IB Mathematics A&A (SL) Internal Assessment

To what extent do the Life Expectancy rates of developing and

developed countries depend on their respective Healthcare
Expenditure Per Capita?

Topic: Statistical Analysis of “Life Expectancy at Birth”

Candidate Code: jdk323
Session: May 2021
Page Count: 20 (excluding bibliography, appendix and cover page)
May 2021 Candidate code: jdk323

INTRODUCTION

Around this time last year, when we were studying the HDI Index in geography class, I got
surprised by some of the extreme average life expectancies listed in the book, such as 53 for the
Central African Republic and 84 for Japan, or 54 for Nigeria and 82-83 for most of the European
countries. Then I wondered what were the main factors creating such huge gaps between the life
expectancy levels of developing and developed countries. After doing some research on the World
Bank database, I found that the developed countries on average spend $2 000-$10 000 per capita
on health, whereas developing countries spend $30-$700 with an exception of some more robust
economies spending up to $1 000. Just by looking at the values, I noticed that there was a clear
correlation between the health expenditure per capita and the life expectancy of countries.
Therefore, for this Internal Assessment, I wanted to investigate the correlation of these variables in
exact values and try to come up with a formula that can be applied to estimate the life expectancy of
countries with their given health expenditure per capita values.

To carry out my investigation, I will use univariate and bivariate statistic data analysis in multiple
steps:

Initially, for the univariate data analysis, I will explore the distribution of my independent (Health
Expenditure per Capita, $US) and dependent (Life Expectancy at birth, years) variables one by one
first for developing and then developed countries using box-whisker plots, values from five-number
summaries, and test for outliers. In case there are outliers, I will identify them and try to uncover
the reasons by doing online research. Lastly, I will create parallel box-whisker plots displaying the
data from both developing and developed countries collectively on a common scale, which I expect
will give a clear understanding of the gap between them, if there is.

Next, for the bivariate data analysis, I will investigate the correlation between the two variables for
developing and developed countries individually, using Pearson’s Correlation Coefficient test,
scatter diagrams, and least-squares regression method to come up with a formula to estimate the life
expectancy of a country with its given health expenditure per capita. If I succeed in generating the
formulas, I will test their validity by applying them to some countries and estimating their life
expectancy values with their given health expenditure per capita values. I will then compare the
obtained results with the real values from the World Bank Data, and calculate their percentage
errors to check the extent to which the estimations were correct. Hence, the percentage error tests
and the formulas to be created along with the results from Pearson’s correlation test will help me to
come up with an exact value for the correlation of two variables both for developing and developed
countries and ultimately I will be able to answer my research question “To what extent do the Life
Expectancy rates of developing and developed countries depend on their respective Health
Expenditure per Capita?”.

| Page 2
May 2021 Candidate code: jdk323

Key Term Definitions:

Below are the definitions for key terms used throughout the whole investigation. Note that all these
definitions were taken from reliable sources such as WHO, OECD Data, and World Bank:

★ Life expectancy at birth - the average number of years a newborn can expect to live if current
death rates do not change. (WHO)

★ Current health expenditure per capita - the total amount of money each country spends on
healthcare divided by the population number, in current US$. (OECD Data)

★ Developing countries - a country-classification terminology referred to low and

middle-income countries, whose GNI per capita is less than $12,535 as of the 2021 fiscal
year. (World Bank)

★ Developed countries - a country-classification terminology referred to high-income

countries only, whose GNI per capita is more than $12,535 as of 2021 fiscal year. (World
Bank)

★ Gross National Income per capita, GNI/capita - the country’s annual total income in
current US$ divided by its midyear population. (World Bank Data)

Data Obtaining Method:

To carry out the investigation within the IA scope 50 samples were to be chosen in total out of 218
provided by the World Bank Data as representatives from both developing and developed
countries. Therefore, to remove the bias while choosing and to give every country an equal chance
of being picked, www.random.org/lists - an online random sampling generator has been used to
randomly choose 25 countries from each category. To do this, all 135 developing countries
provided were placed on the online generator www.random.org/lists which then assorted them in
random order, and then the first 25 were selected to be the investigation representatives. The same
method was repeated to randomly choose 25 developed countries out of 83, provided by the World
Bank Data.

