Dose–response effects of reported meditation practice on mental‐health and wellbeing: A prospective longitudinal study

Abstract

The popularity of meditation has surged in recent years, driven by the accessibility of digital platforms. In this context, shorter sessions have become common, often accompanied by claims of substantial benefits. The vast differences in practice intensity—from traditional monastic training and residential retreats to multi‐week Mindfulness‐Based Programs and infrequent digital home practice—raise the question of how much practice is necessary to see meaningful benefits. Our previous analysis of lifetime practice history suggested that 160 hours were required for clinically meaningful improvements in psychological distress and life satisfaction, with more needed for stable changes in affect. However, those findings could not address the effects of newly undertaken practice, the best ways to accumulate experience, or how these effects vary by practice history. This study fills these gaps by examining dose–response relationships in a diverse sample of meditators engaging in self‐directed practice in ecologically valid settings, while testing the moderating effects of practice history, personality traits, and meditation goals. One thousand fifty‐three participants provided data across nine surveys over a two‐month period of prospectively monitored, self‐directed meditation practice, followed by a 2‐ to 4‐year follow‐up. Using a longitudinal design, we examined associations between meditation practice dose and outcomes including positive and negative affect, psychological distress, and life satisfaction. Meditation practice dose was significantly associated with improvements in well‐being, affect, and distress, with practice frequency being a stronger predictor of beneficial outcomes than session duration. During the 2‐month prospective period, after controlling for prior lifetime practice, 35 to 65 minutes daily practice was required for meaningful improvements in well‐being, and 50 to 80 minutes daily was needed for meaningful improvements in mental health outcomes. Dose–response effects were moderated by lifetime practice experience across all outcomes, while negative emotionality moderated the relationship for mental health‐related outcomes. Additionally, valuing mental health as a practice goal moderated dose–response effects for mental health outcomes, and cumulative practice from baseline to follow‐up predicted increased valuation of spiritual growth as a practice goal. Our findings indicate that practitioners with varied practice histories, personality traits, and practice goals/motivations benefit from meditation on outcomes measuring mental health and well‐being, with such benefits maintained over a 2–4 year follow‐up period.

INTRODUCTION

Meditation and mindfulness have gained widespread appeal in recent decades—an estimated two‐fold increase in the United States from 2002 to 2022 (Davies et al., 2024)— due to reasonably robust evidence of health and wellbeing benefits (Galante et al., 2023; Goldberg et al., 2022). While the most established formal practices are typically undertaken in group settings (i.e., Crane et al., 2017), an increasing number of practitioners meditate on their own (Davies et al., 2024; Lake & Turner, 2017; Nagata et al., 2022). Many individuals use digital platforms, which have enabled meditation instruction on a one‐to‐many basis (Gleig, 2019; Goldberg et al., 2022; Schultchen et al., 2021). In this context, shorter sessions (approximating 10 minutes) have become standard with claims of pronounced benefits (Puddicombe, 2012). The extreme divergence of practice intensity between traditional monastic practice (Gethin, 1998), residential retreats (King et al., 2019), multi‐week Mindfulness‐Based Programs (MBPs; Crane et al., 2017), and infrequent digital home practice (Baumel et al., 2019) begs the question of how much practice is necessary for one to benefit (Bowles et al., 2023).

Critically, if one were to take up a new meditation practice, the question of how long and how often that individual would need to practice to see benefits remains unanswered. Our prior examination of lifetime practice history suggested 160 hours of practice were needed for clinically meaningful change in psychological distress and satisfaction with life (~270 hours for stable changes in affect; Bowles et al., 2023). The results also indicated that there may be differences in dose–response patterns over time, with slower rates of change at higher amounts of practice. These analyses, however, were unable to speak to newly undertaken practice (being confounded with practice history), could not provide advice regarding the best way to accumulate experience (frequency vs duration of practice), nor could they speak to differences for meditators with varying histories of practice. The present work begins to address these issues.

While there is reasonably robust evidence for MBPs (Galante et al., 2023) and preliminary evidence for online offerings (Sommers‐Spijkerman et al., 2021) and apps (Gál et al., 2021), relatively limited research has been undertaken directly comparing different doses. Such examination is critical as increasingly popular digital programs often recommend considerably less practice with claims of potentially similar benefits to their more traditional counterparts (Bowles et al., 2023). Despite an abundance of early‐stage research on intervention generation (see e.g., Dimidjian & Segal, 2015), examination of potential side effects and dosing has largely been absent within mindfulness and meditation research (Bowles et al., 2023) despite it normally being an early step in intervention development.

Dose–response effects have most commonly been investigated based on natural variation in practice amounts across participants or, in the case of meta‐regression models from MBPs, variation in home practice (recommended and self‐reported) across programs. Most (e.g. Bostock et al., 2019; Fredrickson et al., 2017), though not all (e.g. Dobkin & Zhao, 2011) studies have found evidence for a positive dose–response relationship for psychological conditions such as depression and anxiety, where more time practicing was associated with improved outcomes. Notably, in a meta‐regression of MBPs, evidence for dose–response effects for psychological conditions such as depression and anxiety was entirely lacking (Strohmaier, 2020). In the few studies that have experimentally manipulated dose, evidence for a dose–response effect for psychological outcomes is also lacking (Berghoff et al., 2017; Fincham et al., 2023; Palmer et al., 2023; Strohmaier et al., 2020). Though methodological issues, such as the brief nature of these interventions (i.e. a single session to daily sessions for two weeks) and small sample sizes, may contribute to inadequate statistical power (Bowles et al., 2023). Absence of evidence is not, however, evidence of absence, as it is sometimes taken to be (see e.g., Bowles et al., 2023).

Importantly, historical precedent and contemporary recommendations indicate the likelihood of an optimal way to maximize potential benefits and minimize risks, with dosage being an important consideration (Galante et al., 2023). At the extreme end of the dose in meditation is research on practitioners who have been practicing for several decades; those individuals have shown evidence of extraordinary outcomes, albeit causally entangled with various confounds and biases (Davidson & Kaszniak, 2015). On the other extreme, meditation naïve individuals have shown modest benefits from brief “mindfulness inductions” (Leyland et al., 2019). Best practice guidelines behoove us to undertake research that enables research‐informed recommendations of how much practice is needed over a given period of time, and how it interacts with individual factors to achieve specified goals.

The present study addresses this gap by testing how much a large sample of meditators with varied practice histories benefit from practice across a wide range of naturalistic practice doses over a two‐month prospective longitudinal period. We additionally collected data in a subsequent follow‐up period of 2–4 years to examine whether potential relationships were upheld. These analyses allowed us to examine whether a dose–response relationship exists, controlling for lifetime history of practice. It also allowed us to determine the amounts of individual practice necessary within short time windows (as often undertaken by individuals in the world) to achieve meaningful outcomes, as well as to decompose the relative contributions of session frequency and length (i.e., duration). Furthermore, we examined whether benefits differ for those with varying lifetime meditation practice histories, personality types, and motivational objectives.

