Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Oct 30.
Published before final editing as: Addict Behav. 2026 Oct 18;185:108883. doi: 10.1016/j.addbeh.2026.108883

Daily affect profiles during addiction recovery: Transition dynamics and links to acute recovery outcomes

Joseph H Lancaster 1, Timothy R Brick 1,2, Hannah B Apsley 3, H Harrington Cleveland 1
PMCID: PMC13718751  NIHMSID: NIHMS2218483  PMID: 42913146

Abstract

Introduction:

As individuals in recovery from substance use disorder (SUD) find initial stability, the next milestone is developing the emotional wellbeing that will offer greater quality of life, making affect a critical target. Many investigations into emotion during recovery, however, fail to account for the significant heterogeneity in daily affective experiences. The present study leverages advanced person-centered methods to capture that heterogeneity by identifying types of affect days in recovery and examines their place in daily recovery.

Methods:

129 individuals in SUD recovery were recruited remotely to complete a 10-day daily diary protocol (1,177 days). A mixed-indicator latent profile analysis was conducted on PANAS items to identify distinct profiles of daily affect. Latent profiles were examined in relation to prior-day meaning, next-day binary, and conditional craving, using robust standard errors to account for the repeated measures data structure.

Results:

A 5-profile solution was selected: Prevalent NA (14% of affect days), Prevalent NA-Mild PA (30%), High Arousal Affect (11%), Intense PA (17%), and Mixed Affect (28%). Compared to other profiles, Intense PA and High Arousal Affect days were associated with higher prior-day meaning, lower odds of next-day craving, and lower levels of next-day craving among craving days.

Conclusion:

Not only do results demonstrate the prevalence of NA-PA mixed days in recovery, but a second dimension – arousal – was also uncovered in affect patterns. Types of affect days differentially predicted acute recovery outcomes, with High Arousal Affect promoting recovery, warranting consideration of arousal in future studies.

Keywords: Addiction recovery, daily, affect, meaning, craving, latent profile analysis, intensive longitudinal data

1. Introduction

With the growing recognition that addiction is a chronic condition that demands continued care to achieve positive outcomes (see White & Kelly, 2011), researchers have increasingly focused on understanding the factors involved in supporting substance use disorder (SUD) recovery. As the twelve-step expression “One Day at a Time” suggests, much of the work involved in building and sustaining SUD recovery occurs at the daily level. Regardless of the specific circumstances or context, the common element in recovery experiences that all recovering individuals must successfully negotiate each day is their emotions (Zemore et al., 2023). Emotion is tied to both struggling and thriving in SUD recovery in that negative affect (NA) tends to pose risks and positive affect (PA) is generally protective of recovery. Further, once individuals find the initial stability that comes with overcoming the immediate life challenges associated with addiction, the next milestone is developing the emotional wellbeing that will offer greater quality of life, making emotion regulation a critical target for recovery support services (Kelly et al., 2018; Gutierrez et al., 2022). Most of the studies investigating affect in recovery, however, fail to address the significant within-person heterogeneity in daily affective experiences, limiting the insight their findings can provide. Recent methodological advancements available to intensive longitudinal data can capture that heterogeneity by incorporating into measurement the within-person covariance structure amongst specific emotional states. This study applies person-centered methods to intensive longitudinal data to identify types of affect days in SUD recovery. Prior to introducing the methodological details, this paper reviews theory and findings related to daily affective experiences in SUD recovery and elaborates on their complex and heterogeneous nature. We then discuss the dynamic elements of the identified latent profiles and how these profiles differentially relate to key phenomena in daily recovery.

1.1. Daily Affective Experiences in SUD Recovery

Several theories posit emotion as a critical factor in SUD recovery outcomes, with most distinguishing emotion by valence. Based in positive psychology, broaden-and-build theory (Fredrickson, 2001) suggests PA can broaden one’s mindset and build recovery capital (Granfield & Cloud, 1999), while negative reinforcement models (e.g., self-medication hypothesis; Khantzian, 1997) suggest NA is a primary driver of lapse (e.g., Shiffman & Waters, 2004) via coping motives (for a review, see Votaw et al., 2022), though findings from a meta-analysis challenge such theories for alcohol (see Dora et al., 2023). Although risk of lapse (that is, return to any substance use) is highest in early recovery for interpersonal and emotional reasons (Marlatt & Witkiewitz, 2005), distressing emotions pose risk even at later stages of recovery because chronic substance use permanently alters the brain’s mesocorticolimbic circuitry (and hypothalamic-pituitary-adrenal or HPA axis), impairing the prefrontal cortex’s ability to regulate emotions (see Stewart, 2008 for a review of the neurological underpinnings of relapse). In addition to valence, the circumplex structure of core affect suggests affect maps onto a second axis – arousal, organizing all emotional experiences into a two-dimensional space: pleasure-displeasure (valence) and high-low activation (arousal) (Russell & Barrett, 1999). The circumplex model is understudied in addiction research where most studies focus on positive and negative affect (Tovmasyan et al., 2022; Votaw et al., 2022); however, there are reasons to consider the arousal dimension. For example, chronic low arousal mood (e.g., depressive symptoms) has long been linked to lapse risk (e.g., Curran et al., 2000) – likely due to reward deficiency syndrome (File et al., 2022) – and has been used as a day-level outcome in recent addiction recovery research (Stull et al., 2025). High-arousal emotions may also contribute to lapse risk though, given these states trigger hyperactivity in the amygdala and dampen activity in the prefrontal cortex, leading to an amygdala hijack (Goleman, 1995) where rational thought is compromised (see Hand et al., 2024 for a review). Such effects have been shown during active use where high arousal PA (e.g., excitement) was linked to higher chances of drinking (via enhancement or social motives) (Jones et al., 2021). An amygdala hijack may pose similar risks during addiction recovery, where individuals might use substances in response to high-arousal emotions, driven by factors such as impulsivity and a lack of forethought.

Daily emotional experiences are inherently complex for a variety of reasons. Although PA and NA appear to be on opposite ends of a spectrum (and may negatively correlate in certain situations; Blaxton et al., 2021; Dejonckheere et al., 2021), they are generally considered independent systems that can operate in tandem (Dejonckheere et al., 2021), with the latest neuroscience supporting the occurrence of mixed-valence emotions (Vaccaro et al., 2024). Although there may be days in recovery characterized by only PA or only NA, their simultaneously presence (i.e., mixed affect; Vaccaro et al., 2024) – whether in the same moment or across the course of a day – represents emotional complexity that may be difficult to navigate, making these internal experiences critical to recovery. Indeed, there may be several combinations of PA and NA that fall between the two extremes (PA only to NA only) in varying degrees of intensity of each. Furthermore, the degree of arousal may add to the complexity into which affective patterns unfold, expanding the potential typology of affect days in recovery. Considering both the daily nature of recovery and the mixture of emotions that arise each day, the day is an optimal unit of analysis to examine affect in SUD recovery.

