Inventory of Dating Stress: Initial psychometric evaluation and construct validity

Abstract

There is limited research on dating stress, or stressful experiences related to the dating process (e.g., inconsistent communication, sudden lack of response, and rude behavior). Little attention has been given to classifying stressors involved with the pursuit of potential partners from initial contact to relationship formation. In the current study, we developed a novel measure for such experiences, the Inventory of Dating Stress (IDS). We investigated the factor structure and preliminary construct validity for the IDS in an online sample of adults (n = 478). Results revealed a reliable four-factor structure (Mixed Signals, Mismatch, Ambiguous Rejection, and Harassment) across 18 items and the IDS demonstrated initial construct validity. Overall, the current study offers evidence of preliminary psychometric support for the IDS as a measure of dating stress.

Keywords

Relationship initiation stressful life events romantic relationships dating

Navigating relationships can be stressful. Many people seek psychotherapy for relationship problems (Veroff et al., 1981) and 24.2% of clinicians at college counseling centers rate relationship issues among the top problems their clients face (Center for Collegiate Mental Health, 2019). Although many studies document how such difficulties are related to mental health (Whisman & Baucom, 2012), most studies focus on relationships longer than 3 months (Bühler et al., 2021) thereby limiting our understanding of relationship problems to longer-term and/or committed relationships. As more individuals use online dating sites and mobile applications (“dating apps”; Pew Research Center, 2020; Rosenfeld & Thomas, 2012), people are increasingly likely to encounter many potential partners with whom stressful interpersonal behavior may occur, even when a committed relationship may not be established.

Individuals face a variety of potentially stressful experiences throughout the dating process, spanning from initial contact to forming a romantic relationship, that we term dating stress. On the more overt side, such experiences could include fielding rude or harassing behavior, being lied to, or facing persistent rejection based on individual characteristics or identities. More covertly, such experiences could include inconsistent communication, ambiguously canceled plans, or perceived lack of options from which to choose.

To our knowledge, empirical literature in this domain is limited. There is some research on dating anxiety (Chorney & Morris, 2008), which appears to be a manifestation of social anxiety in the dating context. Although there is a review on online dating (Finkel et al., 2012), it focuses more on a comparison to in-person dating than interpersonal experiences when dating. One study examined how various dating behaviors are perceived, yet focused on how such experiences are deemed acceptable prior to establishing a committed relationship (Taylor et al., 2013). Some studies have documented negative effects involved with dating: 45% of past-year online daters endorsed frustration with their dating experiences (Pew Research Center, 2020); dating rejection increased hostility among men (Andrighetto et al., 2019); strong beliefs in soulmates (i.e., predestined compatibility between partners) were associated with more “ghosting” behavior (e.g., ambiguous rejection without explanation; Freedman et al., 2019); and uncertainty about partner motivations and playing “hard-to-get” were related to lower perceived attractiveness of potential dating partners (Birnbaum et al., 2018, 2020; Birnbaum & Reis, 2012; Jonason & Li, 2013). Some qualitative studies have also explored how daters may deal with conflict through avoidant communication (James-Kangal & Whitton, 2019), how narratives about early stages of relationships are often marked by confusion and uncertainty (Banker et al., 2010), and how ghosting can be a problematic relationship dissolution strategy (LeFebvre et al., 2019; LeFebvre & Fan, 2020). Taken together, these studies suggest a need to characterize events that could be construed as dating stressors and to determine if they are meaningfully related to outcomes such as mental health.

Dating stressors may be particularly relevant in the study of relationship experiences among younger adults. The landscape of dating has rapidly evolved for younger generations in particular, who increasingly rely on technology for dating purposes (Finkel et al., 2012; Rosenfeld & Thomas, 2012; Pew Research Center, 2020). For example, 48% of people ages 18–29 have used an online dating website or application, compared to 38% of people ages 30–49 and 19% of people ages 50–64 (Pew Research Center, 2020). Scholars have noted that, developmentally, this age period (18–29) can be explicitly characterized by the exploration, evaluation, and initiation of romantic relationships (Clark & Beck, 2011; Stanley et al., 2011). Existing work has highlighted marked ambiguity and uncertainty about the process of relationship exploration as a major challenge in this age period (Banker et al., 2010; James-Kangal & Whitton, 2019; Stanley et al., 2011), which suggests that further study of dating experiences and potential negative correlates is warranted. Notably, although early relationship experiences can predict future relationship functioning (Collins et al., 2009), the literature has mainly focused on the initiation of committed romantic relationships rather than a range of relationship-related experiences, such as dating stressors, that could occur before a relationship is established.

Thus, to fill this gap in the literature, the purpose of this study was to develop a measure of potentially stressful dating experiences (i.e., dating stressors), called the Inventory of Dating Stress (IDS). Such a measure expands the range of dating-related experiences prior to beginning a romantic relationship that can be examined in empirical research. We intended this measure to encompass dating broadly, whether online or in-person. There were two primary aims to examining the psychometric properties of the IDS: (1) to establish the latent factor structure of an initial pool of items that capture a variety of dating experiences that could be deemed stressful, and (2) to investigate initial construct validity. Secondarily, we conducted some initial exploratory measurement invariance analyses to investigate possible differences in item functioning across key demographic variables (age, sexual orientation, gender identity, and race/ethnicity).

To provide evidence of construct validity, we tested whether the IDS was related to weekly dating app usage, mental health, fear of intimacy, and relationship decision-making skills. We pulled from theories of interpersonal stress and mental health (Hammen, 2005), attachment (Bowlby, 1982), and social cognition (Bandura, 1969) to evaluate the IDS. First, at a minimum, we expected that more time spent using dating apps would mean more engagement with potential partners and thus more opportunity for dating stressors to occur. Second, given that stress, particularly of an interpersonal nature, is a common and well-studied contributing factor to internalizing disorders (Hammen, 2005), we expected that the IDS would be related to greater depressive and general anxiety symptoms. Such associations would be consistent with previous studies on relationship functioning (Whisman & Baucom, 2012) and dating experiences, discussed above, that document negative effects of rejection and uncertainty, which are also correlated with depressive and anxiety symptoms (Einstein, 2014; Slavich et al., 2010). Third, in line with attachment theory, we expected that a measure related to stressful experiences in the exploration of potential close relationships (e.g., rejection) could substantially overlap with attachment-related processes such as fear of intimacy. Attachment theory posits that avoidance of intimacy is a key aspect of insecurity in relationships, which can result in people experiencing, as stressful, behaviors that involve potential closeness with others and/or behaviors that involve approaching others. Because prior studies have shown that fear of intimacy is correlated with misperceptions of potential partners (Birnbaum & Reis, 2012), decreased self-disclosure, and shorter romantic relationship duration (Descutner & Thelen, 1991), we wanted to examine the degree to which self-reports of dating stressor experiences could be reflective of avoidance or discomfort with getting close to new people in the context of potential rejection. Fourth, in line with social cognitive theory, we expected the IDS to be related to self-efficacy in making decisions about healthy relationships (i.e., confidence in potential future relationships, recognition of relationship warning signs, and deliberate relationship decision-making; Vennum & Fincham, 2011). Social cognitive theory posits that people form beliefs about themselves and their behavior (i.e., self-efficacy) based on the experiences they have with others. We expected that more frequent difficult interpersonal experiences when dating—which requires evaluation of potential partners—would be related to lower confidence in one’s ability to make effective choices about potential partners and more negative expectations about future relationships.

