Yoga motivation sources

According to Brems et al. [44], the most common motivation sources in yoga are the desire to exercise, increased flexibility, better posture and balance, relaxation, emotional balance, and improving one’s health. Penman et al. [45] found that yoga participants in Australia generally started doing yoga for health and fitness purposes and continued to do yoga for stress management, and Quilty et al. [46] similarly found that participants in Texas did yoga for general health, physical exercise, stress management, and to alleviate a health problem. Among the reasons for yoga participants in the United States, flexibility ranked first, followed by stress reduction, fitness, health improvement, strength, mental health, spiritual development, weight loss, physical therapy, and health during pregnancy [47]. While the initial motivations of yoga participants in Germany were relaxation and protection (prevention), their continuing motivations were established as prevention and then spirituality [48]. Data from China show that yoga participants are motivated by improving their physical condition, improving psychological condition and appearance, establishing social connections and creating social isolation [37]. Cartwright et al. [36] reported that the motivations of yoga participants in the United Kingdom were initially general health, fitness and flexibility, while some of those who started yoga due to a physical or mental health problem focused on spirituality, stress management and social interaction over time. Cagas et al. [39] found that positive affect, health-fitness and agility motivations were generally prominent in participants of different types of yoga, but compared to exercisers, postural yogis and yogis were more motivated by spirituality, mind-body integration, coping and stress management, and that some people were interested in yoga for its health and fitness benefits alone, while others saw yoga as more than just physical exercise.

In Turkey, where this study was conducted, academic studies on yoga began in 2015 [49], and academic studies addressing the motivations of yoga participants are generally qualitative studies. The results of qualitative studies conducted with small samples show that yoga participants in Turkey have motivations such as health, reference sources, curiosity, economic reasons, aesthetics and having fun, health spirituality and professionalization [50]; psychological relaxation, curiosity/giving it a try, professional expectation/contribution, spiritual/spiritual, calming the mind, physical exercise/physical good appearance, body health [51]; financial concerns, physical and mental health, the contraction in leisure activities during the pandemic [52]; healthy life, creating awareness, reducing pain, staying calm, coping with stress and increasing positive thinking [49].

Although it is a relatively new field of practice [53], the consensus of this research team is that yoga is becoming increasingly widespread in Turkey based on observations such as the opening of many yoga centers, the increase in advertisements of yoga centers on social media, celebrities sharing their yoga practices on social media, and the increasing preference of some individuals to quit their current professions and become yoga instructors.

Yoga activities in Turkey were institutionalized with the organization of the Yoga Federation in 2013 [54]. It is thought that there has been a rapid increase in the number of yoga participants and yoga centers [55]. It is also a fact that the number of people who do yoga in Turkey has not yet been determined exactly [49,56], as in other studies reporting that they could not access numerical data on yoga participants, yoga instructors, and yoga training centers [36,45]. In this context, although it is known that there has been an increasing interest in yoga participation in Turkey over time, it is seen that the behavior of yoga consumers has not been investigated much.

In the literature review, it was seen that exercise, sports or physical activity-based scales such as Exercise Motivation Scale-2 [39], Physical Activity and Leisure Motivation Scale [37] and Sport Motivation [57] were used in the measurement of yoga motivations. It was seen that the studies conducted using a scale were conducted with yoga participants in developed markets where yoga had been practised for a long time [58–60] or in Eastern countries. Undoubtedly, the motivations and responses of consumers in markets where yoga has been experienced for a long time may differ from those of consumers in markets where it has just begun to be experienced [46]. In addition, consumers’ purchasing motivations are subjective [61]. Different groups may have different perceptions about the importance of consumer motivation dimensions [62]. In this context, yoga motivations in countries that have been practicing yoga for a longer time may differ from Turkey.

In the enquiries conducted, no scale developed for yoga motivations in Turkey was found. It was determined that the scale studies that could be reached were scales for behavioral elements other than motivation, such as the yoga self-efficacy scale adapted by Tetik Küçükelçi [63] from Birdee et al. [64]; and the yoga belief scale adapted by Tetik Küçükelçi [65] from Sohl et al. [66].

Based on this literature and in the light of the information, the aim of our research is to develop a scale to determine the motivations of yoga consumers (yoga participants) to participate in yoga in Turkey, where yoga has just begun to develop.

Phase 1. Item development

At this stage, the determination of the items to be included in the scale and the content validity procedures for the items [68] were carried out.

