Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies

Samantha Joel; Paul W. Eastwick; Colleen J. Allison; Ximena B. Arriaga; Zachary G. Baker; Eran Bar-Kalifa; Sophie Bergeron; Gurit E. Birnbaum; Rebecca L. Brock; Claudia C. Brumbaugh; Cheryl L. Carmichael; Serena Chen; Jennifer Clarke; Rebecca J. Cobb; Michael K. Coolsen; Jody Davis; David C. de Jong; Anik Debrot; Eva C. DeHaas; Jaye L. Derrick; Jami Eller; Marie Joelle Estrada; Ruddy Faure; Eli J. Finkel; R. Chris Fraley; Shelly L. Gable; Reuma Gadassi-Polack; Yuthika U. Girme; Amie M. Gordon; Courtney L. Gosnell; Matthew D. Hammond; Peggy A. Hannon; Cheryl Harasymchuk; Wilhelm Hofmann; Andrea B. Horn; Emily A. Impett; Jeremy P. Jamieson; Dacher Keltner; James J. Kim; Jeffrey L. Kirchner; Esther S. Kluwer; Madoka Kumashiro; Grace Larson; Gal Lazarus; Jill M. Logan; Laura B. Luchies; Geoff MacDonald; Laura V. Machia; Michael R. Maniaci; Jessica A. Maxwell; Moran Mizrahi; Amy Muise; Sylvia Niehuis; Brian G. Ogolsky; C. Rebecca Oldham; Nickola C. Overall; Meinrad Perrez; Brett J. Peters; Paula R. Pietromonaco; Sally I. Powers; Thery Prok; Rony Pshedetzky-Shochat; Eshkol Rafaeli; Erin L. Ramsdell; Maija Reblin; Michael Reicherts; Alan Reifman; Harry T. Reis; Galena K. Rhoades; William S. Rholes; Francesca Righetti; Lindsey M. Rodriguez; Ron Rogge; Natalie O. Rosen; Darby Saxbe; Haran Sened; Jeffry A. Simpson; Erica B. Slotter; Scott M. Stanley; Shevaun Stocker; Cathy Surra; Hagar Ter Kuile; Allison A. Vaughn; Amanda M. Vicary; Mariko L. Visserman; Scott Wolf

doi:10.1073/pnas.1917036117

Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies

Samantha Joel^*, Paul W. Eastwick, Colleen J. Allison, Ximena B. Arriaga, Zachary G. Baker, Eran Bar-Kalifa, Sophie Bergeron, Gurit E. Birnbaum, Rebecca L. Brock, Claudia C. Brumbaugh, Cheryl L. Carmichael, Serena Chen, Jennifer Clarke, Rebecca J. Cobb, Michael K. Coolsen, Jody Davis, David C. de Jong, Anik Debrot, Eva C. DeHaas, Jaye L. DerrickJami Eller, Marie Joelle Estrada, Ruddy Faure, Eli J. Finkel, R. Chris Fraley, Shelly L. Gable, Reuma Gadassi-Polack, Yuthika U. Girme, Amie M. Gordon, Courtney L. Gosnell, Matthew D. Hammond, Peggy A. Hannon, Cheryl Harasymchuk, Wilhelm Hofmann, Andrea B. Horn, Emily A. Impett, Jeremy P. Jamieson, Dacher Keltner, James J. Kim, Jeffrey L. Kirchner, Esther S. Kluwer, Madoka Kumashiro, Grace Larson, Gal Lazarus, Jill M. Logan, Laura B. Luchies, Geoff MacDonald, Laura V. Machia, Michael R. Maniaci, Jessica A. Maxwell, Moran Mizrahi, Amy Muise, Sylvia Niehuis, Brian G. Ogolsky, C. Rebecca Oldham, Nickola C. Overall, Meinrad Perrez, Brett J. Peters, Paula R. Pietromonaco, Sally I. Powers, Thery Prok, Rony Pshedetzky-Shochat, Eshkol Rafaeli, Erin L. Ramsdell, Maija Reblin, Michael Reicherts, Alan Reifman, Harry T. Reis, Galena K. Rhoades, William S. Rholes, Francesca Righetti, Lindsey M. Rodriguez, Ron Rogge, Natalie O. Rosen, Darby Saxbe, Haran Sened, Jeffry A. Simpson, Erica B. Slotter, Scott M. Stanley, Shevaun Stocker, Cathy Surra, Hagar Ter Kuile, Allison A. Vaughn, Amanda M. Vicary, Mariko L. Visserman, Scott Wolf

