Assessing Dyadic Reciprocity in Triads

From Individual Tendencies to Dyadic Chemistry

In the last lesson, we defined generalized reciprocity as the correlation between a person's perceiver effect ($\alpha$) and their target effect ($\beta$). It speaks to a person's general social signature. Dyadic reciprocity, in contrast, is the correlation between the two relationship effects ($\gamma$) within a dyad. It answers a fundamentally different question:

• Generalized Reciprocity: Do people who generally feel a lot of guilt also tend to make their partners generally feel a lot of guilt?
• Dyadic Reciprocity: If the wife feels a level of guilt unique to her relationship with the bull (above and beyond her general tendency and the bull's general effect), does the bull feel a level of guilt unique to his relationship with her?

Dyadic reciprocity, therefore, isolates the mutuality of the relational component itself. It is the statistical operationalization of relationship-specific chemistry. For a clear academic definition, we can turn to the paper on stepfamilies we have used before.

Formalizing and Applying Dyadic Reciprocity

Given your background in statistics, we can formalize this concept precisely. Recall the SRM equation which decomposes an outcome $Y$ for person $i$ towards person $j$ in group $k$:

Here, ${\gamma }_{ijk}$ is the relationship effect—the part of the outcome that is unique to the $i\to j$ dyad. Dyadic reciprocity is simply the correlation between ${\gamma }_{ijk}$ and ${\gamma }_{jik}$. This correlation, often denoted as ${\rho }_{\gamma \gamma }$ or found in model outputs as the "relationship correlation," quantifies the mutuality of the unique dyadic bond.

The formal statistical model defines this as a covariance term within a bivariate normal distribution of the two relationship effects.

In your triadic study of a husband (H), wife (W), and bull (B), you have a complete "round-robin" data structure, where every person reports their emotional state regarding every other person.

This structure allows you to estimate three distinct dyadic reciprocity coefficients for each emotion you measure:

1. Husband-Wife Dyad (${\rho }_{{\gamma }_{HW},{\gamma }_{WH}}$): Is the husband's unique jealousy toward his wife mirrored by the wife's unique jealousy toward him? Or, is his unique guilt met with her unique guilt?
2. Wife-Bull Dyad (${\rho }_{{\gamma }_{WB},{\gamma }_{BW}}$): Is the wife's unique feeling of emotional connection to the bull reciprocated? A strong positive correlation here would be direct evidence of mutual "chemistry."
3. Husband-Bull Dyad (${\rho }_{{\gamma }_{HB},{\gamma }_{BH}}$): What is the nature of their unique dynamic? You might hypothesize zero or even negative reciprocity for an emotion like "comfort," where one person's unique ease might correspond to the other's unique unease.

Interpreting Dyadic Reciprocity Estimates

The value of this analysis lies in the interpretation of the resulting correlation coefficient. You will assess its sign, magnitude, and statistical significance.

• Positive ($\rho >0$): Indicates mutuality. What one person feels uniquely in the dyad, the other also tends to feel.
• Negative ($\rho <0$): Indicates a compensatory or inverse dynamic. What one feels, the other tends not to feel.
• Zero ($\rho \approx 0$): Indicates independence. The unique feelings within the dyad are not mutual.

Let's examine how to interpret these estimates from a model's output, using the speed-dating example from the SRM tutorial.

For a concrete example of a significant finding, the stepfamily study is again very useful.

A Practical Consideration for Your Research

Your study may involve dichotomous outcomes (e.g., "Was guilt present? Yes/No"). Estimating dyadic reciprocity for such variables presents a statistical challenge that is important to be aware of. Standard frequentist mixed-effects modeling packages often fail to converge or underestimate the effect. The modern solution involves a Bayesian approach.

Conclusion

Today, we have added a critical layer to our analysis of triadic dynamics. While generalized reciprocity speaks to individual roles, dyadic reciprocity allows us to peer into the unique, mutual "chemistry" of each relationship. By estimating this for the husband-wife, wife-bull, and husband-bull dyads, you can directly test which relationships are characterized by mutual feelings and which are not.

Key Takeaways:

• Dyadic reciprocity is the correlation between the two relationship effects in a dyad (${\rho }_{{\gamma }_{ij},{\gamma }_{ji}}$), capturing the mutuality of the unique relational dynamic.
• It allows you to test for "chemistry" by assessing whether one person's unique feelings toward a partner are reciprocated, after accounting for their individual dispositions.
• In your triadic study, you will estimate three separate dyadic reciprocity coefficients for each emotion, one for each pair (H-W, W-B, H-B).
• Interpretation focuses on the sign (mutuality vs. compensatory), magnitude, and statistical significance (does the CI exclude zero?) of the correlation.
• For binary outcomes, estimating dyadic reciprocity often requires a Bayesian modeling approach.

In our first three lessons on SRM, we have learned how to decompose emotional variance and how to measure two types of reciprocity. In the next lesson, we will synthesize these ideas by learning how to partition the total variance in an emotion across all SRM components. This will allow you to make a definitive statement about whether something like jealousy in the triad is primarily an individual-driven phenomenon or a relationship-driven one.