Abstract
Theory and prior research suggests that decentering— an objective, distanced perspective on one’s internal experiences— may vary based upon characteristics such as age, gender, race/ethnicity, and meditation experience. However, little is known about whether decentering measures are comparable in their meaning and interpretation when administered to individuals with different group membership (e.g., men or women; younger or older adults, etc.). The current study examined the measurement invariance of the Experiences Questionnaire (Fresco et al., 2007), a commonly-used measure of decentering, evaluating age, gender, race/ethnicity, and meditation experience in three samples (students, community members, and clinical participants). Each sample was tested separately to assess the generalizability of results. The Experiences Questionnaire demonstrated full or partial measurement invariance in all cases, suggesting that scores are not biased based upon group membership and may be compared across individuals who vary in age, race/ethnicity, gender, and meditation experience. The current study also examined mean differences in decentering by groups, finding some evidence that decentering scores are higher for men, racial/ethnic minorities, older adults, and individuals with more meditation experiences. Implications are discussed for assessing decentering in diverse samples.
Keywords: Experiences Questionnaire, decentering, measurement invariance, meditation
Interest in mindfulness and related constructs has seen a dramatic increase in recent years, leading to a critical need for measures that are refined and psychometrically-sound (Van Dam et al., 2018). Mindfulness is a broad construct with numerous distinguishable components (e.g., Van Dam et al., 2018). The current study focuses on decentering1, which is often defined as the capacity to take an objective, third-person perspective on one’s ongoing thoughts and feelings (e.g., Bernstein et al., 2015; Naragon-Gainey & DeMarree, 2017). Decentering is particularly important for mental health research and practice, as improved decentering is a key mechanistic target in numerous therapies (Bernstein et al., 2015) that is associated with altered neural connectivity (e.g., in the salience, default mode, and frontal parietal control networks; King & Fresco, 2019). The current study seeks to contribute to the assessment of decentering by evaluating the measurement invariance of the Experiences Questionnaire (EQ; Fresco et al., 2007), which is the most frequently-used measure of decentering in research and clinical contexts (currently cited between 200 and 500 times). It is important to note that the EQ reflects trait decentering, which is distinct from an assessment of decentering as a dynamic, temporal process, and thus the EQ is primarily used to measure individual differences and/or large-scale changes in decentering over time. In the current study, we aim to determine whether the EQ differs in its meaning and interpretation across demographic variables and meditation experience.
Theory and empirical research suggest possible group differences in decentering for several characteristics and experiences. First, decentering and mindfulness-related constructs may change over the lifespan. Older adults are relatively more motivated than younger adults to focus on their present-moment experiences and wellbeing—consistent with a mindful perspective— due to a more limited time perspective later in life (e.g., Carstensen, 2006). Accordingly, numerous studies found that decentering and mindfulness are positively associated with age in adulthood (e.g., Prakash, Hussain, & Schirda, 2015; Shook et al., 2017). Few studies have addressed how decentering and related constructs may differ by gender or race/ethnicity. However, cultural differences associated with race and ethnicity—including distinct assumptions about and familiarity with the Buddhist roots of mindfulness—may impact one’s understanding of and receptivity to mindfulness (DeLuca, Kelman, & Waelde, 2018). There is also some evidence that women benefit more from mindfulness interventions than men, perhaps because of differential engagement (Katz & Toner, 2012; Laurent et al., 2013). Last, meditation experience is an important variable to consider when assessing decentering. Consistent with the close conceptual link between mindfulness and meditation, individuals with more meditation experience tend to report higher levels of decentering and related constructs, relative to those with little or no meditation experience (e.g., Franquesa et al., 2017; Isbel & Mahar, 2015).
Critically, interpretations of or responses to decentering items could differ systematically across members of these groups, which would lead to erroneous conclusions regarding observed scale scores. For example, there is initial support for systematic group differences when responding to mindfulness-related scales, as the interpretation of some mindfulness items appears to be dependent upon one’s meditation experience (e.g., Goh, Marais, & Ireland, 2015; Gu et al., 2016), and a lack of measurement invariance is also plausible for other groups.