In support of convergent validity, we hypothesized that the IDS would be positively correlated with weekly dating app usage, mental health symptoms, and recognition of relationship warning signs. Further, we hypothesized that the IDS would be negatively correlated with relationship confidence and relationship decision-making. In support of discriminant validity, although we expected some overlap given the nature of these constructs, we hypothesized that the IDS would be positively but not strongly correlated with fear of intimacy (i.e., correlations >.7 that would indicate substantial redundancy).

Method

Participants and procedure

Participants were part of an online study focused on attitudes and experiences related to sexuality, dating, and relationships among people of various sexual identities. This study, conducted from August to October 2019, recruited participants via online and social media platforms (e.g., Facebook, Reddit, and Tumblr). Participants completed a 1-minute screener and, if eligible, completed self-report questionnaires via Qualtrics. Participants were eligible if they were 18 years of age or older, resided in the United States, and provided consent. Participants who completed the survey were eligible to enter a raffle for one of ten $50 gift cards. The Institutional Review Board at Stony Brook University approved all study procedures.

We screened responses based on minimum survey completion, attentive responding, and current identification as an active dater. Overall, 2458 responses were recorded. From there, 315 were not eligible, 396 did not consent, and 128 did not make progress beyond demographics. Of the 1619 eligible responses, 13 were excluded due to inattentive responding, defined as three or more incorrect responses on the Directed Questions Scale (Maniaci & Rogge, 2014) that consists of seven items placed randomly throughout the survey (e.g., “Please skip this question”) as validity checks. To select for those who were more likely to have had recent dating experiences, we only displayed the IDS questions to participants if they indicated that they had dated within the past year. From there, 23 participants were excluded because they had indicated in a free-response question after the IDS questions that the scale was not applicable for them (e.g., they had misunderstood the stem question and did not date over the past year). We then excluded 556 participants who indicated that they did not date in the past year and 139 participants who did not complete more than 70% of IDS items. Finally, we excluded 410 participants who indicated that they were not an active dater, leaving us with a final analytic sample of 478 participants.

Sample demographic information is displayed in Table 1. Participants were generally in their late twenties (M = 29.6, SD = 10.03, range = 18–68), LGBTQ+ (78.5%; i.e., lesbian, gay, bisexual, transgender, or queer), and mostly non-Hispanic White (76.4%). About one-third of the sample consisted of students (26.4%) and people in the college age range (29.1%). Participants were geographically diverse, with the highest percentages living in New York (11.7%), California (9.6%), Washington (5.4%), and Pennsylvania (5.2%). The majority of the sample was recruited from Reddit (42.3%, n = 202) and Facebook (39.7%, n = 190).

Table 1.

Study demographic information.

Variable
N	478
Age in years, M (SD), range ^a	29.58 (10.03), 18–68
Sexual orientation, n (%)
Straight	103 (21.5%)
Gay	31 (6.5%)
Lesbian	29 (6.1%)
Bisexual	109 (22.8%)
Pansexual	101 (21.1%)
Queer	59 (12.3%)
Fluid	6 (1.3%)
Asexual	13 (2.7%)
Unsure/questioning	11 (2.3%)
Other	16 (3.3%)
Gender identity, n (%)^b
Woman	231 (48.3%)
Man	106 (22.2%)
Transmasculine	28 (5.9%)
Transfeminine	28 (5.9%)
Gender diverse	85 (17.8%)
Race/ethnicity, n (%)^a
Non-Hispanic White	365 (76.5%)
Hispanic White	23 (4.8%)
Hispanic/Latinx	3 (0.6%)
Black/African American	24 (5.0%)
Asian	17 (3.6%)
Native American/Alaska Native	4 (0.8%)
Multiracial	41 (8.6%)
Annual household income, n (%)^c
Under $12,000	67 (14.0%)
$12,000–$23,999	72 (15.1%)
$24,000–$44,999	99 (20.7%)
$45,000–$74,999	99 (20.7%)
$75,000–$119,999	79 (16.5%)
$120,000 or more	60 (12.6%)
Employment^a
Full-time work	228 (47.7%)
Part-time work	69 (14.4%)
Part-time work/full-time student	68 (14.2%)
Temporary/occasional work	12 (2.5%)
Disabled and not working	16 (3.3%)
Unemployed, student	58 (12.1%)
Unemployed, other	26 (5.4%)
Education level, n (%)
Did not finish high school	6 (1.2%)
High school diploma/GED	53 (11.1%)
Some college	147 (30.8%)
Associate’s degree	45 (9.4%)
4-year college degree	129 (27.0%)
Some graduate school	22 (4.6%)
Master’s degree	56 (11.7%)
Advanced graduate degree	20 (4.2%)

^aOne participant was missing data on these variables.

^bInvestigator-created categories based on a multiple-choice and write-in options. Transmasculine/transfeminine reflects gender identities defined by trans experience. Gender diverse reflects gender identities outside the male/female binary (e.g., nonbinary).

Two participants were missing data on this variable.

Measures

All measures, except the IDS and dating app usage, are commonly used and have shown good psychometric properties in prior research.

Dating stress

IDS items were generated from initial pilot work on dating experiences. Through informal polling of friends, students, and colleagues of our research team members consisting of undergraduate research assistants and graduate students, we asked people to share common scenarios and events encountered when dating that they considered to be stressful or hard. We used these open-ended responses to generate item content and then consulted with colleagues who conduct research on romantic relationships for additional items. Overall, 51 items were generated from this process and covered potential stressors related to the dating process from initial potential partner contact to the initiation of a committed relationship (see Supplemental Material Appendix S1). The stem for each item was “Taking into account all of your dating experiences over the past year, how many times have the following dating experiences happened to you?” and was rated on a 6-point Likert scale (0 = None, 5 = 10 or more times). Items referred to a potential partner, defined as anyone considered for a date or romantic relationship whether this was online, through text, or in person. Broadly, question content spanned digital communication (e.g., messaging through text or dating platforms), relationship expectations, ambiguous or unclear rejection (e.g., sudden communication drop, “ghosting”), overt rejection, rude or mean behavior from potential partners, and decision points about relationships (e.g., establishing a casual or committed relationship). We created mean composite scale scores (possible range = 0–5), where higher scores reflect more frequent past-year dating stressor experiences.

