|Year : 2017 | Volume
| Issue : 3 | Page : 237-240
Multiple regression in fertility and family formation
Anushri Pradip Patil, Naresh Kumar Tyagi
Department of Epidemiology and Biostatistics, KLE Academy of Higher Education and Research, KLE University, Belagavi, Karnataka, India
|Date of Web Publication||5-Sep-2017|
Anushri Pradip Patil
Department of Epidemiology and Biostatistics, KLE Academy of Higher Education and Research, KLE University, Belagavi, Karnataka
Source of Support: None, Conflict of Interest: None
Background: In the present study, attempt has been made to study the determinants of total fertility rate (TFR), by developing the model for TFR so to use TFR determinants for Family Welfare Planning and Programme implementation.
Methodology: Principal component analysis was carried out to study the correlates of TFR of Indian states. Further, regression analysis was carried out to estimate TFR, using relevant determinants by studying its correlates and principal components.
Results: Two principal components were: (i) “Social Status of Women” explaining 58% of variation comprised “infant mortality rate” with correlation coefficient (−0.95), “Percentage of Literacy of Female” (0.94), “Life Expectancy at Birth” (0.93), “Age at Marriage of Women” (−0.90), and (ii) “Fertility Index” with 20% of variation explained, comprised “Employment Status of Female” (0.762) and “Desire of no more child after Two Living Children” (−0.759). The regression model for TFR with the coefficient of determination (0.466), using desire of no more child after Two Living Children was arrived, with utility to enhance health education and communication to achieve ideal family formation behavior. All Indian states were classified by the regression model (TFR = 3.58–0.019* “Desire of no more child after Two Living Children”) within 68% confidence interval except two states Mizoram and Meghalaya.
Conclusions: 'Desire of no more children after two living Children's' has come out the main stay of family formation behavior. Hence, making realize the value of small family size with health education will stabilize the population.
Keywords: Multiple regression analysis, principal component analysis, total fertility rate
|How to cite this article:|
Patil AP, Tyagi NK. Multiple regression in fertility and family formation. Indian J Health Sci Biomed Res 2017;10:237-40
| Introduction|| |
Development is an outcome of three basic events such as quality and magnitude of birth, death and migration of population, affecting the population size, and structure and culture of an area. Study of births and deaths is referred as fertility and mortality, respectively, in population studies. Population science as such is the study of population size, structure, distribution and their consequences.
Total fertility rate (TFR) in India in 1971 was 5.2 and decreased to 2.4 in 2011. India can be divided using TFR in two groups; empowered action group (EAG) states and non-EAG states. In EAG state's TFR for 2011 were Bihar (3.6), Chhattisgarh (2.7), Jharkhand (2.9), Madhya Pradesh (3.1), Odisha (2.2), Rajasthan (3.0), Uttar Pradesh (3.4), and Assam (2.4). Whereas, in two of the best non-EAG states; Andhra Pradesh and Tamil Nadu, the similar figures “TFR” were 1.8 and 1.7, respectively.
Hence, in the present study, attempt has been made to study determinants of TFR, by developing the model for TFR so to use TFR determinants for Family Welfare Planning and Programme implementation.
| Source of Data and Methodology|| |
In this study, the “Age at Marriage of Women,” (AMW) “Infant Mortality Rate” (IMR), “Life Expectancy at Birth,” (LEB) “Desire of no more child after Two Living Children,” “Percentage of Literacy of Female,” and “Employment Status of Female” (ESF) have been used as determinants of “TFR.” The state-wise data were taken from the National Family Health Survey (NFHS) 2005–2006 and from future population of India.
Principal component analysis was carried out to study the correlates of TFR. Thereafter, using these components, regression analysis was carried out to estimate TFR.
The data analysis has been carried out using IBM SPSS Statistics 22.
The study has been carried out from January 2017 to April 2017 as part of M. Sc. Biostatistics Dissertation, using data from NFHS 3rd and from Future Population Survey of India, 2013.
| Results|| |
[Table 1] shows that “Correlation Matrix” explains the structural relationship of TFR with its covariates; “Desire of no more child after Two Living Children”, “Percentage of Literacy of Female” (%LitF), “AMW,” “IMR,” “TFR,” “ESF,” and “LEB.”
The objective variable (TFR) was highly correlated with “Desire of no more child after Two Living Children” with correlation coefficient (0.68) followed by “IMR” (0.0.32), “Percentage of Literacy of Female” (0.321), “AMW” (0.300), “ESF” (0.14), and “LEB” (0.107). Further, “Desire of no more child after Two Living Children” as determinant of TFR did not have any multicollinearity with other covariates. “Percentage of Literacy of Women” has multicollinearity with AMW (0.84), IMR (0.86), and LEB (0.83). The other correlate of TFR, i.e., IMR (0.32) has very high multicollinearity with “Percent Literacy of Women” (0.86); hence, there is a need to find a component containing these two variables to estimate TFR.
[Table 2] shows that “Rotated Component Matrix” reveals two principal components as; “Social Status of a Women” comprising IMR with correlation coefficient (−0.95), “Percentage of Literacy of Female” (0.94), “LEB” (0.93), and “AMW” (−0.90) explaining 58% of total variation. The second principal component, namely, “Fertility Index” comprised “ESF” (0.762) and “Desire of no more child after Two Living Children” (−0.759) should be used as determinants of TFR.