Formal tests of measurement invariance are necessary to evaluate whether a measure in fact functions similarly across groups. If measurement invariance is supported in a given sample, this indicates that the meaning of the score is consistent across groups, and so score comparisons and interpretations may be valid. Like any other psychometric property, measurement invariance is not an immutable feature of a measure, but its support in some samples increases confidence that this same property may hold in other samples. Importantly, scales that demonstrate measurement invariance are needed if the benefits and applications of decentering research are to be distributed equally to different groups and tailored to their needs (e.g., Waldron et al., 2018).
Thus far, the sole analysis of measurement invariance in the decentering literature was conducted on the decentering subscale of the Toronto Mindfulness Scale. Analyses supported partial or full measurement invariance across meditation experience groups, and participants with more meditation experience had higher latent mean levels of decentering (Ireland, Day, & Clough, 2018). The primary aim of the current study is to extend this line of research by evaluating the measurement invariance of the EQ across gender, age, race/ethnicity, and meditation experience. A secondary aim was to test mean latent group differences in decentering, if partial or full measurement invariance was established for that grouping variable. Based on prior research, we hypothesized that older adults (relative to younger adults) and individuals with more meditation experience (relative to those with less) will have higher levels of decentering. There was not an adequate empirical basis to form specific hypotheses for differences based upon gender or race. We conducted invariance analyses for the above characteristics separately in college student, community, and clinical samples (each of which was drawn from different populations with different study inclusion criteria) to assess finding generalizability.
Method
Participants and Procedure
The current study employed secondary data analyses from several larger studies to examine measurement invariance of the EQ, though none of the analyses or aims of this study overlap with prior publications. The student and community samples were drawn from studies conducted at Kent State University and the University at Buffalo, whereas the two clinical sample datasets were collected at the University at Buffalo, only. All studies were approved by institutional review boards at Kent State or the University at Buffalo. As a preliminary analysis, partial or full measurement invariance was demonstrated across recruitment sites within the student sample and the community sample, and across studies for the clinical sample (see Appendix 1 of the online supplement for details). This establishes that datasets from different recruitment sources could be combined in the subsequent primary measurement invariance analyses for each sample. Table 1 shows demographic characteristics of participants, broken down by site and sample. Participants completed questionnaires in a randomized order on Qualtrics. The specific questionnaires differed across datasets, and only the EQ was analyzed in the current study.
Table 1.
Participant Descriptive Statistics
| Student | Community | Clinical | |||
|---|---|---|---|---|---|
| UB | Kent State | UB | Kent State | UB | |
| N | 2181 | 641 | 437 | 229 | 365 |
| Age M (SD) | 19.2(1.8) | 19.8 (2.6) | 46.6 (12.5) | 33.9 (10.8) | 33.1 (12.6) |
| Race n (%) | |||||
| White | 1011 (46.1%) | 527 (82.2%) | 273 (62.5%) | 171 (74.7%) | 281 (75.9%) |
| Asian | 776 (35.4%) | 19 (2.9%) | 59 (13.5%) | 10 (4.4%) | 19 (5.1%) |
| Black | 165 (7.5%) | 46 (7.2%) | 59 (13.5%) | 28 (12.2%) | 31(8.4%) |
| Hispanic | 130 (5.9%) | 10 (1.6%) | 10 (2.3%) | 10 (4.4%) | 16 (4.3%) |
| Other | 97 (4.4%) | 39 (6.1%) | 36 (8.2%) | 10 (4.4%) | 18 (4.9%) |
| Gender n (%) | |||||
| Male | 1077 (49.1%) | 105 (16.4%) | 157 (35.9%) | 109 (47.6%) | 82 (22.2%) |
| Female | 1100 (50.2%) | 535 (83.6%) | 277 (63.4%) | 120 (52.4%) | 280 (75.7%) |
| Other | 4 (0.2%) | N/A | 2 (0.5%) | N/A | 3 (0.8%) |
| Meditation Experience n (%) | |||||
| ≤ 6 months | 1892 (86.7%) | N/A | 252 (57.7%) | N/A | 201 (54.3%) |
| > 6 months | 287 (13.2%) | N/A | 185 (42.3%) | N/A | 164 (44.3%) |
| Receiving Treatment n (%) | |||||
| Therapy | N/A | N/A | N/A | N/A | 258 (70.7%) |
| Medication | N/A | N/A | N/A | N/A | 268 (73.4%) |
Note. UB = University at Buffalo; Kent State = Kent State University.