^*Corresponding author for this work

Psychology

Research output: Contribution to journal › Article › peer-review

118 Scopus citations

Abstract

Given the powerful implications of relationship quality for health and well-being, a central mission of relationship science is explaining why some romantic relationships thrive more than others. This large-scale project used machine learning (i.e., Random Forests) to 1) quantify the extent to which relationship quality is predictable and 2) identify which constructs reliably predict relationship quality. Across 43 dyadic longitudinal datasets from 29 laboratories, the top relationship-specific predictors of relationship quality were perceived-partner commitment, appreciation, sexual satisfaction, perceived-partner satisfaction, and conflict. The top individual-difference predictors were life satisfaction, negative affect, depression, attachment avoidance, and attachment anxiety. Overall, relationship-specific variables predicted up to 45% of variance at baseline, and up to 18% of variance at the end of each study. Individual differences also performed well (21% and 12%, respectively). Actor-reported variables (i.e., own relationship-specific and individual-difference variables) predicted two to four times more variance than partner-reported variables (i.e., the partner's ratings on those variables). Importantly, individual differences and partner reports had no predictive effects beyond actor-reported relationship-specific variables alone. These findings imply that the sum of all individual differences and partner experiences exert their influence on relationship quality via a person's own relationship-specific experiences, and effects due to moderation by individual differences and moderation by partner-reports may be quite small. Finally, relationship-quality change (i.e., increases or decreases in relationship quality over the course of a study) was largely unpredictable from any combination of self-report variables. This collective effort should guide future models of relationships.

Original language	English (US)
Pages (from-to)	19061-19071
Number of pages	11
Journal	Proceedings of the National Academy of Sciences of the United States of America
Volume	117
Issue number	32
DOIs	https://doi.org/10.1073/pnas.1917036117
State	Published - Aug 11 2020

Keywords

Ensemble methods
Machine learning
Random Forests
Relationship quality
Romantic relationships

ASJC Scopus subject areas

General

Access to Document

10.1073/pnas.1917036117

Cite this

Joel, S., Eastwick, P. W., Allison, C. J., Arriaga, X. B., Baker, Z. G., Bar-Kalifa, E., Bergeron, S., Birnbaum, G. E., Brock, R. L., Brumbaugh, C. C., Carmichael, C. L., Chen, S., Clarke, J., Cobb, R. J., Coolsen, M. K., Davis, J., de Jong, D. C., Debrot, A., DeHaas, E. C., ... Wolf, S. (2020). Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies. Proceedings of the National Academy of Sciences of the United States of America, 117(32), 19061-19071. https://doi.org/10.1073/pnas.1917036117

@article{95bd9c0a4e124119b1cc47d5635ad5b3,

title = "Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies",

abstract = "Given the powerful implications of relationship quality for health and well-being, a central mission of relationship science is explaining why some romantic relationships thrive more than others. This large-scale project used machine learning (i.e., Random Forests) to 1) quantify the extent to which relationship quality is predictable and 2) identify which constructs reliably predict relationship quality. Across 43 dyadic longitudinal datasets from 29 laboratories, the top relationship-specific predictors of relationship quality were perceived-partner commitment, appreciation, sexual satisfaction, perceived-partner satisfaction, and conflict. The top individual-difference predictors were life satisfaction, negative affect, depression, attachment avoidance, and attachment anxiety. Overall, relationship-specific variables predicted up to 45% of variance at baseline, and up to 18% of variance at the end of each study. Individual differences also performed well (21% and 12%, respectively). Actor-reported variables (i.e., own relationship-specific and individual-difference variables) predicted two to four times more variance than partner-reported variables (i.e., the partner's ratings on those variables). Importantly, individual differences and partner reports had no predictive effects beyond actor-reported relationship-specific variables alone. These findings imply that the sum of all individual differences and partner experiences exert their influence on relationship quality via a person's own relationship-specific experiences, and effects due to moderation by individual differences and moderation by partner-reports may be quite small. Finally, relationship-quality change (i.e., increases or decreases in relationship quality over the course of a study) was largely unpredictable from any combination of self-report variables. This collective effort should guide future models of relationships.",

keywords = "Ensemble methods, Machine learning, Random Forests, Relationship quality, Romantic relationships",