Many of the studies investigating daily emotional experiences in SUD recovery employ variable-centered methods that examine relationships between variables while assuming homogeneity across individuals. Person-centered approaches, such as latent profile analysis (LPA), do not hold this homogeneity assumption and instead examine relationships between individuals, classifying them into subgroups within a population based on similarity in their response patterns (Lanza, 2016). Recent methodological advancements in LPA have expanded the application of LPA from distinguishing types of individuals to distinguishing types of days. The present study applies LPA via a marginal model with robust standard errors to identify daily affect profiles (i.e., types of affect days) based on response patterns for specific affect items. As will be elaborated on below, this method provides an ideal opportunity to capture the patterns of emotional experiences that characterize days and place these patterns on days within the context of individuals who experience them.

1.2. The Dynamic Elements of Daily Affect Profiles

Affect is a fundamentally dynamic construct that changes across the course of hours and days, showing significant intraindividual variability (Eid & Diener, 1999). Given the dynamicity of affect, we are interested in whether the uncovered affect profiles also exhibit intraindividual variability, demonstrated by individuals switching between profiles from one day to the next. In addition, because the affect profiles were extracted from nested repeated measures data, it is possible the profiles are an artifact of between-person differences. Specifically, if transitions between profiles do not occur (i.e., participants tend to remain in one profile for the duration of the study), then the extracted profiles might better represent different types of individuals in recovery rather than different types of days in recovery. Examining the patterns of affect profile transitions will provide information regarding the volatility or stability of specific affect-day types.

1.3. Daily Outcomes in Recovery

Investigating associations between affect profiles and acute recovery outcomes will shed light on the importance of each affect profile in daily recovery processes. Drug craving is an important aspect of SUD recovery, not only because it is a robust predictor of lapse (Serre et al., 2015), but it is an aversive state and dysphoric aspect of daily recovery (Kleykamp et al., 2019). Craving is an important outcome to consider in research (Tiffany & Wray, 2012), having been formally adopted as a clinical trial endpoint by the Federal Drug Administration (FDA) (Volkow et al., 2018). Indeed, craving is a known within-person correlate of affect, showing positive associations with NA (depending on type of substance used; see Votaw et al., 2022) and negative associations with PA in variable-centered analyses (Huhn et al., 2016), and is expected to relate to some of the affect-day types at the day level.

Sense of meaning was also examined as a daily recovery outcome given the increasing attention it is receiving in the recovery literature (e.g., Knapp et al., 2021; Stull et al., 2025; Lancaster et al., 2025). Not only is sense of meaning and purpose an important aspect of psychological wellbeing, but it is a key motivator in remaining sober in the face of difficulty. This is because finding meaning and purpose increases the price of relapse by giving individuals more to lose in the event of a return to substance use (Wagner & Sanchez, 2002; Kim et al., 2020; Bickel et al., 1993). Furthermore, a sense of meaning is an important psychological resource that can be vital to sustaining recovery. Research shows that finding meaning is associated with better emotional and cortisol recovery following negative stimuli (Schaefer et al., 2013; Fogelman & Canli, 2015), lower stress (Sutin et al., 2024), greater pain tolerance (Smith et al., 2009), and lower impulsivity (Burrow & Spreng, 2016), reducing the likelihood problematic substance use (Kim et al., 2020). Unsurprisingly, sense of meaning is also linked to higher likelihood of recovery (Roos et al., 2015; Krentzman et al., 2015) and predicts longer durations of recovery (in alcohol use disorder; McInerney et al., 2025). Holistic metrics of recovery, such as perceived meaning, are important to consider alongside traditional outcomes like lapse risk because, for recovery to become long-term, goals may need to shift from merely not using substances to building a fulfilling life (Pettersen et al., 2023). Regarding daily affect, research has found that meaning in life is positively associated with PA at the within-person level (e.g., Miao et al, 2017). Because assigning meaning or subjective significance increases personal investment, perceived meaning is thought to trigger high arousal emotional states (Bayer et al., 2017). Like affect, both perceived meaning and craving are dynamic in that they vary from day to day (Knapp et al., 2021), making them excellent criteria to examine against types of affect days in recovery.

Drawing on data collected from over 150 people in recovery who reported up to ten days of surveys on their daily experiences and emotional states, this study sought to identify and characterize the daily patterns of affect experienced by people in recovery and relate those daily patterns to acute recovery outcomes. It is organized around three analytic steps. First, LPA is used in an exploratory manner to identify and classify types of days based on patterns of affect. Second, the across-day transition probabilities between types of days are examined to determine how these days are distributed across time within people. Third, days are predicted within-person to link these days to clinically recognized daily outcomes, namely sense of meaning and drug craving, to illustrate their role in daily recovery.

2. Methods

2.1. Participants and Procedure

Participants were individuals who self-reported as in recovery from SUD (Nparticipants = 184), sampled across the state of Pennsylvania using remote methods to carry out the study. The majority were in stable recovery, with more than half (56.59%) reporting being in recovery for longer than 5 years. Recruitment leveraged the support of recovery organizations (e.g., state-funded regional “recovery hubs”) to increase reach into this community. These organizations distributed information about the study via online flyers and organizational newsletters. Interested participants contacted the study team and were then directed to complete a short eligibility survey via Qualtrics. Eligibility criteria included: 1) being at least 18 years of age, 2) ability to speak English, and 3) being otherwise able to complete the study protocol. The protocol involved completing a baseline survey (via Qualtrics) followed by 10 daily diary surveys (ndays = 1,236). At the onset of the study, a virtual meeting was conducted with each participant to review and provide informed consent and then receive instructions on how to install and use the Wear-IT data collection app (Brick et al., 2020) through which the daily surveys were delivered. The ~15-minute survey was available from 8:30pm to 10:30pm each evening. Participants could receive up to $70 in compensation ($10 for completing the baseline and $6 per daily diary survey), paid through an Amazon gift card. All study protocols were approved by a university internal review board. Table 1 contains more descriptive information on the sample.

Table 1.

Descriptive information and demographic characteristics of sample.

