Dissecting Emotions with the Social Relations Model

From Dyads to Groups: The Core Idea of SRM

Imagine you are measuring the husband's jealousy in response to a specific interaction between his wife and the bull. That jealousy score is a single data point, but it's a composite of several underlying factors. The Social Relations Model provides a statistical method to disentangle these factors.

At its heart, SRM posits that any interpersonal perception or feeling from person A towards person B can be broken down into three components:

1. The Perceiver Effect (or Actor Effect): This is person A's general tendency. Do they tend to feel high levels of jealousy, regardless of who the partner is? This component is about the perceiver's disposition. In your context, it's the husband's baseline level of jealousy.
2. The Target Effect (or Partner Effect): This is person B's general effect on others. Do they tend to elicit high levels of jealousy from others? This component is about the target's characteristics. It captures the degree to which the bull, for instance, is perceived as a threat or elicits jealousy in general.
3. The Relationship Effect: This is the unique, private feeling between A and B that isn't explained by their general tendencies. It’s the "chemistry" or specific dynamic of that dyad. For example, the specific jealousy this husband feels towards this particular bull might be higher (or lower) than what you'd expect based on the husband's general jealousy and the bull's general effect on others.

Given your background in Design of Experiments, you can think of this as a form of variance decomposition, analogous to partitioning the total variance in an outcome into main effects and interaction effects in an ANOVA. SRM partitions the total variance in interpersonal emotions into variance due to perceivers, variance due to targets, and variance due to relationships.

The image below provides a clear visual metaphor for this decomposition within a group.

Formalizing the Decomposition

To formalize this intuition, let's explore a foundational paper that applies SRM to a family triad. The context is different, but the statistical structure of a three-person group is identical to your research scenario.

As the reading highlights, for a three-person group (husband H, wife W, bull B), a full "round-robin" data collection would involve measuring six perceptions for any given emotion (e.g., "connection"):

• H's connection to W; H's connection to B
• W's connection to H; W's connection to B
• B's connection to H; B's connection to W

SRM takes these six scores (across many triads) and decomposes the variance. The underlying statistical model is quite elegant. For any perception from perceiver i towards target j ($Y_{ij}$), the model is:

Where:

• $\mu$ is the grand mean perception across all people.
• ${\alpha }_i$ is the perceiver effect for person i. It represents how much person i's perceptions deviate from the mean, on average.
• ${\beta }_j$ is the target effect for person j. It represents how much person j is perceived differently from the mean, on average.
• ${\gamma }_{ij}$ is the relationship effect, capturing the unique deviation of i's perception of j.
• ${ϵ}_{ij}$ is the measurement error for that specific perception.

The primary goal of SRM analysis is not to find the specific values of $\alpha$, $\beta$, and $\gamma$ for one triad, but to estimate the variances of these effects across a sample of many triads: ${\sigma }_{\alpha }^2$, ${\sigma }_{\beta }^2$, and ${\sigma }_{\gamma }^2$. The relative sizes of these variance components answer crucial questions. For example, if we are studying jealousy:

• If ${\sigma }_{\alpha }^2$ (perceiver variance) is large, it means most of the variation in jealousy comes from people's stable dispositions to be jealous.
• If ${\sigma }_{\beta }^2$ (target variance) is large, it means some individuals consistently elicit more jealousy than others.
• If ${\sigma }_{\gamma }^2$ (relationship variance) is large, it suggests that jealousy is highly specific to particular pairings, independent of the individuals' general tendencies.

An Intuitive Application

The following resource provides a very clear, verbal walkthrough of how these components play out in an applied setting. While the example is from speed-dating, the logic of decomposing behavior into eagerness (perceiver effect), attractiveness (target effect), and unique chemistry (relationship effect) is directly transferable to your research.

Let's apply this logic directly to your study. Suppose the outcome is the wife's feeling of guilt after an encounter. The SRM decomposition would be:

• Wife's Guilt =
  • Grand Mean (average guilt level for all wives in the study)
  • + Wife's Perceiver Effect (her personal tendency to feel guilt)
  • + Bull's Target Effect (his tendency to elicit guilt in partners)
  • + Wife-Bull Relationship Effect (the unique guilt dynamic in this specific pairing)

By estimating the variance of each component, you can quantitatively determine whether the wife's guilt is driven more by her own disposition, the bull's specific behavior/attributes, or the unobserved "chemistry" of their particular dyad.

Conclusion

Today's lesson shifted our perspective from isolated dyads to the triad as an interconnected system. The Social Relations Model provides the precise methodological tool to dissect the complex emotional data that arises from such a group.

Key Takeaways:

• The Social Relations Model (SRM) decomposes interpersonal perceptions (e.g., emotions, attitudes) from a round-robin design into three main components.
• Perceiver (Actor) Effect: An individual's general tendency to perceive or feel a certain way towards others.
• Target (Partner) Effect: An individual's general tendency to be perceived in a certain way by others.
• Relationship Effect: The unique dynamic between a specific perceiver and target that is not explained by their general tendencies.
• SRM estimates the variance attributable to each of these components, allowing you to determine the primary sources of variation in emotions like jealousy or guilt within the triad.

You are now able to conceptualize how an emotional response can be partitioned into these distinct sources. This lays the groundwork for a more sophisticated analysis. In our next lesson, we will build upon this foundation by exploring the correlations between these components, introducing the concepts of generalized reciprocity (do people who feel strong emotions also tend to elicit them?) and dyadic reciprocity (are unique feelings mutual?).