Therefore, this method enabled me to investigate nearly ¼ of all the world countries provided by
the World Bank Data to study the correlation between the “Life Expectancy at Birth” of a country
and its “Health Expenditure per Capita” both for developing and developed countries.

| Page 3
May 2021 Candidate code: jdk323

UNIVARIATE DATA ANALYSIS

Given below are the randomly selected countries for investigation, 25 from each category:

Table 1. Randomly selected 25 developing and 25 developed countries

Developing Countries Developed Countries

1. Thailand 1. New Zealand

2. Sudan 2. Germany
3. Philippines 3. United Kingdom
4. Georgia 4. Belgium
5. Burkina Faso 5. Italy
6. Belarus 6. Denmark
7. Tajikistan 7. Israel
8. Algeria 8. Canada
9. Egypt 9. Singapore
10. Paraguay 10. Australia
11. Uzbekistan 11. United States
12. Madagascar 12. Luxemburg
13. Timor-Leste 13. Japan
14. Azerbaijan 14. Sweden
15. Peru 15. Norway
16. Liberia 16. Iceland
17. Cameroon 17. Ireland
18. El Salvador 18. Finland
19. Kyrgyzstan 19. Greece
20. Cuba 20. Saudi Arabia
21. Venezuela 21. Netherlands
22. Turkmenistan 22. Switzerland
23. Kenya 23. Cyprus
24. Tanzania 24. Spain
25. Haiti 25. Austria

| Page 4
May 2021 Candidate code: jdk323

After generating the random country samples, data values of Life Expectancy at Birth (total) and
Health Expenditure per Capita (in current $US) were extracted from the World Bank Data using
2018 statistics for both variables:

Table 2. Current Health Expenditure per Capita and Life Expectancy at Birth values, developing countries

Developing Countries Health Expenditure per capita Life expectancy at birth, total
(current US$) (years)

1. Thailand 275.92 77
2. Sudan 60.17 65
3. Philippines 136.54 71
4. Georgia 312.75 74
5. Burkina Faso 40.25 61
6. Belarus 356.25 74
7. Tajikistan 59.84 71
8. Algeria 255.87 77
9. Egypt 125.55 72
10. Paraguay 400.39 74
11. Uzbekistan 82.27 72
12. Madagascar 22.05 67
13. Timor-Leste 93.69 69
14. Azerbaijan 165.77 73
15. Peru 369.08 77
16. Liberia 45.42 64
17. Cameroon 54.14 59
18. El Salvador 288.52 73
19. Kyrgyzstan 85.74 71
20. Cuba 986.94 79
21. Venezuela 256.95 72
22. Turkmenistan 460.18 68
23. Kenya 88.39 66

24. Tanzania 36.82 65

25. Haiti 64.25 64

| Page 5
May 2021 Candidate code: jdk323

Table 3. Current Health Expenditure per capita and Life Expectancy at Birth values, developed countries

Developed Countries Current Health Expenditure per capita Life expectancy at birth, total
(current US$) (years)

1. New Zealand 4037.46 82

2. Germany 5472.2 81
3. United Kingdom 4315.43 81
4. Belgium 4912.7 82
5. Italy 2989 83
6. Denmark 6216.77 81
7. Israel 3323.65 83
8. Canada 4994.9 82
9. Singapore 2823.64 83
10. Australia 5425.34 83
11. United States 10623.85 79
12. Luxemburg 6227.08 82
13. Japan 4266.59 84
14. Sweden 5981.71 83
15. Norway 8239.1 83
16. Iceland 6530.93 83
17. Ireland 5489.07 82
18. Finland 4515.68 82
19. Greece 1566.9 82
20. Saudi Arabia 1484.59 75
21. Netherlands 5306.53 82
22. Switzerland 9870.66 84
23. Cyprus 1954.41 81

24. Spain 2736.32 83

25. Austria 5326.44 82

| Page 6
May 2021 Candidate code: jdk323

Step 1:

