The Effects of Multifaceted Introversion and Sensory Processing Sensitivity on Solitude‐Seeking Behavior

ABSTRACT

Introduction

The state of solitude may be desirable and beneficial particularly for individuals who are highly sensitive and introverted.

Methods

To test these predictions, we surveyed a nationally representative US sample of 301 adults and a sample of 99 undergraduates on their levels of sensory processing sensitivity and assessed introversion with the Big Five Inventory and the multifaceted STAR Introversion Scale. Participants then reported the frequency and duration of their volitional solitude, stress levels, and subjective well‐being across 10 consecutive days.

Results

Results revealed that Social Introversion and sensitivity significantly predicted higher motivations for solitude, both self‐determined and not. Thinking Introversion also predicted higher self‐determined solitude, but BFI introversion showed no relationship with either motivation. Social Introversion and sensitivity predicted higher frequency of solitude in daily life and longer duration of solitary episodes; BFI Introversion and Restrained Introversion showed the opposite pattern for both outcomes. Finally, stress was positively associated with daily solitude frequency, and in turn, solitude frequency was negatively associated with same‐day well‐being; there were no interaction effects with personality traits.

Discussion

These findings suggest that introversion, as measured by the STAR Introversion Scale, and sensitivity contribute significantly to solitary motivation; however, solitude appears to be sought after by people in times of stress regardless of their scores on these traits.

1. Introduction

Over the past few decades, researchers have mapped multiple benefits associated with solitude, especially for those who spend time alone by choice. Such volitional solitude has been associated with improvements in mood regulation and identity development (Goossens 2014; Larson 1990) and shown to facilitate experiences of insight, creativity, freedom, and self‐connection (Long and Averill 2003; Storr 1988). In particular solitude appears to have a self‐regulating effect by inducing a calm, low‐arousal state (Nguyen, Ryan, and Deci 2018), as well as a flourishing effect, as evidenced by enhanced levels of eudaimonic well‐being such as personal growth and self‐acceptance (Ost Mor, Palgi, and Segel‐Karpas 2021; Thomas and Azmitia 2019; Weinstein, Nguyen, and Hansen 2021).

Despite these gains in understanding the potential affordances of solitude in everyday life, the research is less clear on why these benefits are not uniformly experienced. Some of this may be attributed to development timing (Coplan, Ooi, and Baldwin 2019), with time alone becoming more rewarding as we age (Larson 1990; Lay, Fung et al. 2020; Lay, Pauly et al. 2020), or to the acquisition of skills that enable a person to utilize solitude constructively (Thomas 2023a). More commonly, it is speculated that individual differences play a role in who seeks and benefits from solitude, with introversion as a key variable.

1.1. Introversion and Solitude

While it may seem self‐evident that introverts seek out and enjoy solitude more than extraverts, findings are mixed. On the one hand, studies have revealed that introversion is indeed correlated with a preference for solitude in adults (Burger 1995) and adolescents (Teppers et al. 2013), and with frequency of solitary experiences among young adults (Leary, Herbst, and McCrary 2003). On the other hand, research has found that introverts and extraverts do not differ significantly in their enjoyment of time alone (Hills and Argyle 2001; Srivastava, Angelo, and Vallereux 2008), or in their likelihood of being intrinsically motivated to engage in solitude (Nguyen, Weinstein, and Ryan 2022; Thomas and Azmitia 2019). Moreover, in a recent study extraverts were more likely to engage in solitude than low‐functioning (i.e., depressed and lonely) introverts (Thomas et al. 2021).

How can these contradictory findings be explained? One factor may be the variability in which introversion is defined and subsequently measured. In his seminal work on personality types, Jung (1923) described introversion as the tendency to be more energized by the contents of one's internal world than by the outer world of people and activities. Later, Hans Eysenck defined introversion as having a lower threshold for cortical arousal (Eysenck and Eysenck 1985), and adjacent work has correlated introversion with heightened sensitivity to environmental stimuli (Aron and Aron 1997; Geen 1984).

Over the past two decades, the vast majority of studies measuring introversion use the Five Factor Model of personality, often referred to as the “Big 5.” This model defines introversion largely as an absence of the qualities that define extraversion, with items in the Big Five Inventory (BFI; John, Naumann, and Soto 2008) often focusing on sociability, and items in the NEO‐PI‐3 (McCrae, Costa Jr., and Martin 2005) focusing on facets such as positive emotionality, assertiveness, and gregariousness.

Conceptually, this assumes that low scores on extraversion represent introversion. However, only one of these traits is defined in these inventories, which use Likert scales to measure extraversion on a continuum from low to high. Items that are reverse coded (e.g., shy, reserved, and inhibited in the BFI) are all negatively valenced, and the presence of these characteristics is assumed to represent introversion. The act of assigning solely negative characteristics to a personality trait is problematic because it creates a deficit view. Rather, we join other scholars in viewing introversion as more than the lack of extraversion, conceptualizing it instead as a trait with its own properties (Block 1995; Carrigan 1960; Grimes, Cheek, and Norem 2011; Guilford and Guilford 1934; Thorne 1987).

We therefore posit that conclusions reached about the relationship between introversion and solitude are opaque due to these conceptual issues in measurement. For instance, studies that utilized the Big 5 to investigate relationships between extraversion and social interaction may assume that introverts are seeking or enjoying solitude due to their lower frequencies of social interaction (Srivastava, Angelo, and Vallereux 2008; Wilt and Revelle 2019), but evidence for this is weak because such studies do not include measures about solitary behaviors, motivations, or preference for solitude.

Moreover, because measures of extraversion may be conflating shyness or social anxiety with introversion (Briggs 1988), it is unclear if individuals are indeed seeking solitude when they avoid social interaction. This is important because intentionally chosen solitude has distinct outcomes from solitude that is engaged in by default (Thomas et al. 2021). For example, shy individuals may be alone because they hesitate to enter social situations, not because they desire solitude. Thus, low frequency of social interaction does not imply high frequency of what has been termed intrinsic (Nguyen, Ryan, and Deci 2018), self‐determined (Thomas and Azmitia 2019), or positive solitude (Ost Mor, Palgi, and Segel‐Karpas 2021), all of which have been described as states of solitude that are conducive rather than detrimental to psychological health.

If introversion is in fact correlated with solitude‐seeking behaviors, we lack research on what is driving such behaviors or what more introverted people may gain from it. In contrast, the literature on extraversion has resulted in ample evidence showing that extraverts seek social interaction because it is experienced as a reward (Lucas et al. 2000), they receive pleasure from it (Berenbaum 2002), they enjoy the social attention (Ashton, Lee, and Paunonen 2002), and they experience gains in positive affect when engaging in it (Srivastava, Angelo, and Vallereux 2008). Therefore, in the present study, we aim to address this gap in the relationship between introversion and solitude seeking.