author = "Samantha Joel and Eastwick, {Paul W.} and Allison, {Colleen J.} and Arriaga, {Ximena B.} and Baker, {Zachary G.} and Eran Bar-Kalifa and Sophie Bergeron and Birnbaum, {Gurit E.} and Brock, {Rebecca L.} and Brumbaugh, {Claudia C.} and Carmichael, {Cheryl L.} and Serena Chen and Jennifer Clarke and Cobb, {Rebecca J.} and Coolsen, {Michael K.} and Jody Davis and {de Jong}, {David C.} and Anik Debrot and DeHaas, {Eva C.} and Derrick, {Jaye L.} and Jami Eller and Estrada, {Marie Joelle} and Ruddy Faure and Finkel, {Eli J.} and {Chris Fraley}, R. and Gable, {Shelly L.} and Reuma Gadassi-Polack and Girme, {Yuthika U.} and Gordon, {Amie M.} and Gosnell, {Courtney L.} and Hammond, {Matthew D.} and Hannon, {Peggy A.} and Cheryl Harasymchuk and Wilhelm Hofmann and Horn, {Andrea B.} and Impett, {Emily A.} and Jamieson, {Jeremy P.} and Dacher Keltner and Kim, {James J.} and Kirchner, {Jeffrey L.} and Kluwer, {Esther S.} and Madoka Kumashiro and Grace Larson and Gal Lazarus and Logan, {Jill M.} and Luchies, {Laura B.} and Geoff MacDonald and Machia, {Laura V.} and Maniaci, {Michael R.} and Maxwell, {Jessica A.} and Moran Mizrahi and Amy Muise and Sylvia Niehuis and Ogolsky, {Brian G.} and {Rebecca Oldham}, C. and Overall, {Nickola C.} and Meinrad Perrez and Peters, {Brett J.} and Pietromonaco, {Paula R.} and Powers, {Sally I.} and Thery Prok and Rony Pshedetzky-Shochat and Eshkol Rafaeli and Ramsdell, {Erin L.} and Maija Reblin and Michael Reicherts and Alan Reifman and Reis, {Harry T.} and Rhoades, {Galena K.} and Rholes, {William S.} and Francesca Righetti and Rodriguez, {Lindsey M.} and Ron Rogge and Rosen, {Natalie O.} and Darby Saxbe and Haran Sened and Simpson, {Jeffry A.} and Slotter, {Erica B.} and Stanley, {Scott M.} and Shevaun Stocker and Cathy Surra and Kuile, {Hagar Ter} and Vaughn, {Allison A.} and Vicary, {Amanda M.} and Visserman, {Mariko L.} and Scott Wolf",