After extracting the data values of both variables for randomly selected developing and developed
countries, the first step of my investigation was to look into the distribution of the independent
variable - Current Health Expenditure per Capita (current US$). To do this, the TI-84 Plus CE
graphic calculator, Google Spreadsheets, and BoxPlotR online boxplot generator at
http://shiny.chemgrid.org/boxplotr/ were used to individually graph the Health Expenditure per
Capita values of developing and developed countries in the form of a box-whisker plot, which
allowed me to visually analyze the distribution of all the data. At the same time, it allowed me to
obtain the five-number summaries, which consist of five values; maximum and minimum, upper
and lower quartiles, and median. Next, the test for outliers was carried out to check if there is an
outlier country from the list of given 25 countries. Hence all this information gave me a better
understanding of health expenditure per capita distributions for each data set, developing and
developed world:

Table 4. Five-number summaries of Health Expenditure per capita in current US$

Five-Number Summaries of Health Expenditure per capita in Current US$

Developing Countries Developed Countries

Minimum 22.05 1484.59

Lower Quartile 60.17 3323.65

Median 125.55 4994.9

Upper Quartile 288.52 5981.71

Maximum 986.94 10623.85

Graph 1. Box-Whisker Plot for Developing Countries’ Health Expenditure per capita in current US$

| Page 7
May 2021 Candidate code: jdk323

Test for Outliers:

The upper limit = Q3 + 1.5×IQR
= 288.52 + 1.5×228.35
= $631.045
★ There is only one value above the upper limit, which is $986.94 belonging to Cuba. So
Cuba is the outlier.

The lower limit = Q1 - 1.5×IQR

= 60.17 - 1.5×228.35
= $ -282.355 US
★ There is no value less than the lower limit, so there is no outlier.

The box-whisker plot above is in a positively skewed distribution, meaning that most of the data
fall around the low values as can be seen from the spread of red data points. Additionally, this
distribution has a median value of $125.55, a range of $964.89, and an IQR of $288.35.

There appears to be only one outlier or an extreme value, $986.94, which belongs to Cuba
exceeding the upper limit of $631.045. I did some research on this matter and found out that
Cuba’s healthcare system is indeed considered one of the best in the world (Warner), which the UN
secretary-general Ban Ki-moon says “a model for many countries” (United Nations). The country
has a GDP per capita of $8,821.8 for its just over 11 million population. In addition to that, the
country runs a current account surplus of $985.4 million, which implies that the country exports
more than it imports indicating high economic productivity. This high economic productivity, in
turn, secures higher incomes and more job opportunities for people in the country. Therefore, the
government and people have more money compared to most of the developing countries to spend
on healthcare, explaining why it appeared as an outlier in our investigation.

| Page 8
May 2021 Candidate code: jdk323

Graph 2. Box-Whisker Plot for Developed Countries’ Health Expenditure per capita in current US$

Test for Outliers:

The upper limit = Q3 + 1.5×IQR
= 5 981.71 + 1.5×2 658.06
= $9 968.8
★ There is only one outlier, $10 623.85, which belongs to the United States.

The lower limit = Q1 - 1.5×IQR

= 3 323.65 - 1.5×2 658.06
= $ -663.44
★ There is no value less than the lower limit, so there is no outlier.

The box-whisker plot in Graph 2 has a negatively skewed distribution with a median of $4 994.9, a
range of $9 139.26, and an IQR of $2 658.06. As can be seen in the box-whisker plot, most of the
data fall around the median, with exceptional extreme values and one outlier.

The outlier identified belongs to the United States, $10 623.85, exceeding the generated upper
limit. After some online research, I found out that the main reasons for this are relatively expensive
pharmaceuticals, medical care, expensive tests, and administrative costs. For example, insulin for
diabetes in the US costs $186 monthly, whereas it costs only a third of that in Canada. Another fact
from the same source, people in the US pay $86 for cholesterol-lowering medication, whereas
people in Germany pay less than half of this price (Dr. Hector Florimon). Hence, these higher
prices add up to the higher expenditure value, explaining why it appeared as an outlier in our
investigation.

| Page 9
May 2021 Candidate code: jdk323

Individually generated box-whisker plots above have different scales, which limits our
understanding of their distribution comparing the two. Therefore, a parallel box-whisker plot
containing both data sets was generated enabling us to visually compare their respective
distribution in relation to each other:

