This
article, A Psychometric Analysis of the Three Gunas, was published in the
June, 1999 issue (vol. 84, pgs. 1379-1390) of Psychological Reports.
A
Psychometric Analysis of the Three Gunas
by David Wolf
Vedic literature contains knowledge about all fields of human endeavor,
from physics and psychology to medicine, art and aeronautics (Goswami, 1977).
Empirical validation of the Vedas could therefore open storehouses of knowledge
in many areas. According to the Vedas, all material elements are infused with
the modes of nature, or gunas- sattva, rajas and tamas.
Dasgupta (1961) describes the gunas as “the universal characteristics
of all kinds of mental tendencies” (p. 468). Broad based confirmation of the
Vedic psychological model would reasonably be grounded on a study of the
gunas, as this idea underlies the Vedic conception of nature. From the
perspective of the Vedas, our psycho-physical disposition consists of mixtures
of the gunas (Prabhupada, 1976).
This article describes development of the Vedic Personality Inventory (VPI),
an instrument that assesses the validity of the three guna constructs.
Others, such as Uma, Lakshmi and Parameswaran (1971), Singh (1971), Rao and
Harigopal (1979), Mohan and Sandhu (1986, 1988), Das (1991), and Pathak, Bhatt
and Sharma (1992), have devised and utilized similar inventories, with inter-guna
correlations that indicate differentiation of the three modes. Since the primary
purpose of the present study is to examine the validity of Vedic constructs, it
incorporates more extensive statistical analyses towards this end than other
research on the gunas.
Also, the items of this inventory are derived from a Vaisnava, or
personalist, understanding of Vedic philosophy. According to this view, the
material personality manifested through the gunas is a covering of the
original spiritual personality of the living entity. Bhagavad-gita [2:12 (Prabhupada,
1986)] asserts that the self is eternally an individual. Prabhupada explains
“Arjuna and all the kings who are assembled on the battlefield are eternally
individual beings ...It is not that they did not exist as individuals in the
past, and it is not that they will not remain eternal persons. Their
individuality existed in the past, and their individuality will continue in the
future without interruption” (p. 89).
The personal emphasis of Vaisnava philosophy can provide a guiding ethic
for social and mental health scientists. According to the Vedas, by removing the
covering of the modes of nature the original spiritual personality of each
individual will be manifest. Without such restoration of our spirituality,
relationships are mechanically conducted on the basis of the gunas.
According to Vaisnava philosophy, fully spiritual relationships can be achieved
only when the gunas are absent, and replaced with the quality of pure
goodness, or suddha sattva, which is the platform of spiritual existence
(Prabhupada, 1976). Consequently, investigating and instituting methods for
extricating the self from the gunas is relevant for the social and mental
health sciences.
To test the validity of the Vedic constructs of the three gunas,
it is best to devise an instrument that covers the domain of personality
described in the Vedas themselves. Apart from studying the gunas as a
universal typology, there has been increased skepticism regarding the
applicability of Western psychological constructs for explicating and
understanding the psychology of indigenous peoples (Gergen, Gulerce, Lock, &
Misra, 1996). Therefore, a cultural perspective also dictates examination of the
gunas as the Vedas describe them, rather than trying to understand Vedic
concepts by reducing them to constructs originating in Western thought. Gergen,
Gulerce, Lock and Misra have suggested exploration of Vedic psychology with
emphasis on the spiritual ontology of personhood. This is an important
perspective, as the Vedas describe the self as a spiritual entity that is
personal and irreducible [Bhagavad-gita 2:12 and 2:24 (Prabhupada, 1986)]. Any
theory based on a reductionistic ontology of personhood would be incompatible
with the Vedic paradigm and persons culturally attuned to that paradigm. The VPI is an attempt to apply the concepts of Vedic personality
while preserving the integrity of guna theory.
Vedic theory asserts that while the classification scheme of the three gunas
comprises all aspects aspects of material, social and psychological reality, the
gunas themselves are related to each other. Specifically, rajas guna
is described as an intermediary between tamas guna and sattva guna
(Prabhupada, 1976). Therefore, it is expected that there will be an inverse
relationship between sattva guna and tamas guna. In terms of
psychometric theory, this type of relationship between constructs meets the
definition of a weak psychometric theory, wherein the factors are non-orthogonal
(Nunnally & Bernstein, 1994).
