Social Relations Research Designs

Research Designs used in Social Relations Modeling

To estimate the parameters of the social relations model, specialized research designs are required.   A detailed discussion of those designs is presented in chapter 3 of Malloy (2018) Below is a brief summary of the research designs used to estimate the parameters of the model.

Round Robin

Round Robin Design with Dyadic and Self Data

Note.   S is a response to the self and X is a dyadic response.  Diagonal elements are not required.

Symmetric Block Design

Actors are on the rows and Partners are on the columns.

S is a response to the self and X is a dyadic response.

Asymmetric Block Design

M is a mother and T is a toddler.  This design is particularly useful when studying intergroup responses.  In the example above, the two groups are mothers (M) and toddlers (T).

For more details go to:

Half-Block and One-with-Many Designs

An X is a dyadic response.

See Chapter 3 of Malloy (2018) for more detail.


Key Person Design with embedded Round Robins

Generations Design: Five Generation Example

The generation design presented below was inspired by that used in an ingenious experiment studying the “perpetuation of an arbitrary tradition” (Jacobs & Campbell, 1961) regarding the autokinetic effect (the perception that a point of light moves when viewed in a dark room) studied by Muzafer Sherif (1936).  In that study, individuals in a control condition viewed a point of light while isolated from others and estimated the length of its movement in inches; the mean estimate was 3.8 inches in this condition.   In experimental conditions, naïve participants made movement judgments after confederates (whose behavior was controlled by the researchers) made judgments that the point of light moved between 15 and 16 inches.  Across different experimental conditions with different numbers of confederates, the point of light was judged as moving 15.5 inches by naïve participants.

As confederates were replaced by real unscripted participants, after 4 to 5 subsequent “generations” were produced, the norm regarding the movement of the point of light maintained.  Eventually, after entirely new generations with unscripted participants, the arbitrary norm that the light moved 15.5 inches diminished and approached that observed in the control condition where there was no confederate involvement.

A variation of the Jacobs and Campbell design is presented below.    This generations design can be viewed as a cross-sequential design because participants at generation 0 are replaced over time (i.e., longitudinally) by new participants at each successive generation.   In the case of a group with 4 people (A, B, C and D), after 4 generations there is an entirely new group of participants.   Considering the number of original group members at generations 0, 1, 2, 3 and 4; in this case, there are 100%, 75%, 50%, 25% and 0% in each generation.   Consideration the makeup of each generation



This design can be used to address different individual, dyadic and group phenomena.   For example, how does the grand mean of a group behave as the original structure is replaced by new members?   When the new members at each generation are out-group members relative to the original members, how do individual, dyadic and group processes change as the in-group/out-group composition of the group change?   The generation design can simulate changes in group composition that occurs naturally.  For example, changes in work groups, neighborhoods, family and friend groups can be addressed using this design.