Graph 3. Parallel Box-Whisker Plot of Health Expenditure per Capita in current US$, developing and
developed countries

The parallel box-whisker plot above shows the difference in Health Expenditure per Capita of
developing (bottom one) and developed (top one) countries on a common scale, enabling us to
visually compare the gap between the two. The median for developing countries is $125.55,
whereas developed countries have a median of $4994.9, nearly 40 times bigger. Therefore, it can be
concluded that developed countries, on average, spend more than developing countries on
healthcare. The reasons can be higher incomes, expensive pharmaceuticals and medical care, higher
government spendings on healthcare, and so on.

Step 2:

From the first step of the investigation, we got a clear understanding of Health Expenditure per
capita distribution for developing and developed countries. Therefore, the second step of my
investigation was to look at the distribution of the Life Expectancy at Birth rates for developing and
developed countries following the same method, first plotting individually and afterward, in a
parallel box-whisker plot to better analyze their distribution in relation to each other. To do this,
the TI-84 Plus CE graphing calculator, Google Spreadsheets, and an online box-whisker plot
generating tool, BoxPlotR, were used. To display the data, again box-whisker plot was chosen,
enabling me to visually analyze the distribution of data at the same time as helping me to obtain the
five-number summaries using the values from Table 2 and Table 3 above:

| Page 10
May 2021 Candidate code: jdk323

Table 5. Five-number summaries of Life Expectancy at Birth

Five-Number Summaries of Life Expectancy at Birth, total (years)

Developing Countries Developed Countries

Minimum (=min) 59 75

Lower Quartile (=Q1) 66 82

Median (=med) 71 82

Upper Quartile (=Q3) 74 83

Maximum (=max) 79 84

Graph 4. Box-Whisker Plot for Developing Countries’ Life Expectancy at Birth, total

Test for Outliers:

The upper limit = Q3 + 1.5×IQR
= 74 + 1.5×8
= 86 years
★ There is no value more than the upper limit, so there is no outlier.

The lower limit = Q1 - 1.5×IQR

= 66 - 1.5×8
= 54 years
★ There is no value less than the lower limit, so there is no outlier.

| Page 11
May 2021 Candidate code: jdk323

The box-whisker plot in Graph 4 is in a negatively skewed distribution, meaning that most of the
data fall around the upper values or on the right side of the median, with some extreme values in
the first and fourth quarters. This can easily be noticed with the distribution of red data points.
Additionally, this distribution has a median value of 71 years, a range of 20 years, and an IQR of 8
years. The lower limit is 54 years whereas the upper limit is 86 years, and since there is no
developing country whose Life Expectancy at Birth value exceeds the limits, there was no outlier
noticed.

Graph 5. Box-Whisker Plot for Developed Countries’ Life Expectancy at birth, total

Test for Outliers:

The upper limit = Q3 + 1.5×IQR
= 83 + 1.5×1
= 84.5 years
★ There is no value more than the upper limit, so there is no outlier on the right side.

The lower limit = Q1 - 1.5×IQR

= 82 - 1.5×1
= 80.5 years
★ There are two outliers with values below the lower limit. These belong to Saudi Arabia
with 75 years, and the US with 79 years.

The box-whisker plot in Graph 5 has a very strong positively skewed distribution, meaning that
most of the data fall in the first half, which also can be seen by the data points on the graph shown
in red colors. The box-whisker plot, surprisingly, has a mean with the same value as the Q1, 82.
Besides, it has a range of 9 years and an IQR of 1 year.