Construct Definition: The creation of the theoretical structure of the scale was guided by the qualitative research conducted by the same research group in 2018 to determine the perception and motivation of yoga consumers in Turkey towards yoga [50]. In this research conducted in 2018, the initial motivations were health, reference sources, curiosity, economy, aesthetics and having fun, while the continuation motivations were health, spirituality, economic orientations, professionalization, promotional activities and the atmosphere of the place. In addition, studies on yoga motivations have also been examined [20,36,37,43–46,57,69–72]. In the context of the qualitative research conducted in 2018 and other studies reviewed, the structural elements of yoga consumers’ motivations were shaped around internal elements such as spirituality, escape, aesthetics, health and achievement, and external elements such as fashion, visibility, reference source, socialization, economy and physical activity/exercise.

Development of Initial Items: As a result of the enquiries, an initial item pool consisting of 118 items was created based on the theoretical basis. In creating the initial item pool, the items to be included in the scale were discussed and prepared by the researchers, taking into account the suggestions made by DeVellis (2017: 73–85), short sentences were used in the item writing, the number of items was increased by writing alternative items in order to find the item that would best express the concept, and negatively/reverse scored items were not written. The use of the “Five-Point Likert Type” measurement tool was preferred for the scale, and the statements were rated between 5 and 1 between “strongly agree” and “strongly disagree”.

Content Validity of Initial Items: The draft scale consisting of 118 items was submitted to expert opinions for content validity. It is generally recommended that the number of experts be between 5 and 40 in content validity studies [73–76]. At this stage, opinions were received from a total of 9 experts from the fields of psychology, sports psychology, and sports sciences, some of whom had previously come into contact with yoga. The experts made evaluations based on whether the scale items represented the characteristic to be measured, and whether the items were sufficiently clear. According to Lawshe’s [74] content validity index, the minimum value required for 9 experts in terms of consensus is 0.75. 67 items of the 118-item draft scale were eliminated because they were below 0.75, and 51 items with a value of 0.77–1 on which the experts reached a consensus remained. After adding questions containing demographic information, the scale was presented to a Turkish language expert for review in terms of writing and Turkish language, and the scale items were reviewed to ensure the linguistic clarity of the measurement tool.

Phase 2: Scale development

Boateng et al. [68] recommend conducting exploratory factor analysis (EFA) within the scope of an appropriate sample size to reveal the factor structure underlying the expressions at this stage. In this direction, the first evaluation of the remaining 51-item scale was made after expert opinions, and the items that did not consistently provide initial reliability and validity were removed, and the optimum number of factors or areas was determined.

Data Collection and Sample: The data for this phase were collected from volunteer participants in the form of an online survey via Google Forms during the Covid-19 pandemic quarantine period between April 15 and July 15, 2021. Data were collected after obtaining ethical approval from the Bolu Abant Izzet Baysal University Human Research Ethics Committee (Protocol No. 2021/92). Participants’ consent was obtained electronically, and participants were asked to select the “I participate in this study voluntarily” option when documenting their consent to participate. This procedure ensured that all participants joined the study voluntarily and with full awareness. Additionally all participants were 18 years of age or older; thus, no parental or guardian consent was required. In order to reach the participants, the managers of yoga centers in Turkey were first contacted via e-mail and informed about the study, and they were asked to share this information with both former and current members. The Google Form was then sent to the members who responded, via e-mail and WhatsApp. The data obtained were stored on a hard drive and were accessible only to the researchers. The first part of the measurement tool is about demographic (age, gender, marital status, income, region of residence, type of yoga, frequency of yoga, duration) information, and the second part is about motivations for participating in yoga. In scale development studies, it is recommended to reach a sample size of 10–15 times [77] or 5–10 times [78] the number of items in the scale. At this stage data was initially collected from 667 individuals using the convenience sampling method. As a result of the examination of missing, incorrect and extreme values, 429 valid data were obtained. The high level of data loss is due to the large number of blank, incorrect and incompletely filled open-ended questions in the demographic information section. Of the 429 participants at this stage, 52.2% live in the provinces in the western region, 94.9% are female and 50.3% are married. 35.2% are between the ages of 26–35, 32.4% are between the ages of 36–45. With 2021 figures, the incomes of the participants are in a wide range between 1000–75,000 TL, and the average income is 6927.35 TL. 17% of the participants have been doing yoga for less than 1 year, 40.6% for 1–3 years, 18.2% for 4–6 years, and 24.2% for more than 7 years. The frequency of yoga participation was 1–2 times a week for 54.8%; 3–4 times for 23.8%; 5–6 times for 12.1%; and every day for 9.3% of the participants. Among the yoga types that the participants answered open-endedly, it was seen that Hatha yoga was the most frequently done type with 52.4%. [79]

