Effect of Meditation on Emotional Intelligence and Perceived Stress in the Workplace: A Randomized Controlled Study

Approval was given on August 31, 2009, by the institutional review board of Maharishi University of Management, Fairfield, IA, before the start of the study. Participants of the study were recruited from Fall 2009 through Spring 2010 from administrators and staff working in the central offices of the San Francisco Unified School District who were interested in being part of a workplace wellness program. Interested participants were asked to attend an informational meeting to learn about the wellness project. Those who wanted to join were then scheduled for baseline testing. Ninety-six supervisors and administrative staff completed written informed consent, followed by baseline testing, and were then randomly assigned to either an immediate start of the TM program (n = 48) or a delayed-start wait-list control group (n = 48; Figure 1).

Group allocation was concealed by an off-site member of the research group, using a computer-generated random number sequence (SPSS 2009, IBM Corp, Armonk, NY). The off-site member assigned the participant to treatment and notified the study coordinator, who informed the participant of his/her allocation. Participants completed psychological measures via the Internet. Final statistical analyses were conducted with blinding to group assignments. Senior investigators and the study statistician were blinded throughout the study.

Inclusion criteria included age 18 years or older, being an employee of the San Francisco Unified School District, attendance at an informational meeting on the TM program, and willingness to be randomly assigned to either active treatment or the control group. Exclusion criteria included having already learned TM.

The Emotional Quotient Inventory (EQ-i, Multi-Health Systems Inc, Toronto, Ontario, Canada) is a widely used measure of emotional intelligence that evaluates a broad range of noncognitive, social-emotional skills, influencing one’s ability to positively cope with environmental demands and pressures.^20,40 The EQ-i has a total Emotional Quotient (EQ) score and 6 composite scales (intrapersonal, interpersonal, stress management, adaptability, reality testing, and general mood) with 14 subscales. Scale scores indicate the degree of effectiveness in understanding oneself, relating well to people, and adapting to and coping with one’s immediate surroundings.^7,20,40 The Cronbach α for the total EQ scale is 0.97, the composite scales range from 0.88 to 0.93, and subscales are all above 0.77.⁴⁰

The Perceived Stress Scale (PSS, 14-item version) measures the degree to which situations in one’s life are appraised as stressful. The Cronbach α is reported to be 0.85. The PSS shows good discriminative validity with meditation.⁴¹

All participants were tested at baseline, after giving written informed consent and before the start of meditation training, and were then randomly assigned to either the TM program or wait-list control groups. Participants were then posttested 4 months later, at the end of the intervention period. Each participant took the EQ-i and PSS online. The EQ-i was scored by Multi-Health Systems, the publisher of the EQ-i. The total EQ, composite, and subscale scores for each participant were then electronically transmitted to the research staff for inclusion in the study database. The PSS was scored by the study statistician.

The sample size was determined using statistical power calculation software (G*Power 3.1.3, available at www.gpower.hhu.de/en.html). We based the sample size on previous studies,^30,40 which reported effect sizes for TM vs control groups on outcome measures similar to our study: Experiential intelligence as measured by the Constructive Thinking Inventory (d = 0.62) and the PSS (d = 0.94). The power calculation showed that a sample size of 48 per group would provide at least 90% power to detect treatment effects similar to those reported in these previous studies, assuming a 2-sided test for between-group differences at the 0.05 significance level.

The main outcomes of the study were the total EQ and perceived stress scores. Secondary outcomes were the EQ composite scales—intrapersonal, interpersonal, stress management, adaptability, reality testing, and general mood. Data were analyzed using repeated-measures analysis of variance according to the intention-to-treat principle; all randomized participants were included. Missing data were imputed using the software programs SAS PROC MI (SAS Institute Inc, Cary, NC), a multiple imputation using the Markov Chain Monte Carlo method, and SAS PROC MIANALYZE. Alpha was set at 0.05, 2-tailed, for the primary outcomes and at 0.01, 2-tailed, for the secondary outcome scales. Additional secondary analyses were conducted for the subscales (comprising the 6 main scales), with the α level set at 0.01, 2-tailed. Effect sizes (Cohen d) were computed for the PSS and total EQ and composite scales, using the difference in mean change scores divided by pooled standard deviation. Effect sizes are: Small (d = 0.20), medium (d = 0.50), and large (d = 0.80).

