Attachment avoidance and amends-making: A case advocating the need for attempting to replicate one's own work

Abstract

Attachment avoidance is typically associated with negative behaviors in romantic relationships; however, recent research has begun to uncover circumstances (e.g., being in high-quality relationships) that promote pro-relationship behaviors for more avoidantly attached individuals. One possible explanation for why more avoidant individuals behave negatively sometimes but positively at other times is that their impulses regarding relationship events vary depending on relationship context (e.g., relationship satisfaction level). An initial unregistered study found support for this hypothesis in an amends-making context. We then conducted three confirmatory high-powered preregistered replication attempts that failed to replicate our initial findings. In our discussion of these four studies we highlight the importance of attempting to replicate one's own work and sharing the results regardless of the outcome.

Introduction

“…As far as a particular hypothesis is concerned, no test based upon the theory of probability can by itself provide any valuable evidence of the truth or falsehood of that hypothesis.”

[ Neyman & Pearson (1933, p. 291)]

In romantic relationships, individuals who are more avoidantly attached tend to eschew closeness and intimacy. Unsurprisingly, then, higher attachment avoidance is often associated with negative relationship outcomes (e.g., Simpson, Rholes, & Neligan, 1992) Recent studies, however, have begun to uncover circumstances in which more avoidant persons desire intimacy and behave in a pro-relationship manner (e.g., Slotter & Luchies, 2014). Why might attachment avoidance be associated with deleterious relationship outcomes in some contexts, but more salutary outcomes in others? We proposed that avoidant persons' responses to relationship-relevant situations reflect distinct impulses that are guided in part by how negatively or positively they view their current partner and relationship (see Campbell, Simpson, Kashy, & Rholes, 2001). Specifically, we believed that less satisfying relationships would foster selfish impulses for more avoidant individuals, whereas more satisfying relationships would foster pro-relationship impulses.

An initial unregistered study in our lab tested and found strong support for this hypothesis by investigating the extent to which persons higher in dispositional attachment avoidance made amends following imagining enacting a transgression against their partner as a function of relationship satisfaction and ego depletion. Armed with this empirical support, we submitted the study for peer review. Although the reviews were sympathetic with our hypothesis and theoretical perspective, the reviewers and the associate editor collectively noted that the study was limited by a small sample size (N = 104) that was perhaps less than ideal for testing our particular hypothesis. In light of the greater focus on confirmatory research and high-powered studies in both the field of relationship science and the field of social/personality psychology in general (see, e.g., Campbell et al., 2014, Finkel et al., 2015, Funder et al., 2014, Nosek et al., 2012), and the sentiments expressed in the opening quote that any given statistical test of a hypothesis does not provide unequivocal evidence of its truth or falsehood, we took the advice to heart and endeavored to replicate and extend our original study with a preregistered replication attempt using a much larger sample (N = 360). We attained a statistically significant pattern of effects in this attempt, but the results were inconsistent with the findings in our original study. We then conducted two additional preregistered replication attempts (N = 399 and N = 329) in order to elucidate the robustness of the effects; in both of the latter replication attempts, the predicted effects were not statistically significant, and when the samples of all four studies were combined, our hypothesized effects did not emerge. In this article, we discuss this research process with the goal of highlighting the importance of (a) attempting to replicate one's own work prior to submitting results for peer review and (b) sharing the results of these attempts regardless of whether or not the replications are successful. We begin by briefly explaining the theoretical rationale underlying our initial hypothesis.

Decades of attachment research suggest that two dimensions tap individual differences in adult attachment (Brennan et al., 1998, Simpson et al., 1996). The anxiety dimension reflects how much individuals worry and ruminate about being rejected or abandoned by their partners, whereas the avoidance dimension reflects how uncomfortable individuals are with closeness and intimacy in relationships. Less avoidant and less anxious persons demonstrate little concern about rejection or abandonment, and comfort with closeness.

In times of need (e.g., when threatened or distressed) the attachment system activates, motivating individuals to seek proximity to significant others (e.g., romantic partners). Whereas less avoidantly and less anxiously attached persons feel their partner will be available when needed, more avoidant and more anxious persons harbor doubts about the responsiveness of their partner, leading them to engage in secondary strategies to cope with the resulting sense of insecurity. More anxious individuals experience hyperactivation of their attachment system, demanding attention and making stronger attempts to maintain proximity to their partner. In contrast, more avoidant individuals experience deactivation of their attachment system, denying attachment needs and distancing themselves from their partner (Mikulincer and Shaver, 2003, Mikulincer and Shaver, 2007). We sought to examine the circumstances in which persons who typically lack motivation to effectively maintain their relationships may engage in relationship maintenance behaviors (i.e., amends-making). Thus, attachment avoidance was our primary focus.

The deactivating strategies employed by more avoidantly attached persons have been linked with a number of deleterious relationship outcomes. For example, more avoidant individuals tend to engage in less self-disclosure (Bradford, Feeney, & Campbell, 2002), fail to support their partner when needed (Simpson et al., 1992), and express more permissive attitudes toward relationship infidelity (DeWall et al., 2011). However, attachment avoidance is not universally associated with negative behaviors. Recent investigations have found that more avoidant individuals behave in pro-relationship ways when their partner engages in “softening” (e.g., accommodating) behaviors during conflict (Overall, Simpson, & Struthers, 2013), when they reflect on positive relationship experiences or engage in intimacy-promoting activities with their partner (Stanton, Campbell, & Pink, 2015), when they are more dependent on their relationship (Campbell et al., 2001), and when they perceive their relationship as high-quality (Slotter & Luchies, 2014).

These divergent findings perhaps suggest that, for more avoidantly attached individuals in particular, different contexts may foster distinct impulses that drive their responses to relationship events, a possibility yet to be systematically investigated. One compelling method of examining impulses is to induce ego depletion (Baumeister, Vohs, & Tice, 2007). Depletion of self-regulatory resources is thought to enhance the “default” response to situations, whether negative or positive. In the relationships domain, researchers have reasoned that ego depletion can yield harmful outcomes when impulses are negative or selfish (e.g., greater interest in romantic alternatives, Ritter, Karremans, & van Schie, 2010; more partner-related aggression, Finkel, DeWall, Slotter, Oaten, & Foshee, 2009), but salutary outcomes when impulses are positive or communal (e.g., greater willingness to sacrifice, Righetti, Finkenauer, & Finkel, 2013; more forgiveness of mild offenses, Stanton & Finkel, 2012). Importantly, the context and cues surrounding an interpersonal situation can determine the valence of an impulse (cf. Fennis, Janssen, & Vohs, 2009). We reasoned, therefore, that in relationships where the negative expectations more avoidant persons harbor are confirmed (e.g., less satisfying relationships), ego depletion should lead these persons to behave especially negatively in contexts that activate the attachment system. Conversely, in relationships where the negative expectations more avoidant individuals hold are counteracted (e.g., more satisfying relationships), ego depletion should lead them to behave particularly positively.

Our original unregistered study aimed to conceptually replicate but also extend prior research in an amends-making context. We hypothesized a three-way interaction such that when depleted (vs. non-depleted), less satisfied avoidant individuals would make fewer amends, whereas more satisfied avoidant individuals would make greater amends.¹ No differences were expected for less avoidant individuals because research suggests that they respond to their partner's distress with appropriate repair attempts (see Mikulincer & Shaver, 2007).