Commitment strength versus commitment bolstering: Uncertainty undermines and promotes relationship success

ABSTRACT

Little research has examined factors that might weaken or strengthen commitment effects on relationship outcomes. The current research integrates attitude strength and investment model perspectives to identify uncertainty as a new moderator of commitment’s predictive ability. Consistent with an attitude strength perspective, having doubt associated with commitment undermines commitment’s predictive power. However, uncertainty can motivate uncertainty reduction achieved by acting in line with commitment. Therefore, uncertainty amplifies commitment effects on relationship maintenance, but only when behaviors are perceived to have implications for uncertainty. Across studies, certainty was found to moderate the link between commitment and relationship outcomes. Whether uncertainty weakened or strengthened commitment’s predictive power depends on an individual’s focus (implication-focused or not) when making relationship judgments.

KEYWORDS:

• Attitude strength
• commitment
• investment model
• relationship stability
• uncertainty

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data described in this article are openly available in the Open Science Framework at https://osf.io/6n4gs/.

Open scholarship

This article has earned the Center for Open Science badge for Open Materials. The materials are openly accessible at https://osf.io/6n4gs/.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

1.  We ran the predicted focal comparison between doubt versus certainty and found a significant difference in commitment slopes. While the control group’s commitment slope fell between doubt and certainty groups, neither doubt nor certainty was significantly different from the control group. See supplementary materials for dummy-coded three-group comparisons.

2.  A t-test was conducted and revealed no differences in self-esteem between the doubt (M = 3.88, SD = 0.83) and certainty conditions (M = 3.89, SD = 1.03), t(157) = −.16, p =.87.

3.  Although original validation research (Petty et al., 2002) included primes of five experiences of (un)certainty, more recent manipulations have reduced the number to two experiences (Briñol et al., 2007) in consideration of retrieval ease.

4.  Study recruitment specified a romantic relationship is necessary for participation. At onset, participants responded to the question “Are you currently in a romantic relationship?” At study conclusion, a similar question was posed allowing participants to privately admit dishonesty about meeting inclusion criteria without jeopardizing research credit. Seven people admitted fabricating a relationship and were omitted from analyses.

5.  In Study 2A, no differences in commitment emerged across doubt (M = 7.54, SD = 1.49) and certainty (M = 7.11, SD = 1.83) conditions, b = −.21, t(185) = −1.73, p =.09. In Study 2B, a t-test revealed unexpected significant differences in commitment across doubt (M = 6.29 SD = 0.95) and certainty (M = 5.84, SD = 1.28) conditions, b = −.23, t(137) = −2.38, p =.02. Surprisingly, commitment was higher in the doubt than the confidence condition. This difference does not replicate across studies and contradicts existing research showing certainty manipulations increasing commitment relative to doubt conditions (Tan & Agnew, 2016).

6.  To fit measures within the experimental period, study measures were reduced in Studies 2A and 2B. The shorter willingness to sacrifice scale (3-items) was included while the longer accommodation scale (16-items) was eliminated. In Study 1, these two indicators of pro-relationship behavior were highly correlated (r = .50) and showed parallel outcomes. Thus, intention to sacrifice is the operationalization of relationship maintenance in Studies 2A and 2B.

7.  The differences in means across Study 2A and 2B are due to the need to standardize manipulation checks in Study 2A but not Study 2B. In Study 2A, checks were standardized because of different numerical anchors across the three items. In Study 2B, numerical anchors were identical, and standardization was unnecessary. Therefore, we report the standardized means for Study 2A and the raw means for Study 2B.

8.  We assessed whether the certainty manipulation influenced self-esteem. Again, there were no mean differences in self-esteem between the doubt (Study 2A M = 3.87, SD = 1.05; Study 2B M = 3.21, SD = 1.34), and certainty conditions (Study 2A M = 3.72, SD = 1.03, Study 2B M = 3.24, SD = 1.28), ps = .33 –.90.

Additional information

Notes on contributors

Vanessa Sawicki

Vanessa Sawicki Ph.D. is an Assistant Professor of Psychology at the Ohio State University-Marion. Her research centers on attitude strength and the link between attitudes and behavior.

Christopher R. Agnew

Christopher R. Agnew, Ph.D., is Professor of Psychological Sciences and Associate Vice President for Research at Purdue University. As a social psychologist, his research focuses on close, interpersonal relationships and the use of relational models to understand social and health processes.