Variable ICC N (%) or Mean (SD)
Person-level variables (N = 129)
Sex (female) 94 (72.87%)
Age (years) 47.36 (10.46)
Race
 White 109 (84.50%)
 Black or African American 12 (9.30%)
 American Indian or Alaskan Native 1 (0.78%)
 Asian 1 (0.78%)
 Multiracial 5 (3.88%)
 Other 1 (0.78%)
Time in recovery
 < 1 year 21 (16.28%)
 1–2 years 15 (11.63%)
 2–5 years 20 (15.50%)
 5+ years 73 (56.59%)
Primary substance used
 Alcohol 31 (24.03%)
 Marijuana 7 (5.43%)
 Cocaine 16 (12.40%)
 Heroin 37 (28.68%)
 Unprescribed other opioids 9 (6.98%)
 Hallucinogens 1 (0.78%)
 Amphetamine 1 (0.78%)
 Methamphetamine 17 (13.18%)
 Benzodiazepines 2 (1.55%)
 Tobacco 2 (1.55%)
 Fentanyl 6 (4.65%)
Day-level variables (1,177 person-days)
Meaning 0.52 75.11 (22.85)
Binary craving (any) 0.53 262 (22.26%)
Conditional craving 0.35 7.68 (16.26)
Open in a new tab

Note. ICC = intraclass correlation coefficient.

2.2. Measures

2.2.1. Daily Affect Items

Eight daily affect items were drawn from the Positive and Negative Affect Schedule (PANAS) (Watson et al., 1988) to be used as indicators in the latent profile analysis. Based on the circumplex structure of core affect (Russell & Barrett, 1999), the items encompassed four quadrants based on valence and arousal: positive/high arousal (happy, joyful), positive/low arousal (relaxed, calm), negative/high arousal (irritable, anxious), and negative/low arousal (lonely, sad). Each item was preceded by the stem “this evening, have you felt…” and was assessed on a 100-point continuous visual analog scale (0 = not at all, 100 = very).

2.2.2. Daily Outcomes

Perceived meaning in recovery was assessed with two items on a continuous scale, ranging from 0 (strongly disagree) to 100 (strongly agree). Prefaced with “My day has been…”, the two items were ‘meaningful’ and ‘purposeful’. Using an average of the two items, higher scores on the scale indicate greater perceived meaning in the individual’s recovery that day (α = 0.93).

Drug craving was measured via two items (“How frequent have your alcohol or drug cravings been?” and “How intense have your alcohol or drug cravings been?”) using a visual analog scale (0 = no cravings, 100 = very frequent/intense). Following suit with prior studies (e.g., Huhn et al., 2016; Knapp et al., 2021), craving frequency and intensity scores were multiplied together and then divided by 100 (for rescaling) to reflect each participant’s “total volume” of cravings, such that higher scores indicate more total craving over the course of the day (α = 0.96). Because zero-inflation was detected during data exploration, drug craving was operationalized as two separate outcome variables: binary craving and conditional craving. For the binary craving outcome, all craving scores were categorized dichotomously (0 = no craving, 1 = any degree of craving) to account for the spike at zero in the distribution. The conditional craving outcome was a continuous variable restricted to craving scores greater than zero (262 out of 1,177 days; 22.26%). Examining conditional craving allowed us to assess the severity of craving rather than just its presence. We refer to this outcome as “conditional craving” because the results are conditional on a craving (>0) being reported.

2.3. Analyses

Mplus software version 8.7 (Muthén & Muthén, 2017) was used to perform LPA and test associations with study outcomes. To examine the heterogeneity in daily affective experiences, LPA was conducted on eight daily indicators: happy, joyful, relaxed, calm, irritable, anxious, lonely, and sad. Exploration of item distributions revealed the latter four (i.e., NA) items were zero-inflated; therefore they were dichotomized (0 = no affect, 1 = any degree of affect) to meet the LPA assumption of multivariate normality. Mixing indicators (continuous PA & binary NA) in LPA complicates interpretation of profiles, leading to hybrid profiles that should be interpreted as presence/absence for categorical indicators (i.e., NA items) and varying levels for continuous indicators (i.e., PA items); however, violating the normality assumption by keeping the NA items as continuous would likely lead to spurious profile extraction and biased parameter estimates (Van Horn et al., 2012). Mixing indicators in LPA may also increase the risk of converging on local maxima; therefore, for each model, 250 sets of random starting values were used to confirm model identification. Models with 1–8 profiles were examined in a series of mixed-indicator LPAs to avoid over-extraction (due to redundancy) and compared based on several model fit indices (as found in Collins & Lanza, 2013). Lower values of the Akaike information criterion (AIC) and Bayesian information criterion (BIC) indicate a better fitting, more parsimonious model. The Vuong-Lo-Mendell-Rubin (VLMR) likelihood test and bootstrapped likelihood ratio test (BLRT) are used to compare fit between models with k profiles and k-1 profiles, with a significant p-value indicating the k model fits better. Higher entropy values indicated more distinct profile separation. The optimal profile solution was selected based on model fit indices, profile sizes, and substantive interpretation. Robust standard errors were used to account for nesting in the daily diary data.

Outcomes were regressed onto the daily affect profiles using the three-step “BCH” method to account for uncertainty in daily profile assignment (Bolck et al., 2004). In this method, BCH weights derived from the measurement model are used as training data in the structural model (Vermunt, 2010). To establish directionality in effects, meaning in recovery was lagged one day (t-1) and binary and conditional craving were lead one day (t+1), such that yesterday’s perceived meaning would predict today’s type of affective experience and today’s type of affective experience would predict tomorrow’s odds or degree of drug craving. Because a ceiling effect was detected in the recovery meaning data, the association between daily affect profiles and perceived meaning in recovery was modeled using censored (Tobit) regression, allowing the scale to exceed its maximum score of 100 (Tobin, 1958). The intraclass correlation coefficients (ICCs) for each daily outcome (meaning, binary craving, conditional craving) were 0.52, 0.53, and 0.35, respectively (Table 1). As these values indicate the proportion of between-person variance in a variable, the remaining variation represents the proportion (e.g., .48 or 48% for meaning) of total variance that is attributable to within-person variation. In moving to the second step of the BCH approach, missing data on the outcomes (treated as exogenous variables) – 73 missing craving surveys (6.2%) and 77 missing meaning surveys (6.5%) – reduced the sample size, bringing the final analytic sample to 1,177 days nested within 129 participants. The missing data patterns were evaluated using a chi-square test. The test was not significant, indicating that the data are missing completely at random and the missingness is not systematic. All models were estimated in Mplus using maximum likelihood with robust standard errors to accommodate the repeated measures structure of the day-level data.

In a two-step approach, discrete-time Markov chains were fitted to the assigned daily profile membership data to compute a transition probability matrix, providing exploratory information about the volatility or stability of each affect profile. To examine the profile solution’s sources of variance (between- vs. within-person), summary statistics pertaining to participants’ propensity to switch profiles (e.g., the average number of profile transitions, average dwell time in each profile) were also calculated from this data. The transition probability matrix was estimated using maximum likelihood via the markovchain package in R (Spedicato, 2017).

