Indian Journal of Public Health

: 2021  |  Volume : 65  |  Issue : 2  |  Page : 96--102

Dimension reduction of subjective motivational values toward child gender tool tested in women of reproductive age from a hospital in rural area of Himachal Pradesh, India

Dinesh Kumar1, Shabab Ahmad2, Chirag Goel3, Avi Kumar Bansal4, Shripad Patil5,  
1 Associate Professor, Department of Community Medicine, Dr. Rajendra Prasad Government Medical College, Kangra, Himachal Pradesh, India
2 Assistant Professor, Department of Social Work, School of Social Sciences, Central University of Himachal Pradesh, Dharamsala, Himachal Pradesh, India
3 Scientist-C, Model Rural Health Research Unit, Una, Himachal Pradesh, India
4 Scientist-E, National JALMA Institute of Leprosy and Other Mycobacterial Diseases, Agra, Uttar Pradesh, India
5 Director, National JALMA Institute of Leprosy and Other Mycobacterial Diseases, Agra, Uttar Pradesh, India

Correspondence Address:
Dinesh Kumar
Department of Community Medicine, Dr. Rajendra Prasad Government Medical College, Kangra, Himachal Pradesh


Background: A novel subjective Motivational Value toward Child Gender (MVCG) tool was developed using the theoretical construct of 10 motivational domains described by Shalom H Schwartz. Objective: The study aimed to summarize the pattern of correlations of (MVCG) in women of reproductive age in Himachal Pradesh, India. Methods: A cross-sectional study was conducted from October 2018 to November 2019 among a sample of 355 women. Required data were collected through an interviewer-administered questionnaire. Maximum likelihood exploratory factor analysis (EFA) with oblique rotation was done with Bartlett's test sphericity and Kaiser-Meyer-Olkin test. Results: A total of 28 (53.8%) questions loaded on eight factors explaining maximum variance (68.7%). Reliability analysis of these questions, with high loadings on extracted factors, of the questionnaire, observed with poor Cronbach's alpha of 0.61 and intraclass cluster coefficient (ICC) 0.49. However, selected domains such as tradition, power, achievement, self-direction, and benevolence were observed with a good Cronbach's alpha and ICC. Conclusion: MCVG is novel tool in its kind with well scalable properties in measuring subjective motivational values towards child gender. After EFA, total questions across 10 domains reduced from 52 to 28, across 8 domains, loaded on 8 factors with good reliability and agreement.

How to cite this article:
Kumar D, Ahmad S, Goel C, Bansal AK, Patil S. Dimension reduction of subjective motivational values toward child gender tool tested in women of reproductive age from a hospital in rural area of Himachal Pradesh, India.Indian J Public Health 2021;65:96-102

How to cite this URL:
Kumar D, Ahmad S, Goel C, Bansal AK, Patil S. Dimension reduction of subjective motivational values toward child gender tool tested in women of reproductive age from a hospital in rural area of Himachal Pradesh, India. Indian J Public Health [serial online] 2021 [cited 2021 Aug 2 ];65:96-102
Available from:

Full Text


Male child preference is a predominant reason for the declining sex ratio at birth (SRB) in India and observed to be the area of concern varying across states.[1],[2],[3],[4] Gender bias and resulted sex preference generally arises due to the embedded social values of an individual, these need to be measured for shaping social interventions.[5],[6] Various individual and cultural factors influence subjective values and decision-making processes.[7],[8] Extensive literature review was carried out and no evidence was found wherein application of motivation values was done to the formulation of the questionnaire, therefore, subjective Motivational Value towards Child Gender (MVCG) tool of 52 items across 10 motivational domains based on the theoretical construct by Shalom H Schwartz was developed.[9],[10] The construct has not been tested in the Indian context and not been applied among women to measure values toward the gender of the child. However, it was understood and hypothesized that the mentioned motivational domains can explain a lot about the pattern of values among women towards the gender of the child.[11],[12]

These motivational values are described in 10 domains; (a) Power: Social status and prestige, control or dominance over people and resources; (b) Achievement: Personal success through demonstrating competence according to social standards; (c) Hedonism: Pleasure and sensuous gratification for oneself; (d) Stimulation: Excitement, novelty, and challenge in life; (e) Self-direction: Independent thought and action-choosing, creating, exploring; (f) Universalism: Understanding, appreciation, tolerance, and protection for the welfare of all people and for nature; (g) Benevolence: Preservation and enhancement of the welfare of people with whom one is in frequent personal contact; (h) Tradition: Respect, commitment, and acceptance of the customs and ideas that traditional culture or religion provide the self; (i) Conformity: Restraint of actions, inclinations, and impulses likely to upset or harm others and violate social expectations or norms; and (j) Security: Safety, harmony, and stability of society, of relationships, and of self.