METHOD

Participants

The present work represents a longitudinal extension of a prior cross‐sectional analysis (Bowles et al., 2022). Of the 1,668 participants who provided informed consent and completed the baseline survey, 1,052 participants (67.5%) opted into this study. The subsample was predominantly female (n = 781, 69%), with a mean age of 46.9 years (SD = 15.0) and 6.3 years of active practice experience (SD = 8.0), or 1,172 hours (SD = 2,350). Half the sample resided in Australia (n = 568, 50.4%) and just over one‐quarter (n = 304, 27.0%) in the USA, with most other participants living in Europe (n = 203, 19.3%). Information on the participants' racial and ethnic background was not collected.

Procedure

Longitudinal data were collected in two phases. Phase 1 consisted of 8 weeks of prospectively monitored practice. Participants signed up by completing a baseline survey (at Time 0; T0) open between April 2020 and September 2021. In Phase 2, participants who provided consent in Phase 1 to be re‐contacted were invited to complete another survey (T9) between 2 and 4 years after baseline. Potential participants were recruited via online public forums and social media channels, and emails to several meditation practice communities. Recruitment was supplemented by a Facebook advertising campaign, with an interest‐based targeting strategy using the keywords “meditation” and “mindfulness”. Surveys were hosted on Qualtrics.

During Phase 1, participants were emailed weekly surveys (i.e. T1 survey was sent 7 days post‐baseline, T2 survey was sent 14 days post‐baseline, etc.). A response from at least one survey over the monitored period was required for inclusion in the study. Participants completing all surveys through T4 and T8 were eligible for gift card draws to a total value of AU$2,500 ($100 maximum per participant). Participant numbers through each stage of the study are summarized in Figure 1.

FIGURE 1.

Response count as % of total participants completing baseline, by timepoint. Notes. SPANE = scale of positive and negative affect; K10 = Kessler scale of psychological distress; SWLS = satisfaction with life scale.

In Phase 2, those who consented in Phase 1 were recontacted. Participants completing the survey at T9 were eligible for gift card draws to a total value of AU$2,000 ($100 maximum per participant). Of the 1,052 participants who completed longitudinal measures during Phase 1, 24 did not provide consent to be recontacted, and 46 emails bounced. Of the remaining participants, 578 responded (59%; 405–41% ‐ did not respond).

Measures

Demographic data

Basic demographic data were collected at T0 including age, gender, country of residence, and presence or absence of a diagnosed mental health or neurological condition.

Lifetime meditation practice at baseline

Meditation lifetime practice history was collected at T0. Questions included (1) whether the participant has practiced meditation in the past and, if so, the total period (in months and years) of active practice; (2) whether the participant practiced meditation in the past month and if so; (3) the number of practice days per week on average, and; (4) the average practice length per day of practice; (5) how the amount in the past month compares to prior periods of active practice, with responses ranging from 0 = “A lot less” to 100 = “A lot more”; (6) aids and supports used to support practice (e.g. books, meditation app); and (vii) the relative strength of different practice goals. Additional questions not relevant to the present work can be found in Bowles et al. (2022) and on OSF (osf.io/zbqdh/).

To estimate lifetime practice hours, responses to (3) and (4) were multiplied to estimate practice hours in the prior month, which was then scaled based on the response to (5) to estimate the average amount of monthly practice during periods of active practice. This value was then multiplied by the total period (in months) of active practice from (1). The scaling factor ranged from 1.5 if practice was “A lot less” than prior periods of active practice to 0.75 if practice was “A lot more” than prior periods of active practice. A sensitivity analysis using different scaling factors (ranging from 2 to 0.5) is summarized in the supplementary information, which revealed no major difference in estimations from using other scaling factors.

Practice doses during phases 1 and 2

During Phase 1, participants completed weekly surveys asking whether they practiced meditation in the past week, and if so, the number of practice days and the average practice amount per day. These data were used to calculate a cumulative practice dose. No instructions on how to practice were provided to participants. We validated the cumulative practice dose over the monitored period for a subset of participants (n = 60) who volunteered to provide their practice log from the Insight Timer app. The correlation between our estimated cumulative practice amount for each month and data from Insight Timer (after removing n = 2 outliers whose logs showed extremely high amounts of practice) was r = .76 for month 1 and r = .73 for month 2.

For Phase 2, participants were asked whether they had practiced meditation in the past month and if so, the average number of practice days per week and the practice amount per day of practice. The calculated practice amount for the last month was then scaled based on whether it was more or less than was typical since the end of Phase 1. This value was then multiplied by the total months elapsed to estimate total practice hours between the end of Phase 1 and the Phase 2 survey.

Outcome variables

Satisfaction with Life Scale, Single Item (SWLS; Cheung & Lucas, 2014 ; Diener et al., 1985 ). Satisfaction with life was measured each week at T1 through T9 with a single question: “Over the past month (for T0 and T9) or week (for T1 through T8), how satisfied have you been with your life?”. The question has a four‐point Likert scale ranging from 1 = “very satisfied” to 4 = “very dissatisfied”. Items were reverse‐coded so that high scores indicated higher life satisfaction.

Scale of Positive and Negative Experience (SPANE; Diener et al., 2009). Positive affect (SPANE_P) and negative affect (SPANE_N) were measured monthly at T0, T4, and T8, and T9 using the SPANE, a 12‐item questionnaire assessing the frequency of positive and negative affect experienced in the past four weeks. Each item has a five‐point Likert scale ranging from 1 = “very rarely or never” to 5 “very often or always”. McDonald's omega (ω) in the present sample was 0.93 for positive affect and 0.89 for negative affect.

Kessler Psychological Distress Scale (K10; Kessler et al., 2003). Psychological distress was measured monthly at T0, T4, and T8 using the K10, a 10‐item questionnaire assessing psychological distress with questions about anxiety and depressive symptoms experienced over the prior four‐week period. Each item has a five‐point Likert scale ranging from 1 = “none of the time” to 10 = “all of the time”. McDonald's omega (ω) in the present sample was 0.93.

Goals and motivations

At T0 and T9, participants were asked how much they valued each of six practice goals and motivations on a 10‐point scale, from 1 = “Not important at all” to 10 = “Extremely important”. The goals and motivations were (i) spiritual growth (e.g., pursuing enlightenment), (ii) mental health (e.g., managing stress/anxiety), (iii) physical health (e.g., sleeping better), (iv) cognitive enhancement (e.g., improving focus, productivity), (v) i relationships (e.g., being a better spouse, friend, parent, etc.), and (vi) general well‐being (e.g., becoming more emotionally balanced and calm).

Personality

Big Five Inventory‐2 Short Form (BFI‐2‐S) (Soto & John, 2017b ). Personality traits were measured at T0 (and T8 and T9 for another project) using the BFI‐2‐S with a 30‐item questionnaire measuring five domains: open‐mindedness, conscientiousness, extraversion, agreeableness, and negative emotionality. In the present sample, internal consistency was acceptable for each domain (open‐mindedness: ω = 0.78, conscientiousness: ω = 0.84, extraversion: ω = 0.84, agreeableness: ω = 0.83, negative emotionality: ω = 0.91).