Item reduction: Before the EFA process, it was examined whether the data was normally distributed, and it was determined that the Skewness and Kurtosis values (Skewness −1.106–1.352; Kurtosis −1.741–1.477) were between −2 and +2 [80], which are accepted in the literature. By looking at the correlations between the items in the scale, it was checked whether the correlation was.30 and above [81], then in the 51-item draft scale, 18 items with item-total correlations below.30 were removed from the scale, and the EFA process was continued with the remaining 33 items.

Exploratory Factor Analyses Process: In order to determine whether the data is suitable for EFA at this stage, the Kaiser-Meyer-Olkin (KMO) coefficient (<.70) and the Bartlett sphericity test value (χ2 < .05) were calculated (KMO = .871, χ2 = 9801.905; sd = 528, p < 0.01), and it was determined that the values met the criteria values specified in the literature [82]. In scale development studies, factor analysis was used to examine whether there was a certain order between the responses given by the participants to each item in the draft scale [78] and to determine whether the items were separated into a single factor or more than one factor. In factor analysis, factors with an eigenvalue of 1 or greater than 1 are considered important factors [83]. There was no limitation on the number of factors in the scale, and factors with an eigenvalue greater than 1.00 were included in the scale. In the analysis process that continued with the remaining 33 items, the Oblique (Oblimin) rotation method was used to find conceptually independent but empirically related factors with principal axis factor analysis [84]. The criteria for keeping the items were as follows:

1. (1) The item has a loading of over.40 on its own factor [85]
2. (2) The difference between the item factor loading values in the two factors is more than 0.10 [86,87];
3. (3) The item fits the factor conceptually [88].
4. (4) At least three variables under each factor [89]

After the analyses based on the criteria mentioned above, the overlapping items that were below 0.40 in factor loading and loaded on more than one factor and had a difference of less than 0.10 between them were removed one by one at each stage, and the analyses continued until no overlapping items remained. In the relevant analyses, 6 items (M17, M27, M39, M41, M46, M47) included in the scale were eliminated. In accordance with the criterion of at least three variables under a factor [89], since two variables remained under the relevant factor, items M12 and M13 were also eliminated from the scale. After the items were removed, 25 items remained on the scale. At this stage, it was seen that all items related to spirituality, aesthetics, health, economy, and physical activity/exercise were eliminated from the Yoga Motivation Scale (YMS).

Phase 3: Scale evaluation

In this phase, CFA and reliability analyses were conducted by following the steps suggested by Boateng et al. [68].

Data Collection and Sample: The data for this phase were collected in the form of an online survey via Google Forms between May and August 2024. After the EFA, the remaining 25 items were renumbered by shuffling the order (1–25) and sent to the participants in this manner. The first part of the measurement tool is about demographic information (age, gender, marital status, income, region of residence, type of yoga practised, frequency of yoga practice, duration), and the second part is about motivations for participating in yoga. As a result of the EFA, data was collected from 270 people by using convenience sampling method to verify the validity of the scale consisting of 5 dimensions and 25 items. As a result of the missing, incorrect and extreme value examination, 224 valid data were obtained. 42% of the 224 participants in this phase live in the provinces in the Northeast, 89.3% are female and 51.3% are married. 35.7% are between the ages of 26–35, 29% are between the ages of 36–45. According to the figures for 2024, 66.5% of the participants have a household income of 42,000 TL and above. 34.8% of the participants at this stage have been practicing yoga for less than a year. Those who have been practicing yoga for 1–3 years are 30.8%, those who have been practicing yoga for 4–6 years are 22.3%, and those who have been practicing yoga for more than 7 years are 12.1%. The vast majority of participants (67%) do yoga 1–2 times a week. Among the yoga types that the participants answered open-endedly, it was seen that Hatha yoga was the most frequently performed type with 44.6%.