All domains reflect four major areas; self-enhancement by power, achievement, and hedonism; openness to change by stimulation and self-direction; self-transcendence by universalism and benevolence; and conservation by security, tradition, and conformity.[9] In this context, the present study aimed to extract relatively independent coherent subsets consistent with the theoretical construct. Specific objectives of the study were to identify the pattern of correlations and domain structure of a subjective MVCG tool in reproductive-age women in the rural area of Himachal Pradesh, India.

 Materials and Methods

This was a cross-sectional observational study carried out in villages of Haroli health block of district Una, Himachal Pradesh from October 2018 to November 2019. Women within the age group from 18 to 35 years, with last children of <1 year of age, resident of the rural area, with formal education, willing to spare alone time for the interview, and with informed consent were selected for the study. Women living with issues related to mental health and admitted or to be admitted to hospital with an illness were excluded from the study. The early phase of reproductive age (18–35 years) was chosen as participants engage and contribute significantly in household activities and can share their experiences towards childbirth without any recall bias. Convenience sampling was followed to recruit the subjects. A total of 370 women were recruited from the civil hospital of rural area, attending for health care and a total of 15 women refused to participate. Thus, finally, a sample of 355 that could be studied was considered to be good for factor analysis as guided by Comrey and Lee.[13]

After obtaining informed consent, MVCG tool with 52 questions was administered by the interviewer where the statement was narrated and the respondent was asked to quantify the response on a Likert scale; 1 = completely disagree, 2 = disagree, 3 = agree, 4 = partially agree, and 5 = completely agree. Questions were asked about their approach toward “the event” which was considered childbirth for an interviewed woman.

To develop the questionnaire based on the theoretical construct, a group of a public health expert (2), social science expert (1), and women from local villages (2) was constituted. Exercise for questionnaire formulation was carried out among technical experts of the group. Independently, women from the village were consulted for gathering their opinion about subjective comprehensibility and applicability of questions in contextual settings. Over 3 months, iterative discussions were carried out to discuss and finalize; motivational values, value-specific questions, and method of scoring (quantification) for questions. Firstly, a consensus was built on the utility of motivational values for the assessment of the parental value system toward the gender of a child. Furthermore, the type of questions and its method of scoring was discussed. The value-specific questions were shared with the group to avoid redundancy of responses and restructuring the statement of questions. Finally, in meeting with the women members, finalization of a number and type of questions for each motivational value was done. The questionnaire was translated into Hindi and then back to English and was decided to be interviewer-administered.[10] It was decided to ask questions in the form of statements, that will be measured in the Likert scale from score 1 to 5. It was chosen as questions were of declarative in nature and considered to be more definitive response rather than no and unvarying response. Although, the extent of granularity is debated evidence have showed that 5–7point scale ensures reliability and validity, so recommended often for use.[14] Distribution of responses for each question was also studied and significant difference from normality was assessed with Chi-square test at 5.0% level of significance.

As a dimension reduction process, principal axis factor analysis was done to extract and analyze factors. It was chosen as the factor analysis method takes measurement errors into account that is variance not attributable to the factor. It does not redistribute the variance that is unique to variables producing reliable results. Therefore, estimate communalities with an attempt to eliminate unique and error variance from variables. Kaiser normalization was done to assign equal weight to items and obtain stability across solutions. Thereafter, loadings were rescaled back to their initial size. Data were observed for the issue of missing values and it was not observed. Questions related to Stimulation and Universalism were observed with no correlation and failed to converge, so were dropped from the analysis. Measure for sampling adequacy of the individual question was reflected in the anti-image correlation matrix, where principal diagonal values were observed to be high ranging from 0.80 to 0.90 indicative of good to proceed for factor analysis.

First, factorability was assessed by a correlation matrix of scale questions. For the purpose of the study, inter-question correlations >0.30 considered good to proceed for factor analysis. Second, anti-image matrix was analyzed for the contribution of individual questions and indication toward multicollinearity. Finally, maximum likelihood exploratory factor analysis with oblique rotation was done with Bartlett's test and Kaiser-Meyer-Olkin (KMO) test for sphericity and sampling adequacy respectively. An insignificant Shapiro–Wilk multivariate omnibus test indicated multivariate normality. Since correlation of factors was assumed, oblique rotation was used as it allows nonorthogonality among factors, so as among motivational domains, and improve the interpretability of the solution. Responses were measured on the Likert scale as assumed to be in continuum, as FA produced linear combinations of questions and each linear combination as a factor. Apart from the scree plot, the decision to extract the appropriate number of factors to analyze correlation matrix among questions was done by parallel analysis and questions with factor loadings of >0.45 were considered to be significant and reported. Conventionally, Kaiser rule, factors with eigenvalue of >1 is recommended to be reported, but, the parallel analysis gives a good objective alternative to decide upon the number of factors to be extracted based on iteratively generated percentile eigenvalues from the randomly generated dataset.[15] Parallel analysis is recommended over the Kaiser rule as it reduces the chance of over-identification of factors based on sampling error. Moreover, it identifies numbers of factors before the analysis based on Monte Carlo simulations on randomly generated data. Factors above the 95th percentile generated by the simulations are considered to be beyond chance. Therefore, it improves the decision to extract the number of factors before the analysis.[16] So, FA assessed whether questions of similar nature load on factor and if yes then on how many factors? For the study purpose, factors are considered as the latent variables that are consistent with domains, which are not directly measured, explaining the theoretical construct. Structure and pattern matrix of factors and questions are reported. Internal consistency of the scale is assessed by calculating Cronbach's alpha coefficient. Reliability assessment for domain-specific questions was done using the intraclass cluster coefficient (ICC) for the reproducibility of the tool. The analysis was carried out using R studio (version 3.6.1) for windows by installing “lavaan” package.[17]

Ethical considerations

Women were interviewed in a private room by a scientist after narrating “Participant Information Sheet” and obtaining written informed consent. No personal identifiers were mentioned at any stage during the study and kept strictly confidential with investigators. The study was approved by the Institute Ethics Committee of Dr. Rajendra Prasad Government Medical College (Dr. RPGMC), Kangra, Himachal Pradesh with an identification number 100/2016 dated December 31, 2016.


As all study participants (355) belonged to the rural area and were women of reproductive age ranging from 21 to 35 years (mean: 31.1; standard deviation: ±5.6). About 70% of women were from backward and scheduled caste/tribe class of society, 95.5% were homemakers, and all were educated (primary: 5.6%; secondary: 59.7%; graduate: 20.0%; postgraduate: 14.7%). Women have children ranging from 1 to 4 and about 90% of women have one or two children, whereas 33.2% of women did not have, a male child.

Variation in the average and median score for individual questions was observed. Mean score of questions within domain ranged from 2.8 to 4.8 for Power, 3.4–4.3 for Achievement, 4.0 for Hedonism, 2.5–3.7 for Self-direction, 3.2–4.2 for Benevolence, 2.8–3.3 for Tradition, 1.8–2.3 for Conformity, and 1.6–2.3 for Security [Table 1].{Table 1}

Moreover, the frequency distribution of individual questions observed that statistically there was a significant difference in response categories (P = 0.00). Frequency distribution of responses of individual questions observed that overall majority (~90%–95.0%) of responses was piled up in the center of scale (2–4). The variability in response was observed across domains like conformity, security, and self-direction was different than other domains [Table 2].{Table 2}

Parallel analysis was done for randomly generated percentile eigenvalues and a total of 8 factors were decided to be extracted and explained further. Factor solution found to be adequate as Bartlett's test observed to be significant (8118.8.7; P: 0.000) with KMO value of 0.80 indicating matrix is different from the identity matrix and good sample size for analysis. Total 8 factors were extracted 64.7% of variance; maximum by factor 1 (18.3%) and 2 (12.6%). Structure and pattern matrix for nonorthogonal rotation are presented, mentioning the extracted factor in parenthesis. Since domain-specific questions were assumed to be correlated, nonorthogonal rotation improved the factor loadings. Interpretation of factor loadings was done by interpreting pattern matrix as it reflects loading controlled for intercorrelation among factors. Analyzing the pattern matrix, three questions of self-direction and two questions of power domain loaded on the first two factors. Two more questions of power, two of Conformity, and one of Security domain loaded on factors three and four. A total of 28 (53.8%) questions loaded on eight factors and were explaining maximum variance. After omission of questions of Stimulation and Universalism domain due to no correlation in the matrix, 10 questions failed to load with the cutoff of >0.45 and were further omitted from the questionnaire [Figure 1]. It reduced from 52 questions to 28 questions and reliability analysis of 28 questions, with high loadings on extracted factors, of the questionnaire, observed Cronbach's alpha of 0.61 and ICC 0.49, considered to be not satisfactory. However, questions of selected domains like tradition, power, achievement, self-direction, and benevolence have a good level of Cronbach's alpha and ICC [Table 3].{Figure 1}{Table 3}


Usually, in a dimension reduction process, observed variables/questions reduced to latent (unmeasured) factors, so questions of similar nature are packed together on latent factors. In the current study, questions of similar nature were framed for preidentified domains based on the theoretical construct. In the current study, FA identified factors loaded with questions of similar nature and found to be equivalent to preidentified domains of chosen theoretical construct. So, an inference can be made that questions empirically measure the hypothesized theoretical construct of motivational values. FA was found to be suitable method to summarize the pattern based on the study objective. Other methods of analysis to explain structure among variables like principal component analysis and structured equation modeling are there but they are more suited for exploratory in former and assessing pathway in the later method.[15]

After FA, total of 52 questions across 10 domains reduced to 28 questions across 8 domains with an improved reliability and good agreement, as questions of stimulation and universalism domain were not included due to the absence of correlation in the matrix. In a chosen dimension reduction process, there is no agreed upon loading cut-off for inclusion of question but as a rule of thumb, only loadings of 0.32 and above are generally reported.[18] However, evidence suggested that loading below 0.45 could be considered fair whereas <0.32 as poor, hence the cut-off of 0.45 was opted for the current study.[14] Improvement with better reliability and agreement of questionnaire was observed when the number of questions reduced from 28 to 20 of domains; tradition (3), power (6), achievement (2), self-direction (5), and benevolence (4). These circumscribed questions covered chosen theoretical areas of motivation like self-enhancement (a subset of power and achievement), openness to change (self-direction), self-transcendence (benevolence), and conservation (tradition). Evidence suggests that these broad areas affect the individual decision-making processes that are further influenced by socio-cultural milieu by pervasive social values.[19]

This is a novel attempt wherein the motivational values were measured among women for the gender of the child. Value shapes normative behavior influenced by socio-cultural environment and cues to action. Although evidence suggested that role and possibility of measuring value exist and studied for their role in areas like; societal pressure among women with body weight and shape, carrier value among medical students, positive valuation of life scale, and masculine identity of males.[20],[21],[22],[23] Current study demonstrated that the motivational values can be measured with its distribution and degree toward the female gender. Value plays an underlying decision-making process of women which can influence the pregnancy outcome. Hence, measuring value has potential translational value for country-wide efforts to reduce female feticide and improve SRB. It helps to reorganize communication campaigns with domain-specific focus of motivation values. Once measured, its area/domain wise association with expected variables such as education, ethnicity, economic capacity, and tenacity of family/community can be further studied, therefore helps to formulate and evaluate targeted campaigns of improving SRB.

The study should be viewed with limitations like whether the difference in variables like birth order, gender of children, family type, and other socio-demographic factors among women who refused and who participated in the study? Since the refusal fraction was low (4.0%) but we assume that it might not have greatly affected the dimension reduction process. Another limitation could be an element of questionnaire fatigue which might have affected the responses toward questions. However, variability of responses for different domain-specific questions suggests it might not be a problem [Table 2].

Moreover, the frequency distribution of responses to questions suggested variation in response allowing the respondent to give the differentiating response.[14]

It can also be argued that the application of elicitation interview could have given better in-depth insight into the experiences of the participant but is recommended for the determination of a specific singular experience. It is a detailed and time-consuming technique as it follows the number of iterative phases. Since, for the current study, themes/domains were already specified in the theoretical construct and the elicitation interview technique would be useful as data-driven themes/domains.[24]

Finally, the nonnormal distribution of certain questions could affect the analysis but the frequency distribution of responses to domain-specific questions showed nonnormality is not much of an issue. Response pattern indicated that majority of questions across domains observed with the response for scale value of 2–4 rather than extreme ones (1 and 5) [Table 2]. As multivariate normality is assumed for statistical inference to determine the factors, so violation of univariate normality increases the likelihood for violation of multivariate normality. It is said that even with the presence of univariate normality the assumption of multivariate normality can be violated.[25] Hence, in the current study, an insignificant value of the Shapiro–Wilk multivariate omnibus test indicated multivariate normality.


To conclude, it is feasible to measure values toward childbirth and can serve as a good tool to assess the progress of various social interventions to improve SRB in India. The study tool successfully measured motivational values and its shorter version (MCVG-28) found to be more reliable than the MCVG-52. We recommend application of MCVG-28 tool in field settings to assess the motivational values of women for better consistency and coherence.

Financial support and sponsorship

We thank National JALMA Institute of Leprosy and Other Mycobacterial Diseases (NIJ and OMD), Agra, Uttar Pradesh, India.

Conflicts of interest

There are no conflicts of interest.


1Manchanda S, Saikia B, Gupta N, Chowdhary S, Puliyel JM. Sex ratio at birth in India, its relation to birth order, sex of previous children and use of indigenous medicine. PLoS One 2011;6:e20097.
2Mohanty SK, Rajbhar M. Fertility transition and adverse child sex ratio in districts of India. J Biosoc Sci 2014;46:753-71.
3Halder A. Skewed sex ratio in India. Indian J Med Res 2006;124:583-4.
4Seth S. Skewed sex ratio at birth in India. J Biosoc Sci 2010;42:83-97.
5Bhattacharjya H, Baidya S. Determinants of child sex ratio in West and South Districts of Tripura, India. Indian J Public Health 2017;61:112-7.
6Bylund CL. Mothers' involvement in decision making during the birthing process: A quantitative analysis of women's online birth stories. Health Commun 2005;18:23-39.
7Wagner MG. Diarrheal disease control and family planning: The status of women and community participation. Turk J Pediatr 1986;28:141-3.
8Hodgson EJ, Collier C, Hayes L, Curry LA, Fraenkel L. Family planning and contraceptive decision-making by economically disadvantaged, African-American women. Contraception 2013;88:289-96.
9Schwartz SH. Are there universal aspects in the structure and content of human values? J Soc Issues 1994;50:19-45.
10Kumar 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.
11Latifnejad Roudsari R, Zakerihamidi M, Merghati Khoei E. Socio-cultural beliefs, values and traditions regarding women's preferred mode of birth in the North of Iran. Int J Community Based Nurs Midwifery 2015;3:165-76.
12Parks L, Guay RP. Personality, values, and motivation. Pers Individ Dif 2009;47:675-84.
13Comery AL, Lee HB. A First Course in Factor Analysis. 2nd ed. Hillsdale, NJ: Lawerence Erlbaum Associates; 1992.
14Pearse N. Deciding on the scale granularity of response categories of likert type scales: The case of a 21-point scale. Electron J Bus Res Methods 2011;9:159-71.
15Patil Vivek H, Surendra N. Singh, Sanjay Mishra, and D. Todd Donavan (2017). Parallel Analysis Engine to Aid in Determining Number of Factors to Retain Using R [Computer Software]. Available from: [Last accessed on 2021 Jan 08].
16Wood ND, Akloubou Gnonhosou DC, Bowling J. Combining parallel and exploratory factor analysis in identifying relationship scales in secondary data. Marriage Fam Rev 2015;51:385-95.
17R Core Team. Language and Environment for Statistical Computing. Foundation for Statistical Computing, Vienna, Austria; 2013. Available from: [Last accessed on 2020 Jan 12].
18Tabachnick BG, Fidell LS. Using Multivariate Statistics. 5th ed. Boston: Pearson Education Inc.; 2007.
19Espinoza MA, Manca A, Claxton K, Sculpher M. Social value and individual choice: The value of a choice-based decision-making process in a collectively funded health system. Health Econ 2018;27:e28-40.
20Delaney ME, O'Keefe LD, Skene KM. Development of a sociocultural measure of young women's experiences with body weight and shape. J Pers Assess 1997;69:63-80.
21Murdoch MM, Kressin N, Fortier L, Giuffre PA, Oswald L. Evaluating the psychometric properties of a scale to measure medical students' career-related values. Acad Med 2001;76:157-65.
22Araújo L, Ribeiro O, Teixeira L, Azevedo MJ, Jopp DS, Rott C, et al. Measuring attachment to life in old age: The Portuguese version of the Positive Valuation of Life Scale (Positive VOL). Qual Life Res 2015;24:2385-9.
23Levant RF, Hall RJ, Rankin TJ. Male Role Norms Inventory-Short Form (MRNI-SF): Development, confirmatory factor analytic investigation of structure, and measurement invariance across gender. J Couns Psychol 2013;60:228-38.
24Hogan T, Hinrichs U, Hornecker E. The elicitation interview technique: Capturing people's experiences of data representations. IEEE Trans Vis Comput Graph 2016;22:2579-93.
25Zygmont C, Smith MR. Robust factor analysis in the presence of normality violations, missing data, and outliers: Empirical questions and possible solutions. Quant Methods Psychol 2014;10:40-55.