note = "Funding Information: ACKNOWLEDGMENTS. The work of aggregating the datasets was supported by a Social Sciences and Humanities Research Council of Canada Insight Grant 435-2019-0115 (to S.J.). Collection of the 43 datasets was supported by many separate funding sources. Datasets 1 and 2 were funded by NSF Grant BCS-719780 (to E.J.F.). Dataset 3 was partially supported by a National Research Service Predoctoral Training Grant (to S.L.G.). Dataset 4 was supported by a Social Sciences and Humanities Research Council (SSHRC) Predoctoral fellowship (to D.C.d.J.). Dataset 5 was funded by a grant from the Fetzer Institute (to H.T.R.). Datasets 8 and 9 were supported by an SSHRC Banting postdoctoral fellowship, and Dataset 10 was supported by an SSHRC doctoral fellowship (to A.M.). Dataset 13 was funded by University of Auckland Doctoral Research funds (Y.U.G. and M.D.H.). Dataset 14 was funded by University of Auckland Grants 3626244 and 3607021 (to N.C.O.). Dataset 15 was funded by the National Institute on Alcohol Abuse and Alcoholism under Award F31AA020442 (to L.M.R.). Dataset 16 was funded by National Cancer Institute Grant R01CA133908 (to P.R.P. and S.I.P.). Dataset 17 was funded by NSF Grant BCS-0443783 (to R.C.F.). Dataset 18 was funded by National Institute of Mental Health (NIMH) Grant BSR–R01-MH-45417 (to Caryl E. Rusbult). Dataset 19 was funded by the Clayton Award for Excellence in Graduate Research from the University of Utah (to A.A.V. and M. R. Reblin). Dataset 20 was funded by NIMH Grant MH49599 (to J.A.S. and W.S.R.). Dataset 22 was funded by a joint Open Research Area grant from the Dutch Science Foundation 464-15-093 (to F.R.) and HO4175/6-1 from the German Science Foundation (to W.H.). Dataset 24 was funded by a Utrecht University High Potential grant (to E.S.K.). Dataset 25 was funded by Israel Science Foundation Grant 615/10 (to E..R.) and work on its adaptation to this study was supported by Azrieli Foundation fellowships (to G. Lazarus and H.S.). Datasets 27 and 28 were funded by Texas Tech University{\textquoteright}s Office of the Vice President for Research, Office of Institutional Diversity, Equity, and Community Engagement, and College of Human Sciences (S.N.). Dataset 29 was funded by the National Institute of Child Health and Human Development under Award HD047564 (to S.M.S.). Dataset 30 was funded by SSHRC Insight Development Grant 430-2016-00422 (to C.H., A.M., and E.A.I.). Dataset 31 was funded by the Swiss National Science Foundation and was part of National Center of Competence in Research of Affective Sciences Grant 51A24-104897 (to M.P. and M. Reicherts). Dataset 32 was funded by the National Research University Fund, Division of Research, University of Houston, and a University of Houston CLASS Research Progress Grant (to J.D.) and National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under Award F31AA026195 (to Z.G.B.). Dataset 33 was funded by a Canadian Institutes of Health Research (CIHR) postdoctoral fellowship (to N.O.R.) and a CIHR grant (to S.B.). Dataset 34 was funded by a CIHR grant (to N.O.R.). Dataset 37 was funded by NSF Grant BCS-1050875 (to S.L.G.). Dataset 39 was funded by NSF Grant BCS-0132398 (to Caryl E. Rusbult). Dataset 40 was funded by Templeton Foundation Grant 5158 (to Caryl E. Rusbult). Dataset 41 was funded by an SSHRC Canadian Graduate Scholarship (to J.A.M.) and an SSHRC Insight grant (to G.M.). Dataset 42 was funded by SSHRC Research Grant 410-2005-0829 (to R.J.C.). Funding Information: The work of aggregating the datasets was supported by a Social Sciences and Humanities Research Council of Canada Insight Grant 435-2019-0115 (to S.J.). Collection of the 43 datasets was supported by many separate funding sources. Datasets 1 and 2 were funded by NSF Grant BCS-719780 (to E.J.F.). Dataset 3 was partially supported by a National Research Service Predoctoral Training Grant (to S.L.G.). Dataset 4 was supported by a Social Sciences and Humanities Research Council (SSHRC) Predoctoral fellowship (to D.C.d.J.). Dataset 5 was funded by a grant from the Fetzer Institute (to H.T.R.). Datasets 8 and 9 were supported by an SSHRC Banting postdoctoral fellowship, and Dataset 10 was supported by an SSHRC doctoral fellowship (to A.M.). Dataset 13 was funded by University of Auckland Doctoral Research funds (Y.U.G. and M.D.H.). Dataset 14 was funded by University of Auckland Grants 3626244 and 3607021 (to N.C.O.). Dataset 15 was funded by the National Institute on Alcohol Abuse and Alcoholism under Award F31AA020442 (to L.M.R.). Dataset 16 was funded by National Cancer Institute Grant R01CA133908 (to P.R.P. and S.I.P.). Dataset 17 was funded by NSF Grant BCS-0443783 (to R.C.F.). Dataset 18 was funded by National Institute of Mental Health (NIMH) Grant BSR-R01-MH-45417 (to Caryl E. Rusbult). Dataset 19 was funded by the Clayton Award for Excellence in Graduate Research from the University of Utah (to A.A.V. and M. R. Reblin). Dataset 20 was funded by NIMH Grant MH49599 (to J.A.S. and W.S.R.). Dataset 22 was funded by a joint Open Research Area grant from the Dutch Science Foundation 464-15-093 (to F.R.) and HO4175/6-1 from the German Science Foundation (to W.H.). Dataset 24 was funded by a Utrecht University High Potential grant (to E.S.K.). Dataset 25 was funded by Israel Science Foundation Grant 615/10 (to E..R.) and work on its adaptation to this study was supported by Azrieli Foundation fellowships (to G. Lazarus and H.S.). Datasets 27 and 28 were funded by Texas Tech University's Office of the Vice President for Research, Office of Institutional Diversity, Equity, and Community Engagement, and College of Human Sciences (S.N.). Dataset 29 was funded by the National Institute of Child Health and Human Development under Award HD047564 (to S.M.S.). Dataset 30 was funded by SSHRC Insight Development Grant 430-2016-00422 (to C.H., A.M., and E.A.I.). Dataset 31 was funded by the Swiss National Science Foundation and was part of National Center of Competence in Research of Affective Sciences Grant 51A24-104897 (to M.P. and M. Reicherts). Dataset 32 was funded by the National Research University Fund, Division of Research, University of Houston, and a University of Houston CLASS Research Progress Grant (to J.D.) and National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under Award F31AA026195 (to Z.G.B.). Dataset 33 was funded by a Canadian Institutes of Health Research (CIHR) postdoctoral fellowship (to N.O.R.) and a CIHR grant (to S.B.). Dataset 34 was funded by a CIHR grant (to N.O.R.). Dataset 37 was funded by NSF Grant BCS-1050875 (to S.L.G.). Dataset 39 was funded by NSF Grant BCS-0132398 (to Caryl E. Rusbult). Dataset 40 was funded by Templeton Foundation Grant 5158 (to Caryl E. Rusbult). Dataset 41 was funded by an SSHRC Canadian Graduate Scholarship (to J.A.M.) and an SSHRC Insight grant (to G.M.). Dataset 42 was funded by SSHRC Research Grant 410-2005-0829 (to R.J.C.). Publisher Copyright: {\textcopyright} 2020 National Academy of Sciences. All rights reserved.",