Weekly dating app usage

Among participants who indicated using online or mobile dating applications (n = 382), we assessed weekly amount of time spent on these platforms with a single item: “How many hours per week do you spend on online dating websites or mobile dating applications (e.g., Tinder, Bumble, Grindr, Scruff, Feeld, FetLife, Her, eHarmony, OKcupid, Match.com, etc.)?” rated on a 10-point scale (0 = < 1 hour/week, 10 = 10+ hours/week). Higher scores reflect more time spent using dating apps.

Depressive symptoms

Depressive symptoms were measured with the Patient Health Questionnaire (9-item form; Kroenke et al., 2001), a brief measure of depressive symptom severity widely used as a screener in outpatient settings. The 9-item scale asks participants to rate on a 4-point Likert scale (0 = Not at all, 3 = Nearly every day) how often they have been bothered by symptoms of depression (e.g., “feeling down, depressed, or hopeless”) over the past 2 weeks. Items are summed for a total severity score, Cronbach’s α = .90, and higher scores reflect greater past 2-week depressive symptoms (possible range = 0–27).

General anxiety symptoms

General anxiety symptoms were measured with the Generalized Anxiety Disorder Scale 7-Item version (Spitzer et al., 2006), a brief measure of generalized anxiety disorder severity. The scale contains seven items where participants rate on a 4-point Likert scale (0 = Not at all, 3 = Nearly every day) how often they have been bothered by anxiety symptoms (e.g., “not being able to stop/control worrying”) over the past 2 weeks. Items are summed for a total severity score, Cronbach’s α = .93, and higher scores reflect greater past 2-week anxiety symptoms (possible range = 0–21).

Relationship decision-making

We measured relationship decision-making skills with the Relationship Deciding Scale (Vennum & Fincham, 2011), a 12-item general measure of self-efficacy for making effective choices regarding potential romantic relationship partners. Items are typically rated on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree); however, due to a programming error, in this study, items were rated on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree) and summed to form three subscales: (1) Relationship Confidence (example item: “I feel good about the prospects of making a romantic relationship last.”), Cronbach’s α = .86, where higher scores reflect greater confidence about future relationships (possible range = 4–28); (2) Warning Signs (example item: “I am quickly able to see warning signals in a romantic relationship.”), Cronbach’s α = .83, where higher scores reflect greater recognition of relationship warning signs (possible range = 3–21); and (3) Relationship Deciding (example item: “It is important to make conscious decisions about whether to take each major step in romantic relationships.”), Cronbach’s α = .71, where higher scores reflect greater care in making important relationship decisions (possible range = 5–35).

Fear of intimacy

We measured fear of intimacy with the 35-item Fear of Intimacy Scale (Descutner & Thelen, 1991), which assesses anxiety and trepidation about sharing intimate feelings with a partner. Items are rated on a 5-point Likert scale (1 = not at all characteristic of me, 5 = extremely characteristic of me). All 35 items are averaged for a total score, Cronbach’s α = .91, with higher scores reflecting greater fear of intimacy (possible range = 1–5).

Data analysis

Exploratory factor analysis

We conducted exploratory factor analysis (EFA) with quartimin oblique rotation using MPlus version 8.2 to explore the initial latent structure of past-year dating stressor frequency. All IDS items were positively skewed, so we used robust maximum likelihood estimation to account for non-normality of indicators (Yuan & Bentler, 2000). We examined the scree plot and a parallel analysis with 500 random datasets to determine how many factors to extract. We used the following cutoffs to guide decisions about model fit: the comparative fit index (CFI; Bentler, 1990) and the Tucker–Lewis index (TLI; Tucker & Lewis, 1973) greater than .95 and root mean square error of approximation (RMSEA) less than .06 (Hu & Bentler, 1999) as evidence of excellent model fit, and CFI and TLI greater than .90 and RMSEA between .05 and .10 for acceptable model fit. We also report the χ² goodness of fit test, although it is overpowered with large sample sizes (Brown, 2015).

To construct the scale, we retained items from EFA analyses with standardized factor loadings greater than .40 as a meaningful contribution (e.g., >16% of the variance). Additionally, we relied on substantive interpretation in examining factor solutions and excluded items that formed poorly defined factors. Subsequent EFAs without poorly fitting items were conducted and this process was repeated until we found an interpretable factor structure with adequate fit.

Descriptive statistics and construct validity

Once the scale was finalized, we computed composite subscale scores to examine descriptive statistics and internal consistency reliability with Cronbach’s α. We utilized Pearson’s product-moment correlations to investigate the strength and direction of the relationship between the IDS subscales and construct validity measures. We screened all variables for univariate outliers and winsorized outlying values with z-scores more extreme than ±3.29. We utilized all available data where possible.

Exploratory measurement invariance

Because our sample consisted of people with diverse identities, we wanted to explore whether the psychometric properties of the IDS operated differently among sample sub-groups to inform future research. Typically, this is tested by examining measurement invariance using multiple groups confirmatory factor analysis (CFA). However, moving to a CFA framework at this stage of measurement development would be premature due to strict CFA assumptions of zero indicator cross-loadings and residual error variances, which could bias even exploratory tests of measurement invariance without further information about how to specify these parameters. To circumnavigate this issue, we used exploratory factor analysis in a confirmatory factor analysis framework (E/CFA; Brown, 2015; Jöreskog, 1969). E/CFA can be viewed as a “stepping stone” from EFA to CFA because E/CFA has an identical number of restrictions and identical model fit to EFA (see below; Brown, 2015), which would more closely mirror our initial EFA results. E/CFA allows for the inclusion of covariates that can be regressed onto latent factors and indicator intercepts to test for these two sources of possible measurement invariance, consistent with the multiple indicators, multiple causes modeling approach to invariance testing (Brown, 2015).

Our E/CFA model was conducted in R version 4.0.2 (R Core Team, 2020) using the “lavaan” package (Rosseel, 2012) with robust maximum likelihood estimation. We specified the E/CFA model following recommendations suggested by Brown (2015), whereby factor covariances and most item cross-loadings were freely estimated. Models were identified by fixed-factor scaling of latent variables and one indicator per factor chosen to serve as an anchor item that had cross-loadings on all other factors fixed to zero. Items with the strongest factor loading and lowest cross-loadings were selected as anchor items based on EFA results.

We explored measurement invariance across age, sexual orientation, gender identity, and race/ethnicity in separate models. Age was a continuous variable. Sexual orientation was coded into heterosexual (n = 103) as 0 and sexual minority identity (n = 375; i.e., any other non-heterosexual identity) as 1. Gender identity was coded into cisgender (n = 337) as 0 and gender minority identity (n = 141; i.e., any other non-cisgender identity) as 1. Race/ethnicity was coded into non-Hispanic White (n = 365) as 0 and people of color (n = 112) as 1. To test for measurement invariance, these variables were regressed onto (1) each IDS factor, which tests for differences in latent means across levels of the demographic variable, and (2) each IDS item, which tests for differences in item intercepts across levels of the demographic variable. This model specification allowed us to estimate the unique association of demographic variables with latent means and indicator intercepts. In other words, latent factor differences were estimated when controlling for intercept invariance and intercept invariance was estimated when controlling for latent factor differences. Conceptually, for example, this allowed us to account for the possibility that intercept invariance could be reflective of true group differences in the construct, if present, rather than different item interpretation across groups.