Statistical analysis

Data analyses were conducted using R version 3.6.1 in RStudio version 1.2.5001. Statistical significance was set at p < .05. To control for multiple comparisons, false discovery rate (FDR) correction (Benjamini & Hochberg, 1995) was applied at the overall model level (F‐test). FDR was not applied to individual predictors within each regression, as estimates of beta weights are calculated in relation to one another. We employed linear mixed effects modeling with maximum likelihood estimation in the “lme4” package in R. Repeated measures for dependent and independent variables (level 1) were nested within participants (level 2). Dependent variables, reported at monthly increments during Phase 1, consisted of psychological distress, positive affect, and negative affect, which were analyzed at four time‐points (T0, T4, T8, T9), and satisfaction with life, which was measured weekly during Phase 1 and analyzed at 10 time‐points (i.e. T0 through T9). The main predictor variable was practice dose for the relevant period. Given that practice dose temporally preceded the collection of outcome variable data, no practice dose value was used for T0, and an ‘NA’ value was assigned. Covariate terms for all models were (i) a continuous variable representing time (in months) from baseline to control for the different amounts of time that elapsed for different participants between Phase 1 to Phase 2, and (ii) a continuous variable representing lifetime practice hours, which was log‐transformed due to the original data being positively skewed. We also undertook moderation analyses with lifetime practice hours, each of five personality traits, and each of six practice goals and motivations. Finally, we employed a linear regression model to see whether any of the six goals and motivational objectives change as a function of practice dose, with the cumulative practice dose from T0 to T9 predicting the subjectively assessed value of different practice goals/motivations at T9, controlling for the same at T0.

Data cleaning was conducted to exclude participants who failed to provide basic demographic information (i.e., age, gender, country of residence). Duplicate responses were removed, with the more complete or (if the same) more recent response retained. Participants were excluded from the analysis if they (1) provided inadequate information on their practice history and recent practice (i.e. whether they meditated in the past month and if so, frequency and duration); (2) reported mutually incompatible responses to different questions (e.g. no meditation experience but prior participation in a meditation retreat); (3) reported using only movement or deliberate breathing practices (e.g. yoga, pranayama); or (4) failed to provide any longitudinal data (i.e. T1 through T8).

Outcome variables consisting of multiple items (i.e., affect and psychological distress) were scored per published guidelines and analyzed at the total score level. All dependent variables approximated a normal distribution and therefore required no transformation, with values of skew $\le$ 0.4 for satisfaction with life, positive affect, and negative affect, and skew = 1.12 for psychological distress. Missing data for multi‐item outcome measures were excluded from the analysis if 20% or more items were missing. When missing items totaled less than 20%, data were imputed with the mean substitution method (i.e. mean value of scores for completed items were used to impute the missing item[s]). Missing meditation practice data (i.e. session frequency and average session length for the prior week) was imputed using with the last observation carried forward method (the last item being that provided for the prior week's questionnaire). For inclusion in the relevant month's analysis, practice data for at least three out of four weeks was required (i.e. prior to imputation). For personality, average scores for each domain were calculated, and participants with more than 20% of items missing were excluded from subsequent analyses (i.e. for the relevant domain only).

Prior to the main statistical analyses, we conducted an attrition check by comparing participants who completed enough items to be included in the longitudinal analyses (“Opt‐in”) to those who did not (“Opt‐out”) on several key variables including age, practice history (lifetime practice years and hours), personality (domain level), and outcome measures (psychological distress, affect, satisfaction with life) at baseline. We calculated mean and standard deviation scores for those two groups and compared the results.

Our statistical analysis consisted of three parts. Analysis 1 tested for the presence of dose–response effects. The model included the practice dose for the relevant period as the independent variable of interest, and time elapsed from baseline (in months) and lifetime practice experience (in hours, at baseline) as covariates. Dependent variables consisted of each of the study's outcome measures. All terms were standardized. The models are represented by Equation 1:

$${\text{𝑌}}_{\text{𝑖𝑗}}={\text{𝛽}}_{00}+{\text{𝛽}}_{10}\left(\text{Time}\right)+{\text{𝛽}}_{01}\left(\text{Lifetime Practice Hours}\right)+{\text{𝛽}}_{20}\left(\text{Practice Dose}\right)+{\text{[𝑈}}_{\text{0𝑗}}+{\text{𝑈}}_{\text{1𝑗}}\left(\text{Time}\right)+{\text{𝑒}}_{\text{𝑖𝑗}}],$$ (1)

where Yij = the relevant outcome variable for participants j at time i predicted from the fixed effect for time (𝛽10, time in months, Level 1), the fixed effect for practice dose (𝛽 20, Level 1), the fixed effect for the lifetime practice hours at baseline (𝛽01, Level 2), a participant‐level random intercept (U0j), a participant‐level random slope for time (U1j), and residual error (eij).

Analysis 2 tested whether dose–response effects were moderated by lifetime practice, each of five personality traits, or each of six goals/motivational objectives, that is, whether differences in those variables affected the observed dose–response relationships. Each moderator variable was entered into a separate dose–response model as an interaction term with practice dose. As the lifetime practice variable was positively skewed (skew = 4.76), we log‐transformed it for our analysis (post‐transformation skew = −0.61). All other predictors had an acceptable skew value (<|2.0|). All terms were standardized. For the models, we modified Equation 1 to include an interaction between the relevant moderator (e.g. lifetime practice hours) and dosage, as per Equation 2,:

$${\mathrm{Y}}_{\mathrm{ij}}={\mathrm{\beta }}_{00}+{\mathrm{\beta }}_{10}\left(\text{Time}\right)+{\mathrm{\beta }}_{01}\left(\text{Moderator}\right)+{\mathrm{\beta }}_{20}\left(\text{Practice Dose}\right)+{\mathrm{\beta }}_{21}\left(\text{Moderator}\times \text{Practice Dose}\right)+\left[U_{\text{0j}}+U_{1j}\left(\text{Time}\right)+e_{\mathit{ij}}\right]$$ (2)

Analysis 3 tested whether the reported strength of a selection of practice goals changed over time between T0 to T9 based on how much meditation practice participants engaged in. For this analysis, we used a series of linear regression models, with the practice goal score at T9 predicted by the cumulative practice dose from T0 to T9 (practiceDose), controlling for the practice goal score at T0. The models are represented by Equation 3:

$$\mathrm{Y}={\mathrm{\beta }}_0+{\mathrm{\beta }}_1\left(\text{Goal}\right)+{\mathrm{\beta }}_2\left(\text{Practice Dose}\right)+e,$$ (3)

where Y = subjective rating for one of six practice goals/motivational objectives at follow‐up predicted from the intercept (β₀), the baseline rating for that goal/motivational objective, the cumulative practice dose (in hours) from baseline to follow‐up (β₂), and residual error (e).

RESULTS

Participant characteristics and attrition check

Table 1 summarizes scores for selected variables covering age, lifetime practice experience, mental health, wellbeing, and personality, reported at baseline (T0). We compare the results of this sample (see ‘Opt‐in column) to participants who completed the baseline survey but opted out of completing longitudinal measures and were thus not included in this study. Participants who opted in were slightly older, had more practice experience, better mental health and wellbeing scores, and were higher in trait consciousness than those who opted out. After implementing independent t‐tests and adjusting for multiple comparisons, nine of the 12 differences were statistically significant. These results are summarized in Table 1.