3. Results

3.1. Latent Profiles of Daily Affect

Model fit information and selection criteria for LPA models with 1–8 profiles are summarized in Table 2. The AIC and BIC favored complexity as these fit statistics decreased as the number of profiles increased. VLMR suggested the 4-profile model was preferrable to the 3-profile model but not the 5-profile model, although the BLRT remained significant for the 5-profile model. In moving from the 4- to the 5-profile model, a clear and interpretable profile emerged. The 6th and further profiles were redundant with existing profiles and added very little substantive information; therefore, the more parsimonious 5-profile model was selected.

Table 2.

Model fit information and selection criteria for latent profile models with 1–8 profiles of daily affect in addiction recovery.

Number of Profiles df AIC BIC VLMR p Entropy
1 12 48584.34 48645.20 -- --
2 21 46199.68 46306.18 0.000 0.85
3 30 45157.94 45310.09 0.000 0.87
4 39 44840.83 45038.62 0.001 0.82
5 48 44565.73 44809.17 0.335 0.83
6 57 44340.54 44629.61 0.389 0.84
7 66 44155.62 44490.34 0.617 0.83
8 75 43999.94 44380.30 0.189 0.84
Open in a new tab

Note. df = degrees of freedom; AIC = Akaike information criterion; BIC = Bayesian information criterion; VLMR = Vuong-Lo-Mendell-Rubin likelihood ratio test.

The overall and profile-specific item-response probabilities (NA indicators) and means (PA indicators) for the 5-profile model are displayed in Table 3. Profile 1 (14% of affect days) was labeled Prevalent NA as it was characterized by high probabilities of experiencing NA and low levels of PA. Profile 2, labeled Prevalent NA-Mild PA, was the most common type of affect day (30%), characterized by high probabilities of experiencing NA and levels of PA that were relatively low (lower than the sample mean for the PA indicators; see Table 3), but higher than profile 1. Profile 3 (11%) was characterized by higher probabilities/ means of high arousal affect, regardless of valence, and so labeled High Arousal Affect. Profile 4 (17%) was labeled Intense PA as it was characterized by high levels of PA and low probabilities of experiencing NA. Profile 5 was the second most common type of affect day (28%). This profile was characterized by relatively high probabilities/ means on both NA and PA items and labeled Mixed Affect.

Table 3.

Parameter estimates for the selected 5-profile model of daily affect in addiction recovery.

Latent Profiles
Sample (ndays = 1,177) Prevalent NA 14% (ndays = 170) Prevalent NA-Mild PA 30% (ndays = 353) High Arousal Affect 11% (ndays = 126) Intense PA 17% (ndays = 203) Mixed Affect 28% (ndays = 325)
Daily Affect Indicators Probability / Mean Profile-specific Item Probabilities of ‘Yes’ Response / Means
SadB 0.74 0.88 0.99 0.30 0.25 0.89
AnxiousB 0.82 0.92 0.99 0.63 0.34 0.96
LonelyB 0.70 0.75 0.94 0.21 0.25 0.89
IrritableB 0.81 0.93 0.98 0.61 0.34 0.95
CalmC 61.54 25.89 49.41 58.20 91.28 76.00
HappyC 67.08 26.56 57.54 73.82 93.77 79.04
JoyfulC 64.68 23.05 53.25 71.42 93.11 78.01
RelaxedC 55.95 20.41 43.12 41.92 90.12 73.07
Open in a new tab

Note.

B =

binary variable;

C =

continuous variable.

All binary indicators reflect negatively-valenced affect. All continuous indicators reflect positively-valenced affect.

3.2. Daily Variation and Stability Across Affect Profiles

On average, participants switched between daily affect profiles 4.05 times during the study period (45% of possible transitions), with only 8 participants remaining in the same profile across the 10 days of data collection. The probability of an individual remaining in the same affect profile from one day to the next ranged from .37 to .58, suggesting moderate stability across profiles, with the High Arousal Affect profile showing the most volatility (0.37) and the Intense PA profile showing the strongest inertia (.58) (see diagonal of Table 4). The average duration of time spent in each affect profile before transitioning to another (i.e., sojourn or dwell time) was: 1.72 days for the Prevalent NA profile; 2.14 days for Prevalent NA-Mild PA profile; 1.58 days for the High Arousal Affect profile; 2.39 days for the Intense PA profile; and 2.07 days for the Mixed Affect profile. The most likely transitions between affect profiles to occur were: from Prevalent NA days to Prevalent NA-Mild PA days (30% chance); from High Arousal Affect days to Intense PA days (26% chance); from Prevalent NA-Mild PA days to Mixed Affect days (24% chance); and from Mixed Affect days back to Prevalent NA-Mild PA days (23.5% chance). Individuals were least likely to move from Intense PA days to Prevalent NA days (2.2% chance), from Prevalent NA-Mild PA days to Intense PA days (2.5% chance), from Prevalent NA days to Intense PA days (3.3% chance), and from Prevalent NA-Mild PA days to High Arousal Affect days (4.8% chance). Table 4 contains the full transition probability matrix.

Table 4.

Transition probability matrix for profiles of daily affect in SUD recovery.

Latent Profile To (tomorrow)
From (today) Prevalent NA (170 days) Prevalent NA-Mild PA (353 days) High Arousal Affect (126 days) Intense PA (203 days) Mixed Affect (325 days)
Prevalent NA 0.418 0.301 0.078 0.033 0.170
Prevalent NA-Mild PA 0.152 0.533 0.048 0.025 0.241
High Arousal Affect 0.137 0.154 0.368 0.256 0.085
Intense PA 0.022 0.056 0.151 0.581 0.190
Mixed Affect 0.056 0.235 0.053 0.140 0.516
Open in a new tab

Note. NA = Negative Affect; PA = Positive Affect.

3.3. Associations with Daily Recovery Outcomes

Figures 1–3 present the estimated mean on each outcome (prior-day meaning, next-day binary craving, and next-day conditional craving, respectively) for the five affect profiles. The day-level associations between each of the latent profiles and recovery outcomes, relative to a reference profile, are displayed in Table 5. Intense PA days and High Arousal Affect days were generally associated with better daily recovery outcomes than other types of affect days. Specifically, Intense PA and High Arousal Affect days were predicted by significantly higher meaning in recovery the previous day and predicted significantly lower odds of craving (binary craving) the following day, compared to the other three types of affect days. Further, among days when any craving was reported, Intense PA and High Arousal Affect days predicted significantly lower next-day craving levels (conditional craving), relative to Prevalent NA and Prevalent NA-Mild PA days, but not Mixed Affect days. Although showing weaker effects than Intense PA and High Arousal Affect days, Mixed Affect days were also significantly associated with higher prior-day meaning, lower odds of next-day craving (binary craving), and lower levels of next-day craving among any craving days (conditional craving), relative to Prevalent NA and Prevalent NA-Mild PA days.