Because we did not have hypotheses about which items might be invariant, we followed recommendations by Brown (2015). We first estimated a model where these paths (i.e., demographic variable regressed on latent means and indicators) were fixed to zero and then inspected modification indices to determine which of these paths, if freed, would result in substantial model improvement (modification index >4.0). We freed individual paths with the highest modification index sequentially until no modification indices greater than 4.0 were present.

Missing data

Missing data across IDS items were sparse, with only five missing responses across all items, and were handled through maximum likelihood estimation. Although only 382 individuals endorsed using an online or mobile dating app, there were no missing data on this variable. Missing data for the remaining variables were 5.2% (n = 25) for relationship confidence, 5.4% (n = 26) for warning signs, 5.2% (n = 25) for deciding, 10.5% (n = 50) for fear of intimacy, and 0.2% (n = 1) for depressive symptoms. There were no missing data for general anxiety symptoms. Given little missing data and no significant mean differences in IDS scores between participants missing data on relationship decision-making (all subscales) and fear of intimacy versus those not, we used pairwise deletion for construct validity analyses.

Results

Exploratory factor analyses

No items were candidates for deletion based on low correlations with other items or possible multicollinearity, so we preceded to conduct an initial EFA on all 51 items. Examination of the scree plot suggested a four- or five-factor structure (sample eigenvalues: 12.4, 2.8, 2.5, 1.8, and 1.7). Parallel analysis suggested that extracting more than four factors would not account for more variance than random data. Given this discrepancy, we investigated both four- and five-factor solutions. The four-factor model was not a good fit to the data, χ² (1077) = 2295.16, p < .001, CFI = .81, TLI = .78, RMSEA = .05, nor was the five-factor model, χ² (1030) = 2026.14, p < .001, CFI = .84, TLI = .81, RMSEA = .05, with the exception of the RMSEA for both models. All factors appeared to be substantively interpretable across both solutions. We utilized a conservative approach for adjudicating between candidate items for deletion across four- and five-factor models by only deleting items that did not have a salient factor loading (e.g., greater than .40) on any factor across both models. We deleted 22 items based on this criterion (items 3, 11, 14, 16–20, 22, 25, 29, 30, 32, 33, 35–39, 41, 46, and 47).

We ran a second EFA with the remaining 29 items. Again, examination of the scree plot suggested a four- or five-factor structure (sample eigenvalues: 8.6, 2.5, 2.2, 1.5, and 1.2) while parallel analysis suggested a four-factor structure, so we investigated both solutions. The four-factor model was not a good fit to the data, χ² (296) = 837.04, p < .001, CFI = .87, TLI = .81, RMSEA = .06, nor was the five-factor model, χ² (271) = 617.79, p < .001, CFI = .91, TLI = .87, RMSEA = .05, although the CFI suggested adequate fit for the five-factor model. We deleted items 10 and 51 due to lack of salient factor loadings across both models.

We ran a third EFA with the remaining 27 items. Both examination of the scree plot and parallel analysis suggested four factors (sample eigenvalues: 8.0, 2.5, 2.2, and 1.5); thus, we chose to only investigate the four-factor model moving forward in favor of parsimony and to avoid overextraction. The four-factor model was not a good fit to the data, χ² (249) = 669.24, p < .001, CFI = .89, TLI = .84, RMSEA = .06. We deleted items 23, 24, and 45 due to lack of salient factor loadings.

We ran a fourth EFA with the remaining 24 items. The four-factor model (sample eigenvalues: 7.3, 2.3, 2.0, and 1.3) fit the data adequately, χ² (186) = 403.23, p < .001, CFI = .93, TLI = .90, RMSEA = .05. We deleted items 21, 26, 34, 44, 48, and 49 due to lack of salient factor loadings.

Our fifth and final EFA revealed a four-factor structure (sample eigenvalues: 6.1, 2.0, 1.7, 1.1) with 18 items and excellent fit to the data, χ² (87) = 116.77, p = .02, CFI = .99, TLI = .98, RMSEA = .03 (see Table 2). No items had factor loadings less than .40 and all factors were interpretable. We labeled the first factor (33.8% of the variance; items 1, 2, 4, and 5) as the “Mixed Signals” subscale, reflecting the experience of receiving conflicting messages from potential partners. We labeled the second factor (11.3% of the variance; items 6, 7, 8, 9, 13, and 15) as the “Mismatch” subscale, reflecting discordant expectations and uncertainty regarding potential partners. We labeled the third factor (9.5% of the variance; items 12, 28, 31, 40, and 50) as the “Ambiguous Rejection” subscale, reflecting situations where rejection is not explicitly stated, but implied through a lack of response. We labeled the fourth factor (6.2% of the variance; items 27, 42, and 43) as the “Harassment” subscale, reflecting situations where a potential partner was rude, harassing, or responded to rejection in a poor or argumentative manner. Together, all four factors accounted for 60.8% of item variance and were significantly correlated, rs = .25–.49. The final 18-item IDS is presented in Supplemental Material Appendix S2.

Table 2.

Standardized factor loadings from final exploratory factor analysis.

Item	Mixed signals	Mismatch	Ambiguous rejection	Harassment
1. You were led on by a potential partner	.793	−.118	.067	.031
2. A potential partner was unclear about their dating intentions	.626	.144	.042	.027
4. You got mixed signals from a potential partner	.511	.270	.043	.087
5. Potential partners were not honest about how they felt about you	.487	.331	−.071	.003
6. You had difficulties communicating with a potential partner	−.053	.749	.048	−.029
7. A potential partner was emotionally closed off from you	.139	.609	.003	−.007
8. You experienced clingy behavior from a potential partner	−.152	.575	−.109	.141
9. You had different expectations for frequency of communication from a potential partner	.118	.492	.025	.120
13. It was not clear how a potential partner felt about you	.317	.419	.169	−.042
15. You were not on the same page with a potential partner	.131	.518	.152	.099
12. You were interested in a potential partner who was not interested in you	.130	.087	.564	−.078
28. Potential partners never responded to your initial messages or texts	−.134	.031	.796	−.030
31. You talked with a potential partner online or through text but never met in person even though you wanted to	.009	−.026	.708	.051
40. A potential partner stopped communicating with you suddenly (they “ghosted” you)	.136	−.022	.682	.059
50. A potential partner was not interested even though the date went well	.111	−.022	.449	.078
27. You experienced rude behavior/harassment from a potential partner online, through mobile dating apps, or through text	.033	.022	.220	.463
42. Potential partners put you down when you rejected them	.030	−.044	−.010	.927
43. Potential partners tried to argue with you when you rejected them	−.036	.060	−.005	.863

Although our goal was to establish how dating stressors covary, we conducted a supplementary analysis to examine whether items were perceived as stressful. When participants endorsed an IDS item (i.e., response of “1 = Once” or higher), they received an additional item: “Thinking back over the past year, how bothered were you by this?” (rated from 0 = Not at all bothered to 4 = Extremely bothered). Descriptive statistics for all items showed sufficient variability (i.e., items ranged 0–4) to suggest that IDS items are not inconsequential; participants were, on average, moderately bothered by the stressors represented by the final IDS. The median rating was 2 (“Moderately bothered”) across most items, with a few exceptions. One item (15) had a median rating of 1.5 and five items (12, 28, 31, 50, and 42) had a median rating of 1 (“A little bit or somewhat bothered”).

