A Theory of Interpersonal Goals and Situations

Two Views of the Situation-Goal Relationship

Our theories propose two different views of the connections between situations and goals. These views are not immediately consistent, although this seems not to have been previously noticed. The purpose of this article is to reconcile the two views. They are as follows: (1) situations activate goals, and (2) goals define situations. The inconsistency appears when one tries to work out the causal ordering. If situations stimulate goals, then the situation’s nature and perceived character must exist before the goals do. But if the situation is defined by the goals in play, then the goals must precede the situation.

The first view is common in the message production literature. Greene’s (1997) action assembly theory maintains that goals are stored in people’s cognitive systems so that contact with various situational features activates particular goals. Many theories in addition to Greene’s propose that goal activation is a function of fit, strength, and recency (e.g., Meyer, 1997, 2002; Wilson, 1995, 2002). Strength means that a particular goal has a chronically high level of semi-activation so that little additional stimulation is required to bring it into play (e.g., aggression has strength for a thug with “a chip on his shoulder”). Recency means that a goal has extra residual activation because it has just been in use. But here our interest is mainly about fit, which refers to a match between situational elements and goals. The idea is that goals are cognitively stored along with what we might call relevance indices, lists of conditions that make the goal pertinent.

Wilson (1990) has given detail to this conception of fit. He theorized that we store cognitive rules that control our momentary formation of goals in an immediate situation. Each rule roughly takes the following form: “If the situation has feature X, then activate goal A.” Obviously, situations have more than one feature, and more than one goal can be energized by a particular situational characteristic. Which goals rise to the fore is determined by a sort of energy algebra: The goals with the best fit have the highest activation levels, and so are the ones that assert themselves. Wilson in particular was aware that this theoretical stance presupposes that a situation has some nature that is perceived before it stimulates goals, not only because he specified goal-independent situation features (e.g., “X” in the rules) but also because he has studied how individual differences result in different goals being formed by different people in contact with the same situation.

So the first view can be summarized in this way: Situations cause goals, and the causing is predicted by fit, strength, and recency. This formulation requires that the situations have a perceivable nature that is not already determined by the goals.

The second view is that situations are subjective constructions and are essentially defined by the goals that dominate them. Thus, we speak easily of conflict situations, initial acquaintance interactions, persuasive encounters, or comforting episodes. Each of these phrases takes some goal as the central defining feature of the situation.

This view is also prominent in our literature. The first edition of the Handbook of Interpersonal Communication contained an essay called, “The Situation as a Construct in Interpersonal Communication Research” (Cody & McLaughlin, 1985). When the Handbook went into a second edition, the authors revised their project to have the title “Situations and Goals as Fundamental Constructs in Interpersonal Communication Research” (Miller, Cody, & McLaughlin, 1994). The reason for the expansion in focus was that the authors felt situations and goals were so intertwined that even if they could be distinguished, there was little point. Although Miller, Cody, and McLaughlin (1994) discussed activation of goals in situations (the first view) and acknowledged that situations are configurations of several elements including plans, goals were the core of their most basic understanding of situation:

The particular configuration of goals activated, whether we have achieved them or believe we can achieve them, and the role of others in facilitating or inhibiting goal achievement are apt to play an important role in our constructions that this is an “x” and not a “y” situation. . . . [G]oals are apt to play a central role not only in defining the “situation” but in directing subsequent communicative behaviors, such as devising plans, implementing plans, and employing multiple tactics over time. (p. 175)

This view clearly regards situations as taking their nature from the particular constellation of goals that are activated in the moment. Thus, the goals define the situation.

This second view was elegantly put forward in Dillard and Solomon (2000). They wrote that situations are “social densities.” Planets and stars are gravitational densities in the universe, and should we travel that far, we might well orient to those physical densities because they have a useful defined character that distinguishes them from the “emptier” and less interesting space around them. Similarly, “the phenomenological world is organized into clusters of events with regularly co-occurring qualities” (Dillard & Solomon, 2000, p. 168). These stand out to us socially in the same way that a gravity well stands out in a survey of the sky. Social densities are simultaneously composed of our perceptions and the goals we apply to them. The latter occur in complex configurations of primary and secondary goals. These recognizable goal configurations are the criterial features of communication situations. “Thus, the contexts for message production can be meaningfully defined in terms of the goal structures that arise within social densities” (Dillard & Solomon, 2000, p. 173). Although the idea of social density leaves room for things other than goals to define or identify a particular sort of context for communication, Dillard and Solomon focused mainly on the goals. The idea is that we act within situations based on the matrix of goals we pursue, and therefore this complex purposeful character constitutes the essential phenomenological understanding of our communication context.

The Dillard and Solomon (2000) essay should be read as an expansion of Dillard’s (1990a, 1990b; Dillard, Segrin, & Harden, 1989) earlier notions of primary and secondary goals. The primary goal in a situation is not necessarily the most forceful one. Instead, it is the defining goal. Having a primary influence goal means that one is in an influence situation, for instance. “From the perspective of the actor, the influence goal brackets the interaction. It helps to segment the flow of behavior into a meaningful unit; it says what the interaction is about” (Dillard et al., 1989, p. 21). So even if one were trying to persuade a friend to calm down, if the primary goal were to give emotional support, we would say that the person was involved in a comforting situation, not a persuasive one. Secondary goals modify (and can even end up disallowing) the primary goal, and involve face and resource considerations, among others.

So the core idea in the second view is that we recognize goal constellations in our social worlds and identify these different goal patterns as being different types of situation. In other words, registering the goals serves to define the situation.