| Page 12
May 2021 Candidate code: jdk323

There are two outliers in the distribution in total, belonging to Saudi Arabia (75 years) and the US
(79 years), exceeding the lower limit of the box-whisker plot. Further research was carried out to
find the reasons for this for both countries. Firstly, for Saudi Arabia, it was found that the country
has newly joined the team of developed countries due to its high income from oil exports.
Therefore, the healthcare system of the country is still in the process of improvement. Besides, the
climate of the country is marked by really high temperatures, following a pattern of a desert climate
(Weather Online). Hence this extreme climate links to one of the main reasons for Saudi Arabia’s
death rates - heat stroke (Tyrovolas et al.). Therefore, the level of Saudi Arabia’s life expectancy
hasn’t reached the level of other rich countries yet. Secondly, for the US, it was found that the main
reasons for shorter life expectancy are high levels of smoking, homicides, opioid overdoses, and
suicides (Roser). However, for the case of the US, it is interesting that it also appeared as an outlier
in the Health Expenditure per Capita investigation for spending more on healthcare than other
rich countries. This means that even though people spend much more on healthcare in the US, the
life expectancy is yet to reach the level of other developed countries. Therefore, for the reasons
mentioned above, it is no surprise that Saudi Arabia and the US appeared as outliers in my
investigation.

Again, individually generated box-whisker plots above have different scales, limiting our
understanding of the real distribution of data sets to some extent. Therefore, a parallel box-whisker
plot containing both data sets was generated to visually compare their respective distribution in
relation to each other:

Graph 6. Parallel Box-Whisker Plot of Life Expectancy At Birth for Developing and Developed countries

The parallel box-whisker plot above shows the difference in Life Expectancy at Birth for the total
population of developing (bottom one) and developed (top one) countries on a common scale,
which enables us to visually compare the gap between the two. From the graph, it can be seen that
the median for developing countries is 71 years with a minimum value of 59 and a maximum value
of 79, whereas the median for developed countries is 82 years with a minimum value of 75 and a
maximum value of 84. Hence, it can be concluded that people in developed countries, on average,
live longer than people do in developing countries.

| Page 13
May 2021 Candidate code: jdk323

BIVARIATE DATA ANALYSIS

Step 3 (a):
After exploring the distribution of Health Expenditure per capita and Life Expectancy at birth in
developing and developed countries individually and together, the third step of my investigation
was to analyze the correlation between the two variables. To do this, I generated their respective
scatter diagrams using Google Spreadsheets, and then used Pearson’s Correlation Coefficient test to
check the nature of the correlation.

Graph 7. Two-Variable Scatter Diagram for Developing Countries

Pearson’s Correlation Coefficient formula:

where;
r = correlation coefficient
xi = x-variable values of the sample
𝑥 = mean of the x-variable values
yi = y-variable values of the sample
𝑦 = mean of the y-variable values

The results are obtained by using TI-84 Plus CE and online Pearson Correlation Coefficient
Calculator at www.socscistatistics.com:

| Page 14
May 2021 Candidate code: jdk323

Correlation coefficient, r (25) = 0.6599

The correlation coefficient obtained shows that there is a strong positive correlation between the
Health Expenditure per Capita and Life Expectancy at Birth values of developing countries. The
positiveness of the correlation means that the more the healthcare expenditure, the higher are the
life expectancy levels.

P-value, p = 0.000332, significant at p < 0.01

The probability of occurrence by chance is really small, 0.03%. Therefore, with 99.97% confidence,
we can say that there is a correlation between the two variables.

Coefficient of determination, r2 = 0.4355

The coefficient of determination value, r2, indicates that in developing countries 43.55% of the
variation in Life Expectancy At Birth can be explained by the variation in Health Expenditure per
Capita, leaving out the remaining 56.45%, to link to other factors.

Least squares regression line, y = 0.01649x + 66.82069

The line of best fit, shown in Graph 7, was found using the TI-84 Plus CE, and then tested using
the online Linear Regression Calculator on www.socscistatistics.com/tests/regression/. The
equation can be applied as follows:

Life Expectancy At Birth (years) = 0.01649 ✕ Health Expenditure Per Capita($US) + 66.82069

Therefore, from the calculations above using the obtained equation of the best fit line, it can be
summarized that, in developing countries, for every additional $100 health expenditure the life
expectancy increases by 1.649 years.