Descriptive statistics and construct validity

Based on the factor analyses, we created unweighted mean scores such that each subscale score represents the average frequency of past-year dating stressors. Descriptive statistics and bivariate correlations for the subscale scores are presented in Table 3. All subscales demonstrated adequate internal consistency reliability, Cronbach’s α = .80 for the Mixed Signals subscale, α = .80 for the Mismatch subscale; α = .80 for the Ambiguous Rejection subscale; α = .77 for the Harassment subscale; and α = .88 for the total IDS score.

Table 3.

Bivariate correlations and construct validity.

Variable	1	2	3	4	5	6	7	8	9	10	11	12
1. IDS Mixed Signals	—
2. IDS Mismatch	.61^***	—
3. IDS Ambiguous Rejection	.46^***	.35^***	—
4. IDS Harassment	.34^***	.36^***	.32^***	—
5. IDS Total	.80^***	.82^***	.75^***	.59^***	—
6. Weekly Dating App Usage	.31^***	.18^***	.33^***	.13^*	.32^***	—
7. RDS Confidence	−.12^*	−.10^*	−.07	.02	−.10^*	−.11^*	—
8. RDS Warning Signs	−.02	−.06	.04	.11^*	.01	−.05	.29^***	—
9. RDS Deciding	−.04	−.03	−.03	−.01	−.04	−.16^**	.21^***	.29^***	—
10. Fear of Intimacy	.19^***	.20^***	.08	.09^†	.19^***	.07	−.51^***	−.13^**	−.03	—
11. Depressive Symptoms	.22^***	.18^***	.04	.08^†	.17^***	.12^*	−.15^**	−.16^***	−.05	.24^***	—
12. Anxiety Symptoms	.17^***	.18^***	.01	.10^*	.15^***	.10^†	−.04	−.19^***	.02	.17^***	.77^***	—
M	1.39	1.50	1.46	0.68	1.33	1.30	22.77	14.02	25.31	2.11	9.61	9.08
SD	1.06	1.01	1.15	1.06	0.81	1.90	4.46	4.00	4.96	0.58	6.87	6.08
Range	0–4.5	0–5	0–4.6	0–5	0–4.7	0–9	5–28	3–21	8–35	1–4.7	0–27	0–21

Note. *** p < .001, ** p < .01, * p < .05, † p < .10.

Results generally supported construct validity. All IDS subscales were positively correlated with weekly dating app usage. Relationship confidence was negatively correlated with Mixed Signals and Mismatch, but not Ambiguous Rejection or Harassment. Recognition of relationship warning signs was positively correlated with Harassment, but not Mixed Signals, Mismatch, or Ambiguous Rejection. Relationship deciding was not significantly correlated with any of the IDS subscales. All IDS subscales were correlated with fear of intimacy to a small-to-moderate degree or were not significant. Depressive and anxiety symptoms were positively correlated with the Mixed Signals and Mismatch subscales, but not the Ambiguous Rejection subscale. Anxiety symptoms were positively correlated with the Harassment subscale, but the correlation between depressive symptoms and the Harassment subscale did not reach statistical significance.

Exploratory measurement invariance

Our EFA results showed that items 1, 6, 31, and 42 had the highest factor loadings and lowest cross-loadings. These items were selected as anchor items with fixed cross-loadings in our initial E/CFA without demographic variables. Model fit was identical to the final EFA reported above and the pattern of item cross-loadings from this solution was similar in strength and magnitude. We then proceeded to test separate models with demographic variables regressed onto the latent IDS factors and item intercepts, with all paths fixed to zero in initial models.

Results from each E/CFA model with demographic variables are presented in Table 4. For brevity, we describe the final model results. Models with age showed that older adults showed higher latent mean levels of Ambiguous Rejection, B = .141, p = .005, and, at mean levels of Mismatch, lower average endorsement of item 6 (“You had difficulties communicating with a potential partner”), B = −.09, p = .010. Models with sexual orientation showed that those with a sexual minority identity had higher latent mean levels of Mismatch, B = .312, p = .005, and at mean levels of Harassment, average endorsement of item 43 (“Potential partners put you down when you rejected them.”) was lower, B = −.16, p = .040, and average endorsement of item 27 (“You experienced rude behavior/harassment from a potential partner online, through mobile dating apps, or through text.”) was higher, B = .309, p = .041, among those with a sexual minority identity. Models with gender identity showed that those with a gender minority identity had lower latent mean levels of Mixed Signals, B = −.344, p = .001, and at mean levels of Mismatch, average endorsement of item 13 (“It was not clear how a potential partner felt about you”) was higher among those with a gender minority identity, B = .319, p = .001. Finally, models with race/ethnicity showed that at mean levels of Ambiguous Rejection, average endorsement of item 40 (“A potential partner stopped communicating with you suddenly (they “ghosted” you)”) was higher among people of color, B = .275, p = .015. Supplemental Material Table S1 presents a summary of which latent factors and IDS items exhibited measurement noninvariance across demographic variables examined.

Table 4.

E/CFA exploratory measurement invariance results.

Variable	Model	χ², p	CFI	TLI	RMSEA	Δχ², p	Freed parameters
Age
	Model 1	152.93, p = .002	.987	.969	.031	—	—
	Model 2	144.61, p = .005	.984	.973	.029	8.46, p = .003	η: Ambiguous Rejection
	Model 3	139.52, p = .010	.985	.976	.027	5.78, p = .016	η: Ambiguous Rejection; τ: Item 6
Sexual orientation (n = 103 heterosexual, n = 375 sexual minority identity)
	Model 1	146.90, p = .004	.983	.973	.029	—	—
	Model 2	141.42, p = .009	.985	.975	.027	6.38, p = .012	η: Mismatch
	Model 3	135.72, p = .017	.987	.978	.026	8.97, p = .003	η: Mismatch; τ: Item 43
	Model 4	132.07, p = .024	.988	.980	.025	4.12, p = .042	η: Mismatch; τ: Item 43, 27
Gender identity (n = 337 cisgender, n = 141 gender minority identity)
	Model 1	151.21, p = .002	.981	.970	.030	—	—
	Model 2	142.17, p = .008	.985	.975	.028	12.48, p < .001	η: Mixed Signals
	Model 3	132.91, p = .025	.988	.980	.025	14.58, p < .001	η: Mixed Signals; τ: Item 13
Race/ethnicity (n = 365 non-Hispanic White, n = 112 people of color)
	Model 1	138.57, p = .016	.986	.978	.026	—	—
	Model 2	133.82, p = .026	.988	.980	.025	5.51, p = .019	τ: Item 40

Note. CFI = comparative fit index. TLI = Tucker–Lewis Index. RMSEA = root mean square error of approximation. η = factor mean. τ = intercept. Model fit was compared using the Satorra–Bentler scaled chi-squared difference test.