[Figure 1] shows that “Scree Plot of Principal Components” has been given to understand the variation explained by latent variables of determinants of TFR considered in the study.
However, from the two latent variables; “Social Status of a Women” and “Fertility Index,” TFR's determinants “IMR” and “D2+,” respectively, has least multicollinearity (−0.19). Hence, these two variables have been used to construct the deterministic model for TFR.
[Table 3] shows that regression model for TFR using “D2+” reveals that the “Desire of no more child after Two Living Children” was the sole determinant of “TFR.” All the state's TFR has been estimated within the 68% confidence interval (mean + standard deviation), except two states namely Mizoram and Meghalaya with higher values.
|Table 3: Regression model for total fertility rate using “desire of two plus living children of women” (total fertility rate=3.19-0.019* “Desire of two children”)|
Click here to view
| Discussion/Conclusion|| |
The present study has been planned to study the determinants of fertility, as measured using TFR so to plan for ideal family formation behavior. To justify the objectives, principal component analysis was carried out, resulting in two principal components, namely; “Social Status of Women” explaining 58% of variation, and “Fertility Index” with 20% of variation. “Social Status of Women” was comprised “IMR” with correlation coefficient (−0.95), “Percentage of Literacy of Female” (0.94), “LEB” (0.93), “AMW” (−0.90), and “Fertility Index” comprised of “ESF” (0.762) and “Desire of 2+ living children of Women” (−0.759). The similar covariates of fertility along with “Socioeconomic Status” in the posteconomic reform period have been quoted. Where many studies have demonstrated that India's fertility decline has been driven by major fertility decline among illiterate and poor women, through widespread use of female sterilization.
In mother and child health programs and the population policy, the government of India has made much higher budget allocations to improve population and health indicator of EAG states, which significantly contributed to the decline in fertility rates, accelerating the pace of fertility decline, and convergence in national fertility levels.,,
Multivariate regression analysis resulted in “Desire of no more child after two living Children's” as the sole explanatory variable to explain the TFR, making Health Education a main component for achieving the desired results for ideal family formation behavior. However, similar models with varying explanatory variables have been published by several researchers.
Regression Model for TFR was constructed by “Infant Mortality Rate,” “Percentage of Literacy of Female,” “LEB,” “AMW,” “ESF,” “Desire of no more child after Two Living Children” using backward stepwise, maximum likelihood method regression procedure, resulting “Desire of no more child after Two Living Children” as only the explanatory variable, estimating almost all states within 68% of confidence interval except two states (Mizoram and Meghalaya) with higher TFR. Several researchers have used regression models to estimate TFR with similar explanatory variables and varying coefficient of variations. [Figure 2] explained the estimated TFR and residuals from observed values of TFR, following the normal distribution, with non-significant variations.
|Figure 2: Normal P-P plot of regression standardized residual of total fertility rate, has been given to magnify the goodness of fit of the deterministic model|
Click here to view
Policy implications and conclusions
The model will be useful to achieve ideal family norms by monitoring “Desire of no more child after Two Living Children” through “Health Education and Communication.”
- I am thankful to Dr. V. D. Patil, Registrar, KLE University, Belagavi, for permission and administrative help
- I am very much thankful to Dr. N. S. Mahantashetty, Principal, Jawaharlal Nehru Medical College, Belagavi, for her support and permission
- I am indebted to my husband Dr. Abhijeet Shitole, Asst. Professor, Department of Cardiac Anesthesia, Jawaharlal Nehru Medical College, Belagavi for his help throughout the study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Pandey A, Jayachandran, Sathyanarayana KM, Swain P, Arokiasamy, Rao N, et al
. The Story of India's Population, Technical Report on: Levels, Trends, and Determinants of Fertility in India; Population Projection and Expected levels of Achievement for Spacing and Limiting FP Methods to Strategies and Priorities the Programme in India, Policy Unit, USAID, Health Policy Project, NIHFW, New Delhi; 2013.
International Institute for Population Sciences, Ministry of Health and Family Welfare, Government of India. National Family Health Survey-3. Vol. 1. Mumbai: International Institute for Population Sciences, Ministry of Health and Family Welfare, Government of India; 2005-2006.
Population Foundation of India, Population Reference Bureau. The future population of India: A long-range demographic view. New Delhi: Population Foundation of India, Population Reference Bureau; 2007.
James KS, Nair SB. Accelerated decline in fertility in India since the 1980s: Trends among Hindus and Muslims. Econ Polit Wkly 2005;45:375-84.
Visaria P, Visaria L. Demographic transition: Accelerating fertility decline in 1980s. Econ Polit Wkly 1994;29:3281-92.
Kulkarni PM, Alagarajan M. Population growth, fertility and religion in India. Econ Polit Wkly 2005;45:403-11.
Preston SH. Causes and consequences of mortality declines in less developed countries during the twentieth century. In: Easterlin RA, editor. Population and Economic Change in Developing Countries. Chicago, London: The University of Chicago Press; 1980. p. 289-360.
Goli S, Arokiasamy P. Demographic transition in India: An evolutionary interpretation of population and health trends using 'change-point analysis'. PLoS One 2013;8:e76404.
Tyagi NK. Proceeding of 25th
Annual Conference of Indian Association for the Study of Population (IASP). Indian Institute Population Sciences. Mumbai; February, 2002.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3]