Research Methods for Relationship Science Workshop hosted by Sabanci University, Istanbul, Turkey Lorne Campbell University of Western Ontario
What *IS* a Relationship? Kelley et al. (1983) “A prominent feature of a ‘relationship’ is that events associated with one person are causally connected to those associated with the other person. Indeed, this is a necessary feature of ‘relationship’ as we define it.” (p. 24)
P-O Interaction Person P Person O Affect Thought Action Action Thought Time
Reis and Shaver, 1988: Intimacy Process Model A`s motives, needs, goals and fears B`s interpretive filter A`s disclosure B`s response A`s reaction to B`s response A`s interpretive filter B`s motives, needs, goals and fears
Challenges for Researchers Definition implies that researchers should study partners people influence each other, over time Challenges exist in the analysis of data where N ≥ 2 per unit Non-independence of data-points Resources
Ellen Berscheid (1999) “Reflecting psychology’s individualistic orientation, virtually all of our methodologies and statistics are predicated on the individual as the unit of analysis. As a consequence, relationship scholars often find themselves jerry-rigging old methodologies and statistics to accommodate the dyadic unit of analysis, but some, such as David Kenny and his associates (e.g., Kenny & La Voie, 1984), are creatively creating new ones.” (p. 261)
Advances in Data Analyses A number of advances in data analysis have turned the first challenge into an opportunity E.g., SEM, HLM, pooled regression To what degree is it being taken advantage of?
Survey of Literature Looked at all articles in 5 journals that focused on close relationships JPSP, PSPB, PR, JSPR, and JM&F Looked at 2 time periods to detect any major shifts in the type of data collected, and the types of analyses employed Major question: how is interdependence as discussed by Kelley et al being tested in modern relationships research?
Frequency and Percentage of Studies Collecting Different Data Types in 5 Journals in 1994 and 2002 Indiv Dyad Group PR 22(76) 5(17) 2(7) 28(74) 10(26) 0(0) JSPR 22(63) 13(37) 19(66) 10(34) JM&F 59(82) 10(14) 3(4) 53(78) 11(16) 4(6) JPSP 8(47) 7(41) 2(12) 19(79) 5(21) PSPB 3(75) 1(25) 14(64) 8(36) Totals 114 (73) 36 (23) 7 (4) 133 (74) 44 (24) 4 (2)
Frequency and Percentage of Studies Collecting Different Data Types in 5 Journals in 1994 and 2002 Indiv Dyad Group PR 22(76) 5(17) 2(7) 28(74) 10(26) 0(0) JSPR 22(63) 13(37) 19(66) 10(34) JM&F 59(82) 10(14) 3(4) 53(78) 11(16) 4(6) JPSP 8(47) 7(41) 2(12) 19(79) 5(21) PSPB 3(75) 1(25) 14(64) 8(36) Totals 114 (73) 36 (23) 7 (4) 133 (74) 44 (24) 4 (2)
Frequency and Percentage of Studies Collecting Different Data Types in 5 Journals in 1994 and 2002 Indiv Dyad Group PR 22(76) 5(17) 2(7) 28(74) 10(26) 0(0) JSPR 22(63) 13(37) 19(66) 10(34) JM&F 59(82) 10(14) 3(4) 53(78) 11(16) 4(6) JPSP 8(47) 7(41) 2(12) 19(79) 5(21) PSPB 3(75) 1(25) 14(64) 8(36) Totals 114 (73) 36 (23) 7 (4) 133 (74) 44 (24) 4 (2)
Frequencies of various analytic approaches applied to dyadic and group data in 1994 and 2002 Analysis Strategy 1994 2002 Nonindependence ignored 3 Means for each dyad/group averaged 6 Separate analyses (1) 9 8 Separate analyses (2) 7 4 Standard analyses; dyad/group unit of analysis 13 15 Social Relations Model 2 Structural Equation Modeling Actor-Partner Interdependence Model Hierarchical Linear Modeling 1
Frequencies of various analytic approaches applied to dyadic and group data in 1994 and 2002 Analysis Strategy 1994 2002 Nonindependence ignored 3 Means for each dyad/group averaged 6 Separate analyses (1) 9 8 Separate analyses (2) 7 4 Standard analyses; dyad/group unit of analysis 13 15 Social Relations Model 2 Structural Equation Modeling Actor-Partner Interdependence Model Hierarchical Linear Modeling 1
Dyadic Data ≠ Studying Dyadic Processes Not all studies with dyadic data investigate mutual influence (or partner effects)
Individual Data ≠ Inability to Test Dyadic Processes Studies can invoke mental image of the partner/relationship Perceived partner responsiveness But, is limited—cannot assess mutual influence
Intermission
Three Goals of Research Runkel and McGrath (1972) A) maximize the generalizability of results across populations, B) the precision with which variables are measured, and C) the realism of the context in which variables are assessed.
Three goals cannot be simultaneously maximized In one study
Three goals cannot be simultaneously maximized In one study
Three goals cannot be simultaneously maximized In one study
Three goals cannot be simultaneously maximized In one study
Three goals cannot be simultaneously maximized In one study
Three goals cannot be simultaneously maximized In one study
Three goals cannot be simultaneously maximized In one study
What is Needed? More truly dyadic research More longitudinal research (“over time”) More replication More focus on context Particular cultural context
Thank you for your time