Discussion

The current study reports the development and initial psychometric evidence for a novel measure, the Inventory of Dating Stress (IDS), among a diverse online sample. Exploratory factor analyses supported a reliable factor structure characterized by four domains of dating stressors: Mixed Signals (i.e., conflicting messages from potential partners), Mismatch (i.e., different expectations and uncertainty regarding potential partners), Ambiguous Rejection (i.e., unassertive behavior regarding relationship dissolution), and Harassment (i.e., problematic behavior and issues surrounding how rejection is handled). Although the IDS subscales measure the frequency of dating stressors experienced over the past year and do not, at this point, account for stress appraisal, supplementary analyses indicated that participants were at least moderately bothered by the stressors included in the IDS. Further, the IDS demonstrated initial construct validity.

This study adds to the body of work on romantic relationships by contributing a measure of experiences related to relationship exploration that could occur prior to establishing a committed partnership. Existing models of relationship initiation have focused primarily on processes after initial attraction to a potential partner has already been established (Clark & Beck, 2011; Murray et al., 2006), which may be limited by a relatively narrow focus of relationship exploration processes after this point in time. Our results suggest that there are other, at least moderately stressful relationship experiences that may occur before initial attraction is even properly evaluated. Thus, further research on dating stress is needed and may hold promise for expanding our understanding of how people navigate relationship exploration.

Construct validity

Overall, results showed initial evidence of convergent and discriminant validity for the IDS. Our construct validity hypotheses for the Mixed Signals and Mismatch subscales, which mirrored the Total IDS scale, were largely supported. More frequent dating stressors overall, and those related to Mixed Signals and Mismatch, were correlated with weekly dating app usage, relationship decision-making self-efficacy, fear of intimacy, and mental health symptoms in expected directions. However, with regard to relationship decision-making self-efficacy, more frequent Mixed Signals and Mismatch dating stressors were related to lower confidence in potential future relationships but not recognition of relationship warning signs or decision-making skills. It is worth noting that recognition of relationship warning signs and decision-making skills, as measured by the Relationship Deciding Scale (Vennum & Fincham, 2011), involve choices related to establishing a committed, long-term partnership with one person rather than ascertaining interest in one or more potential partner(s) who may be worthy of exploration. This focus on serious relationships reflects a limitation of existing work that we note above and, importantly, may be why we did not obtain significant results with these two dimensions of relationship decision-making. The experiences captured by the Mixed Signals and Mismatch subscales cover those that happen at relatively early stages of a relationship, which may require a different set of decision-making skills than those involved in committing to a single partner.

In contrast, most of our construct validity hypotheses for the Ambiguous Rejection subscale were not supported, contrary to expectations and literature on rejection and mental health (Einstein, 2014; Slavich et al., 2010). Only weekly dating app usage was positively correlated with more frequent Ambiguous Rejection as expected. There may be characteristics of Ambiguous Rejection that are different from more overt social rejection. Notably, items on the Ambiguous Rejection subscale reflected a range of rejection experiences that may be experienced differently (e.g., no initial response to a dating app message vs. being ghosted). Some of these items possibly lack a concrete threat to self-image, potentially due to expectations that this could occur or self-distancing processes (i.e., cognitive coping). Future research is needed to examine these possibilities.

Our construct validity hypotheses for the Harassment subscale were partially supported. More frequent Harassment was positively correlated with weekly dating app usage, relationship decision-making self-efficacy, and anxiety symptoms as expected. Correlations between Harassment and fear of intimacy and depressive symptoms, respectively, were at the trend level and of similar magnitude as the correlation between Harassment and anxiety symptoms. The Harassment subscale was positively correlated with recognition of relationship warning signs, but not confidence in future relationships or relationship decision-making skills. Although this pattern of results related to self-efficacy is opposite to the Mixed Signals subscale, it is conceptually plausible. Items on the Harassment subscale capture hostile experiences with potential partners, such as being put down, that should be considered red flags or warning signs in potential partners. It is possible that such experiences do not relate to other domains of relationship self-efficacy because it may be easier to dismiss Harassment experiences. Potential partners should typically be motivated to present a positive view of themselves at early stages of dating (Finkel et al., 2012), and thus, Harassment may be more easily attributed to an individual potential partner than the self.

Generally speaking, the overall pattern of construct validity results was in accord with theoretical predictions and suggests implications for future research. Results showed that those who reported more frequent dating stressors reported greater fear of intimacy, in line with attachment theory (Bowlby, 1982). The dating stressors captured by the IDS were characterized by ambiguity about where one stands with a potential partner and include experiences that could lead to rejection, which could possibly activate attachment-related fears of abandonment in a context where people are seeking to establish close relationships with new people. Results also showed that those who reported more frequent dating stressors reported lower confidence in their abilities for successful future relationships, in line with social cognitive theory (Bandura, 1969). The more frequently that dating stressors marked by mixed signals and partner mismatch are encountered, the more likely it could be that people will come to learn that such experiences are likely to happen in future relationships. Although future research is needed to test such possibilities, both frameworks lead to questions for further exploration. For example, future research could explore whether and how more frequent dating stressors relate to relationship avoidance and instability (e.g., through avoidance of potential rejection, in line with attachment theory, or based on past dating experiences, in line with social cognitive theory), as well as how they may contribute to working models of love and relationships.

Exploratory measurement invariance

In preliminary analyses exploring measurement invariance, we also found that the IDS generally functioned similarly across various demographic groups. We did not find any IDS items that consistently demonstrated a pattern of intercept noninvariance across models with age, sexual orientation, gender identity, and race/ethnicity. Items exhibiting intercept noninvariance varied across factors and models. Accordingly, future researchers should qualify interpretations of potential group differences in dating stressors tentatively, depending on the factor and demographic group considered. However, these results should be regarded as exploratory and interpreted with caution due to the limitations involved with our approach. Analytically, we could only explore two potential sources of invariance (i.e., latent means and intercept invariance) while assuming that other potential sources (e.g., structure and factor loadings) were the same across groups. Further, use of modification indices to identify invariant parameters could capitalize on sampling error (MacCallum, 1986), so replication is needed. Although this was unavoidable, we hope these preliminary results inform future work.