The two views are not immediately consistent, but this does not mean they are necessarily contradictory. There are several ways that they can both be right at the same time. One possibility is that people cycle through sequences of situation to goal to situation to goal in some sort of feedback loop. The loop might have an iteration counter that allows more cycles for more important conversations, or might have some satisficing criterion that tells people when to stop worrying about it. Another theory might be that situations instigate goals in routine encounters, and more delicate matters are defined by the goals that arise. But there is a third possibility, which is outlined next.

A Coalescence of the Two Views

Due to Dillard, we have become comfortable thinking of goals as being primary or secondary. The present article proposes a similar distinction among the non-goal attributes of situations. Some situational features are what will be called “reasonably apparent,” and the others are considered to be “subjective.” The key theoretical move is to distinguish these two sets of descriptors, and then to propose that they take different roles in the situation-goal relationship.

Reasonably apparent characteristics of a situation are those that would be fairly obvious to an observer, and those that an interactant would be unlikely to make a mistake about. In other words, these are uncontroversial and do not require serious interpretive work for their identification. These include own and other’s biological sex, the physical setting of a conversation (e.g., apartment, workplace), the historical and current relationship between the communicators, the apparent topic of the conversation, its medium (e.g., telephone, face-to-face), the type of interaction (e.g., social, professional), and the number of people present. Reasonably apparent features of a situation are registered at a glance, and they are expected to persist throughout an interaction. They can change, of course. One might suddenly realize that one is talking to a woman and not a man, or an apparent topic might eventually be understood as a Trojan horse for another issue, or one might apprehend in the midst of a conversation that a cross-sex friend now wants to have a romantic relationship. But these recognitions come with little jolts, and it is the jolt itself that gives proof that the original formulation was supposed to be secure, something that could be taken for granted.

Subjective situational features are those that are not as well defined in the first place, and they emerge and develop as an interaction moves along. These are elements of context that are always interpreted, and they require genuine interpersonal perceptiveness to get right. Only real social engagement affords much chance for the other person’s motives and plans to be accurately inferred, or for the interaction’s climate to be jointly felt. Subjective elements of a situation include its politeness, whether it has an argumentative quality, how self and other feel about things, and whether circumstances (including the other person) imply obstacles to what one might prefer. Notice that all these things emerge as an interaction grows. They can all persist or evaporate in a moment. They are dynamic, not necessarily apparent to an observer, and never to be safely taken for granted. None are authored: They are coauthored and emergent in the interactive space between the participants. Only the participants have privileged access to them. Here, they are called subjective because the present research strategy involves interrogating only one person at a time, but they are really intersubjective, or at least are supposed to be in a smooth interaction.

This article’s essential theoretical proposal involves connecting this new distinction with the old one between primary and secondary goals. The idea is that reasonably apparent features activate primary goals (Hypothesis 1 [H1]); that the primary goals give rise to the subjective elements of a situation (Hypothesis 2 [H2]); and that the subjective elements energize the secondary goals (Hypothesis 3 [H3]). This is the sequence investigated here in both studies (see Figure 1). H1 expresses the first view: Situations stimulate goals. The latter parts of the process explore the second view by showing how the goals can then define the character of the situation.

OPEN IN VIEWER

Figure 1.

Theory of interpersonal goals and situations.

This simple description of the process is probably incomplete. The secondary goals may later modify the subjective elements, and those might cycle back to affect the primary goal in play, until some equilibrium between what is mutually wanted and what is going on is reached. Figuring out the timing of all this in order to apply the proper statistical methods is daunting, and so here only the base linear sequence will be tested. Collecting data after the episode is complete, as is done in Study 1, gives the settled homeostatic recollections after the feedback loops have been finished. Regrettably, it also restricts analysis to testing associations rather than directional causality, but at least it will be possible to falsify the hypotheses if they are fundamentally wrong.

The essential rationale for the theory does in fact have to do with timing. Something has to come first. Only one portion of the model (reasonably apparent features) can be accomplished with simple unreflective observation. Everything else needs some level of interpretation or interpretation-informed action, and sophisticated interpretation requires nonobvious information. As a person lives through the process of a conversation, more data of different sorts become available. The simplest interpretive information is one’s personal reaction to what is going on, and that should result in a primary goal: What do I want to achieve here? Implementation of such a defining goal will establish the situation’s initial character, generating a new uncertainty: Given my framing goal, what is going on? Answering this question gives additional information that permits adaptation so that secondary considerations (e.g., politeness, anxiety, resources) can be pursued, either in service of the primary goal or in escape from it.

Study 1

Results

All of the measures were categorical, making χ² the appropriate test for bivariate relationships among the groups of variables. Many such tests were conducted, generating considerable detail about the associations between the theorized phases (see Figure 1). In principle, logit analysis (Knoke & Burke, 1980; Menard, 2010) could be used for more complex tests, but the data set had too many categories relative to its sample size to make tests of many variables at once feasible. In common with logistic regression, the χ² test becomes unreliable when expected frequencies for individual cells are often less than five (Koehler & Larntz, 1980). Therefore, small-frequency and “other” categories were omitted from the analyses. Given the number of contingency tables involved in the results to follow, the report focuses on the gross results, neglecting the specifics of particular associations. Those details are available from the author. The key result reported here will be Cramer’s V, an effect size measure that ranges from 0 to 1, and which has significance levels based on the χ² test. Cramer’s V does not assume linearity and cannot take negative values, but can otherwise be nontechnically read as a correlation coefficient.

The first hypothesis was that the reasonably apparent features of the situation would activate the primary goals. The recent and memorable situations were analyzed separately, since these data were not statistically independent. Results of the analyses are in Table 1.

OPEN IN VIEWER

Table 1.

Cramer’s Vs for Associations Between Reasonably Apparent Features and Primary Goals, Study 1.

