# Basic Antecedents of Life Expectancy at Birth. Linear Regression Modelling

## 27 Pages

Excerpt

Index of Figures

Index of Tables

1 Introduction

2 Population and Sample

3 Variables

4 Methodology

5 Analysis of Results

6 Discussion

7 Conclusion

## Index of Figures

Figure 1 - Scatter Plot Matrix of all used variables

Figure 2 - Scatter Plot Unstandardized Residuals vs. Unstandardized Predicted Values

## Index of Tables

Table 1 - Correlation Matrix

Table 2 - Curve Estimates for all dependent variables

Table 3 - Comparison of mean and variance of sample and population

Table 4 - Model Summary

Table 5 - Analysis of Variances

Table 6 - Calculation of Coefficients

## 1 Introduction

For many years, researchers have shown profound interest in human average life expectancy at birth, in this article simply named life expectancy, its causes and antecedents as well as its development over time. In fact, life expectancy is a phenomenon that has been examined by many researchers, primarily in the field of medical research (Lin et al., 2012). The great interest in lifetime expectancy and its influential factors is mainly based on the fact that life is the most valuable asset that every human being values and strives to sustain. The focus of life expectancy research lies on the relationships of effective determinants that characterize a society as a whole and the average expected life time of a human being. For instance, Messias (2003) identified illiteracy rate as well as income disparity as significantly strong determinates for average human lifetime. According to their research results, both indicators are negatively associated with life expectancy in the case of Brazil. On another note, economic performance measured in GDP per capita was found to have a positive impact of life expectancy in their research. Social status and its effects on life expectancy has also been under closer attention. White and Edgar (2010) noticed a significant difference in life expectancy depending on belongingness to social class.1 In their analysis, they examine the differences of average lifetime between professional labor class and unskilled manual labor class, leading to a statistically significant outcome. According to their results, people with a higher social status are more likely to grow older, enjoy a larger number of so called healthy years and their life quality is superior. The results of Johnson (2011) show great congruency. Their study unveiled an improvement of life expectancy independent of gender, hence male and female lifespan are both increasing in a similar manner, however, they consistently observed inequalities related to social class.2 Even though both male and female lifetime expectancy is increasing, average female life expectancy in general outruns male. This gender based disparity and superiority of women has been reported in numerous cases (Leviatan, Cohen 1985; Hosseinpoor et al. 2012; Ehiemeu, 2014; Yang et al., 2012; Tudora et al., 2015) and can be considered as common knowledge. Despite its prevalence and its obviousness, scholars still struggle to sufficiently explain differences and identify specific determinants, but list instead a whole range of different indicators that potentially cause gender based inequality in life expectancy. While some of those indicators are similar to the ones used by Messias (2003) or White and Edgar (2010) for instance, some scientists argue that the composition of the gens and related body type characteristics are possibly responsible for the gender disparity (Medalia and Chang, 2011; Waldron,1976; Austad, 2006).

However, social class or income level are not the only elements for causing differences in expected lifetime. An enormous amount of research has been conducted on the differences of life expectancy across country borders. In general, scholars suggest that unequal development levels in certain areas such as technology, education, health care systems etc. is at the root cause of those differences. For example, Lin et al. (2012) put the focus of their research on the effect of political and social determinants on life expectancy with the respect to less developed countries. Their findings support the hypothesis that the logarithm of GDP per capita, literacy rate and nutrition status are positively related with life expectancy, while the established political regime does also have a significant influence that is much harder to explain. The effects of the regime might not directly impact on life expectancy but be mediated through common factors such as the health insurance system in place, if any. However, this study does not indent to further elaborate on vague responsible causes of political systems on life expectancy but rather focus on the development of it over time.3

Despite the cross-country differences in level of life expectancy, a common trend can be observed. Time series analyses show evidence that life expectancy has experienced continuous growth over the past decades. Scholars claim that developments such as medical improvements, the increase in educational level or raising standard of living, just to name a few, are the underlying drivers of this tendency (Mazumdar 2001; Greene 2001; Oeppen and Vaupel, 2002). Nevertheless, discrepancy of life expectancy with respect to national borders as well as women and men remains noticeably. The study of Hosseinpoor et al. (2012) has a stake in identifying an international shortfall in lifetime inequality for both gender groups. Their article proves that the shortfall in “life expectancy at birth among men and among women decreased dramatically between 1950 and 1975 but stagnated thereafter.” Hereby, the shortfall for women was higher than the one for men, since life expectancy of women shows greater variance with respect to country based economic level. Hosseinpoor et al. (2012) furthermore predicts a high likelihood that this gap will continue to tighten if greater advancements in living and working conditions, especially for women in lowerincome nations, will be made.

