = conflicting norms. F, M precede female and male modes where distributions differ by sex.
 Event                          Prescriptive{a}:     Reactive{b}:
 (Empty space = woman/man)      Woman/man should     You feel
 _____________________________ ____________________ ___________________
 An auditor jests a  .         annoyed (1.9), at    annoyed (2.0)
                               ease (1.8)
 A bailsman exploits a  .      FURIOUS (2.7)        furious (2.7)
 A bellhop disrespects a  .    annoyed (1.8)        F: annoyed (1.9).
                                                    M: disgusted (2.5)
 A bigamist encourages a  .    annoyed (2.2)        annoyed (1.8)
 A bisexual male embraces a  . nervous (2.1), at    F: nervous (2.1),
                               ease (1.9)           at ease (2.2). M:
                                                    nervous (3.0)
 A boarder asks a              NO EMOTION           NO EMOTION
 about something.
 A bouncer evicts a  .         EMBARRASSED (2.5)    embarrassed (2.6)
 A brat implores a   for       *ANNOYED* (2.1)      annoyed (2.3)
 A chaperon observes a         nervous (1.6)        F: *NERVOUS* (1.4).
  .                                                 M: nervous (2.0)
 A church deacon studies a  .  NERVOUS (1.9)        *NERVOUS* (2.0)
 A  's classmate nudges        NO EMOTION           F: annoyed (1.7).
 her/him.                                           M: annoyed (2.2)
 A clown hypnotizes a  .       NO EMOTION           calm (1.9), nervous
 A coroner distresses a  .     depressed (2.2)      flustered (1.9)
 A critic ridicules a  .       embarrassed (2.4)    F: EMBARRASSED
                                                    (2.6). Md: annoyed
 A desperado idealizes a  .    proud (1.9)          NO EMOTION
 The devil inspires a  .       ASHAMED (2.4)        F: scared (2.9). M:
                                                    nervous (2.3)
 A disciplinarian doubts  .    ANNOYED (2.0)        furious (2.4)
 A dropout manipulates a  .    embarrassed (1.9)    annoyed (2.1)
 A  's employer                ashamed (1.8)        F: ashamed (2.3).
 distrusts her/him.{c}                              M: bitter (2.5)
 An evangelist condemns a  .   ashamed (2.4)        F: annoyed (2.0).
                                                    M: annoyed (2.5)
 A file clerk envies a  .      PROUD (1.9)          PROUD (2.0)
 A flight attendant            pleased (1.9)        contented (1.6)
 assists a  .
 A  's friend forgets          annoyed (2.2)        DEPRESSED (2.8)
 A funeral director            *EMBARRASSED* (2.3)  *EMBARRASSED* (2.5)
 shushes a  .
 A gambler sweet talks a  .    annoyed (2.4)        ANNOYED (2.1)
 A girl rebels against a  .    annoyed (2.0)        F: mad (2.0). M:
                                                    unhappy (2.3), mad
 A  's grandchild              *PROUD* (2.5)        proud (2.7)
 follows her/him.
 A handicapped person helps    PLEASED (2.3)        embarrassed (1.8),
 a  .                                               happy (2.0)
 A has-been congratulates      proud (2.0)          proud (2.2)
 a  .
 A housekeeper blackmails      furious (2.7)        FURIOUS (2.8)
 a  .
 A hypochondriac deludes a  .  annoyed (2.1)        annoyed (1.9)
 An imbecile taps a  .         annoyed (1.7)        *ANNOYED* (2.3)
 An intern observes a  .       nervous (1.7), at    NERVOUS (1.9)
                               ease (2.2)
 A janitor fines a  .          embarrassed (1.8)    mad (2.2)
 A  's landlord evades         annoyed (2.1)        F: furious (2.8).
 her/him.                                           M: bitter (2.7)
 A lawyer trusts a  .          pleased (2.1)        pleased (2.3)
 A library assistant           ANNOYED (2.1)        annoyed (2.2)
 neglects a  .
 A loner misses a  .           depressed (1.8),     contented (1.8)
                               pleased (1.8)
 A maniac worships a  .        scared (2.5)         F: *SCARED* (2.6). M:
                                                    NERVOUS (2.1)
 A neurotic admonishes a  .{c} annoyed (1.6)        annoyed (2.0)
 A night watchman questions    NERVOUS (2.0)        NERVOUS (2.2)
 a  .
 A novice idolizes a  .        *PROUD* (2.2)        *PROUD* (2.4)
 A  's parent denounces        ashamed (2.6)        ASHAMED (2.7)
 A phony respects a  .         NO EMOTION           F: annoyed (1.6),
                                                    proud (1.5). M: NO
 A physician looks at a  .     nervous (1.8), at    F: *NERVOUS* (2.0).
                               ease (1.8)           M: *NERVOUS* (2.2).
 A pickpocket follows a  .     *NERVOUS* (2.5)      *NERVOUS* (2.7)
 A proctor inspects a  .       NERVOUS (2.0)        *NERVOUS* (2.2)
 A prude beseeches a           ANNOYED (2.0)        *ANNOYED* (2.2)
 for something.
 A puritan admonishes a  .     annoyed (1.7)        annoyed (2.1)
 A pusher addresses a  .       nervous (2.1)        F: nervous (2.8).
                                                    M: annoyed (2.2)
 A receptionist accommodates   pleased (2.0)        contented (1.8)
 a  .

 A scholar consults a  .       PROUD (2.1)          proud (2.5)
 A  's sibling governs         annoyed (1.9)        mad (2.3)
 A sinner taunts a  .          ANNOYED (1.8)        ANNOYED (1.6)
 A spinster laughs at a  .     embarrassed (1.8)    embarrassed (2.3)
 A  's spouse ogles her/him.   excited (2.7),       F: EXCITED (2.7).
                               furious (2.6)        M: excited (1.5),
                                                    contented (2.0),
                                                    mad (3.0).
 A stenographer misunderstands annoyed (1.6)        F: annoyed (1.8).
 a  .                                               M: impatient (2.8)
 A  's stepfather quizzes      annoyed (2.1)        annoyed (2.5)
 A stuffed shirt explains      annoyed (1.9)        impatient (2.2)
 something to a  .
 A subordinate debates         annoyed (1.7), calm  annoyed (2.1)
 with a  .                     (2.0),excited (2.2)
 A teetotaler watches a  .{c}  nervous (1.8)        NO EMOTION
 A tightwad agrees with a  .   NO EMOTION           NO EMOTION
 A vagrant idolizes a  .       proud (1.8)          PROUD (2.0)
 A vigilante sentences a  .    furious (2.7)        scared (2.6)
 A   tempts an addict  .       *ASHAMED* (2.7)      ASHAMED (2.4)
 A   dresses an alcoholic.     proud (2.1),         F: annoyed (2.1).
                               ashamed (2.0)        M: annoyed.
 A   pampers her/his           pleased (1.9)        pleased (1.6)
 A   consoles a beggar.        pleased (2.2)        unhappy (1.8)
 A   cuddles her/his           happy (2.2)          excited (2.7),
 blind-date.                                        contented (2.4)
 A   suspects a bookkeeper.    annoyed (2.2)        nervous (2.2)
 A   outwits a bookworm.       PROUD (2.2)          PROUD (2.6)
 A   compliments her/his       contented (2.0)      pleased (2.0)
 A   parodies a chaperon.      ashamed (1.8)        *NO EMOTION*
 A   intimidates her/his       *ASHAMED* (2.5)      ashamed (2.7)
 A   shuns a coroner.          ashamed (2.0)        *NO EMOTION*
 A   comforts a co-worker.     proud (2.1)          CALM (2.2)
 A   convicts a crook.         proud (2.5)          PROUD (2.5)
 A   underestimates a          SCARED (2.4)         nervous (2.1)
 A   insults a dimwit.         *ASHAMED* (2.3)      ashamed (1.7)
 A   prays for an evildoer.    ashamed (2.6),       ashamed (2.8), at
                               proud (2.3)          ease (2.0)
 A   silences an expert.       PROUD (2.5)          F: embarrassed
                                                    (2.0). M: proud
 A   avoids a fanatic.         nervous (2.1),       annoyed (2.2)
                               contented (1.4)
 A   deserts her/his           *ASHAMED* (2.8)      *ASHAMED* (2.8)
 A   reveres a genius.         pleased (1.6)        proud (2.2)
 A   flees a gold-digger.      pleased (1.9)        NO EMOTION
 A   joshes a goody.{c}        happy (1.7),         F: *NO EMOTION*. M:
                               ashamed (1.8)        *HAPPY* (2.0)
 A   hushes her/his            ASHAMED (2.1)        ashamed (2.2)
 A   pesters a grind.{c}       ashamed (2.1)        NO EMOTION
 A   rescues a hero.           *PROUD* (2.6)        PROUD (2.8)
 A   imitates a hippie.        embarrassed (1.8),   happy (1.8)
                               happy (1.6)
 A   serves a houseguest.      at ease (1.5)        nervous (1,7),
                                                    happy (1.7)
 A   sings to an infant.{c}    *HAPPY* (2.3)        happy (2.6)
 A   seduces an innocent.      *ASHAMED* (2.7)      nervous (2.3)
 A   discourages an invalid.   *ASHAMED* (2.5)      ASHAMED (2.0)
 A   apologizes to a judge.    ashamed (2.1)        ashamed (2.7)
 A   sits next to a            nervous (2.4)        annoyed (2.4)
 A   submits to her/his        happy (2.4)          EXCITED (2.9)
 A   corrupts a minister.      *ASHAMED* (2.6)      F: *ASHAMED* (2.6).
                                                    M: ashamed (2.7)
 A   abandons a mourner.       *ASHAMED* (2.4)      ashamed (2.3)
 A   lies to a nursemaid.      *ASHAMED* (2.1)      ashamed (1.9)
 A   ignores an old fogy.      ASHAMED (1.6)        ASHAMED (1.8)
 A   grieves for an orphan.    depressed (2.4)      unhappy (2.5)
 A   chatters to a palm        nervous (1.8), at    nervous (1.8)
 reader.                       ease (1.4)
 A   reassures a patient.      pleased (2.1)        calm (2.0)
 A   swindles a pawnbroker.    ASHAMED (2.1)        nervous (2.2)
 A   nuzzles a pornostar.      EXCITED (2.5)        F: nervous (2.6).
                                                    M: EXCITED (2.8).
 A   flatters her/his          pleased (1.9),       pleased (1.9),
 professor.                    ashamed (1.6)        nervous (1.9)
 A   rehabilitates a           *PROUD* (2.4)        proud (2.7)
 A   cautions a pupil.         contented (1.9),     calm (1.6)
                               annoyed (1.6)
 A   derides a quack.{c}       proud (2.3),         NO EMOTION
                               ashamed (2.0)
 A   bawls out a               ashamed (2.1)        FURIOUS (2.8)
 registered nurse.
 A   contemplates a rival.     nervous (2.0)        nervous (2.1)
 A   challenges a rookie cop.  excited (2.0)        NERVOUS (2.0)
 A   saves a runaway.          *PROUD* (2.6)        proud (2.5)
 A   monitors a servant.       at ease (1.9)        NO EMOTION
 A   calms a shrew.            calm (1.6)           calm (1.5),
                                                    impatient (1.8)
 A   poisons a spy.            ASHAMED (2.4)        F: nervous (2.6).
                                                    M: nervous (2.8),
                                                    proud (2.3),
                                                    joyless (2.0).
 A   appeals to a state        NERVOUS (2.0)        NERVOUS (2.4)
 A   heckles a street          ashamed (2.3)        ashamed (2.0)
 A   begs a supervisor         embarrassed (2.0)    embarrassed (2.0)
 for something.
 A   studies a surgeon.        NO EMOTION           excited (2.3)
 A   nags her/his tenant.      ANNOYED (2.0)        impatient (1.8)
 A   enrages a thug.           scared (2.5)         scared (2.5)
 A   defends a tramp.          proud (2.0)          proud (1.8),
                                                    ashamed (1.8)
 A   flunks a truant.          unhappy (2.0)        F: UNHAPPY (1.9).
                                                    M: unhappy (1.7).
 A   judges a virgin.          *ASHAMED* (2.2)      NO EMOTION
 A   humbles a warden.         proud (2.2)          F: NO EMOTION. M:
                                                    *NO EMOTION.*
 A   converts a witch.         PROUD (2.3)          nervous (2.0)

a    Ns: 88 females, 61 males.
b    Ns: 81 females, 44 males.
c    Missing data for 15-23 percent of the subjects. This is due to
typographical errors on a prescription-norm form in the case of "sings to an
infant" and "employer distrusts."  The other items presumably were skipped
because some subjects did not know the meanings of constituent words.

The middle column of Table 1 shows the prescriptive emotion norm for each event. The given emotion is the modal response or the response that is most central in a set of congruous choices, as defined below. The number in parentheses is the mean prescribed intensity for the emotion on an assumed-interval scale going from 1.0 for "low intensity" to 3.0 for "extreme intensity."

Capital letters with asterisks on the modal emotion means that a majority of respondents chose the modal category. These well defined emotion norms with more than 50 percent of respondents selecting the same category arose for 19 events, or 15 percent of all the events we presented to respondents. A focal character should feel ashamed in ten of these events, and five are happenings that should produce pride. The prominence of these two emotions supports Scheff's (1990) treatment of shame versus pride as a critical axis in affective response to social situations. Embarrassment, nervousness, annoyance, and happiness are other emotions that were strongly modal for some event.

The modal frequency was more than twice the frequency of any other choice for an additional 30 events -- the cases where the emotion is displayed in capitals. Shame, pride, and annoyance each occurred five times in this set, and "no emotion" occurred six times. Adding these events to the previous 19, we find 49 events -- 38% of the total -- in which there is a modal prescriptive norm of emotion.

Emotion choices were congruous for another 63 events (the cases where the emotion has no emphasis). These are cases where secondary modes -- emotions that were chosen at least half as often as the modal choice -- were in the same sector as the primary mode, or else all were in a set of sectors with one sector adjacent to the rest. (We treated "no emotion" as being in the middle of the affective space and thereby adjacent to all sectors; the Embarrassed sector as vertically adjacent to the Unhappy, Annoyed, and Furious sectors; and the Nervous sector as vertically adjacent to the Annoyed and Furious sectors.) Adding the cases of congruous expectations to the modal cases, we get 112 events 88% of the total where an emotion norm prescribed at least a diffuse affective tone. Whereas none of the modal norms involved low arousal, seven of the diffuse norms centered on emotions in the Calm or Unhappy sectors.

Competing and incompatible emotion norms arose in 16 events (12%). Some examples show that these involve major variations both in hedonic tone and activation.




The prompt eliciting prescriptions was "The woman/man should feel," and there is some ambiguity in the word "should." It was supposed to be read in the sense of "ought," but could have been read as "would." Thus some of the incompatible norms may contrast what some respondents thought people ought to feel and what other respondents thought people probably do feel. This interpretation implies that one among a set of conflicting norms will correspond to a reactive norm. However, this is not always the case (as can be seen in Table 1 by comparing column two with column three; the third column shows reactive norms). Thus, it is likely that some cases of incompatible norms identify real conflicts in prescriptions.

To summarize, a central tendency occurred in emotion prescriptions for most of the events that we considered, but conflicting norms appeared with some events. Prescribed emotions in various circumstances included both pleasant and unpleasant emotions, states of arousal and non-arousal, and emotions of vulnerability as well as emotions of dominance.

"No emotion" was the prescribed affective state in six cases, five percent of the total. However, we absorbed "no emotion" as adjacent to other emotional states when defining diffuse norms, so the prevalence of "no emotion" choices is somewhat hidden in Table 1. In fact, "no emotion" was the primary mode or chosen at least half as often as the primary mode in18% of the events. These "no-emotion events" are affectively neutral compared to other events in the sense that some respondents prescribed no emotion as the appropriate affective state and also because the rest of the respondents tended to prescribe low emotional intensity. The number of no-emotion choices correlates with the average intensity for other choices, r=-.63.

Results for Reactive Norms

This section reports results for stimuli that hypothetically involved the respondent in a event and asked the respondent "How do you feel at that moment?"


A multivariate analysis of variance of frequency of usage of the 24 emotion words, as a function of sex and the questionnaire form that the respondent answered, produced a significant effect for sex. Across the 64 events that each respondent considered, females more often said they would feel "embarrassed," "scared," or "ashamed," while males more often said they would feel "calm" or "happy" or "joyless."

For each event, we cross-tabulated sex of respondent with the eight sectors on the emotion spiral plus "no emotion." A lenient criterion of statistical significance was fulfilled when the probability of the likelihood ratio Chi-Square and the probability of Pearson Chi-Square both were less than .10 and the probability associated with one statistic was less than .05. A more stringent criterion required both statistics to be associated with probabilities less than .04.

No significant difference was found for 81 percent of the events, but data for 24 events (19%) met the first criterion of significance, and 13 events (10%) met the second criterion of significance. The percentages are about double what might be expected by chance, indicating that emotional reactions differ by sex for some events. Events with sex differences meeting the stringent criterion of significance are shown in Table 2.

Table 2. Events With Significant Sex Differences in Emotional Reactions

                                  Sectors selected more often by:
Event                             Females                Males
_________________________________ ______________________ ______________________
your employer distrusts you       Furious...             Annoyed ...;
your landlord is evading you      Furious...; Nervous... Unhappy...
a girl is rebelling against you   Furious...; Annoyed... Unhappy...;
                                                         Embarrassed...; Calm...
you're corrupting a minister      Embarrassed...         Unhappy...; No emotion.
you're silencing an expert        Embarrassed...;        Pleased...; Excited...
                                  Nervous...; No emotion.
you're poisoning a spy            Embarrassed...         Pleased...; Calm...
the devil is inspiring you        Embarrassed...;        Nervous...
a critic is ridiculing you        Embarrassed...         Annoyed...; Unhappy...
a bellhop disrespects you         No emotion; Nervous... Furious...; Annoyed...;
a bisexual male embraces you      Calm...                Furious... 
your spouse is ogling you         Excited...; Annoyed... Nervous...; Calm...
you're nuzzling a porno-star      Annoyed...; Nervous... Excited...
a phony respects you              Pleased...             No emotion.

The first three events in Table 2 suggest that females more than males are disposed to rage when a significant or trusted other rejects them. The same pattern shows up with "parent denouncing" though the sex difference is not significant. Similarly, the majority of females respond in the Furious sector to a "housekeeper blackmailing," though in this case a majority of males do, too.

The next five events in Table 2 can be viewed as implicating the respondent in social ruination of another or of self, and in these cases, females are more likely than males to report emotions in the Embarrassed sector. A similar pattern for "humbling a warden" also is significant by our less stringent significance criterion.

The events in Table 2 involving "critic," "bellhop," and "bisexual male" suggest that males more than females may be disposed to dominance emotions when threatened by a socially-distant character. "Evangelist condemning" shows the pattern and is significant by our less stringent criterion. "Vigilante sentencing" also shows the pattern, though the sex difference is not significant. On the other hand, the sexes are not different in dominance emotions for "bailsman exploiting," "bouncer evicting," and "disciplinarian doubting. Moreover, females show significantly more dominance emotions for "spinster laughing at," by our lesser criterion.

The items, "spouse ogling" and "nuzzling a porno-star," suggest that male and female emotional reactions are differentiated in sexual events. However, sexual matters also are intimated in events involving "bigamist," "blind-date," "innocent," "lover," "prude," "puritan," "virgin," and "chaperon," and only the last has gender differences that are significant by our less stringent criterion. Lacking a generalization, we simply note that many males were flabbergasted at the thought of being ogled (about one-third of the males skipped the item), and many females were disgusted by the thought of nuzzling a porno-star.

We quantified Pleasure, Arousal, and Dominance as described in the section on prescriptive norms and conducted an analysis of variance of the mean scores for all 128 events, in a factorial design that varied whether the sub-sample of respondents for computing means was male or female, and whether the respondent was the actor or object of the hypothetical action. No sex factor was significant in a main effect or an interaction. However, we also averaged respondents' Pleasure, Arousal, and Dominance over the 64 events to which they responded, and then conducted an analysis of variance of individuals with sex of respondent and questionnaire form as factors. In this case, respondent's sex significantly influenced Pleasure, Arousal, and Dominance. Overall, females emoted more negatively, with more arousal, and with more vulnerability than males. The difference between the results of these two analyses implies that differences in emotional response between the sexes are small compared to emotion variations induced by different events, but the differences in emotional response between the sexes are an important component of individual differences in emotionality.

We found no evidence that males are more disposed in particular to "outer directed" emotions of anger and contempt. For males and females separately, we dichotomized events into those where the modal response was in the Annoyed or Furious sector versus elsewhere and then cross-tabulated this dichotomy with sex and whether the respondent was actor or object in the hypothetical events. There was no significant interaction in the table.

The intensity of emotional reactions did not differ by sex. For each event, we cross-tabulated sex of respondent with our four levels of intensity, ignoring quality of emotion. A sex difference was significant in eight tables by our lenient criterion and in five tables by our stringent criterion. These numbers of significant results are no more than would be expected by chance. We also computed mean intensity over the 64 events presented to a respondent, and then conducted an analysis of variance of individuals, with sex of respondent and questionnaire form as factors. The respondent's sex was not significantly related to the average intensity of emotional responses.

Reactive Emotion Norms

Combining female and male respondents, we tabulated the number of times each specific emotion was selected for each of the 128 events and computed Chi-Squares to test whether the 24 emotion terms plus the no-emotion category were equally likely to be chosen as a way of feeling in each event. The null hypothesis, that each emotion state was chosen by about 1/25 of the respondents, is rejected at the p.04 level of significance for every one of the 128 events (actually the probability is less than .01 for every event). Thus reactive norms of emotion do exist. Respondents' emotional reactions converged at above-chance levels, despite the fact that there were some significant female-male differences.

Highlights of the data on reactive norms are summarized in the right-most column of Table 1. Majority norms, modal norms, diffuse norms, and conflicting norms were defined in the same manner as with prescriptive norms, and the use of underlines is the same as defined previously. Additionally, the norm for each sex is presented separately wherever there is a significant difference in female and male distributions.

In eleven events, a majority of respondents converged on a single option. In another 29 events, the modal frequency was more than twice the frequency of any other choice. These unequivocal reactive norms occurred for 31 percent of all events. In another 56 events, modes and secondary modes were congruous in affective tone. Adding these to the modal norms, we find 75 percent of the events producing emotion reactions with at least a diffusely defined affective tone. Moreover, both sexes ended up with an identical norm in eight of the 24 events where there were significant gender differences in distributions, so a shared reactive norm actually occurred for 81 percent of all events. Conflicting norms of emotional reaction occurred with 19 percent of the events. Sex differences were implicated in two-thirds of the conflicting norms; eight events (6 percent) involved differences in emotional reaction that were not associated with sex differences.

Respondents converged on a response of "no emotion" for 12 events, or nine percent of the total.


Our student-based questionnaire study was an economical form of research suited to the exploratory nature of this work. However, studies of diverse populations based on both questionnaires and systematic observations will be needed to deepen the study of emotion norms. Some studies should explore the consequences of presenting events with known relevance to the respondent population, and of describing the events in more detail than we did so as to narrow interpretations. The invocation of prescriptions and emotional reactions should be examined with different wording of questionnaire prompts (e.g., using "ought" rather than "should" to evoke prescriptions). Other emotion-recording instruments also should be employed -- such as the Affect Grid of Russell, Weiss, and Mendelsohn (1989) or, in observational studies, the Facial Action Coding system of Ekman and Friesen (1978).

Respondents in both of our studies chose diverse emotions in response to many stimuli. We computed co-occurrence coefficients for the emotion choices across all stimuli, applied non-metric multi-dimensional scaling, and essentially reproduced the emotion circumplex (Russell, 1980; 1983; Fisher et al, 1985; Morgan and Heise, 1988; MacKinnon and Keating, 1989) in each study. The outcome might be expected since the circumplex was used to format our data collection instrument, but the analyses nonetheless demonstrated that alternative emotion choices connote similar Pleasure and Activation, in our studies as in other studies (Russell,1989). Diverse emotion specifications spreading from the modal category corroborate Gordon's (1990: 164) point that "Emotion norms prescribe a range of permissible feeling, not a precise point."

The scaling analyses indicated that the measuring instrument shown in Figure 1 can be improved in future studies. The emotion "scared" actually belongs in the Nervous sector because "scared" is employed congruously with "nervous." ("Scared" could replace "impatient," which was used infrequently by respondents in the prescriptive norms study and which was used congruously with "annoyed" in the reactive norms study. "Crushed" would be an appropriate replacement for "scared" in the Embarrassed sector.)

Granting that there is much to learn from subsequent studies, we now summarize information gained from this one.

The Existence of Emotion Norms

Prescriptive emotion norms exist for all of the social events that we sampled. The norms formulaically focus on a specific emotion about a third of the time and on diffuse affective states otherwise. Normative conflict rather than normlessness characterizes the relatively few events where there is a lack of central tendency in emotion expectations. Since our sample of events systematically varied kinds of interactants, kinds of interpersonal behaviors, and institutional affiliations, we conclude that emotionality is normatively regulated in virtually all social encounters within American middle-class culture.

Reactive emotion norms also exist for all the social events that we sampled. That is, emotions of respondents converged on modal responses as they imagined themselves in particular encounters. Even when respondents did not converge on a single emotional state, their responses converged on several different modes rather than being random. Consequently we conclude that particular social circumstances tend to produce specific emotions, as postulated by social automatism theories within the sociology of emotions.{5}

Prescriptive-Reaction Discrepancies

Prescriptive emotion norms and reactive emotion norms are identical in 38 percent of the events in Table 1. In another five percent of the events the reactive norm is the same as one of the options in a prescriptive-norm conflict, so people's reactions agree with what some people think is appropriate. In another 12 percent of the events, one of several modal reactions agrees with a prescriptive norm. On the other hand, in 45 percent of the events there is a discrepancy between emotion prescriptions and emotional reactions.

These results can be viewed from two perspectives. First, people do not need to work to bring their emotions in line with expectations in every social situation -- they sometimes are disposed to feel what they are supposed to feel. Thus the need for emotion work is bounded.

Second, people sometimes have emotional reactions that have to be masked and over-ridden in deference to prescriptive emotion norms. Such predicaments arise in a variety of social situations, so demands for emotion work go far beyond the commercialized emotional labor that has been the focus of research literature.

Many of the discrepancies between prescriptions and reactions that appear in Table 1 are not large, suggesting that emotion work often is not arduous. Consider a person assisted by a flight attendant. The reactive norm in this case is to feel contented, but the prescribed norm is to feel pleased. The passenger would like to sigh and settle back in her seat with a faint smile, but she feels obligated to face the attendant alertly with a broad smile or a grin and some sparkle in her eyes -- a little bit of extra activation until the attendant moves on.

Six events in our sample involve discrepancies that require wrenching from one pole of an emotional state to another. Four of these involve taking an attitude of dominance or submission when one feels the opposite ("dropout manipulates," "evangelist condemns," "vigilante sentences," and "bawling out a registered nurse). "Consoling a beggar" in the presence of an observer requires forgoing empathic unhappiness in order to seem pleased with one's opportunity for magnanimousness. "Serving a houseguest" requires people who are disposed to nervousness in such a situation to suppress their feelings in order to give the impression of being at ease. The low proportion of situations requiring strenuous emotion work suggests that social definitions generally are organized so as to preclude strenuous emotion work. However, demands for strenuous emotion work do arise, and they may arise more frequently than is suggested by our study, since we sampled events to represent a broad range of things that could occur rather than to reflect the frequencies of everyday happenings.

Gender Differences

Prescribed emotions were not a function of whether the focal character was a woman or a man, or whether respondents were male or female. Our respondents -- both male and female -- prescribed stereotypically "female" emotions for a man just as often as for a woman, and they prescribed stereotypically "male" emotions for a woman just as often as for a man. We got the same results when we selected the most traditional respondents and analyzed just their data. Since the respondents in this study, mostly aged 20 or more, are representative of people who promulgate norms in our society, the results very likely describe the prescriptive norms of emotion that regulate the feelings of American middle-class young adults.

Thus, we found no evidence that emotion cultures separate the qualitative feelings of women and men by prescribing different emotions for the sexes. Rather, prescriptive norms actually inhibit gender differences in emotions by calling for the same feelings from both sexes.

A different story about gender developed from the analyses of emotional reactions, where significant sex differences occurred for 19 percent of the events in our sample. Females more than males described their feelings in terms of the stereotypically female emotions, "embarrassed," "scared," or "ashamed," and overall females emoted with somewhat more displeasure, arousal, and vulnerability. This accords with Brody and Hall's (1993) generalization that females are more inclined toward intropunitive affects. Males more often said they would feel "calm" or "happy" or "joyless". The relative positivity of the male preferences perhaps reflects the Brody and Hall generalization that males show more pride in the self. However, contrary to Brody and Hall's report, males were not more disposed to stereotypically-male "outer directed" emotions of anger and contempt. Indeed, we found situations in which females experienced these dominance emotions more than males.

The sex differences in emotional reactions could be due to female and male differences in emotionality, conceivably with a chromosomal basis. However, assuming that emotional reactions emerge while one participates in social events, as proposed by social automatism theories of emotion (Kemper, 1978; Smith-Lovin, 1990), then the differences in emotional reactions would originate in females' and males' different social structural positions within events, or in different socialization backgrounds that lead to different interpretations of events. We offered several conjectures about the kinds of events that generate sex differences: (1) females more than males are disposed to rage when a significant or trusted other rejects them; (2) in ruination of self or other, females more than males are prone to intropunitive emotions; (3) males more than females may be disposed to dominance emotions when threatened by a socially-distant other.

Our respondents expected more intensity of emotion from females -- especially when interacting with disvalued others, but that prescription did not correspond to reactions. On the average, females and males recording their own feelings did not differ in intensity. Brody and Hall (1993: 449) reported that "the finding of generally greater affect intensity in females is well established," but our results suggest that future studies should make a careful distinction between expectations and actual experience.

Prescribed Emotional Neutrality

If secondary social relationships are more affectively neutral than primary relationships, then prescriptions to feel no emotion should be concentrated in academic, business, justice, and medical relationships, while occurring rarely when one is involved in family or intimacy relationships. (Religious and laity relationships might be either primary or secondary, so no prediction is made in these cases.) We estimated a log-linear model for each type of relationship to see whether it involved an unusual number of events with no-emotion prescriptions.

The number of no-emotion norms was what might be expected for academic, business, family, intimacy, and laity events, given the number of events involving each such relationship and the total number of events with no-emotion norms. Unusually few no-emotion norms occurred for events in the justice system and in religion. An exceptionally large number of no-emotion norms occurred when partners were medically-defined -- about a quarter of all medicine-relevant events in our sample had no-emotion prescriptions.

These results suggest that a trickle of no-emotion prescriptions occurs in diverse circumstances. Lack of emotionality is called for even in family and intimacy situations (though our two instances of this involve marginal characters -- stepfather and chaperon).

On the other hand, affective neutrality is far from a defining feature of secondary relationships. Emotional-neutrality prescriptions were concentrated in one arena of secondary relationships -- medicine, but even there most of the sampled events were expected to be emotional. There was no notable concentration of no-emotion prescriptions in two other arenas of secondary relationships -- business and academia. Events in the justice system were supposed to be less emotionally neutral than elsewhere.

These results indicate that emotionality is a ubiquitous concomitant of social interaction, both in primary groups and in secondary groups. In particular, many commercial and professional transactions are expected to be emotionally charged in orderly ways, just as Hochschild (1983) said. Rather than undermining rational objectivity in social institutions, affectivity must be non-deleterious, or even productive. Affect-control theory offers insight into how affectivity could be productive in social institutions. "[T]he theory supposes that cultural sentiments can regenerate the standard events from which they have evolved, so that the affective system provides the means for storing a society's heritage of socially structured interaction" (MacKinnon, 1994, p.180). The affective meanings of identities and behaviors essentialize the rationality of the past, and emotions arise in social institutions as that stored rationality gets re-applied and tested in current affairs.


  1. We are grateful to participants in the social psychology training seminar, Department of Sociology, Indiana University, for their helpful criticism and suggestions. This work was initiated while DH was Director and CC a trainee in the Affect and Affect Measurement Training Program, PHS T32 MH15789.
  2. Our assessment of prescriptive norms employs the respondent as an observer of others, whereas our assessment of reactive norms elicits the respondent's personal reactions. This re-centering of respondents' viewpoints in the two studies is appropriate to the distinction between prescriptive and reactive norms. However, hedonistic factors could be more prominent when respondents are asked to give their own reactions, and that factor should be kept in mind when comparing prescriptive and reactive norms.
  3. When only one or two behaviors fit the EPA profile for a category, we used behaviors that came close to fitting the profile. For example, "idolize" was the only word in the Heise and Lewis (1988) dictionary that fit the +-+ EPA profile (good, powerless, active) so we used the behavior "sweet talk" as a +-+ behavior when actually its EPA rating is .22, .13, .51 (or +++). The demands of our stratified event sampling system also led us to use some identities that turned out to be unfamiliar to some respondents -- see footnote c in Table 1.
  4. Our sample consisted mainly of white (81%) U.S. citizens (97%) of heterosexual orientation (99%). Parental income ranged from "less than $20,000" to "over $100,000," with 19% of the respondents reporting the median category of $61,000 to $80,000 and 23% in the modal category of $41,000 to $60,000. Age ranged from 18 to 45 with a median of 20.
  5. A reviewer raised the possibility that norms might have emerged because respondents tried to be helpful. The clarity of the normative structures in both studies suggests that even if respondents were making an effort to give answers a sociologist would want, they must have been relying on normative information as they guessed the "right" answers. Thus the conclusion remains: ubiquitous emotion norms exist.