TABLE 1.

Key values at baseline for ‘Opt‐in’ and ‘Opt‐out’ participants.

Variable	Opt‐in (n = 1,052)	Opt‐out (n = 686)	Cohen's d	p
	M (S.D.)	M (S.D.)
Age	47.28 (15.02)	42.52 (15.03)	0.32	<.001*
Lifetime practice years	6.36 (8.16)	4.36 (6.59)	0.27	<.001*
Lifetime practice hours	1091.38 (2267.05)	579.18 (1584.87)	0.26	<.001*
Satisfaction with life	2.90 (0.81)	2.68 (0.82)	0.27	<.001*
Positive affect	21.49 (4.46)	20.35 (4.26)	0.26	<.001*
Negative affect	15.07 (4.34)	16.27 (4.22)	0.28	<.001*
Psychological distress	18.31 (6.25)	20.77 (6.92)	0.37	<.001*
Openness to experience	4.10 (0.70)	4.10 (0.69)	0.00	.847
Conscientiousness	3.72 (0.84)	3.36 (0.85)	0.42	<.001*
Extraversion	3.35 (0.65)	3.30 (0.62)	0.08	.118
Agreeableness	4.05 (0.68)	3.99 (0.70)	0.09	.096
Negative emotionality	2.73 (1.03)	2.96 (1.00)	0.22	<.001*

Notes. * pFDR < .05.

Participants who opted‐in for this study (n = 1,052) had a wide range of active lifetime meditation practice (0 to 50 years, median = 3.5 years), with a median estimated cumulative practice experience of 305 hours. The sample includes practitioners with a range of experience, 30% of whom might be considered beginners (<100 hours).

Practice dose values are summarized in Table 2. During Phase 1, participants practiced for an average of 5.18 days per week, and 28 minutes per day of practice, with the total practice dose calculated at 11.06 hours. Completers, that is, those who completed the survey during Phase 1, practiced an average of 9.19 hours during Phase 1 compared to 7.36 for Non‐completers. During phase 2, participants reported slightly more practice (11.07 monthly hours), although practice was in slightly less frequent (4.82 days per week) but longer (33.07 minutes) sessions.

TABLE 2.

Meditation practice dose values.

	Phase 1 M (SD)	Phase 1 M (SD)	Phase 1 M (SD)	Phase 2 M (SD)
	Completers n = 578	Non‐completers n = 474	All n = 1,052	Completers n = 578
Practice days (mean days per week)	5.42 (1.70)	4.87 (1.80)	5.18 (1.76)	4.82 (2.10)
Practice length (mins per practice day)	30.92 (19.87)	25.87 (18.25)	28.77 (19.35)	33.07 (27.66)
Practice time (hours per month)	9.19 (9.65)	7.36 (8.77)	8.41 (9.37)	11.07 (11.88)

Notes. ‘Completers’ refers to n = 578 participants who provided data in both Phase 1 and Phase 2. ‘Non‐completers’ refers to participants who provided data in Phase 1 only.

Dose–response models

Marginal R‐squared values for the four linear mixed effects models range from 0.023 for negative affect to 0.071 for psychological distress, and conditional R‐squared values range from 0.402 for satisfaction with life to 0.819 for psychological distress. Analysis 1 tested dose–response effects over the two‐month monitored period and 2–4 year follow‐up period (i.e. T0 to T9). Practice dose was a statistically significant predictor of beneficial changes for all four variables (p's < .05), with standardized beta coefficients ranging from |β| = 0.05 to 0.08. All four models controlled for lifetime practice experience, which was a statistically significant covariate term for all outcomes (p's < .001) with standardized beta coefficients ranging from|β| = 0.13 to 0.25. All four regression models were statistically significant at the level of the overall F‐test (p's < .05), and remained significant after applying FDR correction (p's < .001, FDR‐adjusted). Results are summarized in Table 3, and dose–response curves are represented in Figure 2.

TABLE 3.

Estimates of lifetime practice and prospectively monitored meditation practice.

	Satisfaction with life		Positive affect		Negative affect		Psychological distress
	β (CI)	p	β (CI)	p	β (CI)	p	β (CI)	p
Intercept	0.05 (−0.00–0.11)	.062	0.04 (−0.03–0.10)	.256	−0.09 (−0.03 ‐ ‐0.15)	.005**	−0.07 (−0.02 ‐ ‐0.13)	.013*
Time	−0.00 (−0.00–0.00)	.517	−0.00 (−0.00–0.00)	.291	0.00 (0.00–0.01)	.001**	0.00 (0.00–0.01)	.008**
Lifetime practice	0.14 (0.09–0.19)	<.001***	0.17 (0.11–0.23)	<.001***	−0.13 (−0.07 ‐ ‐0.18)	<.001***	−0.25 (−0.20 ‐ ‐0.30)	<.001***
Practice dose	0.06 (0.02–0.11)	.008**	0.05 (0.01–0.09)	.017*	−0.08 (−0.04 ‐ ‐0.12)	<.001***	−0.06 (−0.02 ‐ ‐0.10)	.002**

Notes:

^*** p < .001.

^** p < .01.

^* p < .05.

FIGURE 2.

Dose–response models for practice dose during 2 month prospective monitoring. Notes. All outcomes represented on the y‐axis are standardized. Phase 2 data (i.e., 2–4 year follow‐up) excluded to improve interpretability of x‐axis. Black regression line represents the relationship between practice dose and each outcome. Shaded area represents the 95% confidence interval. Horizontal dashed red line is a reference line, used to graphically represent whether the relationship is statistically significant (i.e. by whether or not it overlaps with the shaded area).

Given the evidence for dose–response effects for all variables, we reran models from analysis 1 with the practice dose variable unstandardized to allow for an estimation of the number of practice hours required to achieve a meaningful amount of change. With the practice dose variable being unstandardized, the resulting beta coefficient could be used to calculate the practice hours needed to achieve a meaningful amount of change in outcomes for one month. To define meaningful change, we adopted the threshold value of a SMD of 0.24 from Cuijpers et al. (2014), which they equate to a clinically relevant effect. This analysis necessitated restricting the dataset to Phase 1 only, which consists of 2 months of monitored practice. These results indicate meaningful change was achieved with 24.6 hours monthly practice (95% C.I. 15.1 to 66.0 hrs) for satisfaction with life, 41.0 hours monthly practice (95% C.I. 19.1 to 667.3 hrs) for positive affect, 17.8 hours monthly practice (95% C.I. 12.2 to 33.3 hrs) for negative affect, and 32.8 hours monthly practice (95% C.I. 18.2 to 165.8 hrs) for psychological distress. All four regression models were statistically significant at the level of the overall F‐test (p's < .01), and remained significant after applying FDR correction (p's < .001, FDR‐adjusted). Results are summarized in Table 4.

TABLE 4.

Estimates of lifetime and prospective monitored unstandardized meditation practice dose.