year = "2020",

month = aug,

day = "11",

doi = "10.1073/pnas.1917036117",

language = "English (US)",

volume = "117",

pages = "19061--19071",

journal = "Proceedings of the National Academy of Sciences of the United States of America",

issn = "0027-8424",

publisher = "National Academy of Sciences",

number = "32",

}

Joel, S, Eastwick, PW, Allison, CJ, Arriaga, XB, Baker, ZG, Bar-Kalifa, E, Bergeron, S, Birnbaum, GE, Brock, RL, Brumbaugh, CC, Carmichael, CL, Chen, S, Clarke, J, Cobb, RJ, Coolsen, MK, Davis, J, de Jong, DC, Debrot, A, DeHaas, EC, Derrick, JL, Eller, J, Estrada, MJ, Faure, R, Finkel, EJ, Chris Fraley, R, Gable, SL, Gadassi-Polack, R, Girme, YU, Gordon, AM, Gosnell, CL, Hammond, MD, Hannon, PA, Harasymchuk, C, Hofmann, W, Horn, AB, Impett, EA, Jamieson, JP, Keltner, D, Kim, JJ, Kirchner, JL, Kluwer, ES, Kumashiro, M, Larson, G, Lazarus, G, Logan, JM, Luchies, LB, MacDonald, G, Machia, LV, Maniaci, MR, Maxwell, JA, Mizrahi, M, Muise, A, Niehuis, S, Ogolsky, BG, Rebecca Oldham, C, Overall, NC, Perrez, M, Peters, BJ, Pietromonaco, PR, Powers, SI, Prok, T, Pshedetzky-Shochat, R, Rafaeli, E, Ramsdell, EL, Reblin, M, Reicherts, M, Reifman, A, Reis, HT, Rhoades, GK, Rholes, WS, Righetti, F, Rodriguez, LM, Rogge, R, Rosen, NO, Saxbe, D, Sened, H, Simpson, JA, Slotter, EB, Stanley, SM, Stocker, S, Surra, C, Kuile, HT, Vaughn, AA, Vicary, AM, Visserman, ML & Wolf, S 2020, 'Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies', Proceedings of the National Academy of Sciences of the United States of America, vol. 117, no. 32, pp. 19061-19071. https://doi.org/10.1073/pnas.1917036117

Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies. / Joel, Samantha; Eastwick, Paul W.; Allison, Colleen J. et al.
In: Proceedings of the National Academy of Sciences of the United States of America, Vol. 117, No. 32, 11.08.2020, p. 19061-19071.

Research output: Contribution to journal › Article › peer-review

TY - JOUR

T1 - Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies

AU - Joel, Samantha

AU - Eastwick, Paul W.

AU - Allison, Colleen J.

AU - Arriaga, Ximena B.

AU - Baker, Zachary G.

AU - Bar-Kalifa, Eran

AU - Bergeron, Sophie

AU - Birnbaum, Gurit E.

AU - Brock, Rebecca L.

AU - Brumbaugh, Claudia C.

AU - Carmichael, Cheryl L.

AU - Chen, Serena

AU - Clarke, Jennifer

AU - Cobb, Rebecca J.

AU - Coolsen, Michael K.

AU - Davis, Jody

AU - de Jong, David C.

AU - Debrot, Anik

AU - DeHaas, Eva C.

AU - Derrick, Jaye L.

AU - Eller, Jami

AU - Estrada, Marie Joelle

AU - Faure, Ruddy

AU - Finkel, Eli J.

AU - Chris Fraley, R.