Student sample.
Data were obtained from 2822 college students (641 at Kent State and 2181 at the University at Buffalo) over the age of 18, using an online, secure research pool that compensated participation with course credits. Participants completed the questionnaire battery online remotely, or in small groups in the laboratory.
Community sample.
Data were obtained from a total of 666 adult participants. 229 people provided informed consent to participate from Amazon’s Mechanical Turk platform (MTurk). These participants received compensation of $1.00 for a 30 minute survey, which is consistent with similar studies on MTurk. An additional 437 participants were recruited through ResearchMatch, a national health volunteer registry that was created by several academic institutions and supported by the U.S. National Institutes of Health. ResearchMatch participants were entered into a lottery to win either a new iPad or one of 10 Amazon giftcards worth $50 each. All participants completed the questionnaire battery online from their homes.
Clinical sample.
Data were obtained from 365 adult participants in total. One sample (n = 211) was recruited from ResearchMatch, and they were invited to participate only if they reported, upon first joining ResearchMatch, that they had been diagnosed with an anxiety disorder and have continued to experience anxiety symptoms. These participants received a $10 Amazon giftcard as compensation for their time and effort. The second sample (n = 159) completed a larger study at the University at Buffalo and were eligible to participate if they were currently seeking or receiving treatment for a psychological concern. They received $40 for their participation in the 3–4 hour laboratory baseline assessment where the EQ was collected.
Measure.
Decentering was assessed with the 11-item EQ Decentering scale (Fresco et al., 2007). Participants select a response on a 5-point Likert scale ranging from 1 (Never) to 5 (All the Time) that best describes their general experiences for each item. The EQ has demonstrated good internal consistency (alpha = .83) and showed expected patterns of convergent and discriminant associations with related constructs (Fresco et al., 2007). For further detail about research using the EQ, see Naragon-Gainey, McMahon, Park, and Fresco (2019).
Data Analysis
Confirmatory factor analysis in Mplus 8.0 was used to examine several multigroup models (i.e., race/ethnicity, gender, age, and meditation experience) separately in each sample type. As a preliminary analysis to establish the basic factor structure, a single factor model was specified in each sample, with the inclusion of several theoretically-grounded error covariances as outlined in the initial development of the EQ (Fresco et al., 2007). Robust maximum likelihood estimators (MLR) were used with continuous indicators to adjust for missing data without biasing parameter estimates, as well as to account for non-normality. Model fit was evaluated by examining the chi-square test statistic, comparative fit index (CFI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). According to recommendations for interpreting approximate fit indices proposed by Hu and Bentler (1999), CFI > .90 indicates acceptable fit and > .95 indicates excellent fit. RMSEA < .10 suggests acceptable fit and < .06 indicates good fit, and SRMR < .08 indicates acceptable fit.
We evaluated a series of increasingly stringent nested model comparisons to test several aspects of invariance (i.e., configural, metric, scalar). The configural model tests whether the same factor structure can be imposed across groups (with parameters freely estimated), and is evaluated based on model fit. The metric invariance model constrains factor loadings to be equal across groups, and invariance is supported when this model does not provide a worse fit to the data than the configural model. Conceptually, metric invariance indicates that the relationships between the items and the underlying latent variable are comparable across groups. The scalar invariance model tests whether there are group differences in item intercepts and is examined by comparing model fit to the metric model. Scalar invariance indicates that a response to a given item corresponds to the same latent score across groups (i.e., the response scales are equivalent).
The chi-square difference test-statistic is overly sensitive to large sample sizes and may erroneously suggest non-invariance due to small changes in model fit (Chen, 2007). Accordingly, we present chi-square difference tests but rely upon the following cutoffs from Chen (2007) when evaluating non-invariance: a change ≥ −.005 in CFI, in addition to a change ≥ .010 in RMSEA or a change ≥ .025 in SRMR indicates metric non-invariance, and a change ≥ −.005 in CFI, in addition to a change of ≥ .010 in RMSEA or a change of ≥ .005 in SRMR suggests scalar non-invariance. In cases where change in approximate fit indices indicated a lack of full measurement invariance, we tested partial invariance by iteratively freeing parameters that accounted for the largest source of misfit. Once partial or full measurement invariance was obtained, establishing that scores are comparable across groups, we evaluated latent mean differences in each sample using Cohen’s d as an estimate of effect size.