	Satisfaction with life		Positive affect		Negative affect		Psychological distress
	Β (CI)	p	Β (CI)	p	Β (CI)	p	β (CI)	p
Intercept	−0.31 (−0.15 ‐ ‐0.48)	<.001	−0.39 (−0.22 ‐ ‐0.55)	<.001***	0.24 (0.08‐0.40)	.004	0.49 (0.34–0.65)	<.001
Time	0.00 (−0.07–0.07)	.954	−0.01 (−0.06–0.03)	.579	0.02 (−0.03–0.07)	.412	0.03 (−0.01–0.07)	.197
Lifetime practice	0.11 (0.05–0.17)	<.001***	0.16 (0.09 – 0.23)	<.001***	−0.08 (−0.15 – 0.01)	0.018*	−0.22 (−0.29 ‐ ‐0.16)	<.001***
Practice dose	0.01 (0.00–0.02)	.002**	0.01 (0.01‐0.02)	.065	−0.01 (−0.02–0.01)	<.001***	−0.01 (−0.01‐0.00)	.015 *

Notes. Unstandardized practice dose variable in these models allows coefficients to be interpreted as temporal rates of change.

^*** p < .001,

^** p < .01,

^* p < .05.

Next, we split the practice dose variable into its two components; practice frequency per week and mean practice length (per day of practice). These two variables were entered into dose–response models as separate terms and interaction terms for each of the four study outcomes to test their relative strength on the dose–response relationship. Again, we controlled for time elapsed from baseline and lifetime practice experience. Practice frequency was a statistically significant predictor in all four models, with standardized beta coefficients ranging from|β| = 0.07 to 0.11 (p's < .05). Practice length was statistically significant for satisfaction with life (β = .05, p = .017) and negative affect (β = −.010, p = .001) but not for positive affect (β = .02) or psychological distress (β = −.03). No interaction effect between practice frequency and length was observed. All four regression models were statistically significant at the level of the overall F‐test (p's < .01), and remained significant after applying FDR correction (p's < .001, FDR‐adjusted). Results are summarized in Table 5, and dose–response curves are represented in Figure 3.

TABLE 5.

Models estimates for meditation dose separated into frequency and duration.

	Satisfaction with life		Positive affect		Negative affect		Psychological distress
	β (CI)	p	β (CI)	p	Β (CI)	p	Β (CI)	p
Intercept	0.02 (−0.04–0.08)	.292	0.00 (−0.06–0.06)	.719	−0.05 (−0.11–0.01)	.119	−0.05 (−0.11 ‐ ‐0.01)	.126
Time	0.00 (0.00–0.01)	.004**	0.00 (−0.00–0.00)	.059	0.00 (−0.00–0.00)	.996	0.00 (−0.00–0.00)	.245
Lifetime practice	0.08 (0.02–0.14)	.006**	0.13 (0.07–0.19)	<.001***	−0.06 (−0.12–0.01)	.084	−0.18 (−0.24 ‐ ‐0.12)	<.001***
Practice length	0.05 (−0.01–0.11)	.017*	0.02 (−0.03–0.08)	.115	−010 (−0.15 ‐ ‐0.04)	.001**	−0.03 (−0.09 ‐ ‐0.02)	.174
Practice frequency	0.09 (0.04–0.14)	.001**	0.07 (0.02–0.11)	.004**	−0.09 (−0.14 ‐ ‐0.04)	<.001***	−0.11 (−0.16 ‐ ‐0.07)	<.001***
Practice length X frequency	0.01 (−0.04–0.07)	.592	0.03 (−0.02–0.07)	.313	−0.01 (−0.06–0.04)	.814	0.01 (−0.04–0.05)	.706

Notes.

^*** p < .001.

^** p < .01.

^* p < .05.

FIGURE 3.

Dose–response curves with meditation dose separated into frequency and duration. Notes. All outcomes represented on the y‐axis are standardized. Black regression line represents the relationship between practice dose and the study's four outcomes. Shaded area represents the 95% confidence interval. Horizontal dashed red line is a reference line, used to graphically represent whether the relationship is statistically significant (i.e. by whether or not it overlaps with the shaded area).

As shown in Figure 3, with the exception of negative affect, change in frequency reliably demonstrated a larger impact (on average, ~2 times larger) on the outcomes than change in practice length. In other words, on average, increasing practice frequency by an extra day per week demonstrates more benefit than increasing practice length by approximately 10 minutes.

Moderation effect of lifetime practice, goals, and personality

For Analysis 2, variables representing lifetime practice experience, practice goals and motivations, and personality traits were entered into the same dose–response models (i.e. covering the 2‐month monitored period and 2–4 year follow‐up) as interaction terms to test whether benefits were different for participants differing on these measures. For all four outcomes, there was a statistically significant interaction effect between practice dose and lifetime practice experience (p's < .05) while the main effects for lifetime practice and practice dose remained statistically significant for all outcomes (p's < .001). All four regression models were statistically significant at the level of the overall F‐test (p's < .01), and remained significant after applying FDR correction (p's < .001, FDR‐adjusted). Results are summarized in Table 6 and presented in Figure 4.

TABLE 6.

Moderation effect of lifetime practice experience on practice dose.

	Satisfaction with life		Positive affect		Negative affect		Psychological distress
	β (CI)	P	β (CI)	p	β (CI)	p	β (CI)	p
Intercept	0.08 (0.02–0.14)	.008	0.08 (−0.01–0.11)	.131	−0.10 (−0.04 ‐ ‐0.16)	.002**	−.09 (−0.03 ‐ ‐0.15)	.006**
Time	−0.00 (−0.00 − 0.00)	.181	‐0.00 (−0.00–0.00)	.147	0.00 (0.00–0.01)	.001**	0.00 (0.00 – 0.01)	.005**
Lifetime practice	0.11 (0.07–0.16)	<.001***	0.15 (0.09‐0.21)	<.001***	−0.14 (−0.09 ‐ ‐0.20)	<.001***	−0.22 (−0.17 ‐ ‐0.28)	<.001***
Practice dose	0.13 (0.07–0.19)	<.001***	0.08 (0.03‐0.14)	.001**	−0.11 (−0.06 ‐ ‐0.16)	<.001***	−0.10 (−0.05 ‐ ‐0.14)	<.001***
Lifetime practice X practice dose	−0.05 (0.02–0.07)	<.001***	−0.03 (−0.01‐ ‐0.05)	.011*	0.03 (0.00 – 0.05)	.023*	0.03 (0.00 ‐ ‐0.05	.022*

Notes. CI = 95% confidence interval.

^*** p < .001,

^** p < .01,

^* p < .05.

FIGURE 4.

Lifetime and prospective meditation practice dose interactions. Notes. All outcomes represented on the y‐axis are standardized. P < .01 for satisfaction with life, p < .05 for positive affect, negative affect, and psychological distress. Phase 2 data (i.e., 2–4 year follow‐up) excluded to improve interpretability of x‐axis. Solid line represents the mean; dotted line represents one standard deviation above the mean; dashed line represents 1 standard deviation below the mean. Beta coefficients for the lines 1 S.D. above and below the mean amount of lifetime practice are summarized in Table S20.