	Recent situation	Memorable situation
Own sex	ns	ns
Other’s sex	.24***	ns
Number present	.24***	.16*
Medium	.17***	.15**
Setting	.28***	.32**
Relationship type	.29***	.34***
Interaction type	.41***	.39***
Topic	.31***	.29***

*

p < .05. **p < .01. ***p < .001.

For the recent situations, primary goals appeared in the following proportions: relational maintenance (50%), fulfill role (11%), manage own face (3%), catharsis (2%), obtain benefit (20%), plan future event (4%), and other (0.4%). For the memorable situations, the distribution was relational maintenance (39%), fulfill role (6%), manage own face (3%), catharsis (4%), obtain benefit (31%), plan future event (2%), and other (1%). Table 1 shows that most of the reasonably apparent features of the situation had significant associations with the distributions of primary goals. The only exceptions were own sex and, for memorable situations, other’s sex. The other features were associated with the primary goal, sometimes at substantial effect sizes. The most important associations were with setting, relationship type, interaction type, and topic. Interestingly, the patterns for recent and memorable situations were quite similar. In sum, the data were consistent with H1.

The second hypothesis was that the primary goals would cause the subjective situational features. The testing procedures were parallel to those for H1. Results are in Table 2.

OPEN IN VIEWER

Table 2.

Cramer’s Vs for Associations Between Primary Goals and Subjective Situation Features, Study 1.

	Recent situations	Memorable situations
Politeness	ns	ns
Rudeness	ns	ns
Argumentative	ns	.19**
Conflict	ns	.18*
Obstacle present	ns	.15*
Emotion–self	ns	.17**
Emotion–other	.16**	ns
Omission reasons	ns	ns

*

p < .05. **p < .01.

Results indicated essentially no support for the hypothesis for recent situations. For memorable situations, however, the primary goal in play was associated with several subjective elements: whether the situation contained arguments, whether it was conflictive, the presence or absence of obstacles, and own emotions. The size of these effects was rather modest. The two emotional measures showed an interesting contrast. For recent situations, the primary goal was associated with own emotions but not other’s; for memorable situations, this gross result was reversed. These results essentially falsify H2 for recent situations but offer support in the case of memorable situations. H3 was that the subjective features of situations would cause the secondary goals. The same analytical strategy was used as for the first two hypotheses. Results are in Table 3.

OPEN IN VIEWER

Table 3.

Cramer’s Vs for Associations Between Subjective Features and Secondary Goals, Study 1.

	% present	Polite	Rude	Argument	Conflict	Obstacles	Emotion–self	Emotion–other	Omissions
Recent situations
% present		10	3	25	15	47	14	11	51
Relational maintenance	81	.13*	ns	ns	ns	ns	ns	ns	ns
Organizational role	15	ns	ns	ns	ns	ns	ns	ns	ns
Own face	40	.21***	.14**	.11*	.17***	.24***	ns	ns	ns
Catharsis	4	ns	.10*	ns	ns	ns	.27***	.24***	ns
Benefit	40	ns	ns	ns	.09*	ns	ns	ns	ns
Plan future	17	ns	ns	ns	.12**	ns	ns	.17**	.09*
Memorable situations
% present		9	9	37	28	48	32	22	50
Relational maintenance	73	.09*	ns	.14**	.13**	.30***	ns	ns	ns
Organizational role	10	ns	ns	ns	ns	ns	ns	ns	ns
Own face	41	.22***	.15***	.20***	.26***	.36***	.22**	ns	.32***
Catharsis	10	.11*	ns	ns	ns	.18*	.24***	ns	.15**
Benefit	50	ns	ns	.13**	.16***	.20*	.19*	.20*	ns
Plan future	11	.10*	ns	ns	ns	ns	ns	ns	ns

*

p < .05. **p < .01. ***p < .001.

Many of these χ² tests were statistically nonsignificant, raising the question of experiment-wise error rates. Table 3 reports 96 tests, and 5 significant results would be expected by chance. In fact, 32 were significant at p < .05 or better. This justifies interpreting the patterns, although it is already obvious that the general effects were not consistent or especially strong. Whether effects occurred or not depended noticeably on which secondary goals were being predicted. Table 3 also reports the percentage that the various subjective features and secondary goals were coded as being present, and this has some bearing on the representativeness of the χ² tests.

For the recent situations, analysis found no associations between subjective features and the goal of keeping up one’s organizational role, and only one significant association for the goals of relational maintenance and obtaining benefit. The most consistent results appeared for maintaining one’s own face, catharsis, and planning future events. The effect sizes were all small or modest. Two of the most common secondary goals (relational maintenance and obtaining benefit) were poorly predicted. The most common subjective feature, reasons for omission, was only predictive for the future planning secondary goal. Therefore, for recent situations, there was technical support for the hypothesis that subjective situational features are associated with the formation of secondary goals, but the effects were inconsistent and weak.

The memorable situations showed a somewhat more supportive pattern. The goal of maintaining one’s organizational role was again completely unassociated with subjective features, as was planning for future events. However, there were good predictive patterns for the goals of relational maintenance, maintaining own face, and obtaining benefits, and several associated predictors for the catharsis goal. The weakest predictions were for the three least-frequent goals (maintaining organizational role, planning future events, and perhaps catharsis). In general, the most common subjective features were also the ones most often involved in associations with the secondary goals. If the lower part of Table 3 had been restricted to the commonly occurring codes, overall results would have been almost uniformly supportive of H3. For memorable situations, the support for the hypothesis is more encouraging than for recent situations.