The discussed examples of studies show that much work has been conducted focusing on life expectancy at birth or health in general with the aim to elaborate significant, reasonable and quantitative relationships. The multiplicity of studies leads to multitude of statistical models that determine both quantitative as well as qualitative relationships of various indicators with life expectancy. The wide range of indicators comprises elements from many areas including but is not limited to economics, health, climate, environmental conditions and education. At the same time, the developed models vary in complexity. Especially the cross-nation disparity is an issue at stake. Commonly, less developed countries are associated with a lower life expectancy (Messias, 2003), however, income itself fails to be a sufficient health antecedent (Biciunaite, 2014).

This analysis aims to elaborate a linear statistical model in order to determine the relationship between life expectancy at birth and basic educational respective health-related independent variables. The results will enable politicians and other decision makers to identify and predict the influence of their decisions towards average life expectancy at birth for their population. This study purposefully picked very basic explanatory variables since they are easier to manipulate than complex indicators. Consequently, this paper mainly puts emphasize on highlighting possible ways on how to impact on life expectancy with relative moderate effort.

## 2 Population and Sample

The population of the analysis are all 249 nations in the world which are according to the World Bank currently existent.4 A sample as large as possible is desirable for statistical significance. In best case, a linear regression for all 249 countries should be conducted to examine dependencies over the whole data set and draw statistical significant conclusions. However, the availability of complete data points is most certainly a constraint that limits sample size. This study follows the approach to eliminate incomplete data points instead of replacing missing values using any of the common replacements methods. The described method is the method of choice because some variables, such as HIV infection rate for example, show a multitude of missing values (39%). Therefore, replacing this amount of missing values would lead to non-justifiable data manipulation and harm the validity of the research. The elimination of missing data points produces a sample consisting of 89 countries (35.7% out of the listed countries). However, it is highly important to acknowledge that unavailability of data can be caused in a systematic manner and hence the elimination process leads to a possible bias of the later elaborated model. It can be observed that nations that might potentially lie in a certain range of an explanatory variable do not publish or maybe not even evaluate that very variable. For instance, some countries with high educational level such as Norway do not measure their domestic literacy rate because it is supposed to be very high and not of major interest. A similar pattern can be observed when countries are said to show very low values in a certain category. For example, HIV prevalence is has not been measured or published in countries with a potentially very low rate because HIV is not an issue in those countries and therefore not of major interest. France for example does not publish or measure HIV rates. Also, there might be political reasons for not publishing data that might be considered sensitive in the respective country. These practices lead to the fact that some advanced countries with expected extreme values in the independent variables will be removed from the sample. However, this does not apply to their entirety, therefore the bias should be limited. An additional limitation is the fact that observed values for literacy or HIV rates are not scattered equally over the whole variable range but tend to be bounded to a certain area. While this is statistically questionable, the data set still represents expectable real life distribution of values, with a tendency to low HIV rates and higher literacy rates worldwide.

## 3 Variables

This paper will examine one dependent variable, namely life expectancy at birth, and five potential explanatory, independent variables: health expenditure per capita, adult literacy rate, immunization rate of children for both DPT and measles and prevalence rate of HIV.

Life Expectancy at Birth [a]

The dependent variable life expectancy at birth (LifeExp), measured in years, is the average “number of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life” over the whole population of each individual country (The World Bank 2013f).

Health Expenditure per capita [US\$]

Health expenditure per capita (HlthExpnd), measured in current US\$, is the “sum of public and private health expenditures as a ratio of total population” (The World Bank, 2013a). The spending on health care per head covers the provision of both preventive and curative health services, family planning activities, nutrition activities and emergency aid designated for health. The spending itself will not have any direct influence on lifetime but it is seen as a representative of the quality of the underlying health system. Exclude from the health expenditure are the provision of water and sanitation. (The World Bank, 2013a).