AU - Gable, Shelly L.

AU - Gadassi-Polack, Reuma

AU - Girme, Yuthika U.

AU - Gordon, Amie M.

AU - Gosnell, Courtney L.

AU - Hammond, Matthew D.

AU - Hannon, Peggy A.

AU - Harasymchuk, Cheryl

AU - Hofmann, Wilhelm

AU - Horn, Andrea B.

AU - Impett, Emily A.

AU - Jamieson, Jeremy P.

AU - Keltner, Dacher

AU - Kim, James J.

AU - Kirchner, Jeffrey L.

AU - Kluwer, Esther S.

AU - Kumashiro, Madoka

AU - Larson, Grace

AU - Lazarus, Gal

AU - Logan, Jill M.

AU - Luchies, Laura B.

AU - MacDonald, Geoff

AU - Machia, Laura V.

AU - Maniaci, Michael R.

AU - Maxwell, Jessica A.

AU - Mizrahi, Moran

AU - Muise, Amy

AU - Niehuis, Sylvia

AU - Ogolsky, Brian G.

AU - Rebecca Oldham, C.

AU - Overall, Nickola C.

AU - Perrez, Meinrad

AU - Peters, Brett J.

AU - Pietromonaco, Paula R.

AU - Powers, Sally I.

AU - Prok, Thery

AU - Pshedetzky-Shochat, Rony

AU - Rafaeli, Eshkol

AU - Ramsdell, Erin L.

AU - Reblin, Maija

AU - Reicherts, Michael

AU - Reifman, Alan

AU - Reis, Harry T.

AU - Rhoades, Galena K.

AU - Rholes, William S.

AU - Righetti, Francesca

AU - Rodriguez, Lindsey M.

AU - Rogge, Ron

AU - Rosen, Natalie O.

AU - Saxbe, Darby

AU - Sened, Haran

AU - Simpson, Jeffry A.

AU - Slotter, Erica B.

AU - Stanley, Scott M.

AU - Stocker, Shevaun

AU - Surra, Cathy

AU - Kuile, Hagar Ter

AU - Vaughn, Allison A.

AU - Vicary, Amanda M.

AU - Visserman, Mariko L.

AU - Wolf, Scott

N1 - Funding Information: ACKNOWLEDGMENTS. The work of aggregating the datasets was supported by a Social Sciences and Humanities Research Council of Canada Insight Grant 435-2019-0115 (to S.J.). Collection of the 43 datasets was supported by many separate funding sources. Datasets 1 and 2 were funded by NSF Grant BCS-719780 (to E.J.F.). Dataset 3 was partially supported by a National Research Service Predoctoral Training Grant (to S.L.G.). Dataset 4 was supported by a Social Sciences and Humanities Research Council (SSHRC) Predoctoral fellowship (to D.C.d.J.). Dataset 5 was funded by a grant from the Fetzer Institute (to H.T.R.). Datasets 8 and 9 were supported by an SSHRC Banting postdoctoral fellowship, and Dataset 10 was supported by an SSHRC doctoral fellowship (to A.M.). Dataset 13 was funded by University of Auckland Doctoral Research funds (Y.U.G. and M.D.H.). Dataset 14 was funded by University of Auckland Grants 3626244 and 3607021 (to N.C.O.). Dataset 15 was funded by the National Institute on Alcohol Abuse and Alcoholism under Award F31AA020442 (to L.M.R.). Dataset 16 was funded by National Cancer Institute Grant R01CA133908 (to P.R.P. and S.I.P.). Dataset 17 was funded by NSF Grant BCS-0443783 (to R.C.F.). Dataset 18 was funded by National Institute of Mental Health (NIMH) Grant BSR–R01-MH-45417 (to Caryl E. Rusbult). Dataset 19 was funded by the Clayton Award for Excellence in Graduate Research from the University of Utah (to A.A.V. and M. R. Reblin). Dataset 20 was funded by NIMH Grant MH49599 (to J.A.S. and W.S.R.). Dataset 22 was funded by a joint Open Research Area grant from the Dutch Science Foundation 464-15-093 (to F.R.) and HO4175/6-1 from the German Science Foundation (to W.H.). Dataset 24 was funded by a Utrecht University High Potential grant (to E.S.K.). Dataset 25 was funded by Israel Science Foundation Grant 615/10 (to E..R.) and work on its adaptation to this study was supported by Azrieli Foundation fellowships (to G. Lazarus and H.S.). Datasets 27 and 28 were funded by Texas Tech University’s Office of the Vice President for Research, Office of Institutional Diversity, Equity, and Community Engagement, and College of Human Sciences (S.N.). Dataset 29 was funded by the National Institute of Child Health and Human Development under Award HD047564 (to S.M.S.). Dataset 30 was funded by SSHRC Insight Development Grant 430-2016-00422 (to C.H., A.M., and E.A.I.). Dataset 31 was funded by the Swiss National Science Foundation and was part of National Center of Competence in Research of Affective Sciences Grant 51A24-104897 (to M.P. and M. Reicherts). Dataset 32 was funded by the National Research University Fund, Division of Research, University of Houston, and a University of Houston CLASS Research Progress Grant (to J.D.) and National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under Award F31AA026195 (to Z.G.B.). Dataset 33 was funded by a Canadian Institutes of Health Research (CIHR) postdoctoral fellowship (to N.O.R.) and a CIHR grant (to S.B.). Dataset 34 was funded by a CIHR grant (to N.O.R.). Dataset 37 was funded by NSF Grant BCS-1050875 (to S.L.G.). Dataset 39 was funded by NSF Grant BCS-0132398 (to Caryl E. Rusbult). Dataset 40 was funded by Templeton Foundation Grant 5158 (to Caryl E. Rusbult). Dataset 41 was funded by an SSHRC Canadian Graduate Scholarship (to J.A.M.) and an SSHRC Insight grant (to G.M.). Dataset 42 was funded by SSHRC Research Grant 410-2005-0829 (to R.J.C.). Funding Information: The work of aggregating the datasets was supported by a Social Sciences and Humanities Research Council of Canada Insight Grant 435-2019-0115 (to S.J.). Collection of the 43 datasets was supported by many separate funding sources. Datasets 1 and 2 were funded by NSF Grant BCS-719780 (to E.J.F.). Dataset 3 was partially supported by a National Research Service Predoctoral Training Grant (to S.L.G.). Dataset 4 was supported by a Social Sciences and Humanities Research Council (SSHRC) Predoctoral fellowship (to D.C.d.J.). Dataset 5 was funded by a grant from the Fetzer Institute (to H.T.R.). Datasets 8 and 9 were supported by an SSHRC Banting postdoctoral fellowship, and Dataset 10 was supported by an SSHRC doctoral fellowship (to A.M.). Dataset 13 was funded by University of Auckland Doctoral Research funds (Y.U.G. and M.D.H.). Dataset 14 was funded by University of Auckland Grants 3626244 and 3607021 (to N.C.O.). Dataset 15 was funded by the National Institute on Alcohol Abuse and Alcoholism under Award F31AA020442 (to L.M.R.). Dataset 16 was funded by National Cancer Institute Grant R01CA133908 (to P.R.P. and S.I.P.). Dataset 17 was funded by NSF Grant BCS-0443783 (to R.C.F.). Dataset 18 was funded by National Institute of Mental Health (NIMH) Grant BSR-R01-MH-45417 (to Caryl E. Rusbult). Dataset 19 was funded by the Clayton Award for Excellence in Graduate Research from the University of Utah (to A.A.V. and M. R. Reblin). Dataset 20 was funded by NIMH Grant MH49599 (to J.A.S. and W.S.R.). Dataset 22 was funded by a joint Open Research Area grant from the Dutch Science Foundation 464-15-093 (to F.R.) and HO4175/6-1 from the German Science Foundation (to W.H.). Dataset 24 was funded by a Utrecht University High Potential grant (to E.S.K.). Dataset 25 was funded by Israel Science Foundation Grant 615/10 (to E..R.) and work on its adaptation to this study was supported by Azrieli Foundation fellowships (to G. Lazarus and H.S.). Datasets 27 and 28 were funded by Texas Tech University's Office of the Vice President for Research, Office of Institutional Diversity, Equity, and Community Engagement, and College of Human Sciences (S.N.). Dataset 29 was funded by the National Institute of Child Health and Human Development under Award HD047564 (to S.M.S.). Dataset 30 was funded by SSHRC Insight Development Grant 430-2016-00422 (to C.H., A.M., and E.A.I.). Dataset 31 was funded by the Swiss National Science Foundation and was part of National Center of Competence in Research of Affective Sciences Grant 51A24-104897 (to M.P. and M. Reicherts). Dataset 32 was funded by the National Research University Fund, Division of Research, University of Houston, and a University of Houston CLASS Research Progress Grant (to J.D.) and National Institute on Alcohol Abuse and Alcoholism of the National Institutes of Health under Award F31AA026195 (to Z.G.B.). Dataset 33 was funded by a Canadian Institutes of Health Research (CIHR) postdoctoral fellowship (to N.O.R.) and a CIHR grant (to S.B.). Dataset 34 was funded by a CIHR grant (to N.O.R.). Dataset 37 was funded by NSF Grant BCS-1050875 (to S.L.G.). Dataset 39 was funded by NSF Grant BCS-0132398 (to Caryl E. Rusbult). Dataset 40 was funded by Templeton Foundation Grant 5158 (to Caryl E. Rusbult). Dataset 41 was funded by an SSHRC Canadian Graduate Scholarship (to J.A.M.) and an SSHRC Insight grant (to G.M.). Dataset 42 was funded by SSHRC Research Grant 410-2005-0829 (to R.J.C.). Publisher Copyright: © 2020 National Academy of Sciences. All rights reserved.