To test its validity, the generated formula was applied to four randomly chosen developing
countries, two from the list in this investigation, Tanzania and Madagascar, and two that are not
being explored in this investigation, Zambia and Bulgaria. Note that the values for Current Health
Expenditure per capita and Life Expectancy at birth of Zambia and Bulgaria were extracted from
the same source as other countries, World Bank Data.

| Page 15
May 2021 Candidate code: jdk323

Table 6. Data from World Bank vs generated values using the formula

Country Current Health Expenditure Life Expectancy at birth Life Expectancy at birth
per capita (US$), World Bank (years), World Bank (years), generated equation

Tanzania 36.82 65 67.428

Madagascar 22.05 67 67.184

Zambia 75.99 64 68.074

Bulgaria 689.91 75 78.197

Using the findings from the generated formula and the data from the World Bank, the percentage
error values were estimated with the formula below:

The generated Life Expectancy at Birth value of Tanzania, using the formula, was 67.428 years
compared to the 65 years given by World Bank Data. Therefore, the estimated percentage error was
found to be just 3.735%. Using the same formula, the estimated percentage errors for the other
three countries are all below 7%, indicating that the least-squares regression line has been an
appropriate tool to predict the Life Expectancy At Birth values of developing countries with the
given Healthcare Expenditure per capita values.

| Page 16
May 2021 Candidate code: jdk323

Step 3 (b):

Below is the scatter diagram displaying the Health Expenditure per capita values in relation to Life
Expectancy at birth values of developed countries:

Graph 8. Two-Variable Scatter Diagram for Developed Countries

The TI-84 Plus CE graphing calculator and online tool Pearson Correlation Coefficient Calculator
at www.socscistatistics.com were used to obtain all the results below:

Correlation coefficient, r (25) = 0.1628

The Pearson’s correlation coefficient obtained shows that there is a weak positive correlation
between the Health Expenditure Per Capita and Life Expectancy At Birth values of 25 given
developed countries.

P-value, p = 0.436843
The probability of occurrence by chance is really big, 43.68% in percentage. Therefore, the
correlation results cannot be considered significant.

Coefficient of determination, r2 = 0.0265

The coefficient of determination value, r2, indicates that in developed countries only 2.65% of the
variation in Life Expectancy At Birth can be explained by the variation in Health Expenditure Per
Capita, which also signifies that the remaining 97.35% of variation links to other factors.

| Page 17
May 2021 Candidate code: jdk323

Least squares regression line, y = 0.00013x + 81.28047

The line of best fit, also shown in Graph 8, was found using the TI-84 Plus CE, and later tested
using the online Linear Regression Calculator on www.socscistatistics.com/tests/regression/ . The
equation can be applied as follows:

Life Expectancy At Birth (years) = 0.00013 ✕ Health Expenditure Per Capita($US) + 81.28047

The calculations above are estimated Life Expectancies of the general population for developed
countries with their differences for each additional $100 using the line regression equation obtained
earlier.

The results signify that for every additional $100 expenditure on health, the Life Expectancy at
Birth values increase by just 0.013 years.

Once again, to test its validity, the generated formula was applied to four randomly chosen
developed countries, two from the list in this investigation, Ireland and Denmark, and two that are
not being explored in this investigation, UAE and Portugal. Note that the values for current Health
Expenditure per capita and Life Expectancy at Birth of UAE and Portugal were extracted from the
same source as other countries, World Bank Data.

| Page 18
May 2021 Candidate code: jdk323

Table 7. Data from World Bank vs generated values using the formula

Country Health Expenditure per capita Life Expectancy at birth Life Expectancy at birth,
(current US$), World Bank (years), World Bank generated equation

Ireland 5489.07 82 81.994

Denmark 6216.77 81 82.089

UAE 1817.35 78 81.517

Portugal 2215.17 81 81.568

Afterward, using the findings from the generated formula and the data from the World Bank, the
percentage error values were estimated with the same formula used earlier:

The calculated Life Expectancy at Birth values of all four randomly chosen countries were really
close to the real values extracted from World Bank Data. Therefore, the percentage error values
were all below 5%, indicating that the least-squares regression line equation has been an appropriate
tool to predict the Life Expectancy at Birth values of developed countries with their given
Healthcare Expenditure per capita values, even though the correlation coefficient was really small

| Page 19
May 2021 Candidate code: jdk323

CONCLUSION

Throughout the investigation, it was found that Life Expectancy at Birth values of countries, to a
certain extent, depend on their Health Expenditure per Capita. To explore this, the investigation
was carried out in 3 steps:

