• Users Online: 383
  • Print this page
  • Email this page


 
 
Table of Contents
ORIGINAL ARTICLE
Year : 2023  |  Volume : 19  |  Issue : 1  |  Page : 19-24

Path analysis of measured variables of subjective motivational values toward child gender tool tested in women of reproductive age in a Northern State of India


Department of Community Medicine, Dr. Rajendra Prasad Government Medical College, Kangra, Himachal Pradesh, India

Date of Submission16-Mar-2022
Date of Acceptance08-Feb-2023
Date of Web Publication28-Mar-2023

Correspondence Address:
Dinesh Kumar
Department of Community Medicine, Dr. Rajendra Prasad Government Medical College, Kangra 176002, Himachal Pradesh
India
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/AMJM.AMJM_15_22

Rights and Permissions
  Abstract 

Background: Structural equation modeling was done for path analysis among domain scores of motivational values toward child gender (MVCG) questionnaire. It was used to assess consistency between observed and estimated covariance matrix as theoretically assumed. The objective of the study is to identify paths to influence the power domain among selected women of Himachal Pradesh. Materials and Methods: A total of 655 women were interviewed and information was collected using the MVCG questionnaire. Responses to 28 questions of eight domains were used for path analysis. The model-building process was started from an assumed theoretical construct and referred to as Model-1 and changes in the model were done gradually by addition of pathway(s) within the construct. The addition of pathways was carried out to observe a good model fit. Results: Fit indices from Model-1 to -4 changed across models as the user-defined χ2/df ratio decreased from 56.8 to 5.8 indicating acceptable fit. Change across models for normed-fit index, comparative fit index, and Tucker Lewis index showed that a large improvement from 0.74 to 0.94, 0.73 to 0.95, and from 0.56 to 0.90 indicating a good fit, respectively. Likewise, root mean square error of approximation reduced from 0.29 to 0.08 and standardized root mean square from 0.07 to 0.03 from Model-1 to -4, indicative of a close fit. In final model (Model-3), significantly, tradition, hedonism, and security had negative whereas, benevolence, conformity, and self-direction had positive influence on power. Conclusion: Path analysis observed the influence of motivational values on power through self-direction, achievement, and tradition. They were observed to be the main drivers toward decision-making process toward gender of child.

Keywords: Gender, motivation, social values, structural equation modeling


How to cite this article:
Kumar D. Path analysis of measured variables of subjective motivational values toward child gender tool tested in women of reproductive age in a Northern State of India. Amrita J Med 2023;19:19-24

How to cite this URL:
Kumar D. Path analysis of measured variables of subjective motivational values toward child gender tool tested in women of reproductive age in a Northern State of India. Amrita J Med [serial online] 2023 [cited 2023 Jun 4];19:19-24. Available from: https://ajmonline.org.in/text.asp?2023/19/1/19/372703




  Introduction Top


Path analysis was carried out with structural equation modeling (SEM) for domains of motivational values toward child gender (MVCG) tool. It is a novel tool as no such measure was developed to assess the motivational values among participants.[1] It was prepared to assess the motivational values of women toward the gender of their girl child using a theoretical construct proposed by Schwartz.[2],[3] As an internal validation exercise, dimension reduction exercise was done by factor analysis of 52 questions over 10 domains. It resulted in a reduction to 28 questions loaded well over eight domains. Maximum likelihood exploratory factor analysis was carried, in which questions with significant factor loadings (>0.45) were included. These domains explained maximum variance (68.7%) with Cronbach’s α of 0.61 and intraclass cluster coefficient of 0.49. Domain specific questions are loaded together on each factor, so domains and factors are considered synonymous for the purpose of study. The distribution of questions across domains is 6 for power (social status and prestige, control or dominance over people and resources), 2 for achievement (personal success through demonstrating competence according to social standards), 2 for hedonism (pleasure and sensuous gratification for oneself), 5 for self-direction (independent thought and action-choosing, creating, exploring), 4 for benevolence (preservation and enhancement of the welfare of people with whom one is in frequent personal contact), 3 for tradition (respect, commitment, and acceptance of the customs and ideas that traditional culture or religion provide the self), 3 for conformity (restraint of actions, inclinations, and impulses likely to upset or harm others and violate social expectations or norms), and 3 for security (safety, harmony, and stability of society, of relationships, and of self).[1] All domain-specific questions were measured over the Likert scale ranging from 1 to 5 and were averaged for each domain and the scores were considered to be continuous.[1],[3]