Figure 1.

Figure 1.

Open in a new tab

Estimated means for prior-day meaning associated with each daily latent profile.

Figure 3.

Figure 3.

Open in a new tab

Estimated means for next-day craving (among days any craving was reported) associated with each daily latent profile.

Table 5.

Associations between daily affect profiles and prior-day meaning and next-day drug craving.

Daily outcomes B (SE) / OR (SE)
Latent Profile Prior-day Next-day Next-day
(reference profile) meaning conditional craving binary craving
Prevalent NA −7.53* 8.33* 1.38*
(Prevalent NA-Mild PA) (3.03) (3.29) (0.33)
Prevalent NA −29.50* 16.55* 8.58*
(High Arousal Affect) (3.70) (2.94) (4.24)
Prevalent NA −41.89* 15.31* 16.14*
(Intense PA) (3.27) (2.97) (7.88)
Prevalent NA −18.70* 14.14* 3.23*
(Mixed Affect) (2.98) (2.93) (0.83)
Prevalent NA-Mild PA −21.96* 8.23* 6.22*
(High Arousal Affect) (3.09) (1.73) (3.04)
Prevalent NA-Mild PA −34.36* 6.99* 11.71*
(Intense PA) (2.48) (1.69) (5.53)
Prevalent NA-Mild PA −11.17* 5.82* 2.35*
(Mixed Affect) (2.25) (1.79) (0.57)
High Arousal Affect −12.40* −1.24 1.88*
(Intense PA) (3.49) (1.29) (1.26)
High Arousal Affect 10.79* −2.41 0.38*
(Mixed Affect) (3.22) (1.27) (0.19)
Intense PA 23.19* −1.17 0.20*
(Mixed Affect) (2.74) (1.41) (0.10)
Open in a new tab

Note. Conditional craving only reflects days any craving was reported. Meaning and conditional craving are on a continuous scale, with their corresponding B representing the mean difference in the outcome when comparing a latent profile with a reference profile. Binary craving was coded as 0 = no craving that day, 1 = any craving that day, with the corresponding odds ratio (OR) indicating the odds of ‘any craving’ for each profile.

*

p < .05.

4. Discussion

Emotional experiences are a critical yet complex part of daily recovery from SUD. This investigation employed advanced person-centered methods to capture the heterogeneity in daily emotional experiences as they naturally unfold, identifying the types of affect patterns that emerge during daily recovery. We found five daily profiles of affect, distinguished by both valence and arousal. In regard to valence, patterns of affect indicators suggested an inverse same-day relationship between levels of PA and odds of any NA within two affect-day profiles (Intense PA and Prevalent NA), supporting a bipolar structure for affect (Russell & Carroll, 1999). Although the odds of NA being present and levels of PA were both high in Mixed Affect days, PA was reported at lower levels in the Prevalent NA-Mild PA profile, showing a weaker relationship to NA. This suggests that not only are NA and PA two independent systems (Watson & Clark, 1997), but they operate simultaneously at the day level (Larsen et al., 2001). Further, these two types of days (Prevalent NA-Mild PA, Mixed Affect) constituted the largest profiles at 30% and 28% of all study days, respectively, demonstrating their prevalence in SUD recovery and warranting attention in future affect-recovery research. Interestingly, a High Arousal Affect day also emerged. This is notable because arousal is usually ignored in affect studies in favor of a valence-only approach. Lastly, these findings partially align with those of a study applying a similar approach to another population – university students at risk of problematic substance use (Van Doren et al., 2024). That study found similar patterns for three of five profiles (Prevalent NA, Intense PA, and Mixed Affect), with a high arousal day also emerging, although that profile was distinguished by high PA.

In linking the daily affect profiles to acute recovery outcomes, the pattern of association for the five affect profiles remained consistent across the three outcomes. Specifically, the Intense PA and High Arousal Affect profiles appear the most favorable for navigating daily recovery as they were associated with greater prior-day meaning, lower odds of any next-day craving (binary craving), and lower levels of next-day craving among days when any craving was reported (conditional craving), relative to the other three profiles. The findings for the Intense PA profile are consistent with theory as well as prior research in other areas. According to Broaden-and-Build Theory, positive emotion increases cognitive and social resources and thus improves adaptive functioning (Fredrickson, 2001) and, in the realm of SUD recovery, decreases lapse risk (i.e., craving) indirectly via abstinence-focused social support and approach-oriented coping as demonstrated in Carrico et al. (2013). The theory also holds that the social and personal resources derived from positive emotions may foster social connections and activities that produce a sense of meaning, as exhibited in a study by Miao et al. (2017) which linked PA to meaning in life (and proactive coping) in a daily diary study on a non-SUD population.

Regarding the High Arousal Affect profile findings, the strong positive association with prior-day meaning makes sense in that perceiving one’s day in recovery as meaningful should predict high emotional arousal the next day (regardless of valence). Assigning meaning increases investment in recovery, thereby triggering high-arousal emotional states (Bayer et al., 2017). Note that these results demonstrate that high levels of meaning predict high arousal affect the next day, and not the reverse (i.e., high arousal affect predicting next-day meaning). The High Arousal Affect profile was also linked to lower odds of any next-day craving and lower levels of next-day craving (among days when any craving was reported). This conflicts with extant research on developing and active addiction, which suggests that high emotional arousal (particularly high arousal PA) is a risk factor for alcohol use, likely due to social or enhancement motives for drinking (Van Doren et al., 2024; Jones et al., 2021). This discrepancy may stem from fundamental differences in the samples. For individuals in stable recovery — such as those in the present study, where 56.59% have been in recovery for over 5 years) — the mechanisms relating emotional arousal to addiction differ. Specifically, high arousal states seem to coincide with a deeper investment in recovery, thereby reducing the risk of a lapse. Indeed, the current findings are based on a sample where the majority has significant recovery time and so, in large part, reflect a group that has been under-researched, bringing a rare perspective into the SUD literature.