The First Step was dedicated to looking at the distribution of the independent variable, Health
Expenditure per Capita ($US), for developing and developed countries individually. For developing
countries, it was found that the data have a positively skewed distribution, indicating that most of
the values fall between the first quartile ($60.17) and the median ($125.55). The box-whisker plot
used to display these values also showed that the distribution has a range of $964.89 and IQR of
$228.3. In the test for outliers, only Cuba’s value ($986.94) appeared to exceed the calculated upper
limit ($631.045), being considered as the only outlier of the 25 randomly chosen developing
countries. This was explained by the economic productivity of the country and the high GDP per
capita ($8,821.8) for its small population. For developed countries, on the other hand, the
box-whisker showed a negatively skewed distribution, with a median of $4994.9, a range of
$9 139.26, and an IQR of $2 658.06. With the test, the United States was found to be the only
outlier out of the 25 with the value of $10 623.85, which was then explained by expensive
pharmaceuticals and medical care. Lastly, to gain a better understanding of the distribution a
parallel box-whisker plot was formed displaying all the data on a single scale.

In the Second Step, the same method was used to analyze the distribution of Life Expectancy at
Birth values for developing and developed countries via box-whisker plots, five-number summaries,
and tests for outliers. For developing countries, it was found that the distribution had a negatively
skewed shape with a median of 71 years, range of 20 years, and IQR of 8 years. There were no
outliers identified for this distribution. For the developed countries a very strong positively skewed
distribution was observed with the median of the same value as the first quartile, 82 years.
Additionally, a range of 9 years and an IQR of just 1 year were observed with the US and Saudi
Arabia being the outliers. Different statistics were given to explain these cases by doing some
research. Lastly, again to better understand the distribution of the data in general, a parallel
box-whisker plot was created displaying all 50 values of Life Expectancy at Birth on a single scale.

In the Third Step, I carried out a bivariate data analysis investigating the correlation of two variables
for both developing and developed countries. To do this, I used scatter plot diagrams, Pearson’s
Correlation Coefficient test, and the least-squares regression line equation. For developing
countries, the results showed a strong positive correlation, r=0.6599, with a highly significant
p-value of 0.000332 and an r2 value of 0.4355, indicating that 43.55% of the variation in Life
Expectancy can be explained by variation in Health Expenditure per Capita. Afterward, using the

| Page 20
May 2021 Candidate code: jdk323

TI-84 Plus CE and online tools, a formula with the equation of y = 0.01649x + 66.82069 was
generated to estimate the life expectancy of any developing country with its given health
expenditure per capita value. This formula then was tested by estimating the life expectancy values
of four developing countries, whose percentage errors were found to be just below 7% indicating
that the formula generated was a valid tool to use. In the second section of this Step, the same
method was used to find the correlation of variables for developed countries. It was then found that
there is a weak correlation between the two variables, with an r-value of 0.1628, p-value of
0.436843, and determination coefficient (r2) of 0.0265. The created formula for the least-squares
regression line for this was y = 0.00013x + 81.28047, which was then tested for estimating the life
expectancy values of 4 developed countries with their given health expenditure per capita values.

With all these findings now I am able to answer my initial research question, “To what extent do
the Life Expectancy rates of developing and developed countries depend on their respective Health
Expenditure per Capita?”. The correlations highly vary between developing and developed
countries. For developing countries, for every additional $100 expenditure on health, the life
expectancy goes up by 1.649 years, whereas this value is 0.013 years for developed countries. which
were obtained with their respective regression line equations. This tells me that for developing
countries there is a strong correlation between these two variables whereas for developed countries
it is really low.

| Page 21
May 2021 Candidate code: jdk323

BIBLIOGRAPHY

Dr. Hector Florimon. “Why US Health Care Costs More, but Isn’t Better than Other Countries:
Study.” ABC News, 13 Sept. 2019,
abcnews.go.com/Health/us-spends-health-care-countries-fare-study/story?id=53710650.
Accessed 5 Apr. 2021.

Haahr, Mads. “RANDOM.ORG - List Randomizer.” Random.org, www.random.org/lists/.