Results
Preliminary Analyses
Online supplement Table A shows the text of each of the EQ items, as well as mean and total scores by sample type. The mean EQ score for the clinical sample (32.8) was significantly lower (p < .001) than mean scores in the student (36.8) and community (37.6) samples, which did not differ. In addition, online supplement Table B presents associations of the EQ with self-reported and interview-assessed internalizing symptoms in the clinical sample.
Prior to examining measurement invariance, we first tested whether the hypothesized single-factor structure of the EQ was a good fit in each sample. All EQ items were specified to load onto a single factor. To be consistent with the measure’s development and prior testing (Fresco et al., 2007), we also specified a priori error covariances for items 1 and 4, items 3 and 5, and items 6 and 9 (see Table A for item text). However, the error covariance for items 3 and 5 was non-significant in the clinical and community samples, and had a small effect size in the student sample (r = .11, p < .001). Therefore, we removed the error covariance for items 3 and 5. This single factor model fit the data well in each sample (see Table 2).
Table 2.
Measurement Invariance Results
| Chi-square (df) | CFI | RMSEA | SRMR | ΔChi-square (df) | ΔCFI | ΔRMSEA | ΔSRMR | Invariant? | |
|---|---|---|---|---|---|---|---|---|---|
| Single factor model for each sample (no test of invariance) | |||||||||
| Student | 300.04*** (42) | .963 | .047 | .031 | |||||
| Community | 89.04*** (42) | .974 | .042 | .033 | |||||
| Clinical | 88.77*** (42) | .952 | .055 | .046 | |||||
| Race | |||||||||
| Student (ns = 1538 White/Non-Hispanic, 1276 Minority) | |||||||||
| Configural | 338.04*** (84) | .963 | .046 | .032 | Yes | ||||
| Metric | 356.48*** (94) | .962 | .045 | .036 | 12.40(10) | −.001 | −.001 | +.004 | Yes |
| Scalar | 395.43*** (104) | .958 | .045 | .040 | 39.22*** (10) | −.004 | .000 | +.004 | Yes |
| Community (ns = 430 White/Non-Hispanic, 213 Minority) | |||||||||
| Configural | 131.90*** (84) | .974 | .042 | .037 | Yes | ||||
| Metric | 152.69*** (94) | .968 | .044 | .058 | 21.80*(10) | −.006 | +.002 | +.021 | Yes |
| Scalar | 172.14*** (104) | .963 | .045 | .058 | 20.51* (10) | −.005 | +.001 | .000 | Yes |
| Clinical (ns = 281 White/Non-Hispanic, 83 Minority) | |||||||||
| Configural | 139.81*** (84) | .946 | .060 | .052 | Yes | ||||
| Metric | 143.89*** (94) | .952 | .054 | .056 | 3.45 (10) | +.006 | −.006 | +.004 | Yes |
| Scalar | 156.52*** (104) | .950 | .053 | .058 | 12.34(10) | −.002 | −.001 | +.002 | Yes |
| Gender | |||||||||
| Student (ns = 1180 male, 1633 female) | |||||||||
| Configural | 339.66*** (84) | .962 | .047 | .033 | Yes | ||||
| Metric | 358.09*** (94) | .961 | .045 | .036 | 12.21 (10) | −.001 | −.002 | +.003 | Yes |
| Scalar | 460.29*** (104) | .947 | .049 | .043 | 118.71*** (10) | −.014 | +.004 | +.007 | No |
| Partial Scalar | 409.98*** (103) | .954 | .046 | .039 | 56.45** (9) | −.007 | +.001 | +.003 | Yes (P) |
| Community (ns = 261 male, 379 female) | |||||||||
| Configural | 165.27*** (84) | .957 | .055 | .041 | Yes | ||||
| Metric | 180.67*** (94) | .954 | .054 | .057 | 14.21 (10) | −.003 | −.001 | +.016 | Yes |
| Scalar | 187.39*** (104) | .956 | .050 | .058 | 2.83 (10) | +.002 | −.004 | +.001 | Yes |
| Clinical (ns = 82 male, 279 female) | |||||||||
| Configural | 153.44*** (84) | .934 | .068 | .054 | Yes | ||||