So, scores for eight domains were used for path analysis where influence of value domains over each other was assumed and a recursive (unidirectional path) basic model was prepared. We intend to study what and how values influence power domain and assumptions were made for the influence of domains on value for power. Benevolence, hedonism, self-direction, and achievement assumed to have positive (solid arrows) whereas tradition, conformity, and security negative influence (dotted arrows) on power. Simultaneously, interinfluence of values was also assumed like positive influence on self-direction by achievement, hedonism, benevolence, and conformity. Also, tradition was also assumed to be positively influenced by hedonism, benevolence, conformity, and security. Since all are measured variables but for path analysis, domains such as power, self-direction, and tradition were considered as dependent variables and their influence variables as independent variables (IVs) [Figure 1]. The theoretical construct is divided into four broad areas representing domains; self-enhancement (power and achievement), openness to change (hedonism and self-direction), self-transcendence (benevolence), and conservation (security, tradition, and conformity).[1]
Figure 1: Basic model (Model-1) for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view


SEM was considered for path analysis because not only it tests out multiple regressions to test multiple relationships at once, but also the direction of paths/relationships across domains. In SEM, relationships were free from measurement error as they are estimated and removed, which allows modifications of the basic model with an attempt to find out the best-fitting model. Additionally, it helps to examine complex and multidirectional relationships.[4] Therefore, in current study, an SEM technique was used to assess consistency between model observed and estimated covariance matrix. The objective of study was to identify the paths of influence of various motivational values on power assessed with the MVCG tool among selected women of rural areas of Himachal Pradesh, India.


  Materials and Methods Top


It was a cross-sectional observational study carried out in villages of Haroli health block of district Una, Himachal Pradesh from October 2018 to 2020. A total of 655 women were interviewed; from that, information was collected using the MVCG tool and responses of 28 questions of eight domains were used for path analysis. Study participants were women of 18–35 years of age having last child of less than one year of age at the time of survey, and usual resident of village. Women with mental health issues, speech and hearing impairment, suffering from illness, and not willing to spare time for interview were excluded from the study. Women of early reproductive age (18–35 years) were selected for study as they are active in their social role and can respond to MVCG tool without any recall bias. Sample size was not estimated for SEM, as it is found that with reasonable normality assumption maximum likelihood (ML) method performs well with sample size of more than 500.[4] Therefore, we assume that a sample of 655 in current study is reasonable to perform SEM.

First villages were selected randomly and then participants were selected from all sections of selected village. The tool was administered by a trained interviewer after obtaining informed consent and ensuring the privacy of women. Respondents were asked to rate their satisfaction over the Likert scale as 1 = completely disagree, 2 = disagree, 3 = agree, 4 = partially agree, and 5 = completely agree. Questions were summated for their respective domains and average score per domain was calculated. As the number of questions per domain varied so the scores were translated to a standard scale of 100 (total score/maximum score × 100). Therefore, domain scores were considered as measured continuous variables and SEM was applied to assess a set of relationships among continuous variable scores—measured variables—of domains of MVCG tool. It was a full SEM model with assessment for a causal relationship between latent variables (power, tradition, and self-direction).

Initially, MVCG tool had 10 domains with 52 questions which were reduced to 28 questions over eight domains with better reliability and internal consistency. Distribution of questions over domains included power (6), achievement (2), hedonism (2), self-direction (6), benevolence (4), tradition (3), conformity (3), and security (3).[1] For SEM, each domain was considered as a measured variable and data were checked for identifiability as there are eight variables with 36 data points [8 * (8 + 1)/2] and 18 parameters to be estimated (15 regression coefficients, and three variances). Proposed model was overidentified, as the number of data points was more than that of the parameters to be estimated and path analysis was proceeded to further estimate the model.