Although H3 was supported at a gross level, the evidence was uneven. Several goals were unassociated (or nearly so) with subjective situational features in one or the other types of situation. However, all of the subjective features were associated with several of the secondary goals. Patterns of association were somewhat similar when recent and memorable situations are compared. In the case of memorable situations, the most commonly encountered elements were usually associated, but this was not the case for recent situations.

Study 2

The second study pursues the same theory and hypotheses as displayed in Figure 1 but with a different design, an experimental one using closed-ended instruments. The cell frequencies for the reasonably apparent situation features were roughly equalized by random assignment. Respondents in Study 2 were then asked to make their own ratings of various goals and subjective situation characteristics, generating systematic data to test relationships that nearly evaporated in Study 1’s cross-tabulations because of small cell frequencies. Use of continuous data for many of the measures permitted structural equation modeling, which moves closer to being able to make the causal claims that the theory proposes. Several of the coding systems developed in Study 1 had to be shrunk in order to make statistical analysis feasible, even with the substantial sample size used here.

Method

Respondents and procedures

Data were provided online by 1,681 students enrolled in communication classes at a large public university in the Mid-Atlantic region of the United States. They received minor extra credit for their participation. Because of missing data, confirmatory factor analyses and structural equation modeling were conducted on a reduced data set of 1,643. The missing data were likely lost due to Internet connectivity problems.

Participants were largely female (65%), were often freshmen (40%), had an average age of 19.5 years (SD = 1.90), and most often self-reported that they were Euro-American (51%), followed by African American and Asian American at 10% each, Asian at 6%, and Hispanic American at 5%, with the remainder scattered among various combinations or national origins. Other’s sex was fairly evenly divided, with 51% male.

Respondents were randomly assigned to conditions designed to implement a reduced set of the reasonably apparent features uncovered in Study 1. Conditions included four topics (academic, relational, workplace, and daily activities), four places (home, before class, at work, and walking), and four relationships (good friend, parent, instructor, and romantic partner). Own sex and other’s sex were self-reported. This was a 4 × 4 × 4 × 2 × 2 design with 256 cells. The sample size was selected so as to provide about 20 respondents for each cell due to the manipulated conditions (i.e., 64 cells).

Stimuli

The situational stimuli were based on the Study 1 corpus. They were revised as necessary to fit the conditions. For example (all stimuli and instruments are available from the author), here is the relational topic stimulus, with its variations:

You are about a month into a new romantic relationship. The two of you have been out together or in a small group of friends about half a dozen times, and it is clear to both of you that you are dating. Things have been progressing modestly and there has been no substantial physical intimacy. You are finding yourself more and more committed to the relationship, but you are concerned that the other person seems to have wandering eyes and seems (you’re not entirely sure about this) to occasionally flirt a little bit with other people.

You are [at home/sitting in class before it begins/at your part-time job/walking in between classroom buildings]. The other person in the conversation is [a good friend/one of your parents, who has just called on the phone/an instructor, who is on a private internet chat set up for a class with you on your very capable cell phone/your romantic partner, who has just run into you] and has just asked how you are doing with life in general. Because it’s on your mind, you mention your relationship concerns.

Participants were asked to pause and imagine the conversation that would follow. After reading the stimulus (each respondent was exposed to only one), students were asked to supply information about own and other’s sex, and own age, ethnicity, and year in school.

Instrumentation

Table 4 contains descriptive statistics for the multi-item scales, including Cronbach’s alphas. Reliabilities were acceptable in all cases. Means in Table 4 were calculated by averaging scores on individual indicators. Several items asked for an assessment of the stimuli’s realism, how personally involving it was, and whether or not the respondent could think of an actual person or real situation that matched the stimulus. Scores on these measures indicate that the stimuli were reasonably familiar and realistic.

OPEN IN VIEWER

Table 4.

Reliabilities and Descriptive Statistics for Multi-Item Scales, Study 2.

	n	Cronbach’s α	No. of items	M	SD
Situation characterizations
Realistic	1,681	.88	5	6.48	2.03
Involving	1,681	.90	5	6.12	2.01
Thought of real	1,681	.77	2	6.45	2.56
Goals
Relational maintenance	1,681	.89	5	6.50	2.05
Role fulfillment	1,681	.93	5	6.90	1.84
Own facework	1,681	.88	5	7.52	1.67
Catharsis	1,681	.91	5	7.30	1.77
Obtain a benefit	1,681	.91	5	6.39	1.93
Plan future actions	1,681	.94	5	7.00	1.96
Secondary features
Polite	1,645	.95	5	7.25	1.88
Rude	1,645	.92	5	3.64	2.16
Argumentative	1,645	.86	5	7.04	1.65
Conflictive	1,645	.91	5	4.20	2.20
Obstacles salient	1,645	.86	5	5.58	2.19
Editing done	1,645	.92	5	5.68	2.31
Self emotions
Anger	1,645	.91	5	4.77	2.31
Anxiety	1,645	.91	5	5.64	2.34
Disappointed	1,645	.84	5	4.20	2.10
Happy	1,645	.91	5	5.77	2.03
Sad	1,645	.86	5	3.64	1.96
Surprised	1,645	.94	5	3.83	2.17
Guilt	1,645	.94	5	3.37	2.13
Fear	1,645	.90	5	4.16	2.22
Grateful	1,645	.84	5	5.58	1.96
Bored	1,645	.93	5	3.22	2.04
Other’s emotions
Anger	1,645	.97	5	3.80	2.53
Anxiety	1,645	.92	5	4.22	2.27
Disappointed	1,645	.90	5	3.64	2.11
Happy	1,645	.92	5	5.41	2.08
Sad	1,645	.90	5	3.70	2.03
Surprised	1,645	.93	5	4.17	2.19
Guilt	1,645	.94	5	3.56	2.18
Fear	1,645	.91	5	3.86	2.08
Grateful	1,645	.86	5	4.85	1.97
Bored	1,645	.94	5	3.34	2.07

Note. All items were measured on 1-10 scales. Means and standard deviations are also on a 1-10 scale.