Limitations

Our findings must be interpreted within the context of study limitations. First, although our sample included people with an array of dating experiences ideal for establishing the initial IDS scale structure, results may not generalize to older, less educated individuals. Our sample was also mostly non-Hispanic White and predominately LGBTQ+ identified, and thus, additional cross-validation is necessary. Further, the dichotomization of demographic variables needed for exploratory measurement invariance analyses may have obscured important sub-group differences in IDS response patterns. Our selection criteria also focused on people who are actively dating, which does not necessarily capture effort put into dating or time spent dating. This criterion may have captured more “experienced” daters, thus limiting generalizability to new or inexperienced daters who may experience dating stressors differently. Second, low mean IDS subscale scores could lead to floor effects, suggesting that different response scales (e.g., more restricted frequencies) could be considered in future iterations of the IDS. Third, items for the IDS were not randomly presented to participants, which could have overinflated endorsement on related items presented together. Fourth, the IDS items require participants to average across a wide array of scenarios to determine the relative frequency of dating stressors, thus potentially introducing substantial recall bias. Additionally, frequency of endorsement on IDS items may stem from the same situation or person, as the IDS items as they are constructed currently cannot distinguish between people and situations (i.e., ghosted by five different people or ghosted five times by the same person). Fifth, all construct validity analyses are cross-sectional, and thus, bidirectional interpretations are possible.

Clinical implications and conclusion

Because dating stressors are associated with mental health difficulties, future research on dating stress could meaningfully inform clinical intervention. Existing interventions for relationship-related issues are almost exclusively focused on established partnerships, thus leaving little to no evidence base for clinicians working with clients having difficulties navigating the dating world. On a practical level, clinicians could consider assessing and exploring dating stressors with clients who are actively dating as these experiences are associated with depressive and anxiety symptoms. Although further empirical validation is warranted, the IDS could be used as a tool to foster discussions about how to pre-emptively cope with frustrations involved in the dating process, such as interpretations people might make about themselves and future relationships that could influence behavior that may be goal-inconsistent (e.g., avoidance). Future studies should focus on the pathways through which dating stressors are associated with mental health symptoms to provide clinicians with useful intervention targets. Further, the mental health correlates of dating stressors support continued empirical attention to relationship skills interventions that are focused on individuals (e.g., Davila et al., 2021), which could help people navigate dating stressors more effectively by teaching techniques for insight-building and emotion regulation that could allow for effective coping with ambiguity, uncertainty, and rejection when dating.

Overall, the current study introduced the construct of dating stress and established the initial factor structure, reliability, and validity of the Inventory of Dating Stress to be used in future research. Our results extend the literature on the mental health correlates of interpersonal relationship difficulties (Whisman & Baucom, 2012) to potential (vs. committed) partners and suggest that existing theories of romantic relationships may need to better accommodate the challenges of modern dating (Pew Research Center, 2020; Rosenfeld & Thomas, 2012). For example, existing work assumes the consideration of just one potential partner at a time and has focused on relationship exploration processes that typically occur after initial attraction has been established. This assumption may no longer accurately reflect the reality of dating with the rapid expanse of technology and the dating app market in recent years. Our understanding of how people enter relationships would therefore be enriched by considering such decisions within the context of dating stress.

Supplemental Material

sj-pdf-1-spr-10.1177_02654075221078295 – Supplemental Material for Inventory of dating stress: Initial psychometric evaluation and construct validity

Supplemental Material, sj-pdf-1-spr-10.1177_02654075221078295 for Inventory of dating stress: Initial psychometric evaluation and construct validity by Timothy J. Sullivan, Joanne Davila in Journal of Social and Personal Relationships

Footnotes

Acknowledgments

The authors would like to acknowledge the helpful input of K. Daniel O’Leary, who provided comments on an earlier draft of this article. The authors would also like to acknowledge the assistance of the following individuals in data collection and study administration: Ellora Vilkin, Howard Huang, and Abigail Houck. Results from this study were presented as a poster at the Association for Behavioral and Cognitive Therapies Annual Conference in 2020.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Timothy J. Sullivan was supported by the National Science Foundation Graduate Research Fellowship under Grant No. 1839287. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

Open research statement

As part of IARR’s encouragement of open research practices, the authors have provided the following information: This research was not pre-registered. The data used in the research cannot be publicly shared but are available upon request. The data can be obtained by emailing tj.sullivan@stonybrook.edu. The materials used in this research are available upon request and can be obtained by emailing tj.sullivan@stonybrook.edu.

ORCID iDs

Timothy J. Sullivan

Joanne Davila

Supplemental Material

Supplemental material for this article is available online.

References

Andrighetto

Riva

Gabbiadini

(2019). Lonely hearts and angry minds: Online dating rejection increases male (but not female) hostility. Aggressive Behavior, 45(5), 571–581. https://doi.org/10.1002/ab.21852

Bandura

(1969). Principles of behavior modification. Holt, Rinehart, & Winston.

Banker

J. E.

Kaestle

C. E.

Allen

K. R.

(2010). Dating is hard work: A narrative approach to understanding sexual and romantic relationships in young adulthood. Contemporary Family Therapy, 32(2), 173–191. https://doi.org/10.1007/s10591-009-9111-9

Bentler

P. M.

(1990). Comparative fit indexes in structural models. Psychological Bulletin, 107(2), 238. http://doi.org/10.1037/0033-2909.107.2.238

Birnbaum

G. E.

Kanat-Maymon

Mizrahi

Barniv

Nagar

Govinden

Reis

H. T.

(2018). Are you into me? Uncertainty and sexual desire in online encounters and established relationships. Computers in Human Behavior, 85, 372-384. https://doi.org/10.1016/j.chb.2018.04.023.

Birnbaum

G. E.

Reis

H. T.

(2012). When does responsiveness pique sexual interest? Attachment and sexual desire in initial acquaintanceships. Personality & Social Psychology Bulletin, 38(7), 946–958. https://doi.org/10.1177/0146167212441028

Birnbaum

G. E.

Zholtack

Reis

H. T.

(2020). No pain, no gain: Perceived partner mate value mediates the desire-inducing effect of being hard to get during online and face-to-face encounters. Journal of Social and Personal Relationships, 37(8–9), 2510–2528. https://doi.org/10.1177/0265407520927469

Bowlby

(1982). Attachment and loss: Retrospect and prospect. American Journal of Orthopsychiatry, 52(4), 664–678. https://doi.org/10.1111/j.1939-0025.1982.tb01456.x

Brown

T. A.

(2015). Confirmatory factor analysis for applied research (2nd ed.). The Guilford Press.

10.

Bühler

J. L.

Krauss

Orth

(2021). Development of relationship satisfaction across the life span: A systematic review and meta-analysis. Psychological Bulletin. Advance online publication. https://dx-doi-org.web.bisu.edu.cn/10.1037/bul0000342

11.

Center for Collegiate Mental Health . (2019). 2018 Annual report (Publication No. STA 19-180).

12.

Chorney

D. B.

Morris

T. L.

(2008). The changing face of dating anxiety: Issues in assessment with special populations. Clinical Psychology: Science and Practice, 15(3), 224–238. https://doi.org/10.1111/j.1468-2850.2008.00132.x

13.