Table 1 shows the baseline characteristics of the study participants. The average age was 45.5 years (standard deviation = 10 years), and 83% of the participants were women. There were no statistically significant differences between groups at baseline in age, sex composition, race/ethnicity, administrator/staff status, baseline emotional intelligence, and perceived stress.

Intention-to-treat analysis indicated a significant improvement in total EQ score in those practicing TM compared with controls: F(1,94) = 5.95, p < 0.02. A significant decrease in perceived stress owing to the TM treatment compared with controls was also found: F(1,94) = 9.12, p < 0.003; Table 2.

Multivariate analysis of variance showed significant improvement in the composite scales comprising the total EQ score (Wilks λ = 0.835, p = 0.009). Univariate tests showed significant effects for intrapersonal [F(1,94) = 8.29, p = 0.004]; stress management [F(1,94) = 5.29, p < 0.03]; adaptability [F(1,94) = 4.54, p < 0.04]; reality testing [F(1,94) = 4.45, p < 0.04]; and general mood [F(1,94) = 11.42, p = 0.001] composite scores during the 4-month intervention period. Table 2 shows the pretest and posttest scores by group for total EQ and the composite EQ scores.

The largest effect sizes for the main scales in the study were for perceived stress (d = −0.51) and EQ general mood composite score (d = 0.51). The effect size for the total emotional intelligence main scale was d = 0.37.

Table 3 shows the post hoc findings for each of the emotional intelligence subscales comprising the composite scores. The largest effect sizes were for happiness (d = 0.49) and stress tolerance (d = 0.45).

Further analyses of both EQ total score and perceived stress showed no significant group by sex interaction (p = 0.21 and p = 0.79, respectively, on the dependent variables perceived stress and total EQ score). Thus, there was no evidence that the treatments varied significantly by sex subgroups. In a separate analysis, the group by administrator/staff status interaction also showed no significant effects (p = 0.97 and p = 0.25, respectively, on the dependent variables perceived stress and total EQ score). Hence, there was no evidence that the treatments varied significantly by administrator/staff status.

Overall, change in perceived stress inversely correlated with change in total EQ (r = −0.613, p < 0.01).

Regularity in TM practice was defined as home practice of the TM program of at least once a day, on average, by self-report at posttest. Compliance with the meditation program was high, with 93% of participants meeting the criteria for study practice regularity. A dose response was observed, with those meditating more regularly scoring lower on perceived stress (r = −0.33, p = 0.032) and higher on total EQ (r = 0.28, p = 0.067) at posttest. No adverse events were reported for either group.

The study presented was a randomized controlled trial of 96 administrators and staff working in the same organizational setting. One major advantage of a wait-list control is that all participants can eventually receive the intervention. This, in turn, can facilitate both recruitment and retention. Use of an active control group with matched treatment sessions and daily home practice would have helped control for time and attention factors. Other possible confounders such as participant expectations, treatment preference, and possible social support influence could be addressed using an active control group. All participants were tested under the same conditions, using self-administered, Internet-based measures although sole use of self-report measures for study outcomes is a possible limiting factor. Future workplace studies should consider adding supervisor and colleague rating scales for emotional intelligence and workplace behaviors. Because of the sex composition of the San Francisco Unified School District administrators and staff in their central offices, the findings are most generalizable to female administrators.

Future research is encouraged to use a larger-designed, multisite study, with a more active control group in additional workplace settings. Future studies also may want to have an equal balance of male and female participants to more adequately evaluate the effects of meditation on sex. In addition, a measure of job performance may be useful to determine relationships between mental health factors (eg, emotional intelligence and perceived stress) and job performance.