1.2. Defining Introversion with the STAR Introversion Scale

To address our research questions, we needed to utilize a psychometric scale that measured introversion directly rather than by proxy (i.e., lack of extraversion). Similar to the argument that extraversion is a multifaceted construct (McCrae, Costa Jr., and Martin 2005), the STAR Introversion Scale (Cheek, Brown, and Grimes 2014; Grimes, Cheek, and Norem 2011) conceptualizes introversion as comprising four domains or meanings: Social, Thinking, Anxious, and Restrained (acronym STAR). All domains with the exception of Thinking correlated with distinct facets of extraversion as measured by the NEO‐PI (Costa Jr. and McCrae 1992); social introversion correlated with low warmth and gregariousness, anxious introversion with low positive emotions and assertiveness, and restrained introversion with low excitement seeking and activity.

Importantly, each domain was construed as having substantive defining properties above and beyond having low facets of extraversion (see Grimes, Cheek, and Norem 2011 for a comprehensive overview). Social introversion identifies a preference for being alone or with a few close friends over large social gatherings. Thinking introversion reflects engagement with introspection and imagination. Anxious introversion involves shyness and rumination. Restrained introversion assesses behaviors associated with being deliberative and serious. The distinctiveness of these four domains of introversion, being to some extent inter‐correlated but meaningfully distinct, make the STAR a promising measure for understanding if and why introverts seek solitude.

1.3. Sensory Processing Sensitivity and Solitude

Individual differences in sensitivity may also play an influential role in solitude seeking and its associated benefits. Research on sensory processing sensitivity (SPS; Aron and Aron 1997) indicates that this trait presents as a normal variation in human temperament, with approximately 20% of the general population classified as “highly sensitive.” SPS is defined as an increased sensitivity to environmental and social stimuli (e.g., noise, social interaction, or multiple demands on one's time), accompanied by high emotional and physiological reactivity to such stimuli, as well as an increased ability for depth of processing (e.g., reflection; Aron, Aron, and Jagiellowicz 2012). This trait was found to be moderately associated with, but not identical with, introversion as measured by the BFI (Aron and Aron 1997). In contrast to the BFI's operationalization of introversion as a lack of sociability (John, Donahue, and Kentle 1991), the construct of high sensitivity appears to be more conceptually consistent with Eysenck's (1981) original description of introversion, which emphasized its high arousal features.

While to date there is a lack of quantitative studies analyzing the relationship between sensitivity and solitude, interviews with highly sensitive individuals have revealed that overstimulation is a strong concern of theirs, and a typical strategy for addressing it is to withdraw into solitude (Bas et al. 2021; Black and Kern 2020). This is a logical move, considering that solitude is typically characterized by the absence of social interaction, the presence of quiet and privacy, and the experience of low‐arousal affective states, such as feeling calm and relaxed (Nguyen, Ryan, and Deci 2018; Pauly et al. 2017). Solitude thus offers a state of understimulation where sensitive individuals can retreat from stressful or overstimulating environments, reduce sensory input, and experience restorative effects—and important process, given that highly sensitive people are more vulnerable to stress‐related problems in response to negative environments (Greven et al. 2019).

1.4. Self‐Determination

We examine the relationships of introversion, SPS, and solitude seeking within the framework of self‐determination theory, which emphasizes the importance of intrinsic motivation when engaging in activities (Deci and Ryan 2008). Behaviors that are motivated from a position of self‐determination, as opposed to a position of feeling forced or pressured, are shown to promote flourishing and well‐being. The Motivation for Solitude Scale Short Form (MSS‐SF; Thomas and Azmitia 2019) is grounded in self‐determination theory and differentiates self‐determined solitude (SDS) from not self‐determined solitude (NSDS). Although both subscales represent motivations to seek solitude, SDS is associated with well‐being, whereas NSDS is associated with poor psychological adjustment (see also Nicol 2006; Van Zyl, Dankaert, and Guse 2018). Three of the SDS items in particular seem to indicate a motivation for a low‐arousal state to offset overstimulation (e.g., quiet, privacy, and restoration of energy) versus the remaining five items which seem to point to purposes less related to arousal (e.g., creativity, insight, and spirituality). Thus, this scale serves as a useful measure for understanding how the individual differences of introversion and sensitivity might impact the motivation to be alone.

1.5. Research Questions and Hypotheses

We designed a multi‐phase study to collect both cross‐sectional and daily diary data to answer our research questions. Findings are mixed regarding how best to capture the frequency and quality of specific behaviors in daily life, with some studies indicating participants over‐report such behaviors using experiencing sampling methods (ESM) and others indicating that participants under‐report these same behaviors (aan het Rot, Hogenelst, and Schoevers 2012). For the present study, we chose the daily diary method (Gunthert and Wenze 2012; Silvia and Cotter 2021) because we reasoned it would be more effective than ESM at capturing daily solitude experiences, given that such experiences can be infrequent and can occur unpredictably and spontaneously at times outside of ESM survey alerts. In addition, we preferred the less intrusive aspect of the daily diary method for inquiring about solitude, which is often sought for the express purposes of avoiding social interaction and obligations to others (i.e., responding to an ESM survey alert).

1.5.1. Phase 1

In Phase 1, we sought to clarify the relationships between the variables of interest (i.e., introversion domains, SPS, and motivation for solitude) using established psychometric scales.

1.5.1.1. H1: Introversion

We hypothesized that introversion would show the following nuanced relationships with motivations for solitude. First, we expected three of the STAR domains to predict SDS (i.e., positive motivation): Social introversion, associated with low gregariousness and a preference for solitude (H1a); thinking introversion, associated with imagination and introspection (H1b); and restrained introversion, associated with deliberate seriousness and low sensation seeking (H1c). Second, we expected two of the STAR domains to predict NSDS (i.e., negative motivation): anxious introversion, associated with shyness, rumination, and low positive emotions (H1d), and social introversion, given their ambivalence about social interaction (H1e). Note that social introversion was the only domain that we expected to be associated with both positive (SDS) and negative (NSDS) motivations for solitude. Finally, consistent with previous findings (Thomas and Azmitia 2019), we expected that introversion as measured by low scores of extraversion on the BFI would show no relationship with SDS but would predict NSDS (H1f).

1.5.1.2. H2: SPS

Working with the assumption that people who are highly sensitive might appreciate solitude as an escape from overstimulating social environments, we hypothesized that sensitivity would predict high scores on SDS when controlling for neuroticism, as recommended by Aron and Aron (2018).

1.5.2. Phase 2

In Phase 2, we explored how introversion and sensitivity interact to predict solitude seeking in everyday life using a daily diary methodology whereby we collected data about the frequency and duration of episodes of volitional solitude, stress levels, and subjective well‐being over 10 consecutive days.

1.5.2.1. H3: Episodes

First, we expected that the duration of solitude episodes would be predicted by sensitivity and social introversion (H3a), given that each of these traits exhibit a heightened sensitivity to social stimulation, and both have expected positive associations with an intrinsic motivation for solitude. Individuals scoring high on either of these two traits may need more time in volitional solitude to recover from overstimulating experiences. Second, we expected the frequency of volitional solitude to be predicted by sensitivity, all four STAR introversion domains, but not introversion as measured by the BFI (H3b).