| Metric | 160.93*** (94) | .936 | .063 | .064 | 7.70(10) | +.002 | −.005 | +.010 | Yes |
| Scalar | 182.37*** (104) | .925 | .065 | .069 | 22.10* (10) | −.001 | +.002 | +.005 | No |
| Partial Scalar | 170.48*** (103) | .936 | .060 | .067 | 8.68 (9) | .000 | −.003 | +.003 | Yes (P) |
| Age | |||||||||
| Community (ns =167 young adults, 305 middle aged, 167 older adults) | |||||||||
| Configural | 198.99*** (126) | .962 | .052 | .045 | Yes | ||||
| Metric | 220.07*** (146) | .961 | .049 | .063 | 19.47 (20) | −.001 | −.003 | +.018 | Yes |
| Scalar | 262.39*** (166) | .949 | .052 | .070 | 45.61*** (20) | −.012 | +.003 | +.007 | No |
| Partial Scalar | 242.45*** (160) | .957 | .049 | .069 | 22.73 (14) | −.004 | .000 | +.006 | Yes (P) |
| Clinical (ns =202 young adults, 161 middle aged/older adults) | |||||||||
| Configural | 133.42*** (84) | .951 | .057 | .055 | Yes | ||||
| Metric | 142.90*** (94) | .952 | .054 | .064 | 9.05 (10) | +.001 | −.003 | +.009 | Yes |
| Scalar | 160.12*** (104) | .945 | .055 | .064 | 17.52(10) | −.007 | +.001 | .000 | Yes |
| Meditation Experience | |||||||||
| Student (ns =1888 with ≤ 6 months, 287 with ≥ 6 months) | |||||||||
| Configural | 326.97*** (84) | .956 | .052 | .035 | Yes | ||||
| Metric | 339.81*** (94) | .955 | .049 | .037 | 5.04(10) | −.001 | −.003 | +.002 | Yes |
| Scalar | 358.93*** (104) | .953 | .047 | .039 | 15.04(10) | −.002 | −.002 | +.002 | Yes |
| Community (ns =238 with ≤ 6 months, 176 with > 6 months) | |||||||||
| Configural | 148.87*** (84) | .956 | .061 | .045 | Yes | ||||
| Metric | 160.39*** (94) | .955 | .058 | .060 | 11.17(10) | −.001 | −.003 | +.015 | Yes |
| Scalar | 183.32*** (104) | .947 | .061 | .063 | 24.18** (10) | −.008 | +.003 | +.003 | Yes |
| Clinical (ns =200 with ≤ 6 months, 164 with > 6 months) | |||||||||
| Configural | 142.78*** (84) | .941 | .062 | .054 | Yes | ||||
| Metric | 146.01*** (94) | .947 | .055 | .059 | 3.21 (10) | +.006 | −.007 | +.005 | Yes |
| Scalar | 153.01** (104) | .950 | .051 | .063 | 6.00 (10) | +.003 | −.004 | +.004 | Yes |
Note.
p < .05;
p < .01;
p < .001. (P) = partial invariance; otherwise, “Yes” indicates that full invariance was established.
Measurement Invariance
Within each sample type, EQ scores by measurement invariance groups are shown in Table C of the online supplement. Fit statistics for all invariance tests are displayed in Table 2.
Race/Ethnicity.
For models evaluating measurement invariance based on racial and ethnic identification (i.e., White/Non-Hispanic or Minority), the configural model was a good fit to the data in each sample type. We next compared the fit indices of the metric model (constraining all factor loadings to be equal across groups) to the configural model. The changes in approximate fit indices revealed no meaningful decrement in fit, based upon the criteria of Chen (2007). Scalar invariance was tested by constraining factor loadings and indicator intercepts to be equal across groups. Comparing the fit of these models to the metric models supported scalar invariance. Thus, full measurement invariance was found based on identification as white or as a racial/ethnic minority group member. In comparing latent means across groups, participants who identified as a racial/ethnic minority had higher decentering than whites in the student sample (Cohen’s d = .14, p < .001), but there was not a significant difference in latent means in the clinical (d = .11, p = .420) or community samples (d = .15, p = .083).2
Gender.