Subjective features of the situation were assessed in three general ways: some overall characteristics of the imagined conversation, own emotional reactions to the conversation, and expectations for other’s emotional reactions.

The general features were assessed on a 1-10 scale ranging from clearly absent to clearly present. These features included both politeness (“The conversation as a whole was conducted politely”) and rudeness (“At least one of us was inconsiderate of the feelings of the other person”) because a conversation can have both or neither of these characteristics. Two other general features were argumentative (“At least one person gave his/her reasons for what she/he said or wanted”) and conflictive (“We disagreed explicitly rather than just implying disagreement”). These are conceptually distinguished in that the argumentative scales referred to reasoning and evidence, while the conflictive items focused on disagreement and goal incompatibility. The difficulty of the conversation was reflected in the last two general features, obstacles (“I wanted to avoid creating or escalating any arguments”) and omissions (“I was carefully editing what I was saying”).

The same list of discrete emotions was used for self and other. As with the items for the general features, these were commonly noticed in the Study 1 results. The same 1-10 scale (clearly absent to clearly present) was used for the emotions. The first was anger (“I was mad”), followed by anxiety (“I felt stressed”), disappointment (“I was disappointed in myself”), happiness (“I was happy with what I thought was going to happen”), sadness (“The other person made me feel sad”), surprise (“The conversation took a turn I wasn’t expecting”), guilt (“I feel some shame about what I did or said during the episode”), fear (“I was worried during the interaction”), grateful (“I am appreciative that I was in this conversation”), and boredom (“My mind was wandering throughout the conversation”). Items for the other person were suitably reworded.

The primary goal was measured with a single item. The stem was,

Usually a conversation is essentially defined by an overall (or framing) goal. It may not be the most important goal in the conversation, but it is the one that essentially characterizes the interaction. Please choose the one goal in the list below that was the defining goal for you in the conversation.

Respondents then chose one of “maintaining the nature of your relationship with the other person,” “fulfilling your role obligations,” “protecting or projecting your own identity (i.e., how you want to be viewed by others),” “expressing your feelings, thoughts, or urges,” “obtaining some benefit or advantage,” or “planning some future event.” The primary goals were endorsed with these frequencies: relational maintenance (16.5%), fulfill role obligations (6.5%), own facework (10.5%), self expression (58.9%), obtain benefit (4.9%), and plan future events (2.8%). Later statistical issues sometimes necessitated combining the last two goals into one, called practical aims (7.7%).

The same list was used to assess secondary goals, but this time with multi-item measures using a 1-10 scale ranging from not important at all to extremely important. Each goal’s items were preceded by a brief orientation. For relational maintenance, the preface was “Consider the relationship you have with the other person in this interaction,” and a sample item was “improving the quality of the relationship between me and the other person.” Role obligation was introduced by “In most circumstances, people have a particular role—maybe as a subordinate, as a student, as a friend, as a romantic partner, or something else. Consider what your role would be in this situation,” and one of the items was “acting in the way that I am supposed to in this sort of circumstance.” Own facework was explained by “People often consider how their actions reflect on their identity, either what they think of themselves or what others might think of them,” and an example item was “acting in accord with what I really value about myself.” Self expression was introduced by “Sometimes, people speak mainly to express their feelings,” and one item was “giving voice to my feelings.” Obtain benefit items were preceded by “Sometimes people engage in a conversation in order to accomplish some particular aim—to get a favor, to make an agreement, or to obtain some substantial benefit,” and one of the items was “finding a way to put myself in an advantageous position.” Finally, plan future event was introduced with “Many conversations are oriented to planning some future event with another person,” and an item was “proposing (or responding to) a plan for some future activity.”

A confirmatory factor analysis (using LISREL 8.8 and maximum likelihood estimation, as was also the case with the structural equation models reported later) was conducted to assess the measurement model for all these multi-item instruments combined. Results were χ²(12,064) = 46,999.26, p < .001; root mean square error of approximation (RMSEA) = .052 (90% confidence interval [CI] [.051, .052]); comparative fit index (CFI) = .98; standardized root mean square residual (SRMR) = .061. The significant χ² is interpreted as being due to the large sample size, and the other fit indices are judged to be satisfactory. No indicators were dropped, no correlations among the indicators’ errors were permitted, and no cross-loadings were freed. Inspection of the indicators’ loadings on their latent variables showed consistent and strong relationships, with R²s almost always above .50.

Results

The theory diagrammed in Figure 1 was tested in two stages. The first substantive question was whether the relatively apparent elements of a communication situation dictated participants’ primary goals, the first link in Figure 1. Here, the relevant situational descriptors were topic, place, relationship with other, own sex, and other’s sex. The framing goals were relational maintenance, managing own face and identity, fulfilling role obligations, personal expression, obtaining some personal benefit, and planning some future event. All of these were nominal variables, and so the appropriate analysis was multinomial logistic regression, with primary goal as the dependent measure. Initial analyses indicated that a substantial number of cells had zero frequencies, so the obtain benefit and plan future event frames were combined into one called practical aims to alleviate this analytic problem.

Setting personal expression as the default category of framing goal because it was most common, logistic regressions were conducted with five-way, four-way, three-way, and two-way interactions, as well as with main effects only. The four-way and five-way analyses did not produce significant results when the models were compared with an intercept-only model, χ²(912) = 708.445, p = 1.00; and χ²(1,000) = 727.563, p = 1.00, respectively. The model including all three-way and lesser effects did produce a significant result, χ²(588) = 950.743, p < .001, but this occurred in the context of quasi-complete separation of the data, making results untrustworthy. These models were rejected.