1.5.2.2. H4: Stress

We expected stress levels to predict frequency of volitional solitude, such that frequency of episodes would increase on high stress days and decrease on low stress days, but only for individuals scoring high on sensitivity or social introversion. For these individuals, volitional solitude may serve as a method of coping with stress (Larson and Lee 1996; Martin and Brantley 2004). We therefore anticipated that these two traits would interact with the predicted relationship of stress and solitude.

1.5.2.3. H5: Subjective Well‐being

Similarly, we hypothesized that the frequency of volitional solitude would predict daily subjective well‐being levels, such that well‐being scores would increase on high‐frequency solitude days and decrease on low‐frequency solitude days, but only for those individuals scoring high on sensitivity or social introversion.

2. Method

2.1. Participants

We recruited a nationally representative sample of adults residing in the United States via Prolific.com, an online recruitment platform, as well as a sample of undergraduate students from a liberal arts college in the northeastern United States. Our total sample (n = 400) included 99 undergraduates and 301 general population adults. We chose these two populations to remain consistent with the original validation of the STAR Introversion Scale (Cheek, Brown, and Grimes 2014), which sampled two waves of participants: undergraduates and a sample of adults recruited from Amazon Mturk.

2.1.1. Demographic Information

The Prolific sample was more diverse in age, ranging from 18 to 89 years (M = 45.51; SD = 16.01), than the undergraduate sample which ranged from 18 to 21 years (M = 18.63; SD = 0.88). Two undergraduate participants did not report their age. Their age was imputed as the mode of the age of the full undergraduate sample (i.e., 18 years old). The samples were also diverse in gender, ethnicity, and parental education—one metric of SES—although this diversity varied by sample (see Table 1). There were no significant differences in demographic characteristics between the full sample analyzed for Phase 1 and the reduced analytic sample for Phase 2 (see Analysis Plan regarding inclusion criteria for Phase 2).

TABLE 1.

Demographic frequencies parsed by samples and study phases.

Variable	College sample	Prolific sample	p value*	Phase 1 full sample	Phase 2 valid sample	p value**
Gender
Female	73%	50%	<0.001	56%	55%	0.855
Male	25%	49%		43%	44$
Other	2%	1%		1%	1%
Race/ethnicity
White	58%	72%	<0.001	69%	68%	0.766
Black	8%	14%		12%	12%
Asian	17%	7%		10%	7%
Latin	10%	5%		6%	10%
Other	6%	2%		3%	3%
Parent education
Graduate degree	59%	18%	<0.001	28%	30%	0.279
College degree	25%	30%		29%	27%
Some college	6%	22%		18%	17%
High school or less	10%	30%		25%	26%
Sample number	99	301		400	302

^* p values are from Pearson chi‐square tests comparing college and prolific samples.

^** p values are from Pearson χ ² tests comparing Phase 1 participants included in Phase 2 analyses from those excluded.

2.2. Procedure

2.2.1. Preregistration

The Stage 1 study design and protocol were preregistered on the Open Science Framework (OSF) repository prior to data collection; 10.17605/OSF.IO/9QY4P. Upon completion of the study, data files and syntax for analysis were made public within the preregistered project; https://osf.io/xzkys/.

2.2.2. Ethics Information

Approval for the study was granted from the first author's affiliated Institutional Review Board. Each participant read and signed a statement of informed consent prior to beginning the study.

2.2.3. Phase 1: Online Survey

Participants who signed up for the study received a link to the online survey, hosted by Qualtrics and delivered through Prolific.com for the nationally representative sample, or through the academic institution's research participation online system (SONA) for the undergraduate sample. Each participant completed the survey on their own device, finishing in approximately 20–30 min. The undergraduate sample was compensated with course credit for participating, and the nationally representative sample was compensated with a modest financial payment.

In addition to demographic questions, the survey asked participants to answer items from the following measures of personality and adjustment. Each scale was treated as a continuous measure by recoding any reverse‐worded items and computing a mean score for each participant. Reliability for each scale was computed separately for each sample; below, we report the coefficient of undergraduate sample first, followed by the coefficient for the nationally representative adult sample.

2.2.3.1. STAR Introversion Scale

This measure asked participants to rate on a 1–5 scale to what extent a list of 40 statements were characteristic of them. The 40 items were divided equally among four domains of introversion, with sample items as follows: “After spending a few hours surrounded by a lot of people, I am usually eager to get away by myself” (social introversion; ⍺ = 0.78; ⍺ = 0.83); “I enjoy analyzing my own thoughts and ideas about myself” (thinking introversion; ⍺ = 0.80; ⍺ = 0.85); “Even when I am in a group of friends, I often feel very alone and uneasy” (anxious introversion; ⍺ = 0.84; ⍺ = 0.91); “I take my time to ‘look before I leap’ into new things” (restrained introversion; ⍺ = 0.78; ⍺ = 0.80) (Cheek, Brown, and Grimes 2014).

2.2.3.2. Big Five Inventory Introversion (i.e., Low Extraversion) and Neuroticism

We used the extraversion subscale (⍺ = 0.91; ⍺ = 0.90), which is an inverse measure of introversion, to compare outcomes with the STAR measure, and the neuroticism subscale (⍺ = 0.84; ⍺ = 0.90) to control for trait negativity when measuring sensitivity, as recommended (Aron and Aron 2018). Both subscales asked participants to rate on a 1–5 scale the extent to which they agreed with various items with the sentence stem, “I am someone who…” A sample neuroticism item was, “gets nervous easily.” A sample extraversion item was, “is outgoing, sociable.” We reverse scored the extraversion subscale to produce a BFI scale that reflected introversion on the high end of the measure (John, Donahue, and Kentle 1991; John, Naumann, and Soto 2008).

2.2.3.3. Highly Sensitive Person Scale

This scale measured SPS with 27 statements, asking participants to rate how much each statement applied to them, using a seven‐point scale ranging from not at all to extremely (⍺ = 0.90; ⍺ = 0.93). A sample item was, “Do you feel unpleasantly aroused when a lot is going on around you?” (HSP; Aron and Aron 1997).

2.2.3.4. Motivation for Solitude—Short Form

This 14‐item questionnaire measured self‐determined (SDS) motivation for solitude (⍺ = 0.77, ⍺ = 0.78) and not self‐determined (NSDS) motivation for solitude (⍺ = 0.88, ⍺ = 0.91). The questionnaire provided participants with the prompt: “When I spend time alone, I do so because…” and then asks them to rate statements on a scale from 1 (not at all important) to 4 (very important). Sample items included: “I can engage in activities that really interest me” (SDS) and “I feel anxious when I'm with others” (NSDS) (MSS‐SF; Thomas and Azmitia 2019).

2.2.4. Phase 2: Daily Diary Study

In Phase 2, participants completed a 10‐day diary study in which they responded to a series of questions about their daily stress levels, subjective well‐being, and solitude experiences. We utilized end‐of‐day surveys (Gunthert and Wenze 2012) that asked participants to recall focal events (i.e., volitional solitude episodes) from the day in addition to completing the stress and well‐being measures described below.