In evaluating measurement invariance based on gender (i.e., male or female), the configural model was a good fit to the data in each sample. Metric invariance was supported in all three samples, and full scalar invariance was supported in the community sample. However, there was a substantial decrement in fit for the scalar model compared to the metric model in the student (ΔCFI = −.014, ΔRMSEA = .004, ΔSRMR = .007) and clinical samples (ΔCFI = −.011, ΔRMSEA = .002, ΔSRMR = .005). Partial scalar invariance was achieved by freely estimating the intercept of item 3 in the student sample, and the intercept of item 1 in the clinical sample. Latent means indicated that females had lower levels of decentering in the student (d = −.42, p < .001) and community samples (d = −.18, p = .045), but there was not a significant gender difference in the latent means for the clinical sample (d = −.22, p = .151).
Age.
We evaluated measurement invariance of the EQ across age groups in the community and clinical samples; there was insufficient variability in age in the college students (see Table 1). We created age groups with roughly similar ns (see Table 2) to represent young adults (18–30 years old), middle aged (31–54), and older adults (55–70) in the community sample. For the clinical sample, we created two age groups— young adults (18–30) and middle/older adults (31–79)— because the middle- and older- adults group sizes were too small to separate. The configural model was a good fit to the data in both the community and clinical samples, and metric invariance was supported in both samples. There was evidence of full scalar invariance in the clinical sample, whereas in the community sample there were differences in indicator intercepts across groups (ΔCFI = −.012, ΔRMSEA = .003, ΔSRMR = .007). After freely estimating indicator intercepts for items 3, 4, and 11, partial scalar invariance was supported in the community sample. There was no difference in latent means between younger and older adults in the clinical sample (d = .038, p = .732). However, there was a significant difference in latent means in the community sample, where both middle (d = .22, p = .039) and older adults (d = .58, p < .001) had higher levels of decentering than young adults, and older adults had higher levels of decentering than middle aged adults (d = .34, p = .003).
Meditation Experience.
Last, we evaluated measurement invariance of the EQ based on experience with meditation/mindfulness practice, where we compared those who reported 6 months or less experience with meditation to those with more than 6 months of meditation experience. Note that meditation experience was only assessed in the studies collected at the University at Buffalo, so ns are smaller (see Table 2). The configural models were a good fit to the data in all three samples, and metric and scalar invariance were fully supported. Based on latent means, individuals with 6 months or less mediation experience had lower decentering in the clinical (d = −.53, p < .001) and student (d = −.26, p = .001) samples. However, meditation experience was not associated with decentering in the community sample (d = −.13, p = .188).
Discussion
We examined the measurement invariance of the Experiences Questionnaire in multiple sample types across four different grouping variables: race/ethnicity, gender, age, and meditation experience. Overall, the EQ appears to assess decentering similarly across groups that differ in these variables, in one clinical and two non-clinical samples. It is notable that full metric invariance was demonstrated across all grouping variables and samples, suggesting that the EQ items are similarly related to latent decentering across people with a variety of characteristics; that is, the meaning of the items is comparable. Full scalar invariance was also met for most analyses, whereas partial scalar invariance was established in the three remaining cases (e.g., for gender in students and clinical participants, and for age in community participants) by allowing one to three item intercepts to differ across groups. Thus, it appears that there is some systematic variability in gender and age with regard to how the response scale for certain EQ items relates to values of latent decentering, but these differences were not widespread and were not consistent across samples. Since the majority of items in the scale demonstrate measurement invariance, the full scale score may be validly used and compared across groups. Taken together, these analyses provide support for the EQ, wherein EQ decentering scores demonstrated a similar meaning, interpretation, and level regardless of the respondent’s gender, race/ethnicity, age, or meditation experience. Given the need for studying and applying mindfulness-related constructs in more diverse samples (e.g., Waldron et al., 2018), the EQ is a good candidate for providing comparable, unbiased measurement across the above individual differences.
