Assignment #5 Laila Arnesen

Sociology166 – Assignment #5 – BIRTH RATES

“One might make a reasonable assumption that nations with large Catholic populations would have birth rates which reflect an absence of birth control measures (less birth control = higher birth rate). And likewise, nations which do not have a significant population that is prohibited from using artificial birth control measures would have lower birth rates (more birth control = lower birth rate).” 

This assignment analyses possible relationships between birth rate (defined here as children per women) and birth control. I will collect data for nine countries. The data material will be used in a multiple regression analysis to see which variables and to what extent they affect birth rates. To be able to make a qualified selection of countries and variables to include in my analysis, I will first do a simple deterministic analysis on birth rate and catholic share of population only.

Deterministic analysis

I will first conduct a simple deterministic analysis of birth rate data from CIA’s fact book (2012 numbers) and combine the birth rate data with data for catholic share of population for each country. The set of data collected contains 191 countries (about 99.93% of total world population). Table 1.1 below shows population-weighted averages of Birth rates and Catholic share of population split by geographical region. These numbers gives us no reason to believe that countries with a high share of catholics (i.e. also birth control) have a higher birth rate. Africa has close to twice the birth rate of the world average, but only 16.7% share of Catholics. Europe with a 38% share of Catholic citizens has only 1.56 in birth rate.

Table 1.1 Birth rates and catholic populations by geographical regions (SOURCE: CIA Factbook 2012)

When graphing the 10 largest countries based on birth rate we see that 9 out of 10 are African countries. The other interesting thing to notice is that it does not seem to be any correlation in this subset between catholic share of population and birth rate.

On the other extreme we graph the top 10 catholic countries together with their respective birth rate. Again, there is difficult to give any clear indication to any correlation between birth rate and catholic share of population.

If we define a catholic country as a country with more than 65% of the population Catholic, we end up with 42 out of 191 countries. Calculating the population-weighted birth rate average in a catholic country I end up with 2.17. The number for non-Catholic countries is 2.46. This again does not support our assumption that Catholic countries have higher birth rates.

A simple regression analysis based our 191 countries (99.93% of total world population) shows that birth rate is negatively correlated by catholic share of population. For each percentage increase in catholic share of population, the birth rate decrease by 0.53. However, using catholic share of population as the only decision variable is by far not enough to explain variations in birth rates. The R-square of the simple regression analysis is only 0.02 and suggests that there are other variables that affect birth rate as well.

Conducting a multiple regression analysis will give me a much better feeling to what and to which extent different (multiple) variables affect birth rates. In this short assignment I select a subset of only nine different countries. A challenge is to find a subset that captures a broad range of the world’s population and to find independent variables that describe birth rates. A problem I ran into with this assignment is the lack of reliable data, especially for countries with high birth rates.

Multiple Regression

I select six catholic countries (Catholics representing more than 65% of the total population) and three non-catholic countries. Because of the very small size of the subset, I have tried to create a set that has a large geographical spread.

·       Italy (80% cath)

·       Argentina (92% cath)

·       Poland (89% cath)

·       Croatia (86% cath)

·       Mexico (82% cath)

·       Equatorial Guinea (87% cath)

·       Sweden (non-catholic)

·       South Africa (non-catholic)

·       Tajikistan (non-catholic)

In order to conduct a significant regression analysis, I need a set of decision variables that I believe influences birth rates. Studies and papers on this issue suggests a whole set of variables that affect birth rate, including the degree of women’s rights, life expectancy, infant mortality, economic tax benefits for families, women in work, access to information (internet and education) etc. Based on these findings, I have chosen the following 6 independent variables:

·       Catholic share of population (percentage) [1]

·       GINI index (income inequalities within a country, 0 being perfect equality) [1]

·       School life female (average number of years in school) [1]

·       Infant mortality (possibility of mortality the first 12 months after birth) [1]

·       Life expectancy (by year) [1]

·       Women’s rights measure [2]

Using Birth rate as my dependent variable and Catholic share of population as my independent variable in a simple regression analysis, I end up with an adjusted R squared of -0.14. The beta coefficient for Catholic share is 0.00025, which means that birth rate will increase by 0.00025 for a 1% increase in catholic share of population. A p-value of 0.99, tells me that this variable is not significant. So, the adjusted R squared and p-values are too low and suggest that other variables should be included in the analysis in order to explain movements in birth rate.

Next, I include my other independent variables to my analysis. First I do a simple correlation analysis for each independent variable. The table below shows the results:

The table shows that increasing life expectancy, school life, and women’s rights will decrease birth rates. An increase in GINI index and infant mortality will on the other hand decrease birth rate.

Putting it all together in a multiple regression in MS Excel gives the following results:

First of all, the adjusted R square increase to 0.91 which tells me that the additional variables help to explain birth rate to a larger extent than with only catholic share as independent variable. However, the infant mortality variable is the variable with the lowest p-value, but still only 0.11. I conclude that the population of only 9 countries is not big enough to yield significant results.

The final result for birth rate versus birth control (catholic countries) shows that our initial hypothesis is not correct. Solely based on this study, we cannot say that there is any significant correlation between birth rate and birth control. Other variables are by far more important in describing variations in birth rate. The most important from this study is the infant mortality rate. For a country to decrease birth rates, it should focus on women’s education and increase its health budgets.

It must be noted that the population (9 countries) is far too low to make this a significant analysis.

[1] = CIA’s factbook

[2] = http://www.scribd.com/doc/65645520/Best-Countries-for-Women-Full-List