Consistent with recommendations for maximizing compliance (Silvia and Cotter 2021), we employed the following guidelines: A survey link was emailed to participants every evening at 7 p.m. in their time zone and remained active until 5 a.m., thus providing flexibility for participants to respond at a time that matched their actual “end of day” while simultaneously limiting response delay; surveys that expired without submission were recorded as missing data; we encouraged compliance with email reminders in the case of missed surveys rather than through increased compensation, as advised by Ohly et al. (2010) as a way to ensure integrity of the data.

2.2.4.1. Daily Stress Levels

The 14‐item Perceived Stress Scale (PSS; Cohen, Kamarck, and Mermelstein 1983) is a widely used stress assessment appropriate for university and nonclinical adult populations; we adapted it to measure participants' perceived stress on a daily basis rather than a monthly basis, as in the original version (see Jiang 2020 for an example). On a scale of 1 (never) to 5 (always), participants rated how often they experienced a variety of stressors throughout that particular day. Sample items included, “Today, how often did you feel confident about your ability to handle your personal problems?” and “Today, how often were you able to control the way you spent your time?” To assess reliability in a daily, multilevel, framework, we used van Alphen et al.'s (2022) example Mplus syntax, which is based on recommendations provided by Geldhof, Preacher, and Zyphur (2014), and uses updated equations suggested by Lai (2021). Between‐person Omega was 0.92 and within‐person Omega was 0.84, indicating strong internal consistency.

2.2.4.2. Subjective Well‐being

Well‐being can be both hedonic (e.g., happiness) and eudaimonic (e.g., meaning) in nature, with some research showing solitude may have stronger associations with the latter form (Larson and Csikszentmihalyi 1978; Ost Mor, Palgi, and Segel‐Karpas 2021; Thomas 2023a; Weinstein, Nguyen, and Hansen 2021). We asked participants to evaluate their well‐being generalized over the course of the day by answering nine questions drawn from the Mental‐Health Continuum Short Form (Keyes et al. 2008), a scale that measures both forms of well‐being not as traits but as states that fluctuate based on a changing environment. On a six‐point scale, ranging from never to always, participants were asked to rate how often they experienced various feelings during the day. A sample hedonic item was “feeling interested in life,” and a sample eudaimonic item was “feeling that you had experiences that challenged you to grow and become a better person.” To assess reliability in a daily, multilevel, framework, we again relied on van Alphen et al.'s (2022) example Mplus syntax. Between‐person Omega was 0.94 and within‐person Omega was 0.84, indicating strong internal consistency.

2.2.4.3. Frequency and Duration of Volitional Solitude

Given that our research question focused on solitude seeking, participants were asked to describe only the times during the day that they intentionally sought out time alone. Thus, we did not collect data on occurrences of solitude that were not by choice or were engaged in by default. Rather, we wished to examine only episodes of solitude that the participants actively chose. To ensure that participants were using a shared definition of solitude, the daily dairy survey provided the following prompt:

Think back over the course of your day. How many times today did you intentionally seek out time to be by yourself? “By yourself” is defined as not interacting with anyone, face‐to‐face or virtually, for a period of time. Please only count the times you were alone for a period of time by choice.

For each solitary episode reported, participants shared the duration in minutes that it lasted.

Using the survey platform's branching logic feature, participants then received a series of follow‐up questions. If they reported zero occurrences of daily volitional solitude, the survey included only two questions: whether they ever wished they could have been alone that day (yes/no), and if so, why (open response). If they reported that they did volitionally seek solitude that day, whether one or more episodes, they were asked a series of questions to qualitatively describe one of those episodes; these data will be analyzed as part of a different component of the study and are not reported here.

3. Results

3.1. Phase 1

3.1.1. Analysis Approach

Descriptive statistics and Pearson correlations for all variables are presented in Table 2. To test the Phase 1 hypotheses, two similar hierarchical regressions were run, one with SDS as the outcome variable and one with NSDS as the outcome variable. Predictor variables were centered to improve interpretability and reduce multicollinearity (Aiken and West, 1991). Preliminary analyses were conducted to then test for multicollinearity of the predictor variables. Anxious introversion had the highest VIF of 3.84, which exceeded our strict threshold value of 2.5. As a result, anxious introversion was removed from the models tested. All remaining predictor variables in the tested models had acceptable VIFs less than 2.5.

TABLE 2.

Descriptive statistics and Pearson correlations among Phase 1 variables.

Variable	1	2	3	4	5	6	7	8	9	10
1. Self‐determined solitude	—
2. Not self‐determined solitude	0.29***	—
3. Age	0.05	−0.23***	—
4. BFI neuroticism	0.02	0.56***	−0.36***	—
5. BFI introversion	0.09	0.37***	0.05	0.26***	—
6. Social introversion	0.28***	0.47***	0.17***	0.23***	0.71***	—
7. Anxious Introversion	0.15**	0.73***	−0.34***	0.71***	0.56***	0.51***	—
8. Thinking introversion	0.49***	0.19***	−0.20***	0.18***	−0.06	0.11*	0.19***	—
9. Restraint introversion	0.14**	−0.05	0.36***	−0.15**	0.38***	0.38**	−0.02	0.02	—
10. Sensory processing	0.40***	0.58***	−0.25***	0.61***	0.30***	0.37***	0.66***	0.43***	0.01	—
M	2.83	1.84	38.79	2.86	3.12	3.52	3.01	3.77	3.53	4.39
SD	0.58	0.79	18.02	0.97	1.01	0.73	0.94	0.68	0.63	1.04
Range	1–4	1–4	18–89	1–5	1–5	1.5–5	1–4.9	1.1–5	1.5–5	1.6–6.9

Note: N = 400.

^* p < 0.05.

^** p < 0.01.

^*** p < 0.001.

Variables were entered into the two hierarchical regressions in four blocks: (1) age and neuroticism as initial control variables, (2) BFI introversion, (3) the four STAR introversion facets, and (4) SPS. In Stage 1, we had planned to analyze SPS in separate regressions from the STAR introversion facets, but we decided to add SPS in a fourth block after removing anxious introversion as a predictor model. This approach had three advantages. First, we were able to control for the positive relationship between neuroticism and our key variables of introversion and SPS (e.g., Aron and Aron 1997), as well as control for age changes in personality that have been linked to maturation and social investment (Roberts, Walton, and Viechtbauer 2006). Second, this analysis allowed us to examine whether the four STAR introversion domains explain a significant amount of variance in solitude seeking above and beyond introversion as more traditionally measured by the BFI. Third, it also allowed us to test if SPS explains a significant amount of variance in solitude seeking above and beyond the various facets of introversion. According to G*Power (Faul et al. 2009), these regressions required a sample size of 244 to detect a relatively small effect size (f ² = 0.05) with an alpha of 5% and Power of 80%.