After establishing that EQ scores were comparable across groups, we compared each group’s standing on the latent decentering factor; this method is superior to comparing observed scale scores because the factor adjusts for item-level error variance. Results were largely consistent with hypotheses, as individuals with more meditation experience had higher latent levels of decentering than individuals with less than six months experience (or none) (see also Ireland et al., 2018). Although the difference was not statistically significant in the community sample, the effect size was consistent with the other samples in direction. As expected, older adults had higher levels of decentering than younger adults in the community sample, with a positive linear association across the three age groups. However, there was not an age difference in decentering in the clinical sample; it is possible the large proportion of young adults in this sample or heightened symptoms in this relatively small sample size overall may have obscured the effect. In more exploratory analyses, females consistently had lower levels of decentering than males, as indicated by small effect sizes (though the difference was not significant in the clinical sample). This is consistent with the tendency of women to ruminate or perseverate on negative thoughts to a greater degree than men (Johnson & Whisman, 2013), as negative perseverative thought is incompatible with a decentered stance. Last, there was a very small effect across samples suggesting that racial or ethnic minority individuals may score higher in decentering than white individuals, though it only reached significance in the large student sample. This marginal effect may be due to greater familiarity of some East Asian individuals with mindfulness, though it is difficult to draw strong conclusions without more fine-grained groupings by race and ethnicity.
A number of limitations should be considered when interpreting these results. The size of the clinical sample was relatively small to conduct measurement invariance analyses, and therefore these analyses may not have had adequate power to detect minor non-invariance in this sample. In addition, our assessment of meditation experience was limited to one self-reported item asking about years of meditation experience, which does not allow for a nuanced characterization of meditation practice and skill. Furthermore, the samples were not selected based upon meditation experience, so there were likely few expert meditators and the mean proficiency of the meditator group was probably not high. We examined one specific measure of decentering and several specific characteristics likely to impact decentering, but these findings should not be generalized to other measures or grouping variables, and results may differ in other samples. In addition, sample sizes were not large enough to examine fine-grained groupings, which is particularly relevant to the race/ethnicity analyses. Finally, trait self-report measures have numerous weaknesses, and other dynamic and non-self-report methods of assessing decentering (e.g., measurement in daily life, state measurement in the lab, behavioral assessment; see Bernstein, Hadash, & Fresco, 2019) should be considered or used in conjunction with self-report when possible. These limitations notwithstanding, the current study bolsters confidence in the EQ as a widely-applicable measure of decentering, and provide an initial characterization of how decentering levels differ across demographic features and meditation experience.
Supplementary Material
Public Significance Statement:
This study found that one of the primary measures of decentering—the Experiences Questionnaire—functions similarly across various groups of people (i.e., groups that differ by gender, race/ethnicity, age, and meditation experience). As such, these results suggest that this measure can be validly used and scores can be compared across people in these different groups.
Acknowledgments
Kristin Naragon-Gainey was supported by the National Center for Complementary and Integrative Health Grant R21AT009470. David M. Fresco was supported by National Heart, Lung, and Blood Institute Grant R01HL119977, National Institute of Nursing Research Grant P30NR015326, National Center for Complimentary and Integrative Health Grant R61AT009867, and National Institute of Child Health and Human Development Grant R21HD095099. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
1 Given limited research on decentering specifically, we draw on results from mindfulness studies when available in this literature review. However, mindfulness is a broader construct, of which decentering is only one component (Bernstein et al., 2015). As such, studies of mindfulness may be suggestive of findings for decentering but should not be interpreted as equivalent.
2 The student sample was sufficiently large to perform a more fine-grained analysis of different racial groups (i.e., White, Black, Asian, Other/Multiple). Full metric invariance and partial scalar invariance were supported. A comparison of latent means revealed that participants who identified as Black and Other/Multiple had higher levels of decentering than Asian or White participants.