For personality, after controlling for the effects of each of the five personality domains, the main effect of practice dose remained statistically significant for all outcomes (p's < .05). The only statistically significant interaction effect was for negative emotionality, which moderated the dose–response relationship for negative affect (p < .001) and psychological distress (p < .001) as represented in Figure 5. Therefore, people higher in negative emotionality at baseline benefited more in terms of lowering negative affect and psychological distress than those lower in negative emotionality. Full results are summarized in Table S1.

FIGURE 5.

Moderation effect of negative emotionality personality trait on practice dose. Notes. All outcomes represented on the y‐axis are standardized. P < .01 for both displayed variables. Phase 2 data (i.e., 2–4 year follow‐up) excluded to improve interpretability of x‐axis. Solid line represents the mean; dotted line represents one standard deviation above the mean; dashed line represents 1 standard deviation below the mean.

After controlling for each of the six practice goals and motivations, practice dose remained a statistically significant predictor of benefits for all outcomes except for satisfaction with life and positive affect, which was no longer statistically significant after controlling for spiritual growth (p's < .05). All four regression models were statistically significant at the level of the overall F‐test (p's < .01), and remained significant after applying FDR correction (p's < .001, FDR‐adjusted). The mental health practice goal/motivation score had a statistically significant interaction effect with practice dose for satisfaction with life (β = 0.02, p = .004), positive affect (β = 0.03,, p = .014), and negative affect (β = −0.03, p = .011), but not for psychological distress (β = −0.02, p = .097). Therefore, participants who placed a higher value on mental health as a practice goal experienced enhanced practice outcomes on measures of mental health (i.e. negative affect and psychological distress) than participants who valued mental health less. Interaction plots for the mental health goal are represented in Figure 6, with full results summarized in Table S2.

FIGURE 6.

Moderation effect of mental health goal on practice dose. Notes. All outcomes represented on the y‐axis are standardized. P < .05 for satisfaction with life, p < .01 for positive affect and negative affect, p > .05 for psychological distress. Phase 2 data (i.e., 2–4 year follow‐up) excluded to improve interpretability of x‐axis. Solid line represents the mean; dotted line represents one standard deviation above the mean; dashed line represents 1 standard deviation below the mean.

Change in practice goals/motivations

Finally, we tested whether the relative strength assigned to six practice goals/motivations changed in the 2–4 year period from baseline (T0) to Phase 2 (T9) as a function of the practice dose over that period. For this analysis, we used linear regression models, with the practice goal score at T9 was predicted based on the cumulative practice dose between T0 and T9, controlling for the equivalent practice goal score at T0 and lifetime practice history. All four regression models were statistically significant at the level of the overall F‐test (p's < .01), and remained significant after applying FDR correction (p's < .001, FDR‐adjusted). The only statistically significant change in goals/motications observed (at p < .05) was for spiritual growth (β = 0.15, p < .001); for every standard deviation increase in practice time, the subjective rating assigned to spiritual growth at follow‐up is expected to increase by approximately 0.147 standard deviations (holding baseline spiritual growth constant). Figure 7 shows the partial residuals plot illustrating the relationship between practice dose (T0 to T9) and spiritual growth, after controlling for baseline spiritual growth (T0) and lifetime practice hours.

FIGURE 7.

Spiritual growth at follow‐up (controlling for baseline spiritual growth). Notes. Standardized partial residuals plot showing the relationship between the spiritual growth practice goal at follow‐up and cumulative practice dose through phases 1 and 2, controlling for the spiritual growth at baseline.

DISCUSSION

Summary of findings relevant to the study's aims

With a sample of 1,052 meditators in a prospective longitudinal analysis (578 meditators provided data out to 2–4 years), we examined the dose–response effects of meditation practice on mental health and wellbeing outcomes, controlling for lifetime practice experience over a prospective monitored period of two months and subsequent follow‐up of several years. We found moderate evidence for a dose–response relationship, with more practice over the two‐month monitored period predicting improved outcomes. After controlling for prior lifetime practice experience, 25–41 monthly hours of practice was associated with meaningful improvement in wellbeing, as represented by positive affect and satisfaction with life (i.e. ~50 minutes to 1 hour 20 minutes daily), and 18–33 hours of practice was linked to meaningful improvement in mental health outcomes, represented by negative affect and psychological distress (i.e. ~35 minutes to 1 hour daily). These results were maintained for the 2–4 year follow‐up period. Having found a positive dose–response relationship, we examined which of the two main components of practice dose (i.e. session frequency and practice length per day) had a greater impact on outcomes. We found that across three of the four outcome variables, practice frequency had greater predictive power than mean practice length. On average, practice frequency had 2.5 times the impact on outcomes relative to practice duration. We tested the moderation of the dose–response relationship by lifetime practice experience, personality, and motivational goals/objectives. Lifetime practice experience had a significant interaction effect in the dose–response models, with a greater benefit per equivalent amount of practice among less experienced practitioners. The negative emotionality personality trait moderated the dose–response relationship for negative affect and psychological distress; those with greater negative emotionality showed greater benefit per equivalent amount of practice relative to those with lesser negative emotionality. Mental health practice goal moderated the dose–response relationship for satisfaction with life, positive affect, and negative affect; those who placed a higher value on mental health as a practice goal benefited more on those who placed a lesser value on mental health. Finally, we tested whether the relative strength of a selection of practice goals and motivations changed longitudinally as a function of how much people practiced. We found evidence that practitioners who engaged in more practice over 2–4 year period from baseline through follow‐up, on average, increased the value they ascribe to spiritual growth as a practice goal than participants who practiced less.

First, we tested for dose–response effects on four outcomes after controlling for lifetime practice experience: satisfaction with life, positive and negative affect, and psychological distress. Results indicated benefits of greater practice with moderate evidence of a dose–response effect, a result that was maintained for the follow‐up period. In our earlier work, we estimated that clinically relevant changes over a lifetime of practice were attainable with 160 to 650 hours of practice. Based on the strength of the dose–response relationship, we estimate it may take approximately 18–33 hours of self‐directed practice over a month to achieve a meaningful change (SMD = 0.24) in negative outcomes, and 25 to 41 hours of practice for positive outcomes, after controlling for the effects of prior practice experience. This equates to approximately 35 minutes to 60 minutes per day to experience meaningful change in negative outcomes, and 50 to 80 minutes per day for positive outcomes. In other words, on average, meditators need to engage in 35 to 80 minutes of practice per day, every day, for the entirety of a month, to see noticeable changes within that timeframe.

While the time commitment required to complete a MBP varies, MBSR and MBCT traditionally consist of approximately 26 hours of class sessions (Carmody & Baer, 2009) and 36 hours of recommended home practice (based on six 45‐minute sessions per week, see Parsons et al., 2017). That equates to a total of 62 hours over 8 weeks. Meta‐analysis indicates that actual recommendations and typical average engagement are less, totaling 39.2 and 25.7 hours respectively (Strohmaier, 2020). Given reasonably robust evidence for efficacy of MBPs (Goldberg et al., 2022) and the fact that average MBP engagement sits within the range of practice observed herein, there seems to be converging evidence around the dose amount indicated here (18 to 41 hours monthly) for meaningful change. These results do not account for how that experience is acquired (e.g., interaction of session frequency and duration). It is also noteworthy that meditation‐naïve participants may benefit at a greater rate than those who are more experienced (see Bowles et al., 2023), reducing the amount of practice necessary for initial benefits.