Confirmatory Factor Analysis Process: Confirmatory Factor Analysis (CFA) was applied to the 5-factor and 25-item scale obtained after EFA to verify the construct validity. This process was performed on a sample different from EFA. In order to test the pre-CFA analysis assumptions, the corrected item-total correlation values of the items were first examined. Since the item-total correlation of the 5^th item was below 0.30 [81], it was removed from the scale, and the CFA process was carried out with the remaining 24 items. Tabachnick and Fidell [85] stated that a sample size of 200 would not violate normality in multivariate analyses. Accordingly, no normality examination was performed, and only Mahalanobis distance analysis was performed to control extreme values. After the analysis, 46 extreme values were removed from the data set and the analyses were continued with 224 data. Since the data set to be analyzed was a Likert-type scale [90], CFA was performed using the maximum likelihood estimation procedure. CFA was performed both at the first-level multidimensional and second-level multidimensional levels. In the examination of the fit values related to the structure in CFA, the following fit index values were taken as criteria: Chi-square (χ2) and Chi-square/degree of freedom (χ2/df < 3), comparative fit index (CFI > 0.95–0.90), root mean square error of approximation (RMSEA<0.05–0.08) [91] and standardized root mean square error of approximation (SRMR<0.05–0.10) (Brown and Cudeck, 1993; as cited in: [92] recommended for a sample size less than 250. The analyses were made using the AMOS 21 program.

Findings of exploratory factor analysis (Study 1)

After the analyses conducted within the scope of Study 1, the KMO and Bartlett values of the remaining 25 items were re-examined in terms of their suitability for factor analysis, and it was determined that the number and distribution of the obtained data were suitable for EFA (KMO = .849, χ² = 7951.437; sd = 300; p < 0.01) [82]. Five sub-dimensions with eigenvalues greater than 1 were identified in the 25-item scale. In EFA, factors with an eigenvalue of 1 or greater are considered as important [83]. According to the Scree Plot graph in Fig 3, five breakpoints with an eigenvalue of 1 or greater indicate that the scale forms a five-factor structure. The sub-dimensions, respectively, are named as “Visibility” (8 items), “Escape” (6 items), “Socialization” (4 items), “Reference” (3 items) and “Achievement” (4 items) (Table 1). The total variance explained by the five factors is 62.696 percent. 27.414% of the total variance is explained by the 1^st factor, 12.820% by the 2^nd factor, 10.086% by the 3^rd factor, 6.545% by the 4^th factor and 5.831% by the 5^th factor (Table 1). The total variance explained by the scale is above 50% [84] indicates that it is at the desired level of adequacy. Factor loadings of the items were determined to be between 0.96 and 0.60, and above 0.40 [85]. Cronbach’s Alpha internal consistency coefficient was used to determine the reliability of the scale in EFA analyses. The Cronbach Alpha internal consistency coefficient calculated for 25 items in the obtained data was.847. This value indicates that the scale has an acceptable internal consistency. The Cronbach’s Alpha coefficient for the sub-dimensions of the scale is.944 for the 1^st factor (Visibility),.861 for the 2^nd factor (Escape),.848 for the 3^rd factor (Socialization),.851 for the 4^th factor (Reference), and.784 for the 5^th factor (Achievement) (Table 1). Internal consistency coefficients at the sub-dimensions are also within acceptable ranges [93]. Table 2 shows that the item means range from 1.221 to 3.476 and the mean of all items is 2.039. Büyüköztürk [81] states that items with an item-total correlation of 0.30 and above are highly discriminatory, and items between 0.20 and 0.30 can be included in the test if deemed necessary. The item-total correlation of all items on the scale, except M25, is above 0.30, and M25 is within the acceptable range stated by Büyüköztürk [81].

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Results of Exploratory Factor Analysis of the YMS.

https://doi.org/10.1371/journal.pone.0359209.t001

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Mean, Standard Deviation, Cronbach’s Alpha (α) and Item-Total Correlation Values.

https://doi.org/10.1371/journal.pone.0359209.t002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Scree plot graph obtained as a result of EFA.

https://doi.org/10.1371/journal.pone.0359209.g003

Findings of confirmatory factor analysis (Study 2)

In Study 2, confirmatory factor analysis of the 5-dimensional scale resulting from EFA was conducted twice: first-level multifactorial and second-level multifactorial. Table 3, Figs 4 and 5 show the modifications for the first and second level multifactorial confirmatory factor analysis and the goodness of fit values resulting from the modification.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. First and Second Level Multifactor Confirmatory Factor Analysis Goodness of Fit Values Before and After Modification.

https://doi.org/10.1371/journal.pone.0359209.t003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Single-Level Multifactor Model.

https://doi.org/10.1371/journal.pone.0359209.g004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Double-Level Multifactor Model.