3.1.2. Self‐determined Solitude

With respect to SDS, results of the hierarchical regression revealed that the model was significant only in the third and fourth steps (see Table 3). At Step 3, the STAR introversion facets added significantly to the model and explained 30.5% of the SDS variance after controlling for age, BFI neuroticism, and BFI introversion. At Step 4, SPS added significantly to the model, explaining an additional 6.5% of the SDS variance. The final model accounted for 38.1% of the variance in SDS.

TABLE 3.

Hierarchical regression analysis of factors predicting self‐determined solitude.

Predictor variables	Step 1 β	Step 2 β	Step 3 β	Step 4 β
Step 1
Age	0.064	0.050	0.063	0.078
BFI neuroticism	0.042	0.016	−0.090	−0.269***
Step 2
BFI introversion		0.081	−0.075	0.102
Step 3
STAR social introversion			0.290***	0.222***
STAR thinking introversion			0.482***	0.361***
STAR restrained introversion			0.013	0.17
Step 4
Sensory processing sensitivity				0.374***
F model	0.77	1.32	30.15***	34.40***
R 2 change	0.004	0.006	0.305	0.065
F change	0.77	2.40	58.41***	41.32***

Note: N = 400. All predictor variables were centered. β's are standardized.

^*** p < 0.001.

In support of Hypothesis 1, higher scores on social introversion (1a) and thinking introversion (1b) were both significant predictors of SDS, but unexpectedly restrained introversion (1c) was not. As expected, BFI introversion did not predict SDS (1f). We also noted that, although not originally predicted, high neuroticism was a significant predictor of SDS, but in the opposite direction. In other words, people were less likely to be self‐determined in their solitude if they scored high on neuroticism. Finally, in support of Hypothesis 2, SDS was indeed predicted by higher SPS.

3.1.3. Not Self‐determined Solitude

With respect to NSDS, results of the hierarchical regression revealed that the model was significant at each step (see Table 4). At Step 3, the STAR introversion facets added significantly to the model and explained 8.9% of the NSDS variance after controlling for age, BFI neuroticism, and BFI introversion. In Step 4, SPS added significantly to the model, explaining an additional 3.2% of the variance in NSDS. The final model accounted for 49.9% of the variance in NSDS.

TABLE 4.

Hierarchical regression analysis of factors predicting not self‐determined solitude.

Predictor variables	Step 1 β	Step 2 β	Step 3 β	Step 4 β
Step 1
Age	−0.029	−0.070	−0.100*	−0.90*
BFI neuroticism	0.554***	0.473***	0.396***	0.271***
Step 2
BFI introversion		0.253***	0.033	0.014
Step 3
STAR social introversion			0.409***	0.362***
STAR thinking introversion			0.057	−0.027
STAR restrained introversion			0.122**	−0.119***
Step 4
Sensory processing sensitivity				0.262***
F model	93.00***	80.06***	57.28***	55.68***
R 2 change	0.319	0.058	0.089	0.032
F change	93.01***	37.20***	21.86***	25.05***

Note: N = 400. All predictor variables were centered. β's are standardized.

^* p < 0.05.

^** p < 0.01.

^*** p < 0.001.

In support of Hypothesis 1, higher social introversion (1e) predicted greater NSDS. Although not predicted, lower age, higher neuroticism, lower restrained introversion, and higher SPS all predicted greater NSDS. Of note, BFI introversion was a significant predictor in Step 2, but was not a significant predictor after the STAR introversion facets were entered into the model in Step 3.

3.1.4. Phase 1 Results Summary

In sum, analyses from Phase 1 revealed three key findings regarding the relationships between motivation for solitude and the traits of introversion and sensitivity. First, introversion as conceptualized by the Five Factor Model was not a significant predictor of motivation for solitude in either its positive (SDS) or negative (NSDS) dimensions. To be precise, BFI introversion did predict negative motivation (NSDS) but only until the STAR introversion facets were added to the model, after which it showed no relationship to NSDS.

Second, of the three facets of introversion that we retained from the STAR measure, social introversion was the only one that predicted higher motivations for both motivations of solitude. In contrast, thinking introversion was uniformly positive, being a predictor solely of SDS and unrelated to NSDS, while restrained introversion was rather benign, given that it predicted lower NSDS but had no relationship with SDS.

Third, the results demonstrated that similar to social introversion, the trait of SPS had a dual relationship with both positive (SDS) and negative (NSDS) motivations for solitude, emerging as a distinct contributor above and beyond the BFI and STAR facets.

3.2. Phase 2

3.2.1. Data Preparation

We engaged in a four step process to validate the diary data set: (1) the participant completed that day's survey, (2) solitude experiences were reported in a way that was consistent with our operationalization of voluntary solitude, (3) the duration of reported solitude episode was not an outlier, and (4) the person completed at least 5 days of surveys. Of the possible 4000 days of surveys, 77%% (f = 3075 days) were completed. Participants as a whole reported having at least one voluntary solitude episode on 59% of the survey days (f = 1801 days).

With respect to whether the solitude experience reported was consistent with what we asked for, we provided a prompt at the start of each daily survey that defined solitude as “not interacting with anyone, face‐to‐face or virtually;” we asked them to then count the times they were alone “for a period of time by choice.” As a validity check, the branching logic of the survey led them to a yes/no question after they indicated they had been alone one or more times that day; the question asked whether they were physically alone during their solitude episode. We examined each daily survey to verify whether they checked “yes” to this question, as well as their qualitative descriptions of the episodes to determine whether they had been communicating with others (i.e., texting or talking on the phone were examples of failing this validity check). Daily surveys that included episodes that failed these validity checks were removed from analysis (25%, f = 339 of daily surveys with an episode of solitude). We also removed daily surveys that included episodes with a time duration more than 5 h based on an IQR analysis of solitude duration outliers across all daily surveys (Tukey 1977; f = 126 of daily surveys).

Last, most participants completed at least eight daily surveys (70%). To balance the strength of our repeated diary data method with the inclusion of participant data, we included participants with at least 5 days of daily surveys (50% of the full 10 days) that had been validated with the steps described above. This resulted in 302 participants (75.5% of Phase 1 participants) and 2422 days of full survey data. The common rule of including participants with approximately four of seven completed survey days has recently been found to be generally robust, and even fewer days are sometimes needed (Griffiths, Williams, and Brohan 2022). Full information maximum likelihood was used for running analyses in Phase 2.

We acknowledge that participants who were more reliable in completing the daily surveys and more careful in reporting a valid volitional episode of solitude might be different than participants who were not included in Phase 2 of the analyses. Little's (1988) MCAR test revealed that the missing diary data were not missing completely at random (p < 0.001). Follow‐up analyses found no differences in the gender, race/ethnicity, or SES between participants who were included in Phase 2 and those who were not. Furthermore, independent‐samples t‐tests revealed no differences in age, neuroticism, BFI introversion, social introversion, thinking introversion, SPS, and SDS between participants who were included in Phase 2 and those who were not. Participants who were included in the Phase 2 data sample, however, were significantly higher in restrained introversion (M = 3.58) compared with participants who were not included (M = 3.38), t(398) = −2.80, p = 0.005.