Clark

M. S.

Beck

L. A.

(2011). Initiating and evaluating close relationships: A task central to emerging adults. In Fincham

F. D.

Cui

(Eds.), Romantic relationships in emerging adulthood. (pp. 190–212). Cambridge University Press.

14.

Collins

W. A.

Welsh

D. P.

Furman

(2009). Adolescent romantic relationships. Annual Review of Psychology, 60(1), 631–652. https://doi.org/10.1146/annurev.psych.60.110707.163459

15.

Davila

Zhou

Norona

Bhatia

Mize

Lashman

(2021). Teaching romantic competence skills to emerging adults: A relationship education workshop. Personal Relationships, 28(2), 251–275. https://doi.org/10.1111/pere.12366

16.

Descutner

C. J.

Thelen

M. H.

(1991). Development and validation of a Fear-of-Intimacy Scale. Psychological Assessment, 3(2), 218–225. https://doi.org/10.1037/1040-3590.3.2.218

17.

Einstein

D. A.

(2014). Extension of the transdiagnostic model to focus on intolerance of uncertainty: A review of the literature and implications for treatment. Clinical Psychology: Science and Practice, 21(3), 280–300. https://doi.org/10.1111/cpsp.12077

18.

Finkel

E. J.

Eastwick

P. W.

Karney

B. R.

Reis

H. T.

Sprecher

(2012). Online dating: A critical analysis from the perspective of psychological science. Psychological Science in the Public Interest, 13(1), 3–66. https://doi.org/10.1177/1529100612436522

19.

Freedman

Powell

D. N.

Williams

K. D.

(2019). Ghosting and destiny: Implicit theories of relationships predict beliefs about ghosting. Journal of Social and Personal Relationships, 36(3), 905–924. https://doi.org/10.1177/0265407517748791

20.

Hammen

(2005). Stress and depression. Annual Review of Clinical Psychology, 1(1), 293–319. https://doi.org/10.1146/annurev.clinpsy.1.102803.143938

21.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. http://doi.org/10.1080/10705519909540118

22.

James-Kangal

Whitton

S. W.

(2019). Conflict management in emerging adults’ “nonrelationships.” Couple and Family Psychology: Research and Practice, 8(2), 63–76. https://doi.org/10.1037/cfp0000118

23.

Jonason

P. K.

N. P.

(2013). Playing hard-to-get: Manipulating one’s perceived availability as a mate. European Journal of Personality, 27(5), 458–469. https://doi.org/10.1002/per.1881

24.

Jöreskog

K. G.

(1969). A general approach to confirmatory maximum likelihood factor analysis. Psychometrika, 34(2), 183–202. https://doi.org/10.1007/BF02289343

25.

Kroenke

Spitzer

R. L.

Williams

J. B. W.

(2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16(9), 606–613. https://doi.org/10.1046/j.1525-1497.2001.016009606.x

26.

LeFebvre

L. E.

Allen

Rasner

R. D.

Garstad

Wilms

Parrish

(2019). Ghosting in emerging adults’ romantic relationships: The digital dissolution disappearance strategy. Imagination, Cognition and Personality, 39(2), 125–150. https://doi.org/10.1177/0276236618820519

27.

LeFebvre

L. E.

Fan

(2020). Ghosted?: Navigating strategies for reducing uncertainty and implications surrounding ambiguous loss. Personal Relationships, 27(2), 433–459. https://doi.org/10.1111/pere.12322

28.

MacCallum

R. C.

(1986). Specification searches in covariance structure modeling. Psychological Bulletin, 100(1), 107–120. https://doi.org/10.1037/0033-2909.100.1.107

29.

Maniaci

M. R.

Rogge

R. D.

(2014). Caring about carelessness: Participant inattention and its effects on research. Journal of Research in Personality, 48, 61-83. https://doi.org/10.1016/j.jrp.2013.09.008.

30.

Murray

S. L.

Holmes

J. G.

Collins

N. L.

(2006). Optimizing assurance: The risk regulation system in relationships. Psychological Bulletin, 132(5), 641–666. https://doi.org/10.1037/0033-2909.132.5.641

31.

Pew Research Center . (2020). The virtues and downsides of online dating. Pew Research Center. https://www.pewresearch.org/internet/2020/02/06/the-virtues-and-downsides-of-online-dating/.

32.

R Core Team . (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org

33.

Rosenfeld

M. J.

Thomas

R. J.

(2012). Searching for a mate: The rise of the internet as a social intermediary. American Sociological Review, 77(4), 523–547. https://doi.org/10.1177/0003122412448050

34.

Rosseel

(2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. https://doi.org/10.18637/jss.v048.i02

35.

Slavich

G. M.

O’Donovan

Epel

E. S.

Kemeny

M. E.

(2010). Black sheep get the blues: A psychobiological model of social rejection and depression. Neuroscience and Biobehavioral Reviews, 35(1), 39–45. https://doi.org/10.1016/j.neubiorev.2010.01.003

36.

Spitzer

R. L.

Kroenke

Williams

J. B. W.

Löwe

(2006). A brief measure for assessing generalized anxiety disorder: The GAD-7. Archives of Internal Medicine, 166(10), 1092–1097. https://doi.org/10.1001/archinte.166.10.1092

37.

Stanley

S. M.

Rhoades

G. K.

Fincham

F. D.

(2011). Understanding romantic relationships among emerging adults: The significant roles of cohabitation and ambiguity. In Fincham

F. D.

Cui

(Eds.), Romantic relationships in emerging adulthood. (pp. 234–251). Cambridge University Press.

38.

Taylor

A. C.

Rappleyea

D. L.

Fang

Cannon

(2013). Emerging adults’ perceptions of acceptable behaviors prior to forming a committed, dating relationship. Journal of Adult Development, 20(4), 173–184. https://doi.org/10.1007/s10804-013-9169-3

39.

Tucker

L. R.

Lewis

(1973). A reliability coefficient for maximum likelihood factor analysis. Psychometrika, 38(1), 1–10. http://doi.org/10.1007/bf02291170

40.

Vennum

Fincham

F. D.

(2011). Assessing decision making in young adult romantic relationships. Psychological Assessment, 23(3), 739–751. https://doi.org/10.1037/a0023287

41.

Veroff

Kulka

R. A.

Douvan

E. A. M.

(1981). Mental health in America: Patterns of help-seeking from 1957-1976. Basic Books.

42.

Whisman

M. A.

Baucom

D. H.

(2012). Intimate relationships and psychopathology. Clinical Child and Family Psychology Review, 15(1), 4–13. https://doi.org/10.1007/s10567-011-0107-2

43.

Yuan

K.-H.

Bentler

P. M.

(2000). Three likelihood-based methods for mean and covariance structure analysis with nonnormal missing data. Sociological Methodology, 30(1), 165–200. https://doi.org/10.1111/0081-1750.00078

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.22 MB