The choice between a main-effects and a two-way interaction model was not so clear-cut. Both were significant improvements over the intercept-only model, χ²(44) = 218.517, p < .001; and χ²(228) = 438.255, p < .001, respectively. The two-way interaction model gave a significantly better fit to the data than did the main-effects model, Δχ² (184) = 219.738, p < .05. Overall, the topic × relationship and relationship × own sex interactions were significant, and the relationship × other’s sex nearly so (p = .054). However, inspection of the detailed effects (e.g., predicting the relational maintenance goal in the context of a self male and other close friend condition) showed that only 12 of the 228 nonredundant tests of interactions were statistically significant, a rate of only 5%. Consequently, the two-way interaction model was rejected, and the main-effects-only model was judged the best.

Besides the overall significant fit mentioned above, the main-effects-only model produced a Nagelkerke’s pseudo-R² of .136, indicating that the main-effects model is about 14% better than the intercept-only model. Place did not have a significant main effect, χ²(12) = 8.033, p = .783, but the other relatively apparent situation variables did: topic, χ²(12) = 119.724, p < .001; relationship, χ²(12) = 44.729, p < .001; own sex, χ²(4) = 27.997, p < .001; and other’s sex, χ²(4) = 15.633, p < .01. Of the 44 cell-level significance tests, 11 were significant, a rate of 25%.

The other goals were each analyzed from the perspective of the self expression goal, which had the largest frequencies. So a significant cell effect for another goal means that the result is significantly different from that for self expression. Details of the relevant cross-tabulations are in Table 5. The tests reported in text are from the multinomial logistic regressions.

OPEN IN VIEWER

Table 5.

Framing Goals by Topics, Relationships, Own Sex and Other’s Sex, Frequencies and (Percentages), Study 2.

	Academic	Relational	Workplace	Daily activity
Relational maintenance	38 (8.7)	99 (24.9)	51 (12.5)	91 (20.8)
Fulfill role obligations	30 (6.8)	19 (4.8)	33 (8.1)	27 (6.2)
Face identity work	38 (8.7)	25 (6.3)	84 (20.6)	28 (6.4)
Self expression	272 (62.1)	231 (58.2)	223 (54.7)	261 (59.6)
Practical aims	60 (13.7)	23 (5.8)	17 (4.2)	31 (7.1)

Note. Cramer’s V = .16, p < .001.

OPEN IN VIEWER

	Good friend	Parent	Instructor	Romantic
Relational maintenance	57 (13.6)	80 (18.1)	72 (17.7)	70 (16.9)
Fulfill role obligations	32 (7.7)	31 (7.0)	24 (5.9)	22 (5.3)
Face identity work	36 (8.6)	49 (11.1)	46 (11.3)	44 (10.6)
Self expression	263 (62.9)	248 (56.2)	210 (51.6)	266 (64.1)
Practical aims	30 (7.2)	33 (7.5)	55 (13.5)	13 (3.1)

Note. Cramer’s V = .09, p < .001.

OPEN IN VIEWER

	Self male	Self female
Relational maintenance	119 (20.4)	152 (14.3)
Fulfill role obligations	50 (8.6)	56 (5.3)
Face identity work	72 (12.4)	100 (9.4)
Self expression	295 (50.7)	673 (63.4)
Practical aims	46 (7.9)	80 (7.5)

Note. Cramer’s V = .13, p < .001.

OPEN IN VIEWER

	Other male	Other female
Relational maintenance	167 (19.5)	112 (13.6)
Fulfill role obligations	51 (6.0)	58 (7.0)
Face identity work	83 (9.7)	92 (11.2)
Self expression	476 (55.5)	511 (62.0)
Practical aims	80 (9.3)	51 (6.2)

Note. Cramer’s V = .11, p < .001.

Table 5 shows that the relational maintenance goal was particularly affected by topic, own sex, and other’s sex. Academic topics were especially unlikely to stimulate the relational maintenance goal (b = −.79, p < .001), but it was more salient when either the respondent (b = .58, p < .001) or the other (b = .45, p < .01) was female. Interactions were also more likely to be framed as fulfilling one’s role obligations when respondents were female (b = .72, p < .001). Facework in support of one’s own identity was especially likely in the context of workplace conversation (b = 1.27, p < .001) and for respondent women (b = .51, p < .001). Finally, situations were framed as having practical aims (obtaining a benefit or planning a future event) when the topic was academic (b = .76, p < .01), when the other person was a good friend (b = 1.02, p < .01), a parent (b = 1.18, p < .001), or an instructor (b = 1.78, p < .001), and when the other person was female (b = .56, p < .01).

These results together support H1, the expectation that the relatively apparent features of a situation affect the primary framing goal for it. This offers support for the first step in the theorized model of goals and situations (see Figure 1).

The next major question, combining H2 and H3, was how the remaining three panels of variables (framing goal, subjective situational features, and secondary goals) were related (see Figure 1). Two models were tested. The first expressed the two hypotheses: framing goal → subjective features → secondary goals. The alternative model reversed the order of subjective features and secondary goals: framing goal → secondary goals → subjective features. In both cases, modification indices suggested that errors among each panel be freed to correlate (i.e., all the secondary goals’ errors were freed among themselves, and all the subjective features’ errors were freed within that panel). This was theoretically plausible (i.e., the goals probably relate to one another, and the situational features—for example, politeness and rudeness—ought to covary as well), and so those error covariances were freed.