Recent reviews, however, show a trend toward modified programs with fewer hours in‐class and less home practice (Strohmaier, 2020). Despite this reduction, meta‐analysis indicates that MBPs yield standardized mean differences of 0.56 for anxiety, 0.53 for depression, 0.45 for distress, and 0.33 for well‐being (Galante et al., 2021). By comparison, our study exhibited an effect size for distress of approximately d = 0.30. While the time commitment for MBPs (13–20 monthly hours, see Strohmaier, 2020) is comparable to the practice dose required to achieve meaningful relevant changes in our study (18–41 hours), practice in MBPs is relatively less intensive than what is indicated here and spread over a longer duration of time (i.e. 7.5 weeks compared to the 4‐week periods used for our analysis). Given that many studies of MBPs have been undertaken among meditation‐naïve participants, the results are largely consistent with our observation that relatively less practice may be required early on for individuals to benefit (see e.g., Bowles et al., 2023).

In a meta‐analysis of meditation apps, similar effect sizes to the benchmark used in the present study for meaningful outcomes were reported; a SMD = 0.26 for positive affect, SMD = 0.41 for satisfaction with life, SMD = 0.21 for negative affect, and SMD = 0.10 for distress (Gál et al., 2021). These interventions had an average overall duration of just over 4 weeks and featured daily practice approximating 10 minutes (total recommended time over the intervention 5 hours, with participants engaging with approximately 43% of recommended practices (Gál et al., 2021). These effects would seem to suggest benefits with total overall practice hours approximating 2–3 hours, which neither aligns with our results nor the requirements of MBPs. The study authors note that these results should be interpreted with caution due to a high risk of bias, and we would add that concerns with attrition in app‐based mindfulness programs may undermine their effectiveness (see e.g., Baumel et al., 2019). Thus, while our own results and those of MBPs would seem to indicate something like 20 hours a month may be necessary to see benefits, studies of meditation apps seem to boldly assert comparable effects in one‐tenth of the time, which seems unlikely. In other words, it's more likely that studies of app‐based practice have yielded an inflated estimate of the true effect size (perhaps due to study bias and placebo effects, amongst other explanations) than that our study underestimated the true effect size.

Practice frequency was independently predictive of benefits when practice frequency and length were entered as separate terms into the dose–response models from Analysis 1. Practice length (per day of practice) was a statistically significant predictor for satisfaction with life and negative affect, but not for positive affect and psychological distress. Across three of the four models, practice frequency had stronger predictive power than practice length, although this was not the case for negative affect, where the effects were approximately equal (and both statistically significant). This finding supports common advice of meditation teachers who emphasize the importance of daily practice, particularly in the early stages of practice (Bowles et al., 2023). While a single study by Riordan et al. (2024) compared massed (once daily) and stacked (twice daily) practice for two weeks, where no effect was found, research that directly investigates the impact of daily versus less frequent practice regimes is lacking. In the domain of skill acquisition and memory retention, research suggests an inverted U‐shaped relationship between practice spacing and performance (Smith & Scarf, 2017). Initially, increased spacing improves performance as it allows a sufficient time interval to facilitate the consolidation of the skill or task. However, exceeding the optimal spacing interval leads to performance decline due to forgetting. The optimal time interval varies based on factors like task complexity and prior experience. There is analogous data for psychotherapy indicating weekly sessions being more effective than fortnightly (Erekson et al., 2015; Robinson et al., 2020). Insofar as mindfulness is analogous to the process of skill acquisition and/or the types of learning that occur within psychotherapy, a similar dynamic may apply. The optimal spacing for mindfulness sessions, however, may be influenced by other factors that are not relevant to skill acquisition or psychotherapy, such as motivations and goals for self‐actualization or self‐transcendence. When practice is in the service of goals and motivations that go beyond the kind of psychological processes included in this study, periods of intensive practice such as in the context of a residential retreat may offer unique benefits. Therefore, while the optimal spacing of meditation practice sessions remains unclear, our findings (along with expert recommendations) suggest regular (daily or near‐daily) practice offers benefits over less frequent regimes (i.e. when total practice amount is held constant). Controlled experiments are needed to confirm this effect and investigate the degree to which more frequent practice sessions within a day (including in a residential retreat setting) may confer additional or unique benefits.

Next, we found evidence that prior practice experience (measured as lifetime practice hours) moderated the observed dose–response relationships. Steeper dose–response curves were observed for participants with fewer lifetime practice hours, and flatter curves for those with more experience. These results are consistent with our previous finding that practice benefits accumulate non‐linearly over a lifetime of practice, with faster gains in earlier stages of practice and slower gains or a plateauing effect at later stages. To continue experiencing benefits of a similar magnitude to the early stages of practice, therefore, experienced practitioners may have to increase their practice dose. Our findings also show evidence of longitudinal changes in spiritual growth as a practice goal, but not other goals such as mental health, relationships, and general well‐being. That is, practitioners engaging in more practice during the several‐year period between baseline and follow‐up increased the value they place on spiritual growth more than those who practice less (after controlling for the baseline value). Several cross‐sectional studies have found that experienced meditators tend to value more spiritually‐aligned goals, such as self‐liberation and transformation, while relative novice practitioners place a higher value on more pragmatic goals like self‐regulation (Shapiro, 1992). Additionally, Sedlmeier and Theumer (2020) found that beginner meditators primarily practice to reduce negative aspects of life (e.g. alleviating mental health symptoms) while experienced meditators were motivated to enrich their lives (including but not limited to spiritual growth). The cross‐sectional nature of these studies, however, precludes the determination of whether spiritual motivation changes over time with more practice or whether survivor bias accounts for the result, with people placing a higher value on spiritual goals being more likely to persist (and supply data). Our longitudinal finding suggests that continued practice over time, particularly at higher doses, is associated with a shift toward more spiritual goals. Combined, the flatter dose–response curve for more experienced practitioners and the change in emphasis toward spiritual growth with more practice suggest that detecting changes for more experienced practitioners may require different measures that capture spiritual or religious dimensions of practice. Such measures may include equanimity, decentering, and compassion, all of which are considered normative within classical religious traditions such as Buddhism (Desbordes et al., 2014).

