<<
>>

Moderation and Mediation

In model testing, moderation refers to the interaction of two or more independent variables in predicting a dependent variable. (For more details on moderation, see Kenny, 2011b.) An interaction means that the effect of one independent variable on a dependent variable differs according to the level of one or more other independent variables.

Sometimes such moderation is predicted, in which case the data analysis needs to test for the presence of the predicted interaction. This test may be easily done using ANOVA and may also be done using regression or related techniques, such as SEM.

An example of an interaction is found in Cai, Fink, and Xie (2012): Chinese and American participants were asked, in one condition, about the likelihood of spend­ing money to assist another, or, in another condition, the likelihood of spending time to assist another. In addition, participants responded with regard to the kind of relation­ship involved: a close other, an acquaintance, an other known through a close friend, or an other known through an acquaintance. Among other findings, Chinese were more likely than Americans to spend money to assist someone who was a close other or an acquaintance, which is a two-way interaction of Country ? Relationship Type. On the other hand, Americans were more likely to spend time talking with someone who was known through a close friend or someone who was known through an acquaintance, which is also a Country ? Relationship Type interaction.

Mediation refers to a variable that is pre­dicted by at least one variable and also pre­dicts at least one other variable in a statistical model. One current description of the issues involved in evaluating mediation is found in Kenny (2011a).

Most conflict communication theories spec­ify processes, which are often best understood by determining the possible mediators that account for the relationship between an exog­enous variable and the variable that represents its ultimate effect.

For example, Yao, Fink, and Cai (2010) examined how a person who offends someone in an interaction accounts for his or her behavior through various inter­actional strategies. From the perspective of the offender, it was found that the perception of the offense’s outcome severity had a direct positive effect on the expected defensiveness of the account selected by the offender; in addi­tion to this direct effect, there was an indirect effect in that the perception of the offense’s outcome severity caused an increase in the expected responsibility for the offense, which caused an increase in the expectation of anger on the part of the offended person, which then caused a decrease in the expected defensive­ness of the account selected:

Thus, perceived outcome severity is a double­edged sword: To understand its effects, the relative magnitudes of the direct and indirect paths need to be considered. Furthermore, this complexity may explain the contradic­tory results reported in the literature for the effect of outcome severity. (p. 27)

Understanding mediation and moderation clarifies the way communication and conflict relate in an elaborated theoretical model.

<< | >>
Source: Oetzel John, Ting-Toomey Stella. The SAGE Handbook of Conflict Communication: Integrating Theory, Research and Practice. SAGE Publications,2013. — 912 p.. 2013

More on the topic Moderation and Mediation:

  1. Moderation and Mediation
  2. This chapter focuses on quantitative meth­ods for the study of conflict communi­cation for several reasons.
  3. Third-Party Intervention
  4. Subject Index
  5. Oetzel John, Ting-Toomey Stella. The SAGE Handbook of Conflict Communication: Integrating Theory, Research and Practice. SAGE Publications,2013. — 912 p., 2013
  6. References
  7. Public Hearings
  8. SHAPING STATE-ETHNIC RELATIONS