Results indicated a clear preference for the hypothesized sequence: framing goals → subjective features → secondary goals. For that model, χ²(12,600) = 47,806.59, p < .001; RMSEA = .051 (90% CI [.050, .051]); CFI = .98; SRMR = .060; Akaike information criterion (AIC) = 67,652.73. Results for the alternate model, framing goals → secondary goals → subjective features, were χ²(12,680) = 47,877.68, p < .001; RMSEA = .051 (90% CI [.050, .051]); CFI = .98; SRMR = .061; AIC = 67,880.06. The fit indices for the two models were comparable. Decisively, the AIC index shows that the first model was quite clearly better (a difference of 10 or more is a strong decision criterion; Burnham & Anderson, 2004).

For the preferred model, modification indices suggested adding these paths between the framing goals and the secondary goals: from the relational maintenance frame to relational maintenance and self expression; from the role obligation frame to self expression and planning future events; from the own facework frame to relational maintenance, role obligation, and self expression; and from the practical aims frame to self expression and seeking benefits. These direct paths indicate that the framing goals’ effects on the secondary goals were not completely mediated by the particular list of subjective situational features studied here. Freeing these relationships resulted in minor improvements in the fit indices, although the AIC measure indicates that this version of the model was better: χ²(12,591) = 47,627.43, p < .001; RMSEA = .051 (90% CI [.050, .051]); CFI = .98; SRMR = .060; AIC = 67,439.66. Most of the R² results for the endogenous variables were also unchanged. Nonetheless, this is the final model for the framing goals → subjective features → secondary goals relationship system.

Because of the number of variables, drawing a diagram of the model is unhelpful. Instead, Tables 6 and 7 show the significant paths (the nonsignificant paths were also included in the model fit results given above).

OPEN IN VIEWER

Table 6.

Significant Paths into Endogenous Variables in the Framing Goals → Subjective Features → Secondary Goals Model: Part I, Effects of Framing Goals, Study 2.

	Unstandardized path coefficient	Predictor
Secondary goals
Relational maintenance	.42	Relational frame
Role obligation	—
Own face	—
Self expression	−.35	Relational frame
	−.50	Role obligation frame
	−.46	Own face frame
	−.48	Practical aims frame
Seek benefits	.36	Practical aims frame
Plan future events	—
Subjective situational features
Polite	−.27	Relational frame
	−.63	Role obligation frame
	−.88	Own face frame
	−.20	Practical aims frame
Rude	.31	Relational frame
	.68	Role obligation frame
	.82	Own face frame
Argumentative	−.43	Relational frame
	−.60	Role obligation frame
	−.35	Own face frame
Conflictive	.31	Relational frame
	.54	Role obligation frame
	.64	Own face frame
Obstacles	.18	Relational frame
	.33	Own face frame
	.24	Practical aims frame
Omissions	.18	Relational frame
	.18	Own face frame
Own emotions
Anger	.41	Own face frame
	−.22	Practical aims frame
Anxiety	−.16	Relational frame
	−.23	Role obligation frame
Dissatisfied	.32	Relational frame
	.40	Role obligation frame
	.59	Own face frame
Happy	−.19	Relational frame
	−.32	Role obligation frame
	−.70	Own face frame
Sad	.24	Relational frame
	.44	Role obligation frame
	.50	Own face frame
Surprised	.27	Relational frame
	.44	Role obligation frame
	.48	Own face frame
	.20	Practical aims frame
Guilty	.23	Relational frame
	.42	Role obligation frame
	.53	Own face frame
Fearful	.19	Relational frame
	.28	Role obligation frame
	.37	Own face frame
Grateful	−.32	Relational frame
	−.47	Role obligation frame
	−.51	Own face frame
Bored	.26	Relational frame
	.82	Role obligation frame
	.47	Own face frame
	.26	Practical aims frame
Other’s emotions
Anger	.26	Relational frame
	.48	Role obligation frame
	.67	Own face frame
Anxiety	.24	Relational frame
	.39	Role obligation frame
	.44	Own face frame
Dissatisfied	.34	Relational frame
	.56	Role obligation frame
	.51	Own face frame
Happy	−.36	Own face frame
Sad	.27	Relational frame
	.45	Role obligation frame
	.44	Own face frame
Surprised	.21	Relational frame
	.34	Role obligation frame
	.50	Own face frame
Guilty	.32	Relational frame
	.59	Role obligation frame
	.54	Own face frame
Fearful	.26	Relational frame
	.45	Role obligation frame
	.34	Own face frame
Grateful	−.38	Own face frame
Bored	.18	Relational frame
	.53	Role obligation frame
	.35	Own face frame
	.20	Practical aims frame

OPEN IN VIEWER

Table 7.

Significant Paths Into Endogenous Variables in the Framing Goals → Subjective Features → Secondary Goals Model: Part II, Effects of Subjective Features on Secondary Goals, Study 2.

	Unstandardized path coefficient	Predictor
Relational maintenance	.37	Polite
	.13	Rude
	.20	Obstacles
	−.07	Anger (self)
	.17	Disappointed (self)
	−.19	Bored (self)
	.13	Grateful (other)
Role obligation	.40	Polite
	.17	Rude
	.14	Argumentative
	.18	Obstacles
	.21	Disappointed (self)
	−.10	Surprised (other)
Own face	.38	Polite
	.21	Rude
	.12	Argumentative
	−.12	Conflictive
	.29	Obstacles
	.09	Omissions
	.22	Disappointed (self)
	−.08	Bored (self)
Self expression	.29	Polite
	.16	Rude
	.20	Argumentative
	.10	Obstacles
	−.10	Omissions
	.10	Anger (self)
	.14	Grateful (self)
	−.17	Bored (self)
	−.12	Grateful (other)
Obtain benefits	.17	Polite
	.15	Argumentative
	.19	Obstacles
	.10	Happy (self)
Plan future events	.32	Polite
	.16	Argumentative
	.19	Obstacles
	.27	Disappointed (self)
	−.08	Bored (self)
	.13	Anger (other)
	−.14	Disappointed (other)
	.09	Grateful (other)

Table 6, which shows the effects of the primary goals, makes clear that every subjective situational feature was predicted by the framing goals. The practical aims frame was least often involved in predicting the subjective situational features, but even that goal had a statistically significant path for 6 of the 26 subjective measures. The ubiquity of the role obligations frame in these predictions may be worthy of note, since this goal is so rarely studied.