Conversely, the Prevalent NA profile, along with the Prevalent NA-Mild PA and Mixed Affect profiles, was associated with high recovery risk based on the craving and meaning metrics. These findings are generally consistent with prior within-person research that suggests NA is a predictor of craving (see Votaw et al., 2022 for a review) and that perceived meaning in life can reduce distress (e.g., Hill et al., 2018). The present study extends this work by bringing a more nuanced understanding of recovery risks by including NA and PA simultaneously. Specifically, the present findings suggest that on days marked by NA (i.e., high probability of NA), lapse risk (i.e., craving) appears to decrease proportionally following days more PA is experienced, such as on Prevalent NA-Mild PA days (relative to Prevalent NA days) and Mixed Affect days (relative to Prevalent NA and Prevalent NA-Mild PA days) – the most common types of affect days found in this study (28% and 30%, respectively). A similar pattern emerges for meaning. Specifically, level of meaning appears to be proportional to the level of PA following NA marked days, such that the highest levels of PA on NA marked days (i.e., Mixed Affect days) are followed by the most meaning of the three NA marked days (i.e., Prevalent NA, Prevalent NA-Mild PA, Mixed Affect).

4.1. Limitations and Future Work

The findings presented here are subject to some important caveats. First, the pool of items used as indicators in the LPA was limited to only eight affect items. A wider array of affect terms (the full PANAS scale contains 20 items) might offer a more nuanced profile solution, although the present investigation selected multiple LPA indicators for each quadrant of the circumplex model to adequately represent core affect. Next, the present sample was 84.5% white and 72.87% female, causing concerns about potential race and gender bias; replication with a more balanced composition of race and sex would address these generalizability concerns. Additionally, the analytic methods used here to model daily affect profiles as predictors of acute recovery outcomes could not support covariates, limiting the ability to address other potential influences on outcomes, including interactive effects. In preliminary analyses, length of time in recovery and demographic characteristics were examined separately from the affect profiles against the day-level craving and meaning outcomes. Results indicated longer durations in recovery were indeed associated with better outcomes, and age was negatively related to binary craving, but not conditional craving or meaning; these influences should, therefore, be considered in future investigations. The purpose of testing these daily outcomes against the affect profiles was to further characterize each type of affect day, elaborating on each’s unique role (relative to others) as dynamic risk or protective factors in daily recovery. Further, recovery duration is also theoretically pertinent to affect processes; however, it was not accounted for in the models used to recover the types of affect days. We also want to point out that affect processes and lapse risk differ meaningfully across substances. However, due to the craving instrument’s broad nature and insufficient statistical power, we are unable to draw substance-specific conclusions. Indeed, more light can be shed on the examined relationships by incorporating covariates and interaction terms into the LPA with distal/proximal outcomes framework, and we encourage methodologists to pursue this.

Lastly, our investigation considered transitions from one profile to the next only as a basic index of variation and stability. A closer examination into the transition patterns between types of affect days may yield rich results for future intensive longitudinal studies. Use of latent transition or Markov models could not only reveal the probabilities of transitioning from one type of affect day to another type of affect day, but also provide insights unique to affect dynamics, such as the inertia or volatility of a given emotion-day type. These examinations might also shed light on transition sequences that may be maladaptive such as sudden shifts to emotion-day types associated with risk. Although there are certain methodological challenges to implementing these types of models (e.g., only available in certain statistical software or in Bayesian inference framework), we believe they can provide significant contributions to our understanding of affect and SUD recovery.

Figure 2.

Figure 2.

Open in a new tab

Estimated probability of any next-day craving associated with each daily latent profile.

Declarations of interest:

This study was supported by grant R24 DA051988 from the National Institute on Drug Abuse (Cleveland, H.H.; MPIs: Kelly, J. & Hoeppner, B.). Authors were supported by the Prevention and Methodology Training Program (T32 DA017629; Trainee: Lancaster, J.H.; MPIs: Maggs, J. & Lanza, S.) with funding from the National Institute on Drug Abuse. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Drug Abuse or the National Institutes of Health. The authors declare that they have no conflicts of interest.