Health expenditure is supposed to be positively associated with life expectancy, meaning that an increase in expenditure will lead to an increase in quality in the prevalent health system and subsequently extend life expectancy.

Immunization Rate of Children against DPT and Measles [%]

Immunization rates measure the percentage of children aged between twelve and 23 months “who received vaccinations before 12 months or at any time before the survey” (The World Bank 2013b, 2013c).This study will take two different types of vaccinations into account. DPT immunization rate (ImmunDPT) refers to the rate of infants between twelve and 23 months who received three rates of vaccine against pertussis (or whooping cough), and tetanus (The World Bank, 2013b). According to WHO, a child is considered adequately immunized after receiving three doses of vaccine (The World Bank, 2013b). Measles immunization rate (ImmunM) refers to the rate of infants between twelve and 23 months who received one dose of vaccine against measles (The World Bank, 2013c). According to WHO, a child is considered adequately immunized against measles after receiving one doses of vaccine (The World Bank 2013c).

Vaccination is said to actively prevent deaths, especially early child deaths (Anekwe et al., 2015). But this study does not only take vaccination rate into account because of its immanent, death preventing effect, but perceives it also as an indicator of the prevalence and the focus of medical child care. It is believed that medical assistance in the early years will have a large stake at life quality and also life expectancy (Oeppen, Vaupel 2002). Additionally, immunization is a relative simple medical procedure and be easily implemented and conducted, also in un- or underdeveloped countries, therefore it will be an easy measure to impact directly on life expectancy.

In this study, immunization rates and life expectancy are assumed to have a positive relationship.

Prevalence of HIV [%]

Prevalence of HIV (HIV) is the percentage of people aged 15 to 49 who are infected with HIV (The World Bank 2013e). High HIV rates are a major issue at stake especially in African countries, however, the deadly virus is present all over the world. The consequences of HIV are severe and eventually deadly, hence affecting life time (Salmon-Ceron et al. 2005; Soriano et al. 1999; Gandhi et al., 2006). Yet prevention of HIV transmission is quite simple. HIV rates are chosen as an antecedent of lifetime due to their life-threatening character but also due to the fact that they reflect educational level and health sensitivity of a society. HIV rates become an essential part of the linear model and are supposed to be negatively correlated with life expectancy.

Literacy rate [%]

Literacy rate (Ltrcy) as the only non-health related variable is measured in percent and describes the total “percentage of the population age 15 and above who can, with understanding, read and write a short, simple statement on their everyday life” (The World Bank 2013d). Noticeably, not being able to read will most likely not be a main cause to end the life of a human being. However, it is representing the quality of life in a country and it can also be seen as an indicator to draw conclusions on individual life conditions. Also, literacy is a basic skill that has been proven as correlated with life expectancy by former research (Messias, 2003). Similarly, literacy rate is supposed to have a positive effect on life expectancy.

[...]

1 Antonovsky (1967) also conducts research on the dependencies of social class and life expectancy.

2 Lynch et al. (2004) discuss in addition, if income inequality that goes along with social status is a determinant of population health.

3 The interested reader is referred to Mackenbach (2013) and Chen et al. (2012) who also examine the influence of political and social determinants on life expectancy in Europe respectively in less developed countries.

4 Based on the number of countries listed in the data set of life expectancy (The World Bank, 2013f)

Excerpt out of 27 pages

Details

Title
Basic Antecedents of Life Expectancy at Birth. Linear Regression Modelling
College
Tongji University  (Sino-Germany College)
Author
Year
2015
Pages
27
Catalog Number
V311417
ISBN (eBook)
9783668101685
ISBN (Book)
9783668101692
File size
551 KB
Language
English
Tags
basic, antecedents, life, expectancy, birth, linear, regression, modelling
Quote paper
Patrick Glasen (Author), 2015, Basic Antecedents of Life Expectancy at Birth. Linear Regression Modelling, Munich, GRIN Verlag, https://www.grin.com/document/311417