Accessed 5 Apr. 2021.

IMathAs. “Boxplot Grapher.” Www.imathas.com, www.imathas.com/stattools/boxplot.html.

Accessed 4 Apr. 2021.

OECD Data. “Health Resources - Health Spending - OECD Data.” Data.oecd.org, 2019,
data.oecd.org/healthres/health-spending.htm. Accessed 4 Apr. 2021.

Roser, Max. “Why Is Life Expectancy in the US Lower than in Other Rich Countries?” Our World
in Data, 29 Oct. 2020, ourworldindata.org/us-life-expectancy-low. Accessed 6 Apr. 2021.

Stangroom, Jeremy. “Pearson Correlation Coefficient Calculator.” Socscistatistics.com,

www.socscistatistics.com/tests/pearson/Default2.aspx. Accessed 5 Apr. 2021.
---. “Quick Linear Regression Calculator.” Www.socscistatistics.com,
www.socscistatistics.com/tests/regression/. Accessed 5 Apr. 2021.
---. “Quick P Value from Pearson (R) Score Calculator.” Www.socscistatistics.com,
www.socscistatistics.com/pvalues/pearsondistribution.aspx. Accessed 5 Apr. 2021.

The World Bank. “Life Expectancy at Birth, Total (Years) | Data.” Data.worldbank.org, 2019,
data.worldbank.org/indicator/SP.DYN.LE00.IN. Accessed 4 Apr. 2021.
---. “World Bank Country and Lending Groups – World Bank Data Help Desk.”
Www.worldbank.org, 2021,
datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-len
ding-groups. Accessed 4 Apr. 2021.

Tyrovolas, Stefanos, et al. “The Burden of Disease in Saudi Arabia 1990–2017: Results from the
Global Burden of Disease Study 2017.” The Lancet Planetary Health, vol. 4, no. 5, 1 May
2020, pp. e195–e208. The Lancet,
www.thelancet.com/journals/lanplh/article/PIIS2542-5196(20)30075-9/fulltext,
10.1016/S2542-5196(20)30075-9. Accessed 6 Apr. 2021.

| Page 22
May 2021 Candidate code: jdk323

United Nations. “Secretary-General Hails Cuba for Training Medical ‘Miracle Workers’, Being on
Frontlines of Global Health | Meetings Coverage and Press Releases.” Www.un.org, 28 Jan.
2014, www.un.org/press/en/2014/sgsm15619.doc.htm. Accessed 6 Apr. 2021.

Warner, Rich. “Is the Cuban Healthcare System Really as Great as People Claim?” The
Conversation, 30 Nov. 2016,
theconversation.com/is-the-cuban-healthcare-system-really-as-great-as-people-claim-69526.
Accessed 5 Apr. 2021.

Weather Online. “Climate of the World: Saudi-Arabia | Weatheronline.co.uk.”

Weatheronline.co.uk, www.weatheronline.co.uk/reports/climate/Saudi-Arabia.htm.
Accessed 5 Apr. 2021.

WHO. “Life Expectancy at Birth.” World Health Organization.

World Bank Data. “Current Health Expenditure per Capita (Current US$) | Data.”
Data.worldbank.org, 2018, data.worldbank.org/indicator/SH.XPD.CHEX.PC.CD.
Accessed 4 Apr. 2021.
---. “Glossary | DataBank.” Databank.worldbank.org,
databank.worldbank.org/metadataglossary/health-nutrition-and-population-statistics/serie
s/NY.GNP.PCAP.CD. Accessed 4 Apr. 2021.

| Page 23
May 2021 Candidate code: jdk323

APPENDIX

Appendix 1. Pearson’s Correlation Coefficient, r, for developing countries from www.socscistatistics.com

| Page 24
May 2021 Candidate code: jdk323

Appendix 2. Least Squares Regression Line for developing countries from www.socscistatistics.com

Appendix 3. Pearson’s Correlation Coefficient, r, for developed countries from www.socscistatistics.com

| Page 25
May 2021 Candidate code: jdk323

Appendix 4. Least Squares Regression Line for developed countries from www.socscistatistics.com

| Page 26