In moderation analyses with practice goals/motivations and personality traits, we found the strength of mental health as a practice goal moderated the dose–response relationship for satisfaction with life, positive affect, and negative affect, but not psychological distress. And the negative emotionality personality trait moderated the dose–response relationship for negative affect and psychological distress (p < .001), while weak interaction effects were observed for positive affect (p < .05) and satisfaction with life (p < .10). Combined, these results suggest that practitioners for whom alleviating mental health symptoms is a primary practice motivation or who are higher in trait negative emotionality may achieve greater benefits (particularly on mental health outcomes) from equivalent amounts of practice than practitioners with lower concern for mental health or who are lower in trait negative emotionality. We also found that when the spiritual growth practice goal was entered into dose–response models as a moderator, the main effects of practice dose were no longer statistically significant for the study's two wellbeing measures (i.e. satisfaction with life and positive affect). No moderation effect was observed. This suggests that if someone is practicing with a strong motivation to promote spiritual growth, the actual practice dose may be less important than the attitude this practice goal entails. It may also indicate a ceiling effect for wellbeing measures in those who place a higher priority on practicing for spiritual growth.

Strengths, limitations, and future directions

The principal strength of the study is the large sample size (n = 1,052), which permits the detection of relatively small effects that are nonetheless relevant to dose–response investigations for meditation. This has allowed for an estimation of the practice dose that is required to achieve a meaningful change in outcomes over a month, as well as an estimation of how benefits accumulate at different stages of a practitioner's meditation journey. A second strength of our study is the inclusion of participants engaging in self‐directed practice, that is, outside the context of formal programs, and therefore, in ecologically valid practice settings. This is important given the shift in recent years toward more independent practice, with many practitioners engaging in meditation practice with little more support than a meditation app or other online materials. Another strength is the inclusion of a range of measures that cover both wellbeing (life satisfaction and positive affect) and mental health (negative emotions and distress). This approach contrasts with a bias within the scientific literature of mindfulness meditation on negative states, particularly anxiety, depression, and stress, and avoids problems associated with self‐reported measures of mindfulness (Grossman, 2019; Grossman & Van Dam, 2011).

Our study also has several limitations. First, because participants were not randomly assigned to different doses, causation about the effects of larger and smaller doses of meditation cannot be inferred. While it could be that more practice caused improved outcomes, it is possible that people who felt better (and thus reported better outcomes) practice more than those who felt worse. Though given the variation in the dose–response curves by lifetime practice hours (see Figure 2), we have some confidence that at least part of the reported benefits may have been coincident with differences in practice doses. Future studies where the dose is experimentally manipulated and randomly assigned to participants are needed to allow for more confident causal inferences about different doses. Second, our attrition check revealed that participants that opted‐out of the longitudinal analysis did not opt‐out at random, with statistically significant differences found based for age, lifetime practice experience, and baseline values for satisfaction with life, psychological distress, and positive/negative affect. Such differences may have resulted in some bias in the participants included in the longitudinal analysis. Despite differences being statistically significant, the magnitude of differences was relatively small, with values for participants who opted out versus those that opted in being less than 0.5 standard deviations. Therefore, while we believe the scale of bias is relatively small, future studies should encourage and incentivize participants to continue responding to questionnaires regardless of whether they are still actively practicing meditation. Third, meditation practice dose was self‐reported, and therefore relied on a participant's accurate recall of their prior week's practice. Furthermore, session length data were collected in bands (0–15 min, 15–30 min, 30–60 min, more than 60 mins) with the midpoint used to calculate total practice dose for the relevant week. Estimates of weekly practice dose were therefore subject to several sources of potential error and bias. Although we validated practice dose for a selection of participants with a log of their practice data from the Insight Timer app and found a relatively high correlation (r = .72), future studies would benefit from the direct collection of participant meditation time without relying on self‐reports, for instance, via a mobile application. Fourth, although our sample is international, it is unlikely to be representative of meditators globally given the heavy weighted toward English‐speaking countries (93% of participants reside in Australia, the United States, United Kingdom, European Union, and Canada). The sample also featured a high proportion of secular (i.e. non‐religious) practitioners, which may not accurately represent meditators globally (Bowles et al., 2022). We do, however, have some confidence that our sample is broadly representative of engaged meditation practitioners within the countries where our recruitment was targeted, as specified in Bowles et al. (2022). Future research of meditation practitioners in countries and practice traditions that are not well represented in our sample is necessary to confirm the generalizability of our findings. Fifth, for the calculation of monthly practice required for meaningful change, some estimated (particularly positive affect) have wide confidence intervals, and as such, estimates should be treated with some caution. And sixth, the measures we used in this study may not capture the full range of benefits of meditation practice, particularly for more experienced practitioners, for whom spiritual rather than psychological goals may be more important (Bowles et al., 2022; Sedlmeier & Theumer, 2020). Furthermore, while our results suggest a plateauing effect may occur in more experienced practitioners, floor and ceiling effects for measures of affect and psychological distress may also be evident. Future studies may benefit from the use of measures that may be more relevant and sensitive to change for experienced practitioners, such as equanimity and insight (Desbordes et al., 2014; Eberth et al., 2019).

CONCLUSION

The present study contributes to the literature on the dose–response effects of meditation by estimating the strength of dose–response effects and the required dose to achieve a meaningful change in measured outcomes over a defined period of self‐directed practice. The results indicate that there are likely benefits that are achievable from larger practice doses, and the frequency of practice may be more important than longer sessions, particularly in the early stages of practice.

AUTHOR CONTRIBUTIONS

NIB wrote the first draft of the manuscript, led the data analysis, and finalized the manuscript. NVD helped with the design and execution of the study and revisions to the manuscript.

CONFLICT OF INTEREST STATEMENT

We have no known conflicts of interest to disclose.

ETHICS STATEMENT

Approval has been obtained from the Human Ethics Team at the University of Melbourne's Office of Research Ethics and Integrity under number 13834.

Supporting information

Table S1. Base dose–response model of lifetime practice experience only.

Table S2. Moderation effect of openness personality trait on practice dose.

Table S3. Moderation effect of conscientiousness personality trait on practice dose.

Table S4. Moderation effect of extraversion personality trait on practice dose.

Table S5. Moderation effect of agreeableness personality trait on practice dose.

Table S6. Moderation effect of negative emotionality personality trait on practice dose.

Table S7. Moderation effect of mental health goal on practice dose.

Table S8. Moderation effect of physical health goal on practice dose.

Table S9. Moderation effect of relationships goal on practice dose.

Table S10. Moderation effect of performance enhancement goal on practice dose.

Table S11. Moderation effect of wellbeing goal on practice dose.

Table S12. Moderation effect of spiritual growth goal on practice dose.

Table S13. Alternative scaling factors for lifetime practice experience calculation.

Table S14. Lifetime practice hours for different scaling scenarios.

Table S15. Dose–response models with different scaling scenario (i).

Table S16. Dose–response models with different scaling scenario (ii).

Table S17. Dose–response models with different scaling scenario (iii).

Table S18. Dose–response models with different scaling scenario (iv).

Table S19. Dose–response models with different scaling scenario (v).

Table S20. Practice dose regression lines for different amounts of lifetime practices.

Bowles, N. I. , & Van Dam, N. T. (2025). Dose–response effects of reported meditation practice on mental‐health and wellbeing: A prospective longitudinal study. Applied Psychology: Health and Well‐Being, 17(4), e70063. 10.1111/aphw.70063

DATA AVAILABILITY STATEMENT

Data are available at the Open Science Framework (https://osf.io/bawkp/).