Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A cautionary tale of two statistics: partial correlation and standardized partial regression.

Duncan Cramer1

  • 1Department of Social Sciences, Loughborough University, Leicestershire, England. d.cramer@lboro.ac.uk

The Journal of Psychology
|November 25, 2003
PubMed
Summary

Partial correlation and standardized partial regression can reverse or inflate effect sizes, impacting study interpretations. Researchers must carefully consider these statistical properties when analyzing relationships between variables.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Mutuality of Rogers's therapeutic conditions and treatment progress in the first three psychotherapy sessions.

Psychotherapy research : journal of the Society for Psychotherapy Research·2014
Same author

Psychological predictors of high-risk sexual behaviour and drug use among adolescents in Georgia.

Current HIV research·2010
Same author

Effect of the destructive disagreement belief on satisfaction with one's closest friend.

The Journal of psychology·2005
Same author

Emotional support, conflict, depression, and relationship satisfaction in a romantic partner.

The Journal of psychology·2004
Same author

Satisfaction with a romantic relationship, depression, support and conflict.

Psychology and psychotherapy·2004
Same author

Effect of the destructive disagreement belief on relationship satisfaction with a romantic partner or closest friend.

Psychology and psychotherapy·2004

Area of Science:

  • Social Sciences
  • Psychology
  • Sociology
  • Statistics

Background:

  • Partial correlation and standardized partial regression are common statistical methods for controlling extraneous variables in research.
  • These techniques are frequently applied in both cross-sectional and longitudinal study designs.
  • Misinterpretation of these statistics can significantly alter the understanding of data.

Purpose of the Study:

  • To highlight two critical aspects of partial correlation and standardized partial regression that can alter data interpretation.
  • To examine how the sign and magnitude of these statistics can be unexpectedly affected by other variables.
  • To question previous interpretations of findings based on these statistical properties.

Main Methods:

Related Experiment Videos

  • Analysis of the mathematical properties of partial correlation and standardized partial regression.
  • Illustration of the first phenomenon using data on marital satisfaction and conflict behavior (Gottman & Krokoff, 1989).
  • Illustration of the second phenomenon using similar data (Heavey, Layne, & Christensen, 1993).

Main Results:

  • The sign of partial correlation/regression can invert compared to zero-order correlation under specific conditions (correlation sign matches, but magnitude is smaller than the product of other correlations).
  • The magnitude can be substantially larger than the zero-order correlation when correlation signs are opposite to the product of other correlations.
  • These effects can lead to misleading conclusions if not properly understood.

Conclusions:

  • The interpretation of partial correlation and standardized partial regression requires careful attention to the interplay between multiple variables.
  • Researchers should be aware that these statistics can produce counterintuitive results regarding the strength and direction of relationships.
  • Re-evaluation of previous findings in marital satisfaction studies is suggested due to potential statistical interpretation issues.