Data were assessed for missing values, multivariate normality, and outliers. Frequency distribution of variables was studied to look into missing values. Linearity was assessed using scatter plots among pairs of measured variables. The absence of multicollinearity and singularity was inspected from the determinant of covariance matrix. Since data were devoid of above-mentioned issues and SEM analysis proceeded with model testing and stepwise building. The model-building process was started from an assumed theoretical construct and referred to as Model-1 and changes in the model were done gradually by addition of pathways within the construct. It was carried on and decision to build next model was based on fit indices values till all indices indicated a good fit. An ML technique was used for the estimation of parameters of all models. It is a one normal theory estimation technique that has the ability to simultaneously provide estimation of model parameters. The first fit index was ratio of chi-square (χ2) to a degree of freedom (df) and lowest was chosen as an indicator of a good fit. The χ2 test statistic is sensitive to sample size and is an absolute fit index with assumption of multivariate normality. The ratio of χ2 to df is considered to be a parsimonious and standalone fit index. Model consistency is considered to be better with small value of ratio and ratio gives information about fit between data and model. Other sets of indices are normed-fit index (NFI), comparative fit index (CFI), and Tucker Lewis index (TLI); of which high values represent a better fitting model. For these indices, a cutoff of ≥0.90 represented a good fitting model. CFI is an incremental fit index calculated from χ2 and df. It is a corrected version of non-centrality index and tests the extent to which model is superior to an alternative model based on covariance matrix. TLI also is an incremental fit index and is calculated from χ2: df. It is also called as non-normed fit index (NNFI) and developed against the disadvantage of NFI which is affected by small sample size. Apart from these, we considered root mean square error of approximation (RMSEA) and standardized root mean square (SRMR) with lowest values to represent good fit. RMSEA measures difference between observed and hypothesized covariance matrix per df representing model. Whereas, SRMR is an index of average standardized residuals of observed and hypothesized covariance matrix. Other fit indices such as Akaike and Bayesian information criteria (AIC and BIC) are also reported where lowest values were considered to be an indication of a good fitting model. Both AIC and BIC derived from likelihood function of model and resulting maximum likelihood estimate. AIC is derived by minimizing the divergence between true and estimated predictive distributions. Both AIC and BIC asymptotically select model with smallest population minimum discrepancy function value regardless of nested and non-nested model.[5],[6] Information criteria and analysis of variance (ANOVA) test was done to assess the significant difference between the models. Statistically insignificant difference between two recent models was considered to stop SEM and a more parsimonious model was adopted as final one.[6] Data were entered in Excel and imported to R studio (version 1.3.959) for analysis using “lavaan” and “semPlot” packages.[7]

Participants were interviewed in a private room after introduction of “participant information sheet.” Personal identifiers were kept confidential and not disclosed to anyone at any stage of study. Participants were free to withdraw themselves from the study without giving any explanation. The study was approved by the Institute Ethics Committee of Dr. Rajendra Prasad Government Medical College (Dr. RPGMC), Kangra, Himachal Pradesh, India with identification number 100/2016 dated December 31, 2016.


  Results Top


All variables were measured one and their dependence as per the theoretical construct was checked using SEM. Firstly, the theoretical construct was analyzed in Model-1, where standardized estimate showed a significant negative influence of tradition (–0.19, 0.00), security (–0.19, 0.00), and hedonism (–0.08, 0.01) on power. Achievement showed insignificant negative influence on power (–0.02, 0.48) whereas, domains found with a significant positive influence on power were self-direction (0.29, 0.00), benevolence (0.12, 0.00), and conformity (0.12, 0.00). Simultaneously, influence of domains was assessed on self-direction where benevolence and hedonism had insignificant influence, achievement (0.31, 0.00) had positive influence, and conformity (–0.28, 0.00) had negative and statistically significant influence. Apart from power and self-direction, influence on tradition was also observed where only benevolence (0.35, 0.00) and conformity (0.09, 0.00) were observed with a statistical significance. Security and hedonism were observed to be insignificantly influencing over tradition [Figure S1]; [Table S1]. Fit measures of Model-1 indicated a poor fit; therefore, this model was adjusted further by changing the path of theoretical construct [Table 1].
Figure S1: SEM for basic theoretical construct (Model-1) for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view
Table S1: Parameter estimates and variance of basic and modified models for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view
Table 1: Fit indices of SEM for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view


