Securing your relationship: Quality of intimate relationships during the COVID-19 pandemic can be predicted by attachment style

Abstract

The COVID-19 pandemic presents a global medical stressor and brought about unprecedented governmental regulations. The present study investigates the romantic relationships of 350 participants from Austria, Poland, Spain and Czech Republic as movement restrictions were implemented and phased out in those countries. Participants filled in questionnaires over a period of seven weeks, and we predicted relationship using machine learning models that include a variety of potential predictors related to psychological, demographic and environmental factors.On average, our models predict 32 percent of the variance of relationship quality, the most important predictors being attachment security and amount of fights. We demonstrate that attachment security is a major predictor of relationship quality in these unstable times, whereas sexual behavior, fear, pathogenic threat and the severity of governmental regulations do not influence these variables. Statement of relevanceThe COVID-19 pandemic along with the curfews that were introduced in Europe in spring 2020 presents a unique opportunity to study the effects of environmental factors on intimate relationships and identify factors that contribute to relationship quality under stressful conditions. The present study took place precisely as strict measures in response to the rising spread of the virus were adopted in Austria, Poland, Spain and Czech Republic and ends as the main restrictions were dropped. We examine predictors of relationship quality and sexual behavior in these unstable times, and found that attachment security is one of the main predictors, whereas fear of the virus or environmental factors such as the stringency of curfews do not hold any predictive value.The cross-cultural composition of our sample and the fact that our models are evaluated by their performance on unknown subjects makes these findings relatively generalizable for western cultures.

Attachment anxiety predicts worse mental health outcomes during COVID-19: Evidence from two studies

“…This may be because anxiously attached individuals are preoccupied with their partner's availability because of their fear of being abandoned ( Brennan & Carnelley, 1999 ; Cassidy, 1994 ; Shaver et al, 2005 ), thus being unable to focus on taking care of themselves. Furthermore, mounting research has shown that attachment anxiety is one of the most important predictors of poor relationship outcomes during the pandemic ( Eder et al, 2021 ). This is particularly concerning because in the moderator analyses in Study 1, we found that while people lower in attachment anxiety became less anxious and depressed over time, individuals higher in attachment anxiety remained highly anxious and depressed.…”

Food insecurity, hoarding behavior, and environmental harshness do not predict weight changes during the COVID-19 pandemic

“…Details concerning procedure, analysis and variables have been described previously and are accessible online (see https://osf.io/db4px/; Eder et al, 2020, August 22;Eder et al, 2020, September 14).…”

Dangers and Strangers: Pathogenic threat, fear, and perceived vulnerability do not predict ethnocentric orientations during the COVID-19 pandemic in Europe

“…To do so, we fit linear and non-linear machine-learning models aiming to predict inter-individual variation in identification with all humanity. These models can capture potentially complex interaction between different input variables (including if-then decision structures) and have previously been shown to make accurate predictions of psychological constructs in such samples (Eder et al, 2020(Eder et al, , 2021. We hypothesize that between-subject variation in ethnocentrism can be predicted better than by a trivial predictor.…”

“…Further, the effects may have been too weak to be detected at our sample size, or participants may have answered randomly. The same methods using the same sample were however sensitive enough to detect other associations (see Eder et al 2020Eder et al , 2021, thus ruling out random answering, and effects of viral threats on ingroup binding norms have been observed at 95% power in smaller samples (Makhanova et al, 2019).…”