With respect to power, according to Silvia and Cotter (2021), simulating power for all but the most simple of hypotheses for diary studies requires the estimation of many parameters; much of this is unknown for the current study given the novel facets of introversion measured. More Level 2 (i.e., persons) data points in comparison with Level 1 (i.e., days) data points are expected to have a stronger effect on increasing power (Bolger et al. 2012). Hox and Maas (2002) have found that at least 50 participants (i.e., Level 2 in the current study) are needed to avoid biased estimates; our final sample size of 302 participants for Phase 2 significantly exceeded this.

Level 1 variables that were collected in each daily survey included frequency and duration of volitional solitude episodes, and ratings of perceived stress and well‐being. All Level 1 predictors were group mean centered in the person across days as the mean (i.e., annotated as “person centered” for clarity in Tables 5, 6, 7), which allowed us to analyze how a person's day‐to‐day variation (e.g., if a person had more stress today than on average across all days for that person) was related to the other variables. Level 2 person variables that had been originally collected for Phase 1 included age, neuroticism, BFI introversion, the STAR introversion facets, and SPS. All Level 2 predictor variables were grand mean centered using the full sample mean across all days (i.e., annotated as “grand centered” for clarity in Tables 5, 6, 7).

TABLE 5.

Generalized estimating equations (GEE) multiple regression analysis examining correlates of voluntary solitude episode duration.

Predictor variables	β	95% CI	p value*
Age GC	−0.146	−0.518 to 0.225	0.440
BFI neuroticism GC	−1.859	−8.780 to 5.062	0.599
BFI introversion GC	−8.091	−16.354 to 0.171	0.055
STAR social introversion GC	10.362	−1.123 to 21.847	0.077
STAR thinking introversion GC	−0.360	−8.949 to 8.229	0.934
STAR restrained introversion GC	−10.068	−20.194 to 0.058	0.051
Sensory processing sensitivity GC	8.983	1.291 to 16.675	0.022
Stress PC	2.620	−3.354 to 8.593	0.390
Stress PC × STAR social introversion GC	−2.139	−9.276 to 4.999	0.557
Stress PC × sensory processing sensitivity GC	−0.226	−6.187 to 5.735	0.941

Note: β's are unstandardized.

Abbreviations: GC, grand centered; PC, person centered.

^* p values are from Wald Chi‐Square Type 3 tests of model effects.

TABLE 6.

Generalized estimating equations (GEE) multiple regression analysis examining correlates of voluntary solitude episode frequency.

Predictor variables	β	95% CI	p value*
Age GC	−0.004	−0.010 to 0.003	0.250
BFI neuroticism GC	−0.161	−0.296 to −0.026	0.019
BFI introversion GC	−0.189	−0.312 to −0.067	0.002
STAR social introversion GC	0.244	0.065 to 0.424	0.008
STAR thinking introversion GC	0.046	−0.086 to 0.178	0.494
STAR restrained introversion GC	−0.187	−0.346 to −0.027	0.022
Sensory processing sensitivity GC	0.295	0.160 to 0.430	<0.001
Stress PC	0.178	0.073 to 0.283	<0.001
Stress PC × STAR social introversion GC	0.064	−0.048 to 0.177	0.263
Stress PC × sensory processing sensitivity GC	−0.051	−0.160 to 0.059	0.365

Note: β's are unstandardized.

Abbreviations: GC, grand centered; PC, person centered.

^* p values are from Wald Chi‐Square Type 3 tests of model effects.

TABLE 7.

Generalized estimating equations (GEE) multiple regression analysis examining correlates of well‐being.

Predictor variables	β	95% CI	p value*
Age GC	0.001	−0.005 to 0.006	0.863
BFI neuroticism GC	−0.576	−0.714 to −0.153	<0.001
BFI introversion GC	−0.345	−0.519 to −0.221	<0.001
STAR social introversion GC	−0.039	−0.265 to −0.003	0.731
STAR thinking introversion GC	0.182	−0.024 to 0.206	0.024
STAR restrained introversion GC	0.311	0.121 to 0.846	0.001
Sensory processing sensitivity GC	−0.033	−0.170 to 0.186	0.641
Solitude episode frequency PC	−0.042	−0.072 to −0.049	0.006
Solitude episode Freq PC × STAR social introversion GC	−0.020	−0.065 to 0.449	0.394
Solitude episode Freq PC × sensory processing Sens GC	−0.009	−0.050 to 0.068	0.653

Note: β's are unstandardized.

Abbreviations: GC, grand centered; PC, person centered.

^* p values are from Wald Chi‐Square Type 3 tests of model effects.

3.2.2. Analysis Approach

Phase 2 hypotheses were examined with Generalized Estimating Equations (GEE), a common approach to analyzing longitudinal and clustered data (see Zorn 2001). A significant advantage of GEE over traditional multilevel modeling as initially proposed in Stage 1 is that it allowed us to use a linear distribution for modeling episode duration and well‐being, and to use a negative binomial distribution for modeling episode frequency, which was in the form of an overdispersed count outcome (Gardner, Mulvey, and Shaw 1995). In addition, although not specified in Stage 1, we used an AR1 correlation matrix for the analyses. This takes into account how the correlation between days close to each other, such as Day 1 and Day 2, might be larger than the correlation between days farther apart, such as Day 1 and Day 10 (Grady and Helms 1995).

The outcome variables in each GEE analysis were as follows: duration of solitude episodes to test H3a; frequency of solitude episodes to test H3b and H4; well‐being to test H5. See Tables 5, 6, 7 for results. For each GEE model, we regressed the outcome variable (i.e., daily duration of solitude episodes, daily frequency of episodes, or daily well‐being) onto the covariates of age, neuroticism, and BFI introversion, as well as the STAR introversion facts (except anxious introversion due to multicollinearity concerns identified in Phase 1), and SPS (i.e., to test H3a, H3b). For duration and frequency of episodes, we also included stress as a person centered predictor. In addition, we included two interaction term predictors: stress person centered by social introversion and stress person centered by SPS. These interaction terms allowed us to test whether people who are higher in SPS, or higher in social introversion, engaged in longer or more frequent solitude episodes on days of high stress (i.e., to test H4). For our well‐being hypotheses, we included episode frequency as a person centered predictor. In addition, we included two interaction term predictors: episode frequency person centered by social introversion, and episode frequency person centered by SPS. These interaction terms allowed us to test whether people who are higher in SPS, or higher in social introversion, reported higher well‐being on days that they had more frequent episodes of volitional solitude (i.e., to test H5). Exploratory analyses in which hedonic well‐being and eudaimonic well‐being were run separately as outcome variables in the GEE analysis demonstrated similar results to general well‐being, and so are not reported on below.

3.2.3. Duration of Solitude

In support of H3a, higher SPS significantly predicted longer duration of solitude episodes, and higher social introversion was marginally significant (see Table 5). Although not predicted, lower BFI introversion and lower restrained introversion were also marginally significant in predicting longer duration of solitude episodes.