https://doi.org/10.1371/journal.pone.0359209.g005

As stated in Table 3, since the RMSEA and SRMR values in the first-level multi-factor analysis were above the criterion index value range (RMSEA > 0.80 and SRMR > 0.80), modification suggestions were considered. Following the relevant suggestion, covariance was plotted between items 22 and 25, and the analysis was repeated (Fig 4). After the analysis, it was determined that the relevant values were within the acceptable range of the criteria index values. The same process was carried out for the bi-level scale as well. At this stage, after the first confirmatory factor analysis, the RMSEA and SRMR values were again not within the criterion index value range (RMSEA > 0.80 and SRMR > 0.80), so modification suggestions were examined. The covariance paths between Items 22 and 25 and between Items 15 and 19 were incorporated into the measurement model and are presented in Fig 5. Items 22 and 25 belong to the Visibility dimension, whereas Items 15 and 19 belong to the Socialization dimension. Within each dimension, these item pairs reflect conceptually similar aspects of the same latent construct. Therefore, the error terms of these items were allowed to covary, as they represent theoretically related content within their respective dimensions.After the second analysis, it was determined that all values were at an acceptable level within the criterion index value range (Table 3). As a result of these analyses, it was determined that the relevant scale can be used both as a five-dimensional scale and as a two-dimensional scale at the second level. In the first-level multifactorial model, it was determined that the factor loadings of the items varied between.98 and.43 (>0.40), t values varied between 59.661 and 5.161 (>1.96), R² (Square of multiple correlation) values varied between 0.19 and 0.96, and corrected item-total correlation values varied between 0.37 and 0.95 (>0.30) (Table 4). According to the first-level multifactorial analysis, the Cronbach Alpha internal consistency coefficient for the whole scale (24 items) was calculated as 0.90 and Omega as 0.89. These values state that the scale has an accepted internal consistency [84,93,94]. As Table 5 is examined, it is seen that there is a positive moderate relationship between the factors in the first-level multifactorial model (p < 0.01). The analysis results for the verification of the second-level multifactorial model state that factor loadings ranged between.98 to.41 (>0.40), t-values ranged between 59.657 to 5.412 (>1.96), R² (Square of Multiple Correlation) values ranged between 0.171 to 0.96, and corrected item-total correlation values ranged between 0.37 to 0.95 (>0.30) (Table 6). According to the second-level multifactorial analysis, the Cronbach Alpha internal consistency coefficient for the whole scale (24 items) was calculated as 0.90 and Omega as 0.89. These values state that the scale has an accepted internal consistency [84,93,94]. Cronbach’s Alpha coefficient for the sub-dimensions of the scale varies between.63 and.96, and Omega values vary between 0.64 and 0.96. It is seen that all the values specified in Table 6, comply with the criterion values within the literature [77,82,95]. Table 7 shows the correlation between the external and internal dimensions in the second-level multifactorial model and the other sub-dimensions under these dimensions. According to Table 7, there is a positive significant relationship between visibility, socialization and references, which constitute the external dimension, and escape and achievement which constitute the internal dimension, as well as between the internal and external dimensions (p < 0.01). The results of the final form of YMS are presented in Tables 8.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Analysis Results for Validation of First -Level Multi-Factor Model.

https://doi.org/10.1371/journal.pone.0359209.t004

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Correlation Analysis Between Sub-Dimensions of the First-Level Multifactor Model.

https://doi.org/10.1371/journal.pone.0359209.t005

Download:

• PNG
  larger image
• TIFF
  original image

Table 6. Analysis Results for Validation of Second-Level Multifactor Model.

https://doi.org/10.1371/journal.pone.0359209.t006

Download:

• PNG
  larger image
• TIFF
  original image

Table 7. Correlation Analysis between Sub-Dimensions of the Second-Level Multi-Factor Model.

https://doi.org/10.1371/journal.pone.0359209.t007

Download:

• PNG
  larger image
• TIFF
  original image

Table 8. Final Version of YMS: Renumbered items.

https://doi.org/10.1371/journal.pone.0359209.t008

Following all analyses, the final renumbered version of the 24-item scale is presented in Table 8. In addition, the allocation of the renumbered items to their respective subdimensions is provided in detail in Table 9.

Download:

• PNG
  larger image
• TIFF
  original image

Table 9. Final Version of YMS: Factors and items.

https://doi.org/10.1371/journal.pone.0359209.t009