References
- Bernstein A, Hadash Y, & Fresco DM (2019). Metacognitive processes model of decentering: Emerging methods and insights. Current Opinion in Psychology, 28, 245–251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bernstein A, Hadash Y, Lichtash Y, Tanay G, Shepherd K, & Fresco DM (2015). Decentering and related constructs: A critical review and metacognitive processes model. Perspectives on Psychological Science, 10, 599–617. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Carstensen LL (2006). The influence of a sense of time on human development. Science, 312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chen FF (2007). Sensitivity of fit indexes to lack of measurement invariance. Structural Equation Modeling, 14, 464–504. [Google Scholar]
- DeLuca SM, Kelman AR, & Waelde LC (2018). A systematic review of ethnoracial representation and cultural adaptation of mindfulness-and meditation-based interventions. Psychological Studies, 1–13. [Google Scholar]
- Franquesa A, Cebolla A, García-Campayo J, Demarzo M, Elices M, Pascual JC, & Soler J (2017). Meditation practice is associated with a values-oriented life: The mediating role of decentering and mindfulness. Mindfulness, 8, 1259–1268. [Google Scholar]
- Fresco DM, Moore MT, van Dulmen MM, Segal ZV, Ma SH, Teasdale JD, & Williams JG (2007). Initial psychometric properties of the Experiences Questionnaire: Validation of a self-report measure of decentering. Behavior Therapy, 38, 234–246. [DOI] [PubMed] [Google Scholar]
- Goh HE, Marais I, & Ireland MJ (2017). A Rasch model analysis of the Mindful Attention Awareness Scale. Assessment, 24, 387–398. [DOI] [PubMed] [Google Scholar]
- Gu J, Strauss C, Crane C, Bamhofer T, Karl A, Cavanagh K, & Kuyken W (2016). Examining the factor structure of the 39-item and 15-item versions of the Five Facet Mindfulness Questionnaire before and after mindfulness-based cognitive therapy for people with recurrent depression. Psychological Assessment, 28, 791–802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hu L & Bentler P (1999). Cutoff criteria for fit indices in covariance structure analysis: conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. [Google Scholar]
- Ireland MJ, Day JJ, & Clough BA (2018). Exploring scale validity and measurement invariance of the Toronto Mindfulness Scale across levels of meditation experience and proficiency Journal of Clinical Psychology. Advance online publication. [DOI] [PubMed] [Google Scholar]
- Isbel B, & Mahar D (2015). Cognitive mechanisms of mindfulness: A test of current models. Consciousness and Cognition: An International Journal, 38, 50–59. [DOI] [PubMed] [Google Scholar]
- Johnson DP, & Whisman MA (2013). Gender differences in rumination: A meta-analysis. Personality and Individual Differences, 55, 367–374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Katz D, & Toner B (2012). A systematic review of gender differences in the effectiveness of mindfulness-based treatments for substance use disorders. Mindfulness, 4, 318–331. [Google Scholar]
- King AP, & Fresco DM (2019). A neurobehavioral account for decentering as the salve for the distressed mind. Current Opinion in Psychology, 28, 285–293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Laurent H, Laurent S, Hertz R, Egan-Wright D, & Granger DA (2013). Sex-specific effects of mindfulness on romantic partners’ cortisol responses to conflict and relations with psychological adjustment. Psychoneuroendocrinology, 38, 2905–2913. [DOI] [PubMed] [Google Scholar]
- Naragon-Gainey K, & DeMarree KG (2017). Structure and validity of measures of decentering and defusion. Psychological Assessment, 29, 935–954. [DOI] [PubMed] [Google Scholar]
- Naragon-Gainey K, McMahon TP, Park J, & Fresco DM (2019). “Experiences Questionnaire (EQ)” In Medvedev ON, Krageloh CU, Siegert RJ, & Singh NN (Eds.), Handbook of Assessment in Mindfulness. New York, NY: Springer. Under review. [Google Scholar]
- Prakash RS, Hussain MA, & Schirda B (2015). The role of emotion regulation and cognitive control in the association between mindfulness disposition and stress. Psychology and Aging, 30, 160–171. [DOI] [PubMed] [Google Scholar]
- Shook NJ, Ford C, Strough J, Delaney R, & Barker D (2017). In the moment and feeling good: Age differences in mindfulness and positive affect. Translational Issues in Psychological Science, 3, 338–347. [Google Scholar]
- Van Dam NT, van Vugt MK, Vago DR, Schmalzl L, Saron C, … & Meyer DE(2017). Mind the hype: A critical evaluation and prescriptive agenda for mindfulness and meditation research. Perspectives in Psychological Science, 13, 36–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Waldron EM, Hong S, Moskowitz JT, & Burnett-Zeigler I (2018). A systematic review of the demographic characteristics of participants in US-based randomized controlled trials of mindfulness-based interventions. Mindfulness, 1–22. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.