3.2.4. Frequency of Solitude

In support of Hypothesis H3b, higher social introversion and higher SPS both significantly predicted higher frequency of voluntary solitude episodes (see Table 6). Thinking introversion, however, was not associated with frequency of voluntary solitude episodes. Although not hypothesized, lower BFI introversion, lower restrained introversion, and lower BFI neuroticism also predicted higher frequency of voluntary solitude episodes.

3.2.5. Stress

With respect to Hypothesis H4, we found that days with higher stress than normal for a person (i.e., person centered stress) predicted significantly higher frequency of voluntary solitude episodes (see Table 6). We found no significant interactions between stress and the traits of social introversion or SPS.

3.2.6. Well‐being

Our hypothesis for H5 was not supported. Rather, the reverse was found; on days when a person had more frequent solitude episodes compared with their average across all days of the study, their well‐being tended to be lower on that same day (see Table 7). The GEE revealed no significant interactions between solitude frequency, well‐being, and the traits of social introversion or SPS.

3.2.7. Phase 2 Results Summary

To summarize, our findings indicated that personality traits did show a relationship with one's daily solitary behaviors, with higher social introversion and higher SPS contributing to both higher frequency and longer duration of time in solitude. Introversion as measured by the BFI, however, showed the opposite pattern for both outcomes. In addition, higher daily stress levels were associated with more frequent episodes of voluntary solitude for that day, but this pattern was not restricted to those who scored high in social introversion or SPS. In the same vein, person‐specific increases in daily solitude episodes corresponded with decreased well‐being for that day, regardless of scores on the personality traits measured in this study.

4. Discussion

In this study, we sought to clarify the relationships between the traits of introversion and sensitivity with solitude motivation and solitary seeking behavior. Previous research showed mixed findings regarding introversion, which we speculated may be related to operationalizing this trait reductively as low scores on the extraversion factor of the Five Factor Model. In our study, we retained this measure (Big Five Inventory; John, Naumann, and Soto 2008) in order to compare it with the recently developed STAR Introversion Scale that measured the properties of introversion directly and in a multidimensional way (Cheek, Brown, and Grimes 2014). In addition, we included the trait of SPS (Aron and Aron 1997) which had thus far been missing from studies on solitary behavior, but which we hypothesized would play a significant role in solitude seeking.

Overall, our results revealed that motivations for solitude, both self‐determined and not, were indeed driven by sensitivity and various facets of the STAR model of introversion, but showed no relationship to introversion as measured by the BFI. In daily life, higher stress levels and reduced well‐being were both associated with more frequent voluntary solitude episodes on a given day.

4.1. STAR Introversion

Remarkably, nearly all of the variance in SDS was explained by two facets of introversion, social and thinking. We were surprised to find that although thinking introversion was a strong predictor of SDS and showed no relationship to its opposite (NSDS), this positive position did not translate into action; thinking introversion showed no relationship to frequency or duration of voluntary solitude during the daily diary phase of the study. Cheek, Brown, and Grimes (2014) described this type of introvert as focused on introspection and having a rich inner life. Potentially, those needs were not active during the 10‐day window of time in which these introverts were surveyed, or if they were active, did not require periods of social withdrawal during the day.

In contrast, social introversion showed a dual relationship with solitude, significantly predicting both self‐determined and not self‐determined motivations, and these motivations carried over to daily life; this type of introversion strongly predicted a high frequency of solitary episodes and marginally predicted a higher duration of time alone per episode. Cheek, Brown, and Grimes (2014) defined this type of introvert as low in gregariousness and as one who enjoys time alone as well as time with a few close friends and described them as therefore ambivalent about social interaction. The results of this study suggest that social introversion has a correspondingly ambivalent relationship with solitude, seeking it for constructive reasons but just as likely using it to escape from others. We speculate that if social connection is desired with close others who are not available, social introverts may retreat into solitude not because they want to be alone but because they wish to avoid socializing with acquaintances or large groups composed of weak social ties. This “push‐and‐pull” attitude toward the social and solitary worlds may not bode well for social introverts, as demonstrated by past work showing a relationship between negative well‐being and dual solitude motivations (Smith, Thomas, and Azmitia 2023).

Our findings are less clear regarding the solitude pattern of the remaining two types of introversion. The restrained facet appears rather benign, with high scores showing no relationship to SDS and lessening the likelihood that one is negatively motivated for solitude (NSDS). This lack of motivation carried over in daily behavior, where it was correlated with less frequency and lower duration of time alone. Finally, we can only speculate about anxious introversion given that it was removed from the model due to its excessive multicollinearity with other predictors, in particular neuroticism; as such, we suspect its relationship to solitude would be similar to that of the latter trait.

4.2. BFI Introversion

We designed this study in part to test our assumption that the Five Factor Model's limited view of introversion prevents an accurate understanding of its relationship to solitude. Indeed, we found that the BFI measure of introversion showed no relationship with a motivation to enter solitude, self‐determined or not (its association with NSDS early in the model disappeared after the STAR introversion facets were added). Furthermore, low rather than high scores on BFI introversion predicted both frequency and duration of time in daily solitude. In other words, according to the Big 5 paradigm it is extraverts who seek solitude. Conceptually, this is incongruent, given that solitude marks a withdrawal from social contact and the BFI measures extraversion with descriptors such as sociable and outgoing. Moreover, BFI introversion was highly correlated with neuroticism as well as anxious introversion, but not at all related to thinking introversion, the latter of which was the sole STAR facet to correspond exclusively with a positive motivation for solitude.

Taken together, these findings imply that the BFI captures a rather negative, or low‐functioning, version of introversion (see Thomas et al. 2021 for an example of distinguishing between low‐functioning and high‐functioning introversion). As a consequence, we surmise that higher‐functioning introverts who are neither anxious nor shy find themselves scoring high on Five Factor Model measures of extraversion, and it is their motivation for solitude that may be contributing to the mixed findings in previous literature regarding solitude's relationship to the extraversion–introversion dimension of personality (Nguyen, Weinstein, and Ryan 2022; Srivastava, Angelo, and Vallereux 2008; Thomas and Azmitia 2019). In contrast, the inclusion of both low‐ and high‐functioning introversion in the STAR Introversion Scale allows for a more nuanced understanding of this trait and offers a more precise understanding of its associations with solitude seeking.