PY - 2020/8/11

Y1 - 2020/8/11

N2 - Given the powerful implications of relationship quality for health and well-being, a central mission of relationship science is explaining why some romantic relationships thrive more than others. This large-scale project used machine learning (i.e., Random Forests) to 1) quantify the extent to which relationship quality is predictable and 2) identify which constructs reliably predict relationship quality. Across 43 dyadic longitudinal datasets from 29 laboratories, the top relationship-specific predictors of relationship quality were perceived-partner commitment, appreciation, sexual satisfaction, perceived-partner satisfaction, and conflict. The top individual-difference predictors were life satisfaction, negative affect, depression, attachment avoidance, and attachment anxiety. Overall, relationship-specific variables predicted up to 45% of variance at baseline, and up to 18% of variance at the end of each study. Individual differences also performed well (21% and 12%, respectively). Actor-reported variables (i.e., own relationship-specific and individual-difference variables) predicted two to four times more variance than partner-reported variables (i.e., the partner's ratings on those variables). Importantly, individual differences and partner reports had no predictive effects beyond actor-reported relationship-specific variables alone. These findings imply that the sum of all individual differences and partner experiences exert their influence on relationship quality via a person's own relationship-specific experiences, and effects due to moderation by individual differences and moderation by partner-reports may be quite small. Finally, relationship-quality change (i.e., increases or decreases in relationship quality over the course of a study) was largely unpredictable from any combination of self-report variables. This collective effort should guide future models of relationships.

AB - Given the powerful implications of relationship quality for health and well-being, a central mission of relationship science is explaining why some romantic relationships thrive more than others. This large-scale project used machine learning (i.e., Random Forests) to 1) quantify the extent to which relationship quality is predictable and 2) identify which constructs reliably predict relationship quality. Across 43 dyadic longitudinal datasets from 29 laboratories, the top relationship-specific predictors of relationship quality were perceived-partner commitment, appreciation, sexual satisfaction, perceived-partner satisfaction, and conflict. The top individual-difference predictors were life satisfaction, negative affect, depression, attachment avoidance, and attachment anxiety. Overall, relationship-specific variables predicted up to 45% of variance at baseline, and up to 18% of variance at the end of each study. Individual differences also performed well (21% and 12%, respectively). Actor-reported variables (i.e., own relationship-specific and individual-difference variables) predicted two to four times more variance than partner-reported variables (i.e., the partner's ratings on those variables). Importantly, individual differences and partner reports had no predictive effects beyond actor-reported relationship-specific variables alone. These findings imply that the sum of all individual differences and partner experiences exert their influence on relationship quality via a person's own relationship-specific experiences, and effects due to moderation by individual differences and moderation by partner-reports may be quite small. Finally, relationship-quality change (i.e., increases or decreases in relationship quality over the course of a study) was largely unpredictable from any combination of self-report variables. This collective effort should guide future models of relationships.

KW - Ensemble methods

KW - Machine learning

KW - Random Forests

KW - Relationship quality

KW - Romantic relationships

UR - http://www.scopus.com/inward/record.url?scp=85089613576&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85089613576&partnerID=8YFLogxK

U2 - 10.1073/pnas.1917036117

DO - 10.1073/pnas.1917036117

M3 - Article

C2 - 32719123

AN - SCOPUS:85089613576

SN - 0027-8424

VL - 117

SP - 19061

EP - 19071

JO - Proceedings of the National Academy of Sciences of the United States of America

JF - Proceedings of the National Academy of Sciences of the United States of America

IS - 32

ER -

Machine learning uncovers the most robust self-report predictors of relationship quality across 43 longitudinal couples studies

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this