References

  1. Bayer M, Ruthmann K, & Schacht A. (2017). The impact of personal relevance on emotion processing: evidence from event-related potentials and pupillary responses. Social Cognitive and Affective Neuroscience, 12(9), 1470–1479. 10.1093/scan/nsx075 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bickel WK, DeGrandpre RJ, & Higgins ST. (1993). Behavioral economics: A novel experimental approach to the study of drug dependence. Drug and Alcohol Dependence, 33(2), 173–192. 10.1016/0376-8716(93)90059-Y [DOI] [PubMed] [Google Scholar]
  3. Blaxton JM, Nelson NA, & Bergeman CS. (2021). The positive and negative affect relation in the context of stress and age. Emotion (Washington, D.C.), 21(8), 1712–1720. 10.1037/emo0000975 [DOI] [PubMed] [Google Scholar]
  4. Bolck A, Croon M, Hagenaars J, 2004. Estimating Latent Structure Models with Categorical Variables: One-Step Versus Three-Step Estimators. Political Analysis. 12(1), 3–27. [Google Scholar]
  5. Brick TR, Mundie J, Weaver J, Fraleigh R, & Oravecz Z. (2020). Low-burden mobile monitoring, intervention, and real-time analysis using the Wear-IT framework: Example and usability study. JMIR Formative Research, 4(6), e16072. 10.2196/16072 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Burrow AL, & Spreng RN. (2016). Waiting with purpose: a reliable but small association between purpose in life and impulsivity. Personality and Individual Differences, 90, 187–189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Carrico AW, Woods WJ, Siever MD, Discepola MV, Dilworth SE, Neilands TB, Miller N, & Moskowitz JT. (2013). Positive affect and processes of recovery among treatment-seeking methamphetamine users. Drug and Alcohol Dependence, 132(3), 624–629. 10.1016/j.drugalcdep.2013.04.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Collins LM, & Lanza ST. (2013). Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. John Wiley & Sons. [Google Scholar]
  9. Curran GM, Flynn HA, Kirchner J, & Booth BM. (2000). Depression after alcohol treatment as a risk factor for relapse among male veterans. Journal of Substance Abuse Treatment, 19(3), 259–265. 10.1016/s0740-5472(00)00107-0 [DOI] [PubMed] [Google Scholar]
  10. Dejonckheere E, Mestdagh M, Verdonck S, Lafit G, Ceulemans E, Bastian B, & Kalokerinos EK. (2021). The relation between positive and negative affect becomes more negative in response to personally relevant events. Emotion (Washington, D.C.), 21(2), 326–336. 10.1037/emo0000697 [DOI] [PubMed] [Google Scholar]
  11. Dora J, Piccirillo M, Foster KT, Arbeau K, Armeli S, Auriacombe M, Bartholow B, Beltz AM, Blumenstock SM, Bold K, Bonar EE, Braitman A, Carpenter RW, Creswell KG, De Hart T, Dvorak RD, Emery N, Enkema M, Fairbairn CE, Fairlie AM, … King KM. (2023). The daily association between affect and alcohol use: A meta-analysis of individual participant data. Psychological Bulletin, 149(1–2), 1–24. 10.1037/bul0000387 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Eid M, & Diener E. (1999). Intraindividual variability in affect: Reliability, validity, and personality correlates. Journal of Personality and Social Psychology, 76(4), 662–676. 10.1037/0022-3514.76.4.662 [DOI] [Google Scholar]
  13. File D, Bothe B, File B, & Demetrovics Z. (2022). The role of impulsivity and reward deficiency in “liking” and “wanting” of potentially problematic behaviors and substance uses. Frontiers in Psychiatry, 13, 820836. 10.3389/fpsyt.2022.820836. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Fogelman N, & Canli T. (2015). ‘Purpose in Life’ as a psychosocial resource in healthy aging: an examination of cortisol baseline levels and response to the Trier Social Stress Test. NPJ Aging and Mechanisms of Disease, 1, 15006. 10.1038/npjamd.2015.6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Fredrickson BL. (2001). The role of positive emotions in positive psychology: The broaden-and-build theory of positive emotions. American Psychologist, 56(3), 218–226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Goleman D. (1995). Emotional intelligence: Why it can matter more than IQ for character, health and lifelong achievement. Bantam Books. [Google Scholar]
  17. Granfield R, & Cloud W. (1999). Social context and “natural recovery”: The role of social capital in the resolution of drug-associated problems. Substance Use & Misuse, 34(11), 1543–1570. [DOI] [PubMed] [Google Scholar]
  18. Gutierrez D, Goshorn JR, & Dorais S. (2022). An exploration of thriving over time in recovery. Journal of Substance Abuse Treatment, 132, 108612. 10.1016/j.jsat.2021.108612 [DOI] [PubMed] [Google Scholar]
  19. Hand LJ, Paterson LM, & Lingford-Hughes AR. (2024). Re-evaluating our focus in addiction: emotional dysregulation is a critical driver of relapse to drug use. Translational Psychiatry, 14(1), 467. 10.1038/s41398-024-03159-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hill PL, Sin NL, Turiano NA, Burrow AL, & Almeida DM. (2018). Sense of Purpose Moderates the Associations Between Daily Stressors and Daily Well-being. Annals of Behavioral Medicine, 52(8), 724–729. 10.1093/abm/kax039 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Huhn AS, Harris J, Cleveland HH, Lydon DM, Stankoski D, Cleveland MJ, Deneke E, & Bunce SC. (2016). Ecological momentary assessment of affect and craving in patients in treatment for prescription opioid dependence. Brain Research Bulletin, 123, 94–101. 10.1016/j.brainresbull.2016.01.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jones DR, Allen HK, Lanza ST, & Graham-Engeland JE. (2021). Daily associations between affect and alcohol use among adults: The importance of affective arousal. Addictive Behaviors, 112, 106623. 10.1016/j.addbeh.2020.106623 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kelly JF, Greene MC, & Bergman BG. (2018). Beyond abstinence: Changes in indices of quality of life with time in recovery in a nationally representative sample of U.S. adults. Alcoholism: Clinical & Experimental Research, 42(4), 770–780. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Khantzian EJ. (1997). The self-medication hypothesis of substance use disorders: A reconsideration and recent applications. Harvard Review of Psychiatry, 4(5), 231–244. [DOI] [PubMed] [Google Scholar]
  25. Kim ES, Ryff C, Hassett A, Brummett C, Yeh C, & Strecher V. (2020). Sense of Purpose in Life and Likelihood of Future Illicit Drug Use or Prescription Medication Misuse. Psychosomatic Medicine, 82(7), 715–721. 10.1097/PSY.0000000000000842 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kleykamp B, Weiss R, & Strain E. (2019). Time to Reconsider the Role of Craving in Opioid Use Disorder. JAMA Psychiatry, 76(11), 1113–1114. 10.1001/jamapsychiatry.2019.1839 [DOI] [PubMed] [Google Scholar]
  27. Knapp KS, Brick TR, Bunce SC, Deneke E, & Cleveland HH. (2021). Daily meaningfulness among patients with opioid use disorder: Examining the role of social experiences during residential treatment and links with post-treatment relapse. Addictive Behaviors, 119, 106914. 10.1016/j.addbeh.2021.106914 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Krentzman AR, Cranford JA, & Robinson EA. (2015). Long-term increases in purpose in life are associated with remission from alcohol dependence. Alcoholism Treatment Quarterly, 33(3), 252–269. [Google Scholar]
  29. Lancaster JH, Apsley HB, Brick TR, & Cleveland HH. (2025). A within-person investigation of recovery identity following substance use disorder: examining the impact of recovery-focused social contexts. Frontiers in Public Health, 13, 1534432. 10.3389/fpubh.2025.1534432 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lanza ST. (2016). Latent Class Analysis for Developmental Research. Child Development Perspectives, 10(1), 59–64. 10.1111/cdep.12163 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Larsen JT, McGraw AP, & Cacioppo JT. (2001). Can people feel happy and sad at the same time? Journal of Personality and Social Psychology. [PubMed] [Google Scholar]