4.3. Sensory Processing Sensitivity

Although SPS is a trait and not a disorder, people who are highly sensitive are vulnerable to increased stress and maladaptive outcomes when encountering (over)stimulating environments (Greven et al. 2019). Thus, we speculated that the comparatively understimulating environment of voluntary solitude with its lack of social interaction would be a logical and appealing resource for such individuals, and indeed, we found that to be the case. As with social introverts, people who are highly sensitive are dually motivated to enter solitude; when self‐determined (SDS), they enter solitude for the proactive purposes of self‐regulation and self‐discovery, and when not self‐determined (NSDS), they seek solitude in order to escape from uncomfortable or overwhelming social situations. Such dual motivation is consistent with work showing that highly sensitive people often feel conflicted about spending time with others, expressing that they use solitude to recover as well to engage in contemplation, self‐reflection, and emotion regulation (Black and Kern 2020). Not only was this trait a strong predictor of solitary motivation, it was also significantly associated with the frequency and duration of daily time alone. Compared with the trait of introversion, high sensitivity appears to be a robust indicator of solitude seeking, having strong and consistent correlations with both motivation for solitude and solitary behavior, although the STAR introversion facets show a clearer pattern of distinguishing between positive (SDS) and negative (NSDS) motivations for solitude.

4.4. Stress, Well‐being, and Solitude

Our findings revealed that it was not only introverts and the highly sensitive who sought solitude when stressed, and it was not only they who experienced decreased well‐being when they had a higher daily frequency of solitary episodes than was typical; rather, these associations were reflected in the sample as a whole. We speculate that daily increases in stress drove the increase in solitude frequency and were likewise reflected in diminished well‐being levels for that day. Participant reports of well‐being were generalized over the whole day and reported at end of each day, so it is unknown whether reduced well‐being levels remained constant over the course of the day or whether they fluctuated before and after solitude (falling due to stress and then rising after time alone), which is what we would expect if voluntary solitude indeed has a stress‐relieving function, or whether they fluctuated in the opposite direction. In the literature, some experimental (Nguyen, Ryan, and Deci 2018), experience sampling (Larson and Csikszentmihalyi 1978), and qualitative (Thomas 2023a, 2023b) studies have indicated that moderate amounts of volitional solitude have a mood‐regulating effect, allowing people to process difficult problems or feelings in private, after which they emerge feeling emotionally restored and better prepared to socially interact. Still other work has shown that time in solitude does not ameliorate stress and may in fact exacerbate ill‐being (Larson and Lee 1996; Pauly et al. 2017).

The direction of influence may depend on developmental skills more than individual differences in personality, while high sensitivity as well as social and thinking introversion are strong predictors of a motivation to seek solitude in general, their importance recedes when the contextual factor of stress enters the picture. Developmental maturity (Coplan, Ooi, and Baldwin 2019) as well as the capacity to utilize solitude constructively (Thomas 2023a) may be necessary to benefit from solitude during times of stress.

4.5. Limitations and Future Directions

While this study does provide a more comprehensive picture of the relationship of personality traits with solitude seeking, a few limitations should be noted. First, despite our attempt to establish a shared definition of solitude for the daily diary surveys, the content of some participant responses suggested that this understanding was inconsistent. Given some participants' difficulties with identifying and reporting valid solitary episodes, we recommend that researchers who employ this method include stronger validity checks; for example, closer monitoring of surveys as they are completed each day, or creating branching logic that provides feedback if a participant fails a validity check.

Second, retrospective daily diary sampling of solitude experiences as used in this study might yield different results than real‐time experiences captured by periodic alerts throughout the day (e.g., experience sampling). On the one hand, Hurlburt et al. (2022) recently found that people reported engaging in more self‐talk when assessed retrospectively than when assessed in the moment with the use of beepers; on the other hand, a review of ESM studies showed that both under‐reporting and over‐reporting of various mood states can occur with this method (aan het Rot, Hogenelst, and Schoevers 2012). With respect to solitude, if participants felt situational demands to experience solitude (by being in a study asking about solitude), they might have recalled more episodes of solitude than they actually had that day. Such biases in the reporting might have attenuated or inflated the associations we found. We encourage researchers to examine how strongly momentary sampling correlates with end of day diary sampling with respect to the occurrence of solitary episodes.

A third limitation involves the measures of introversion we selected. The STAR Introversion Scale is relatively new and to our knowledge has not been tested apart from the initial validation studies by Cheek, Brown, and Grimes (2014). Based on the results from this study, the STAR appears theoretically sound and promises to be empirically useful, but this measure needs to be utilized more widely to refine our understanding of a multidimensional model of introversion. We encourage researchers to consider utilizing the STAR Introversion Scale when investigating questions of introversion.

Relatedly, there are multiple questionnaires that measure the extraversion–introversion dimension of the Five Factor Model (FFM), and we selected the BFI. We made this choice in part because it provided an opportunity for the STAR Introversion Scale to be compared with an additional FFM measure (the original STAR validation studies had utilized the NEO‐PI‐3). It is possible that a different picture of results would have emerged had we utilized the NEO‐PI‐3, though we anticipate those differences would have been slight.

Although this study provides a clearer picture of the associations between the traits of introversion and sensitivity with solitary motivation and daily behavior, there are likely other individual differences and developmental factors that also play a role, and that moreover may determine the extent to which a person experiences benefits from time alone. Qualitative studies have begun to identify potential candidates, including the capacities for introspection and creativity, the development of a private self, and exposure to the benefits of solitude early in one's life, which may serve as a socializing factor (Ost Mor, Palgi, and Segel‐Karpas 2021; Thomas 2023a; Weinstein, Hansen, and Nguyen 2023). A small but growing literature has begun to examine the role of culture in shaping solitary motivations and behaviors, with most studies focused on comparing populations from individualistic cultures which emphasize personal freedom and independence (e.g., the United States) with those from collectivistic cultures which emphasize social harmony and community (e.g., China); similar to the introversion literature, results have been mixed with some studies showing solitude is experienced more positively in individualistic cultures while other studies show the opposite pattern (Jiang et al. 2019; Lay, Fung et al. 2020; Lay, Pauly et al. 2020; Van Zyl, Dankaert, and Guse 2018). More research is needed in this domain. Finally, experiences of trauma connected to being alone cannot be overlooked as a factor that shapes the solitude experience (Palgi, Hayun, and Greenblatt‐Kimron 2021).

Despite these limitations and the need for further research, the findings from this study mark an important step forward in identifying key individual differences that motivate solitary behavior and providing a deeper understanding of introversion in particular. Although introversion and sensitivity do predict the strength and direction of one's motivation to be alone, and to some extent predict the frequency and duration of time alone, solitude appears to be sought after by people in times of stress regardless of those individual differences. Future work can investigate more closely whether the likelihood of experiencing particular well‐being affordances of solitude, such as restoration or creativity, is associated with stable individual differences, dependent on momentary goals, needs, or moods, or contingent upon the maturation of one's emotional and cognitive capacities.

Author Contributions

Virginia Thomas: conceptualization (lead); methodology (lead); data collection (lead); writing (equal); review and editing (equal). Paul A. Nelson: conceptualization (supporting); methodology (supporting); data analysis (lead); writing (equal); review and editing (equal).

Conflicts of Interest

The authors declare no conflicts of interest.

IRB Status

The Institutional Review Board (IRB) of Middlebury College reviewed this study and approved that it was in compliance with the ethical treatment of human subjects.

Data Availability Statement

The data that support the findings of this study are publicly available in the Open Science Framework (OSF) repository at https://osf.io/xzkys/.