Table 7 shows the effects of the subjective features in forming the secondary goals. Every secondary goal was predicted by a variety of subjective situational elements. The most common predictors included the following characteristics: polite, rude, argumentative, and obstacles. The discrete emotions were involved in predicting every secondary goal, but their record was rather scattered. The most commonly relevant self emotions were disappointment and boredom. The most commonly relevant emotion perceived in one’s partner was grateful to have been in the conversation, but it must immediately be mentioned that other’s emotions were less important than own feelings, and were entirely powerless to predict two of the secondary goals (own face and obtain benefits).

In addition, Table 8 shows the R² values for the endogenous variables. A noticeable result in Table 8 is that the subjective features were not impressively predicted. The stronger statistical relationships in this data set were between the subjective situational features panel and the secondary goals panel.

OPEN IN VIEWER

Table 8.

R² for Endogenous Variables in the Framing Goal → Subjective Features → Secondary Goals Model, Study 2.

	R 2	R 2
Secondary goals
Relational maintenance	.29
Role obligations	.30
Own face	.34
Catharsis	.36
Obtain benefits	.19
Planning future events	.25
Subjective situational features
Polite	.08
Rude	.08
Argumentative	.04
Conflictive	.05
Obstacles	.01
Omissions	.01
Emotions	Own	Other’s
Anger	.02	.05
Anxiety	.01	.03
Dissatisfaction	.04	.04
Happy	.05	.01
Sad	.03	.03
Surprised	.03	.03
Guilty	.03	.05
Fearful	.02	.03
Grateful	.04	.01
Bored	.06	.03

General Discussion

The theory of goals and situations displayed in Figure 1 was consistent with the results of two studies using different methodologies, both having data from substantial numbers of U.S. undergraduates. Figure 1 shows a temporally simple model in which reasonably apparent features of the situation stimulate a framing goal, which leads to subjective detailing of the situation, which in turn activates an array of secondary goals, each with variable importance in the episode.

This model improves on the current theoretical conceptions of goals and situations. Standard literature in the field turned out to be expressing two different views that were not obviously consistent with one another, even though both made sense. One theoretical position was that situations stimulate goals, and the other view was that one’s complex of salient goals essentially defined the situation in its important respects. Considered in concert, these views led to a causal confusion in which situations cause goals (the first view) and in which goals constitute situations (the second view). By distinguishing between relatively apparent (perceptually uncontroversial) and subjective (emergent) features of situations, the present theory was able to show in what respects situations stimulate goals and in what respects goals give rise to situation characteristics. Both studies supported the new theory, which seems to be substantially clarifying.

Figure 1 is offered as a beginning template for basic theory and more specialized investigations of goals and situations. More pointed and focused designs might very well justify qualifications.

For instance, Study 2 found that place had no effects on framing goals. Place simply was not salient to respondents as they worked through the items asking for details about the character of the conversation. It is hard to imagine that places such as confessional, courtroom, or bedroom would have no effect on one’s goals or feelings, however. And for a particular topic, a person planning an interaction might well choose an appropriate place to initiate the conversation. When a conversation is carefully planned, we might suppose that either the topic or the framing goal could initiate the anticipatory thoughts, resulting in either something close to Figure 1 (when topic begins the planned encounter) or something contradictory to it (when the framing goal, such as expressing resentment, results in topics being wildly called into play or chosen arbitrarily).

Both studies might also be reasonably read as dealing with interactions that were fairly static and well established in their outlines. In Study 1, the episodes were recalled in whole, reflecting respondents’ settled memories of those little bits of personal history. In Study 2, a number of situational elements were stipulated in the stimulus materials. Some conversations, especially the routine ones, are like this. But others might involve considerable uncertainty, with participants trying out interpretive hypotheses, revising their sense-makings, changing intentions and directions, and otherwise trying to navigate the immediate social world without a clear map. In those circumstances, we might expect to see revision and feedback loops among the elements of Figure 1 rather than a clear linear development. Hample and Dallinger (1998), for instance, have shown how message production goals change as an interaction becomes more frustrating over time.

In sum, the present theory is offered as having two virtues. First, it provides a more coherent understanding of the ways goals and situations are related, and makes it clear that a situation is something more than a label for the goals active during the interaction. These two studies include lists of relatively apparent and subjective situational features that are not goals, and that would have to be distorted in some measure to be regarded as motivations. Second, the model gives a conceptual baseline from which to study interesting sorts of interactions that are likely to require qualifications to the theory. Confession to a police officer, seductive whispers across a pillow, angry resignations at work, slow and sensitive negotiation of the beginning of a new romantic relationship—any of these could result in more delicate causal relations than have appeared in the present data. And any of them might necessitate revision of the lists of goals or features used here. The present data go beyond the realm of ordinary conversation—recall that half the data in Study 1 concerned memorable encounters—but the conversations were not specialized in the way that a hostage negotiation would be, for example. In fact, part of showing what is specialized about various kinds of conversation might exactly involve detailing how they depart from the general model presented here, which reflects fairly common and well-understood interpersonal interactions.