Furthermore, Model-2 was modified with removal of influence over power, self-direction, and tradition. The influence of achievement was removed from power, along with removal of benevolence and conformity influence on self-direction, and of hedonism on tradition. Effect on all domains on power and tradition remained significant except for self-direction where hedonism showed insignificant influence. Parameter estimates of IVs over power did not change, whereas, influence of achievement over self-direction reduced (0.31–0.25) but stayed statistically significant. As the effect of hedonism on self-direction also reduced and stayed insignificant, but the addition of influence of security was negative and statistically insignificant (–0.36, 0.00). Effect of benevolence (0.35 to 0.36) and conformity (0.09 to 0.10) on tradition increased slightly as compared to Model-1 and was significant [Figure S2]; [Table S1].
Figure S2: SEM for modified theoretical construct (Model-2) for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view


In Model-3, achievement domain was considered as DV along with power, self-direction, and tradition along with independent effect of hedonism. It observed with a strong and significant effect (0.78, 0.00). In Model-3, the influence of hedonism on self-direction was omitted. Like on power and tradition, the effect of IVs remained the same in degree and significance as compared to Model-2 [Figure S3]; [Table S1]. In Model-4, only the influence of benevolence on self-direction was omitted and rest was kept same as that of Model-3. The degree and significance of IVs were same on power, self-direction, achievement, and tradition [Figure S4]; [Table S1]. Overall, across all the models, direction, degree, and significance of IVs on DVs remained same.
Figure S3: SEM for modified theoretical construct (Model-3) for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view
Figure S4: SEM for modified theoretical construct (Model-4) for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view


As compare to Model-1, the Model-2 was observed with improved fit indices but were still out of acceptable limits [Table 1]. Therefore, a further change in influence path was done in Model-3, where influence of Hedonism was assumed to be direct on power. Hence, the influence of hedonism on power was negative and insignificant (–0.06, 0.36), but self-direction (0.44, 0.00) and achievement (0.14, 0.03) had a significant positive influence on power [Figure S3]. Here, degree, direction, and significance of the mentioned domains on tradition were similar to the previous model. Compared to previous models, fit indices of Model-3 improved further closing toward the acceptable ranges [Table 1].

Thereafter, in Model-4, an additional path was assumed, wherein self-direction was considered to influence tradition as well and observed with a negative significant influence (–0.12, 0.04). In comparison to earlier models, the inclusion of this path changed the influence of security (–0.22, 0.00), conformity (0.22, 0.00), and benevolence (0.26, 0.00) on tradition, whereas, the influence of self-direction, achievement, and hedonism on power remained same. Effect of power on tradition changed from –0.36 in Model-3 to –0.32 in Model-4 but stayed significant [Figure S4]. Model-4 was observed with better fit indices scoping an additional change in path of modelling procedure [Table 1]. Lastly, in Model-5, conformity was allowed to influence power and observed with a positive significant (0.16, 0.00) effect. It changed the influence of self-direction, which was positive and significant (0.48, 0.00) on power; however, the influence of achievement (0.10, 0.12) and hedonism (–0.02, 0.80) on power became insignificant. In this model, the effect of power on tradition remained similar in degree, direction, and significance as compared to Model-4 [Figure S5]. Fit indices improved again and were found to be in an acceptable range [Table 1].

Comparison of fit indices from Model-1 to -3 was observed with the change in their values across models. The user-defined χ2 statistic changed from significant to insignificant with χ2/df ratio, decreasing from 56.8 to 5.8 indicating an acceptable fit. Change across models for NFI, CFI, and TFI showed a large improvement, respectively, from 0.74 to 0.94, 0.73 to 0.95, and from 0.56 to 0.90 indicating good fit. Likewise, RMSEA reduced from 0.29 to 0.08 and SRMSR from 0.07 to 0.03 from Model-1 to -3, which is indicative of a close fit. However, AIC and BIC showed an increase of about 5000 points [Table 1]. Moreover, ANOVA found that each model was significantly different than the previous model, except from Model-3 to -4, which allowed Model-3 as an appropriate model for existing dataset [Table 2].
Table 2: Analysis of variance (ANOVA) for SEM for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view


In a final chosen model (Model-3), a statistically significant influence on power, self-direction, achievement, and tradition was observed. Power had positive influence of self-direction, benevolence, and conformity, whereas negative influence of tradition, security, and hedonism was observed. Achievement exerted positive influence and security exerted a negative influence on self-direction. Benevolence and conformity had positive influence on tradition, and independently hedonism had positive influence on achievement [Figure 2].
Figure 2: Final model for ascertaining the influence of motivational values over each other toward girl child among women of reproductive age in a rural area of northern state of India

Click here to view



  Discussion Top


SEM of collected information from women about motivational values toward gender of girl child was observed with the modification of paths in a basic theoretical model. Intuitively, in current analysis, values such as power, self-direction, achievement, and tradition expected to be main drivers for decision-making. It indicates that women with sense of power, self-direction, and achievement expected to be empowered. Theoretically, power and achievement reflects self-enhancement, self-direction mentions openness to change, and tradition indicates conservation. Path analysis was carried out where influence on power along with intradomain influence was observed. In an assumed and identified model, the influence of power on other values was not expected, whereas it was considered as a DV.

Analysis observed benevolence and hedonism to have operated though its positive influence on tradition, which in turn had negative influence in power. However, benevolence and hedonism also had direct positive influence on power instead through tradition. Self-direction had direct positive influence over power but was found as a negative intermediatory to power for security and positive for achievement. Despite self-direction, security directly exerted a negative influence on power as well. Hedonism had positive influence over achievement which in turn operated through self-direction on power. However, hedonism was directly observed with negative influence on power [Figure 2]. It is expected that being fearful (security), pleasureful (hedonism) and tendency to follow laid customs (tradition) reduce decision making capacity in family (power). Whereas, sense of achievement and self-direction along with being conforming and benevolent do improve decision making capacity of women in family.

An application of SEM is often for tools to assess multiple regressions of variables in a single model. Adjustment for relationships between variables is usually done to assess changes in regression coefficients and model fit.[8],[9],[10] In current study, question scores were summed over pre-identified domains and these were further considered as measured variables for path analysis using SEM. Model-based on a theoretical construct (Model-1) was identified to estimate parameters. Current study observed with model modification by changing the paths four times to observe better fit statistics. ANOVA observed a significant improvement from previous to next built-in model with improved fit indices.

Study results could have possible limitations that require discussion for better interpretation. Firstly, although SEM is considered for path analysis hinting a causal relationship between domains, whereas results of SEM required to be considered with caution as a causal relationship is theoretical rather statistical, which depends upon study design. Secondly, it is often recommended that the application of SEM should not merely be exploratory but should have a valid research question with a plausibility. Current study carried out an analysis with theoretical assumptions and kept it more confirmatory rather than exploratory. Thirdly, missing data and sample size could affect the size of parameter estimates. Current study was not observed with missing data values, and the issue of sample size might not have affected the results as it was sufficient to measure estimates. Lastly, model modification procedure could have an inflated type-I error rate.[4] However, it might not be present in current study as no parameter was added or deleted; rather, paths were changed. SEM regresses multiple regression equations at once, so many included variables and cases might have an issue with sample size. For Model-1 (7 predictors), minimum sample size of 655 was expected with type-I error probability of 0.007 (0.05/number of predictors) and 0.82 study power.[11] There is a consideration about the choice of estimation techniques, and the current study chose ML as it is good for medium sample size with dependence among domains and errors. So, a sample of 655 can be considered adequate with a ratio of 82:1 for cases to an observed variable.

Fit statistics helps to detect difference between sample and estimated covariance matrix. In the case of a good model, parameter estimates produce estimated matrix close to sample matrix and were evaluated by χ2 test statistic and fit indices. Indices such as NFI and CFI are comparative fit indices, whereas criteria such as AIC and BIC assess fit that includes parsimony adjustment.[4] Therefore, indices compared target and null models, but criteria such as AIC and BIC were regarded as information theory goodness of fit measure. AIC and BIC do not have any suggested cut-off, but a lower value suggests good-fitting parsimonious model. AIC tends to penalize too complex models, and BIC especially tends to pronounce the complex models.[12] In current study, fit statistics showed improvement suggesting a better model fit but criteria such as AIC and BIC showed an increase indicating the model being less parsimonious. The question remaining unanswered is which are the indices to rely upon in SEM for current study purpose? Usually, it is believed that good fitting models produce consistent results on many different indices than few as observed in current study. Usually, AIC and BIC are found to be helpful to use when models are not nested. In current study, the models were nested as they were converted to one another by adding and removing paths of parameters.[4] We believe Model-3 as a final model as most indices reported better fit and also models were nested and consistent to be examined by SEM.

Financial support and sponsorship

National JALMA Institute of Leprosy and Other Mycobacterial Diseases (NJIL&OMD), Agra, Uttar Pradesh, India for funding.

Conflict of interest

There are no conflicts of interest.



 
  References Top

1.
Kumar D, Ahmad S, Goel C, Bansal AK, Patil S Dimension reduction of subjective Motivational Values towards Child Gender (MVCG) tool tested in women of reproductive age in a northern state of India. Indian J Public Health 2021;65:96.  Back to cited text no. 1
    
2.
Schwartz SH Are there universal aspects in the structure and content of human values? J Soc Issues 1994;50:19-45.  Back to cited text no. 2
    
3.
Kumar D, Kalia R, Kaushal A Formulating a tool to measure motivational values of women towards gender of child. J Compr Health 2019;7:50-6.  Back to cited text no. 3
    
4.
Ullman JB Structural equation modeling. In: Tabachnick BG, Fidell LS. Using Multivariate Statistics. 5th ed. Boston: Pearson Education Inc.; 2007.  Back to cited text no. 4
    
5.
Huang PH Asymptotics of AIC, BIC, and RMSEA for model selection in structural equation modeling. Psychometrika 2017;82:407-26.  Back to cited text no. 5
    
6.
Sengul C, Ilker E Comparison of model fit indices used in structural equation modelling under multivariate normality. J Mod Appl Stat Methods 2015;14:152-67.  Back to cited text no. 6
    
7.
R Core Team. A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, 2013.  Back to cited text no. 7
    
8.
Carlsson M, Hamrin E Evaluation of the life satisfaction questionnaire (LSQ) using structural equation modelling (SEM). Qual Life Res 2002;11:415-25.  Back to cited text no. 8
    
9.
Lu Y, Jiang Y, Gu L Using path analysis to investigate the relationships between depression, anxiety, and health-related quality of life among patients with coronary artery disease. Qual Life Res 2019;28:2695-704.  Back to cited text no. 9
    
10.
Nguyen PH, Scott S, Neupane S, Tran LM, Menon P Social, biological, and programmatic factors linking adolescent pregnancy and early childhood undernutrition: A path analysis of India’s 2016 National Family and Health Survey. Lancet Child Adolesc Health 2019;3:463-73.  Back to cited text no. 10
    
11.
Faul F, Erdfelder E, Buchner A, Lang AG Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behav Res Methods 2009;41:1149-60.  Back to cited text no. 11
    
12.
Vrieze SI Model selection and psychological theory: A discussion of the differences between the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). Psychol Methods 2012;17:228-43.  Back to cited text no. 12
    


    Figures

  [Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6]
 
 
    Tables

  [Table 1], [Table 2], [Table 3]



 

Top
 
  Search
 
    Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
    Access Statistics
    Email Alert *
    Add to My List *
* Registration required (free)  

 
  In this article
Abstract
Introduction
Materials and Me...
Results
Discussion
References
Article Figures
Article Tables

 Article Access Statistics
    Viewed436    
    Printed32    
    Emailed0    
    PDF Downloaded41    
    Comments [Add]    

Recommend this journal