(:toc:) The PERSOC framework allows to investigate a multitude of different research questions. This includes univariate social relations analyses (How much do people differ with respect to perceiver/actor, target/partner, and relationship components? How much are these components interrelated within a variable), bivariate social relations analyses (How much are components interrelated between componential variables?) as well as correlational and multivariate analyses including dispositional variables (How much are dispositions related to perceiver/actor, target/partner, and relationship effects? How can these relations be explained).

To be able to analyze all of these research questions one needs a design in which (a) each perceiver perceives multiple targets / each actor interacts with multiple partners (polydiadic) and (b) each perceiver is also a target / each actor is also a partner (bidirectional). All easier designs have some shortcomings. Look for design considerations and statistical solutions for more informations and tools to get started.

Univariate social relations analyses

For each cue and interpersonal perception variable of interest, one examines the portions of variance that are referable to the perceiver/actor, target/partner, and relationship component (variance decomposition) as well as the interrelations of components within each variable (reciprocities).

Variance decomposition

As described in the concepts of our framework, three components make up each interpersonal perception and behavior. Consequently, the total variance in a given perception or behavior can be decomposed into three independent sources of variance: perceiver variance/ actor variance (e.g., How much do people differ in how they generally see others and in how they generally behave towards others?), target variance/ partner variance (e.g., How much do people differ in how others generally perceive them and in how others behave towards them?), and relationship variance (e.g., How much do people differ with respect to their unique perceptions and actions towards specific other persons?) with perceiver and target variance controlled for.

Results of variance decomposition give us a first interesting glimpse on our interpersonal perception or behavioral data. They are also an important prerequisite for subsequent analyses: Univariate correlations (reciprocities), bivariate correlations, and correlations with external variables should only be analyzed for those components that account for a significant amount of variance.

The amount of perceiver variance of interpersonal perceptions shows the use of different standards of perceivers in generally evaluating others. Analogously, actor variance of behavioral observations can be seen as general differences in how people behave. The amount of target variance of interpersonal perceptions indicates how much people differ in how they are seen by all others. The amount of behavioral partner variance indicates how much people consistently evoke different behaviors by interaction partners. Finally, the amount of relationship variance describes how much people differ in how they idiosyncratically judge and behave towards specific interaction partners.

Reciprocities

Within each variable one is often interested in the reciprocity of interpersonal phenomena. Interestingly, there are two different ways to analyze the reciprocity of interpersonal perceptions and behaviors.

At the individual level, generalized reciprocity can be computed by correlating the perceiver and target effects and actor and partner effects, respectively. For interpersonal perceptions, this indicates how much generally perceiving others in a certain way is correlated with being perceived in the same way. For behaviors, generalized reciprocity is about how much generally acting towards others in a certain way is correlated with others acting in the same way towards oneself.

Dyadic reciprocity can be computed by correlating relationship effects of interpersonal attraction within dyads. For interpersonal perceptions dyadic reciprocity shows how much seeing a specific other person in a particular way is related to being seen in the same way by this person. In the case of interpersonal behavior, dyadic reciprocity indicates how much uniquely behaving towards a particular person is related to this person uniquely behaving towards oneself.

Bivariate social relations analyses

When analyzing two componential variables (e.g., dominance behavior, mate choice) six possible correlations can be computed. On the individual level, there are four possible correlations between actor and partner effects of both variables. The actor-actor correlation indicates how much generally behaving dominant towards others is related to choosing others more frequently. The actor-partner correlation shows how much generally behaving dominant towards others is related to being chosen more frequently. The partner-actor correlation is a measure of the degree to which generally evoking dominant behavior is related to the frequency of choosing others. And the partner-partner correlation indicates how much generally evoking dominant behavior is related to being chosen more frequently. On the dyadic level there are two types of correlations between relationship components. The intraindividual relationship correlation is about how much uniquely behaving dominant towards a particular person is related to uniquely choosing this person. The interindividual relationship correlation indicates how much behaving dominant towards a specific person is related to uniquely being chosen by the same person.

Particularly interesting research questions can be analyzed when combining an interpersonal perception with the respective metaperception (e.g., liking and thinking to be liked). The perceiver-perceiver correlation indicates how much generally liking others is related to generally thinking to be liked by others (perceiver assumed reciprocity). The perceiver-target correlation shows how much generally liking others is related to generally being seen as a liker (perceiver meta-accuracy). The target-perceiver correlation is a measure of how much generally being liked is related to generally thinking to be liked (generalized meta-accuracy). The target-target correlation indicates how much generally being liked correlates with generally being seen as a liker (generalized assumed reciprocity). On the dyadic level, the intraindividual relationship correlation shows how much uniquely liking a specific other person is related to uniquely thinking to be liked by that person (dyadic assumed reciprocity). Finally, the interindividual relationship correlation is about how much uniquely thinking to be liked by a specific person is related to uniquely being liked by that person (dyadic meta-accuracy).

The preceding examples combined two behavioral or two perceptional measures, but of course any combination of perceptional and/or behavioral variables is possible. For example, when looking at the effect of smiling on perceived attractiveness, one would have an actor-perceiver correlation (Do smilers generally perceive others as attractive?), an actor-target correlation (Are smilers generally perceived as attractive?), a partner-perceiver correlation (Do people that others smile at a lot generally perceive others as attractive?), a partner-target correlation (Are people that others smile at a lot generally perceived as attractive?), an intraindividual relationship correlation (Do people uniquely perceive specific others as more attractive, they particularly smile at?), and an interindividual relationship correlation (Are people who uniquely smile at a particular person, uniquely perceived as attractive by this person?).

Correlational dispositional analyses

Dispositional measures (e.g., extraversion) can be related to all different kinds of perceptional (e.g., liking) and behavioral (e.g., smiling) components. On the individual level one might for example correlate extraversion with perceiver effects liking (Are extraverts likers?), target effects liking (Are extraverts liked more?), actor effects smiling (Do extraverts smile more?), and partner effects smiling (Do others smile more at extraverts?).

Specific relations/interactions between characteristics of individuals might moreover predict relationship effects of interpersonal perceptions or behaviors. Perhaps extraverts uniquely like smilers and as a consequence uniquely smile at them. As another example, unique liking might be explained by similar levels of openness (openness similarity) and a unique preference for diverse and intellectual conversational topics.

Complex multivariate analyses

When simultaneously considering multiple dispositional variables and perceptional and/or behavioral components an endless variety of more complex research questions can be examined by means of multivariate analyses. This might include the prediction of perceptional and/or behavioral components by multiple dispositional variables, explaining the effect of dispositions on interpersonal perceptions by behavioral or physical appearance mediators, the longitudinal course of interpersonal perception and behavioral components and its interplay with dispositions, dispositions as moderators of all kind of multivariate dynamics between interpersonal perception and behavioral components and so on.

In our Toolbox you find important design considerations and statistical solutions to help you get started.