Method
Defining Constructs
Descriptions
of each guna were compiled from the Bhagavad-gita As It Is (Prabhupada,
1986), Chapters 14, 17 and 18, as well as Dasgupta (1961). Sattva guna is
characterized by qualities such as cleanliness, truthfulness, gravity,
dutifulness, detachment, discipline, mental equilibrium, respect for superiors,
contentment, sharp intelligence, sense control, and staunch determination.
Dasgupta describes the sattvic quality as being “free from attachment
and vanity and absolutely unruffled in success and failure” (p. 470). Krsna
explains in the Bhagavad-gita [17:15, p. 783] that the speech of a person
predominantly in sattva guna is “truthful, pleasing, beneficial, and
not agitating to others” and that a sattvic person is characterized by
“satisfaction, simplicity, gravity, self-control and purification? [17:16, p.
784].
Attributes
of rajas guna include intense activity, desire for sense gratification,
little interest in spiritual elevation, dissatisfaction with one’s position,
envy of others, and a materialistic mentality. Bhagavad-gita [18:24, p. 815]
describes rajasic activity as “action performed with great effort by
one seeking to gratify his desires, and enacted from a sense of false ego,”
and a person predominated by rajas guna is described as “greedy, always
envious, impure, and moved by joy and sorrow” [18:27, p. 817].
Qualities
associated with tamas guna include mental imbalance, anger, ignorance,
arrogance, depression, laziness, procrastination, and a feeling of helplessness.
Dasgupta (1961) explains that “the quality of tamas overcomes the
illumination of knowledge and leads to many errors. Tamas, being a
product of ignorance, blinds all living beings and binds them down with
carelessness, idleness and sleep” (p. 462).
Scale Design and Item Development
From
the descriptions of the Bhagavad-gita (Prabhupada, 1986) and Dasgupta
(1961), fifty statements were formulated for each guna. These took the
form of sentences with which someone who is predominated by that mode would
strongly agree. Utilizing feedback from a team of five Vedic scholars, each of
whom had more than twenty years experience studying and teaching the Vedas, the
item sample was reduced to ninety items that were considered most representative
of the gunas, and that would best differentiate between the guna
constructs. According to the Vedic experts, these ninety items, comprised of 30
for sattva guna, 28 for rajas guna, and 32 for tamas guna,
adequately covered the domain of attributes for each mode. To further determine
content validity, two Vedic authorities, who had not previously been exposed to
any of the items, each completed three copies of the 90-item questionnaire. On
each copy, they answered as if they were primarily influenced by one of the gunas.
Then their responses were compared to the expected responses for each item. For
example, a sattva item would be expected to receive a response of “very
strongly agree” when the Vedic experts answered as if they were predominated
by sattva, and a response of disagreement or less agreement if they were
predominated by rajas or tamas. Based on this analysis, the
wording of some items was changed to accentuate the differences between the
modes. Also, this procedure indicated that several items, while primarily
representative of one mode, would be expected to load heavily on another mode.
Specifically, many rajas items were predicted to have strong secondary
loadings on tamas.
A
seven-point scale was chosen as a balance between convenience, for the
participant and researcher, and statistical power. Given the relatively large
number of items in the instrument, nine or eleven response choices would not
significantly strengthen statistical determinations. Still, to minimize chances
of spurious outcomes, especially in factor analyses, a seven-point scale was
chosen over a five-point scale.
For
convergent construct validity, two existing scales, the Verbal Aggressiveness
Scale (VAS) (Infante & Wigley, 1986) and the Satisfaction With Life Scale (SWLS)
(Diener, Emmons, Larsen, & Griffin, 1985), were included at the end of the
VPI. Existing personality inventories based on the gunas were not used
for convergent validity because there are no reliability studies for these
instruments. Also, it would have been impractical to administer two lengthy
surveys with similar questions. The scores on the VAS were hypothesized, based
on Vedic theory, to correlate positively with rajas and negatively with
sattva, and the scores on the SWLS were expected to correlate positively
with scores on the sattva scale and negatively with the tamas scale.
Also, a single-item indicator, hours of sleep per day, was included at the end
of the survey. This indicator was hypothesized to correlate positively with tamas,
as excessive sleep is a characteristic of one in tamas. Additionally, six
demographic questions, pertaining to gender, age, years of education, ethnicity,
religion practiced and religion of parents, were asked. Gender was included as a
measure of discriminant validity, as it was not expected to correlate with any
of the modes, and the other demographic questions were included for possible use
in future analysis of the instrument.
Based
on assessment of Cronbach’s alpha reliability, content validity, and
confirmatory, multiple groups centroid factorial analysis (Nunnally &
Bernstein, 1994), each item was evaluated and the 90-item scale was reduced to
an 80-item scale, consisting of 28 sattva items, 24 rajas items,
and 28 tamas items. In the version of the 80-item VPI that was
distributed to subjects, two additional questions, regarding height and number
of siblings, were added to more thoroughly evaluate discriminant construct
validity.
Data Collection
The
author and two research assistants administered the questionnaire to voluntary
participants in a mid-sized Southern city in the United States in November of
1996. Anonymity of subjects was maintained, as surveys were returned to a box,
without the participants’ identity being revealed to the researcher. This was
a nonprobability convenience sample of 247 persons, consisting of 52.5% males
and 47.5% females. Eighty-four percent of the sample was Euro-American, with the
remaining 16% being African-American and Asian-American. University students
comprised 41% of the sample, while nurses and doctors constituted 14%. Other
occupations represented in the sample included clerical work, physical labor,
and social services. More than 93% of the participants were raised in Christian
homes, and more than 67% currently practice some form of Christianity. The mean
years of education for participants was 14.9, and their ages ranged from 18 to
64 years, with a mean of 32.5.
A
combination of Cronbach’s alpha analyses, multiple groups factor analyses (Nunnally
& Bernstein, 1994) and evaluation of content validity applied to the data
from the above sample resulted in an 80-item scale, consisting of 28 sattva
items, 24 rajas items, and 28 tamas items. During January, 1998,
this 80-item scale was administered to 247 persons. For the combined 494 person
sample, there were 257 (52%) (females) and 237 (48%) males. Seventy-two percent
of the 494 persons was Euro-American, 20% was African-American, and 8% was
Asian-American. Of the 494 subjects, 287 (58%) were university students and 124
(25%) were medical professionals. The average number of years of education for
participants was 15.3, and their ages ranged from 18 to 64 years, with a mean of
31.9. For the January, 1998 sample, two items, height and number of siblings,
were added to the survey for assessing discriminant construct validity.
From
July through December, 1998, the 80-item VPI was completed by 125 subscribers to
a magazine dealing with Eastern-style spirituality. This version of the VPI did
not include any demographic items or scales for construct validity.
Results
and Data Analysis
Descriptive Statistics
After
statistical analysis of all 619 surveys, a final 56-item scale was created1.
Table 1 provides descriptive statistics for the revised, 56-item version of the
VPI.
Insert
Table 1 here
Scores
for the gunas were obtained by adding the responses for the items for
each scale corresponding to a guna and then dividing by the number of
items. For each subscale, a higher score indicates greater predominance of that
mode. Scores on the SWLS and VAS are the sums of responses on the respective
instruments, with the VAS having ten out of twenty items with reverse scores.
Sleep was measured in hours per day. Cronbach’s alpha for this sample for the
VAS was .90, and for the SWLS alpha was .82. Designers of the VAS (Infante &
Wigley, 1986) report an alpha of .81, and the creators of the SWLS (Diener,
Emmons, Larsen, & Griffin, 1985) report an alpha of .87. As per
recommendations by the authors of the scales, VAS cases were retained for the
present analysis if at least fifteen of twenty questions were answered, and SWLS
cases were retained if at least four of five questions were answered. Height was
measured as an ordinal scale, with “1” representing “Under 5 feet”,
“2” indicating “Between 5 feet and 5.5 feet”, “3” indicating
“Between 5.5 and 6 feet”, and “4” representing “Over 6 feet”.
Reliability
Each
guna constitutes a subscale, and internal reliability for each subscale
was measured using Cronbach’s alpha. For the 90-item VPI, alpha for the
sattva scale was .85, for rajas it was .92, and for tamas it
was .90.
Using
SPSS statistical software for Macintosh, Alpha if Item Deleted was computed for
each item in each subscale. Alpha if Item Deleted values indicate Cronbach’s
alpha for the subscale if the item was not included in the subscale. If Alpha if
Item Deleted is lower than alpha for the subscale, this means that the item
increases alpha, and thus adds to the reliability of the subscale. If Alpha if
Item Deleted is higher than alpha for the subscale, this indicates that the item
weakens the reliability of the subscale (Nunnally & Bernstein, 1994).
For
the sattva subscale, two items had an Alpha if Item Deleted that was
higher than alpha for the sattva subscale, and these two items were deleted from
the subscale. Alpha for the 28-item sattva subscale was then calculated
as .87.
In
the rajas subscale, four items had an Alpha if Item Deleted that was
higher than alpha for the subscale, and these four items were removed from the
scale. Alpha for the 24-item rajas subscale was then calculated as .93.
For
the tamas subscale, four items had an Alpha if Item Deleted that was
higher than alpha for the subscale, and these four items were removed from the
scale. Alpha for the 28-item tamas subscale was then calculated as .91.
As
described above in the data collection section, the 80-item VPI was administered
to 372 subjects. Reliability analyses were performed for the entire sample of
619 persons. In this analysis Alpha if Item Deleted was calculated for each
item, as was Corrected Item-Total Correlation. Hudson and Faul (1997) explain
that Corrected Item-Total correlations are an assessment of convergent construct
validity at the item level of analysis, and they recommend a criteria of .50 as
a cutoff point for retention of items. This criteria was applied to the
statistical analysis of the 80-item VPI, as was the criteria that each item had
to have an Alpha if Item Deleted lower than the alpha for the subscale.
Using
the criteria described above, the sattva subscale retained 15 items and
had an alpha of .93. The rajas subscale retained 19 items and had an
alpha of .94, and the tamas subscale retained 22 items and had an alpha
of .94. This resulted in a final VPI with 56 items, with each item having a
Corrected Item-Total Correlation greater than .50, and each item adding to the
reliability of its subscale, as measured by Cronbach’s alpha2.
Construct Validity
Correlations
of the subscales with the VAS and SWLS, as well as with hours of sleep per day,
were calculated as measures of convergent validity. All correlations in this
analysis are Pearson product-moment (r) correlations.
Vedic
theory predicts that verbal aggressiveness is positively correlated with
rajas and negatively correlated with sattva, and that life
satisfaction correlates positively with sattva and negatively with tamas
(Dasgupta, 1961). Hours of sleep was predicted to be positively correlated
with tamas. Values of r2 reported
in Table 2 show good preliminary evidence for convergent validity of the VPI.
Mean r2 values are calculated using
only those correlations that were expected to give evidence for convergent
validity. Convergent validity measures do not include the subjects who were
subscribers to the magazine dealing with spirituality, because the VPI surveys
they completed had no convergent validity items. For each measure of convergent
validity, items expected to correlate had a higher r2 than items not expected to correlate. All correlations were in
the expected direction.
Insert
Table 3 here
Factorial Validity
Factorial
validity for the VPI was assessed using the confirmatory, multiple groups
centroid method. Loadings for each item of the 56-item VPI were computed for
each of the three subscales. Factorial validity indicates the extent to which an
item correlates with the subscale with which it is intended, as compared to
other subscales (Nunnally & Bernstein, 1994).
For
the 15 items in the sattva subscale, correlations between the items and
the sattva subscale ranged from +.62 to +.87. Absolute values of the
correlations of the rajas subscale with individual items in the sattva
subscale ranged from .04 to .51, with only 3 of 15 items having a correlation
with an absolute value greater than .23. Absolute values of the correlations of
the tamas subscale with individual items in the sattva subscale
ranged from .43 to .67. All 15 of these correlations had negative values, and
two of them had absolute values greater than .62 (the lowest value for sattva
subscale correlations with sattva items). For these two items,
however, the correlation with the sattva subscale was stronger than the
correlation with the tamas subscale.
For
the 19 items in the rajas subscale, correlations between the items and
the rajas subscale ranged from +.57 to +.80. Absolute values for the
correlations of the sattva subscale with individual items in the rajas
subscale ranged from .02 to .47, with only 3 of 19 correlations having a
correlation with an absolute value greater than .30. Absolute values of the
correlations of the tamas subscale with individual items in the rajas subscale
ranged from .01 to .52. Only 1 of these 19 correlations (-.01) had a negative
value, and 18 had positive values.
For
the 22 items in the tamas subscale, correlations between the individual
items and the tamas subscale ranged from +.55 to +.76. Absolute values
for the correlations of the rajas subscale with individual items in the tamas
subscale ranged from .01 to .53. Two of these correlations had negative
values (-.01 and -.06), and 20 had positive values. Absolute values of the
correlations of the sattva subscale with individual items in the tamas
subscale ranged from .35 to .74, with all 22 items possessing negative
values. Although some items in the tamas subscale had correlations with
the sattva subscale with an absolute value greater than .55, every item
in the tamas subscale had a correlation with the tamas subscale
that was higher than its correlation with the sattva subscale.
These
factor analyses3 demonstrate that
every item in the VPI has a strong correlation with its intended subscale, and
the strongest correlation for every item is with its intended subscale. However,
some items in the sattva and tamas subscales have strong negative
correlations, defined as correlations that are stronger than the weakest
item’s correlation with the intended subscale, with an unintended subscale.
These results suggest that Vedic guna theory can be considered a
“weak” theory for psychometric purposes. Nunnally and Bernstein (1994)
explain that a “weak” theory is suitably tested with a confirmatory,
multiple groups centroid method, as used in this investigation. A “weak”
theory is concerned with whether or not proposed groupings of variables exist.
However, such a theory does not assume that the factor structure underlying
these groupings are the only possible explanatory classification. When assessing
a “weak” theory, independence of factors is not assumed.
With
regards to this investigation of the three gunas, analyses of reliability
and construct validity suggest that the three gunas are a viable
categorization scheme. That is, the analyses indicate that the elements
comprising each of the gunas as described in the Vedas are congeneric,
with a single factor underlying each of the three combinations of attributes, as
delineated by the items of the VPI. Multiple groups centroid factor analysis (Nunnally
& Bernstein, 1994) support this conclusion, though factor analyses also
imply that the constructs are not orthogonal. This is further evidenced by
inter-subscale correlations, which are shown in Table 4. These correlations
support the results of the factor analyses, and demonstrate that sattva
and tamas have a very strong inverse relationship. Similar correlations
between the gunas subscales were found in other research on the gunas
(Pathak, Bhatt, & Sharma, 1992).
Insert
Table 4 here
Discussion
Nunnally
and Bernstein (1994) state that Cronbach’s alpha should be at least .80 for an
instrument to produce interpretable research outcomes. Alpha measures for the
three subscales are .93, .94 and .94. Therefore the 56-item VPI is adequate for
group research, and perhaps also for individual assessment, with regards to
reliability. Each item of the VPI contributes positively to the reliability of
its subscale, and each item also has a corrected item-total correlation with its
subscale greater than .50. These statistics suggest strong subscales, meaning
that the elements of the subscales correlate strongly with each other.
Establishing
face validity involves judgment of item adequacy after an instrument has been
constructed, and establishment of content validity involves the rational process
of defining the domain of a construct and selecting items that accurately and
comprehensively cover that domain (Nunnally & Bernstein, 1994). Face and
content validity for the VPI have been confirmed by a panel of Vedic experts.
This panel checked contents of the inventory at each stage of the item screening
process, and they have affirmed that the items of each subscale of the 56-item
VPI adequately cover the domain of each mode. The three items assessed for
convergent validity- verbal aggressiveness, life satisfaction, and hours of
sleep- all provide evidence for construct validity of the instrument. Further,
the four items evaluated for discriminant validity- gender, age, height, and
number of siblings- also support construct validity of the VPI. Bloom, Fischer,
and Orme (1995) explain that construct validity is a measure of the extent to
which empirically observed relationships between measures of concepts agree with
theoretically predicted interrelationships between concepts. The results of
convergent and discriminant validity measures in this study provide evidence for
the construct validity of the VPI. Multiple groups factor analyses (Nunnally
& Bernstein, 1994) support the categorization scheme of the three gunas,
and they also support the assertion of Vedic theory that sattva and tamas
are inversely correlated, with rajas as an intermediary mode.
Overall,
the hypotheses of the Vedas concerning the gunas are supported by this
research. Items of each subscale array themselves as a unit, and guna theory
is further supported by measures of constructs such as life satisfaction, verbal
aggressiveness and gender.
Test-retest
reliability of the VPI should be conducted. Such testing will indicate the
potential usefulness of the instrument for both group experimental work and
assessment of individuals. Also, to facilitate practical utilization, alternate
forms of the scale can be developed to remove testing effects among subjects.
For further testing of the validity of the instrument, the VPI can be correlated
with additional measures of convergent validity. For instance, the rajas
subscale, which in this investigation was evaluated for only one measure of
convergent validity, can be correlated with a standardized measure of stress,
and tamas can be correlated with a standardized measure of depression.
Future statistical analysis of the VPI should evaluate floor and ceiling effects
of the items and subscales. In addition, the instrument should be tested on a
wide variety of populations, in order to establish norms for the subscales and
also to determine the transcultural value of the VPI.
Currently
the VPI is being used in a group experiment on the effects of chanting the maha
mantra, which is a mantra predicted in the Vedas (Prabhupada, 1976) to
increase sattva and decrease rajas and tamas. There are
many potential applications of the instrument, in group experimentation as well
as individual analysis. In the Vedic tradition guna theory is the basis
for vocational guidance (Dasgupta, 1961), and the VPI could be applied,
experimentally at first, for that purpose. To do this, the evaluators would need
to become acquainted with guna theory as it applies to career
orientation. In individual assessment, the VPI can be used as a measure of
change. For instance, an intervention for a client with a depressive disorder
would be expected to result in an increase in the sattva subscale score
and a decrease in the tamas subscale score, and a client with a
stress-related disorder would be expected to decrease the rajas subscale
score and increase the sattva subscale score if therapy is successful.
References
Bloom, M., Fischer,
J., & Orme, J. G. (1995). Evaluating
practice: Guidelines for the accountable
professional, (2nd ed). Needham Heights, Massachussets: Allyn and Bacon.
Das, R.C. (1991). Standardization of the Gita Inventory of
Personality. Journal of Indian Psychology, 9, 47-55.
Dasgupta, S. (1961). A History of Indian Philosophy. Great
Britain: Cambridge University Press.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985).
The satisfaction with life scale. Journal of Personality Assessment, 49,
71-75.
Faul, A. C., & Hudson, W. (1997). The Index of Drug
Involvement: A partial validation. Social Work, 42 (6), 565-572.
Gergen, K.J., Gulerce, A., Lock, A., & Misra, G. (1996).
Psychological science in cultural context. American Psychologist, 51 (5),
496-503.
Goswami, S. D. (1977). Readings in Vedic literature: The
tradition speaks for itself. Los Angeles: Bhaktivedanta Book Trust.
Infante, D. A. & Wigley, C. J., III. (1986). Verbal
aggressiveness: An interpersonal model and measure. Communication Monographs,
53, 61-69.
Mohan, V., & Sandhu, S. (1986). Development of scale to measure
sattvic, rajasic and tamasic guna. Journal of Indian Academy of Applied
Psychology, 12, 46-52.
Mohan, V., & Sandhu, S. (1988). Samkhyan tri-guna and
Eyesenck’s dimensions of personality. The Vedic Path, 23-38.
Nunnally, J. C., & Bernstein, I.H. (1994). Psychometric
Theory (3rd Ed.). New York: McGraw-Hill.
Pathak, N. S., Bhatt, I. D., & Sharma, R. (1992). Manual for
classifying personality on tridimensions of gunas- An Indian approach. Indian
Journal of Behaviour, 16 (4), 1-14.
Prabhupada, A. C. B. S. (1976). Srimad-Bhagavatam. Hong
Kong: Bhaktivedanta Book Trust.
Prabhupada, A. C. B. S. (1986). Bhagavad-gita as it is. Los
Angeles: Bhaktivedanta Book Trust.
Rao, P. V. K. & Harigopal, K. (1979). The three gunas and ESP:
An exploratory investigation. Journal
of Indian Psychology, 2 (1), 63-68.
Rosenthal, J. A. (1996). Qualitative descriptors of strength of
association and effect size. Journal of Social Service Research, 21 (4),
37-59.
Singh, R. (1971). An inventory from Mahabharata. Indian Journal
of Psychiatry, 13, 149-161.
Uma, K., Lakshmi, Y.S., & Parameswaran, E.G. (1971).
Construction of a personality inventory based on doctrine of 3 gunas. Research
Bulletin, 6, 49-58.