  32. Marlatt GA, & Witkiewitz K. (2005). Relapse Prevention for Alcohol and Drug Problems. In Marlatt GA & Donovan DM. (Eds.), Relapse prevention: Maintenance strategies in the treatment of addictive behaviors (2nd ed., pp. 1–44). The Guilford Press. [Google Scholar]
  33. McInerney K, Best D, & Hodgson P. (2025). A Cross-Sectional Study Investigating the Role of Meaning and Purpose in Life Among Older Individuals (≥50-Years Old) in Recovery from Alcohol Use Disorder/Problem Drinking. Alcoholism Treatment Quarterly, 43(2), 230–252. 10.1080/07347324.2024.2423672 [DOI] [Google Scholar]
  34. Miao M, Zheng L, & Gan Y. (2017). Meaning in life promotes proactive coping via positive affect: A daily diary study. Journal of Happiness Studies: An Interdisciplinary Forum on Subjective Well-Being, 18(6), 1683–1696. 10.1007/s10902-016-9791-4 [DOI] [Google Scholar]
  35. Muthén B, Muthén L, 2017. Mplus. Chapman and Hall/CRC. [Google Scholar]
  36. Pettersen G, Bjerke T, Hoxmark EM, Eikeng Sterri NH, & Rosenvinge JH. (2023). From existing to living: Exploring the meaning of recovery and a sober life after a long duration of a substance use disorder. Nordisk Alkohol- & Narkotikatidskrift: NAT, 40(6), 577–589. 10.1177/14550725231170454 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Roos CR, Kirouac M, Pearson MR, Fink BC, & Witkiewitz K. (2015). Examining temptation to drink from an existential perspective: Associations among temptation, purpose in life, and drinking outcomes. Psychology of Addictive Behaviors, 29(3), 716. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Russell JA, & Barrett LF. (1999). Core affect, prototypical emotional episodes, and other things called emotion: Dissecting the elephant. Journal of Personality and Social Psychology, 76(5), 805–819. [DOI] [PubMed] [Google Scholar]
  39. Russell JA, & Carroll JM. (1999). On the bipolarity of positive and negative affect. Psychological Bulletin, 125(1), 3–30. 10.1037/0033-2909.125.1.3 [DOI] [PubMed] [Google Scholar]
  40. Schaefer SM, Morozink Boylan J, van Reekum CM, Lapate RC, Norris CJ, Ryff CD, & Davidson RJ. (2013). Purpose in life predicts better emotional recovery from negative stimuli. PloS one, 8(11), e80329. 10.1371/journal.pone.0080329 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Serre F, Fatseas M, Swendsen J, & Auriacombe M. (2015). Ecological momentary assessment in the investigation of craving and substance use in daily life: a systematic review. Drug and Alcohol Dependence, 148, 1–20. 10.1016/j.drugalcdep.2014.12.024 [DOI] [PubMed] [Google Scholar]
  42. Shiffman S, & Waters AJ. (2004). Negative Affect and Smoking Lapses: A Prospective Analysis. Journal of Consulting and Clinical Psychology, 72(2), 192–201. 10.1037/0022-006X.72.2.192 [DOI] [PubMed] [Google Scholar]
  43. Smith BW, Tooley EM, Montague EQ, Robinson AE, Cosper CJ, & Mullins PG. (2009). The role of resilience and purpose in life in habituation to heat and cold pain. The Journal of Pain, 10(5), 493–500. [DOI] [PubMed] [Google Scholar]
  44. Spedicato G. (2017). “Discrete Time Markov Chains with R.” The R Journal, 9(2), 84–104. https://journal.r-project.org/articles/RJ-2017-036/index.html. [Google Scholar]
  45. Stewart J. (2008). Review. Psychological and neural mechanisms of relapse. Philosophical Transactions of the Royal Society of London. Series B, Biological sciences, 363(1507), 3147–3158. 10.1098/rstb.2008.0084 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Stull SW, Marsch LA, & Lanza ST. (2025). Daily profiles of nondrug reward and their association with daily outcomes for people in recovery from opioid use disorder: An application of latent profile analysis applied to intensive longitudinal data. Drug and Alcohol Dependence, 271, 112646. 10.1016/j.drugalcdep.2025.112646 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sutin AR, Luchetti M, Stephan Y, Sesker AA, & Terracciano A. (2024). Purpose in life and stress: An individual-participant meta-analysis of 16 samples. Journal of Affective Disorders, 345, 378–385. 10.1016/j.jad.2023.10.149 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Tiffany ST, & Wray JM. (2012). The clinical significance of drug craving. Annals of the New York Academy of Sciences, 1248, 1–17. 10.1111/j.1749-6632.2011.06298.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Tobin J. (1958). Estimation of relationships for limited dependent variables. Econometrica: Journal of the Econometric Society, 24–36. [Google Scholar]
  50. Tovmasyan A, Monk RL, Sawicka I, & Heim D. (2022). Positive but not negative affect is associated with increased daily drinking likelihood in non-clinical populations: systematic review and meta-analyses. The American Journal of Drug and Alcohol Abuse, 48(4), 382–396. 10.1080/00952990.2022.2082300 [DOI] [PubMed] [Google Scholar]
  51. Vaccaro AG, Wu H, Iyer R, Shakthivel S, Christie NC, Damasio A, & Kaplan J. (2024). Neural patterns associated with mixed valence feelings differ in consistency and predictability throughout the brain. Cerebral Cortex, 34(4), bhae122. 10.1093/cercor/bhae122 [DOI] [PubMed] [Google Scholar]
  52. Van Doren N, Bray BC, Soto JA, & Linden-Carmichael AN. (2024). Associations between day-level affect profiles and same-day substance use among young adults. Psychology of Addictive Behaviors, 38(3), 323–333. 10.1037/adb0000979 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Van Horn ML, Smith J, Fagan AA, Jaki T, Feaster DJ, Masyn K, Hawkins JD, & Howe G. (2012). Not quite normal: Consequences of violating the assumption of normality in regression mixture models. Structural Equation Modeling: A Multidisciplinary Journal, 19(2), 227–249. 10.1080/10705511.2012.659622 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Vermunt JK. (2010). Latent class modeling with covariates: Two improved three-step approaches. Political Analysis, 18(4), 450–469. [Google Scholar]
  55. Volkow ND, Woodcock J, Compton WM, Throckmorton DC, Skolnick P, Hertz S, & Wargo EM. (2018). Medication development in opioid addiction: Meaningful clinical end points. Science Translational Medicine, 10(434), eaan2595. [DOI] [PubMed] [Google Scholar]
  56. Votaw VR, Tuchman FR, Piccirillo ML, Schwebel FJ, & Witkiewitz K. (2022). Examining Associations Between Negative Affect and Substance Use in Treatment-Seeking Samples: A Review of Studies Using Intensive Longitudinal Methods. Current Addiction Reports, 9(4), 445–472. 10.1007/s40429-022-00441-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Wagner CC, & Sanchez FP. (2002). The role of values in motivational interviewing. In Miller WR, & Rollnick S. (Eds.) Motivational interviewing: Preparing people for change (2nd ed., pp. 284–298). New York: Guilford Press. [Google Scholar]
  58. Watson D, & Clark LA. (1997). Measurement and mismeasurement of mood: recurrent and emergent issues. Journal of Personality Assessment, 68(2), 267–296. 10.1207/s15327752jpa6802_4 [DOI] [PubMed] [Google Scholar]
  59. Watson D, Clark LA, & Tellegen A. (1988). Development and validation of brief measures of positive and negative affect: the PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. 10.1037//0022-3514.54.6.1063 [DOI] [PubMed] [Google Scholar]
  60. White WL, & Kelly JF. (2011). Recovery Management: What If We Really Believed That Addiction Was a Chronic Disorder? In Kelly JF & White WL. (Eds.), Addiction Recovery Management: Theory, Research and Practice (pp. 67–84). Humana Press. 10.1007/978-1-60327-960-4_5 [DOI] [Google Scholar]
  61. Zemore SE, Ziemer KL, Gilbert PA, Karno MP, & Kaskutas LA. (2023). Understanding the shared meaning of recovery from substance use disorders: New findings from the What is Recovery? Study. Substance Abuse: Research and Treatment, 17, 1–12. doi: 10.1177/11782218231199372 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES