Determinants of Child Mortality: Role of Poverty & Policy - Research Proposal Example

Determinants of Child Mortality: Role of Poverty & Policy I. Introduction Reducing child mortality is one of the eight goals of the United Nations for the current millennium. Reducing poverty is also one of the eight goals. It is a typical view that poverty is a fundamental root of child mortality. Is it possible to focus on addressing poverty alone and solve the problem of child mortality in the process? This is the main concern of this inquiry. At the same time, the analysis also factors in the possible role of strategic government policy to reduce child mortality. II. Review of literature The literature on the causes of child mortality has emphasized on the role of poverty as a principal cause of child mortality. As health is linked with mortality rates, the study of Doyle et al. suggest that income affects health and in doing so affect mortality rates among children less than five years old (21). Both for poverty and for the high mortality rates, Save the Children-United Kingdom has held governments accountable for children’s mortality rates (1). Several studies suggest many other factors influence mortality among children under five years old. Tezcan identified the practice of abortion as an important determinant of child mortality (146). Van den Bosch et al. has claimed that child mortality, particularly infant mortality, is also associated with Roman Catholicism (1031). Wagstaff associated the difference in mortality rates across countries with consumption (19). In Bangladesh, Kabir identified socio-economic status as an important correlate of child mortality (292). Ogunjuyigbe associated mortality among under-five children with people’s beliefs, attitudes, and behavioural practices (43). Maitra and Pal argued that not only are health inputs important for reducing child mortality but effective use practices of the health inputs as well. (28). This inquiry is a continuation of various studies on the determinants of child mortality, particularly on the role of poverty and possible role of strategic government policy. III. Methodology A. Design This inquiry uses quantitative methods. The inquiry can be described as quantitative research based on a dichotomization of research (although a mixed method exists). The inquiry uses the tools of statistics and statistical theory to assess whether certain inferences are warranted. In particular, this study will use multiple regressions to assess whether its hypotheses are supported by regression statistics. In a non-experimental setting, real controls are not possible. However, there are quasi-experimental “controls” as well as statistical controls. For other types of research, such as qualitative research, triangulation is sometimes used instead of controls. The principal statistical method used by this inquiry is regression analysis. Regression is a powerful tool that allows simultaneous address of control and identification of effects that can be associated with specific variables. A regression function is able to isolate the effect of a particular variable on the dependent variable. The coefficient associated with the an independent variable in a valid regression function represents the specific effect of the independent variable on the dependent variable with the other variables on a regression function held constant. The regression function that will be used for analysis has a variable representing poverty as one regressor and a variable representing policy regime as the other regressor with mortality of children under five as the dependent variable. The regression is used to investigate the role of poverty and strategic policy in reducing child mortality under five years old. B. Variables, Data, and Hypotheses The sources of data for this work are the statistical tables found in UNICEF’s Progress for Children 6: 1. Data 1: Prevalence of underweighted children for children under five years old for 2000 to 2006 in terms of percent (50-1); 2. Data 2: Average annual rate reduction in percent for prevalence of underweighted children under five for 1990 to 2006 (50-1) in terms of percent; and 3. Data 3: Children mortality figures per 1,000 live births for 2006 (56-7). Data 1 above refers to the UNICEF proxy variable for measuring or assessing progress in eradicating poverty. The UNICEF is monitoring the variable in relation to the realization of Millennium Development Goal 1 which the eradication of extreme poverty and hunger worldwide. On the other hand, data 3 refers to Millennium Development Goal 4, which is the reduction of child mortality. The Millennium Development Goals refer to eight broad development goals that the United Nations signed in September 2000. The UN hopes to realize most of the eight goals by 2015 (Development Goals 4). Following the UNICEF’s use of proxies in its 2007 report, this study assesses the effect of poverty as the independent variable or cause of child mortality that is treated in this inquiry as the dependent variable. As in the UNICEF use of data 1 or prevalence of underweighted children as proxy for poverty, this inquiry adopted the same: prevalence of underweighted children is used in this inquiry as the proxy variable for poverty. In the case of Data 3, however, this study uses the variable represented by the data as the proxy for government strategic policy. Thus, the regression function being studied by this inquiry revolves on the role of poverty and government strategic policy on reducing child mortality among children under five years old. The protocols for constructing hypotheses are provided by the protocols of the statistics themselves. For example, the applicable test of hypothesis on correlations is that the null hypotheses of the correlation coefficient is zero. The alternative hypothesis depends on the tail of the test, whether it is one-tailed or two-tailed test of hypotheses. In the case of an F-test for an ANOVA, the null hypothesis is that all of the means are equal versus the alternative hypothesis that at least two of the means are not equal. Similarly, other statistics have specific protocols for testing hypotheses (Walpole & Myers 466-467). C. Analysis A positive relation between prev0006 and mortality06 would indicate a positive relation between poverty and mortality among children less than five years old. In contrast, a positive relation between red0006 would indicate that government policy to reduce mortality works counterproductively or that it increases the mortality of children below five years old. However, the latter is not an expected result of the regression. IV. Results and discussion A. Univariate Descriptive Statistics Using Excel, a software of Microsoft Corporation, this inquiry obtained the univariate statistics of variable names prev0006, red9006, and mortality06. The results are given in Table 1. The original data set consisted of 196 cases consisting of countries and territories. However, problematic cases, especially those with unreliable or missing figures were excluded, leaving only 98 cases. Thus, the statistical analysis covered by the statistical analyses under this inquiry covered only the cases without missing data. Based on Table 1, the mean mortality figure for the sample is 81.949 mortality per 1,000 live births of children below five years old in 2006. The median mortality rate per 1,000 live births of the sample is 61.50. Following Walpole (24-27), this means that half of the sample has a mortality rate below 61.50 while the other half has a mortality rate above 61.50. Table 1. Descriptive Statistics Statistics prev0006 red9006 mortality06 Mean 18.701 2.629 81.949 Median 18.500 2.200 61.500 Standard Deviation 13.296 3.838 66.046 Sample Variance 176.782 14.731 4362.131 Kurtosis -1.080 3.078 0.042 Skew 0.446 1.178 0.914 Range 46.800 25.200 263.000 Minimum 0.700 -7.000 7.000 Maximum 47.500 18.200 270.000 Sum 1832.700 257.600 8031.000 Count 98.000 98.000 98.000 The sample standard deviation for the variable 66.046 indicates that around 68.26% of the distribution lies in the range 81.949 minus and plus 66.045 (or from 15.903 to 147.995) mortality per 1,000 live-births of children five years old and below, assuming a normal distribution. Following Gujarati (769-771), the value of Kurtosis suggests that the distribution of mortality06 is platykurtic indicating relative flatness, fatness, and shorter tail compared to a normal distribution. This is because a Kurtosis or K less than 3 indicates a platykurtic distribution. A K>3 indicate leptokurtic distribution or relative slimness and longer-tail distribution compared to the normal distribution. It is possible to employ a statistical procedure known as the Jarque-Bera test normality in which the null hypothesis is that the distribution is normal versus the alternative hypothesis that the distribution is not normal. However, we do not do this in this inquiry because inferences that assume a normal distribution is justified when the population is large. Nevertheless, we note that the distribution is platykurtic and that skew statistics of variable mortality06 suggests a positively skewed distribution or that the distribution has a relatively longer right tail compared to a normal distribution which has symmetrical tails. A similar analysis can be done for variables prev0006 and red9006 but this is no longer done because the implications are obvious or at least similar with our earlier discussion. B. Correlation statistics Table 2. Correlation Statistics   Prev0006 Red9006 Mortality06 Prev0006 1 Red9006 -0.4723 1 Mortality06 0.6328 -0.3705 1 Table 2 assess the correlation between mortality06 and prev0006 and mortality06 and red9006. The correlation coefficients indicate that while mortality06 and prev0006 are positively correlated, mortality06, and red9006 are negatively correlated. This is expected because an increase in the prevalence of underweights among children below five would likely increase the mortality among children below five while a trend of consistently increasing reduction in the mortality among children below five is expected to have negative correlation with the most recent data on mortality. C. Regression analysis Table 3 provides the regression statistics with mortality06 as the dependent variable and prev0006 and red9006 as the independent variables. We note that for prev0006, we can reject the null hypothesis that the value of the coefficient is zero to accept the alternative hypothesis that value of the coefficient is positive at a very significant level of 0.001. Following Moore and McCabe (723-24), this means that the probability the null hypothesis is true is very low or less than 0.1%. Unfortunately, Table 3 suggest that we are unable to reject the null hypothesis that the coefficient of red9006 is zero. This indicates that red9006 is probably not affecting mortality06 significantly, if the former is actually affecting the latter. Table 3. Regression Statistics 1 Variable  Coefficients Standard Error t Stat P-value Intercept 31.38 12.03 2.61 0.010527612948817 prev0006 2.93 0.45 6.57 0.000000002634408 red9006 -1.59 1.54 -1.03 0.306182816798414 Table 4 indicates that the variation in the independent variables of our regression can explain around 39.45% of the variation in the dependent variables. This suggests that other variables are possibly at work that affects mortality rates among children below five years old. Table 4. Regression Statistics 2 Regression Statistics Multiple R 0.6380 R Square 0.4070 Adjusted R Square 0.3945 Standard Error 51.3916 Observations 98 Table 5. ANOVA Table for the Regression  Variable df SS MS F Significance F Regression 2 172,222.58 86,111.29 32.60 0.0000000000165682 Residual 95 250,904.16 2,641.10 Total 97 423,126.74       Table 6. Bounds of the Estimate for the Coefficient at 95% Confidence Level Variable  Coefficients Lower 95% Upper 95% Intercept 31.38 7.51 55.26 prev0006 2.93 2.04 3.81 red9006 -1.59 -4.65 1.48 Table 5 indicates the F-statistics of the regression would allow us to reject the null hypothesis that all of the coefficients of the independent variables are simultaneously equal to zero to accept the alternative hypothesis that not all of the coefficients are zero. Table 6 merely indicate the lower and upper bounds of the values of the coefficient at a confidence of 95%. The estimate for the value of the coefficient in Tables 3 and 5 are merely estimated values. There is a 95% confidence level that true value lies between the upper and lower bounds indicated in Table 6. In Table 6, for instance, the true value of the coefficient for red0006 is between -4.65 to +1.48 at 95% confidence level. This indicates that a 95% confidence level, it is possible that the true value of the coefficient is even zero. In contrast, this is not the case for prev0006 whose coefficient can be from 2.04 to 3.81 at a 95% confidence level, indicating that the coefficient is positive whatever its value at that confidence level. V. Conclusion This inquiry investigated the possible role of poverty and strategic policy in reducing child mortality under five years old. As mentioned on page 3-4 of this inquiry, the regression function being studied by this inquiry revolves on the role of poverty and government strategic policy on reducing child mortality among children under five years old. Based on the regression statistics developed under this inquiry, there is empirical support for the role of poverty in exacerbating the mortality rate among children. However, the role of strategic policy in reducing mortality rate among children is not adequately supported because the regression statistics do not provide evidence that we can reject the null hypothesis that the coefficient associated with strategic policy is equal to zero. Works Cited Doyle, Orla, Colm Harmon, and Ian Walker. “Impact of Parental Income and Education on Child Health: Further Evidence for England”. Warwick Economic Research Papers No. 788 (2007): 1-30. Gujarati, Damodar. Basic Econometrics. 3rd ed. New York: McGraw-Hill, 1995. Kabir, Ahmad, Mohammad Islam, Muhammad Ahmed, and Khalique Barhuiya. “Factors Influencing Infant and Child Mortality in Bangladesh”. The Sciences 1.5 (2001): 292-295. Maitra, Pushcar and Sarmistha Pal. “Early Childbirth, Health Inputs and Child Mortality: Recent Evidence from Bangladesh”. IZA DP No. 2841 (2007). Bonn: IZA. Moore, David and George McCabe. Introduction to the Practice of Statistics. 3rd ed. New York: W.H. Freeman and Company, 2000. Ogunjuyigbe, Peter. “Under-Five Mortality in Nigeria: Perception and Attitudes of the Yorubas towards the Existence of ‘Abiku’”. Demographic Research 11.2 (2004): 43-56. Save the Children Fund-UK. The Child Development Index. London: Save the Children. Tezcan, Akile GYrsoy. “Infant Mortality: A Turkish Puzzle”. Health Transition Review 2.2 (1992): 131-149. Tymicki, Krzysztof. Correlates of infant and childhood mortality: A theoretical overview and new evidence from the analysis of longitudinal data of the Bejsce (Poland) parish reconstituton study of the 18th-20th centuries. Demographic Research 20 (2009): 559-594. UNICEF. Progress for Children 6 (December): A World for Children Statistical Review. New York: UNICEF Office of Strategic Information, 2007. United Nations. “Resolution Adopted by the General Assembly, United Nations Millennium declaration”. Fifty-fifth session, agenda item 60 (b). New York: United Nations, 2000. United Nations. The Millennium Development Goals Report 2007. New York: United Nations, 2007. Wagstaff, Adam. Socioeconomic Inequalities in Child Mortality: Comparisons Across Nine Developing Countries. Bulletin of the World Health Organization 78.1 (2000): 19-29. Walpole, Ronald. Introduction to Statistics. 3rd ed. London: Collier Macmillan Publishers, 1982. Walpole, Ronald, and Raymond Myers. Probability and Statistics for Engineers and Scientists. New York: MacMillan Publishing Company, 1989. Van den Bosch, Judith, Frans Poppel, Caspar Looman, and Johan Mackenbach. “Determinants of infant and early childhood mortality level and their decline in The Netherlands in the late nineteenth century”. International Epidemiological Association, 29 (2000): 1031-1040. Read More