You are on page 1of 16

Introduction

The Human Development Index (HDI) was created to assess the capabilities of citizens in
countries around the world. It is measured using three different indices: life expectancy, GNI
(gross national income) and mean years in school. My aim is to create a HDI of my own to
compare with the current HDI. The three indices I have chosen are: primary school completing
(expressed as a percentage of the relevant age group), overall access to at least basic
sanitation (also as a percentage) and CO2 emissions per person. The 20 countries I have
chosen are equally divided into the GNI groups as listed below, with 5 countries per GNI group
so that I am able to analyse a wide range of data, and not just countries with high GNI, and
therefore high HDI.

Country GNI Group

Burundi L

Mozambique L

Nepal L

Togo L

Burkina Faso L

Djibouti LM

India LM

Kyrgyz Republic LM

Moldova LM

Ghana LM

Mauritius UM

Cuba UM

Jordan UM

Samoa UM

Peru UM

Croatia H

Finland H
Estonia H

Seychelles H

Lithuania H

Hypothesis
I chose primary school completion because it measures the education aspect of the country. I
think that the primary school completion (in percentage of relevant age group) index will have a
positive correlation with the HDI because the more people complete basic education, the more
likely they are to move into a higher education, thus increasing the human development of the
country. I included the overall access to at least sanitation because I predict that it will have a
positive correlation with the HDI. This is because having more access to sanitation would lead
to fewer deaths and higher life expectancy rate. I decided to use CO2 emissions per person for
my last index because it shows the lack of renewable energy a country uses. I hypothesize that
it will have a negative correlation with the HDI because CO2 is usually released as a result of
nonrenewable energy such as the combustion of coal and natural gas. The lower the number,
the higher the human development because it shows that the population of that country is trying
to make a minimal negative impact on the environment.

Univariate Statistics
Below are box plots I created to compare the spread of data of each index I have chosen as
well as the HDI between the year 2000 and 2017.
2000 2017

Upper whisker 0.86 0.94

3rd quartile 0.71 0.8

Median 0.62 0.71

1st quartile 0.44 0.54

Lower whisker 0.29 0.42

The median HDI of the 20 countries increased by 0.9 from 2000 to 2017 and is equivalent to the
3rd quartile in 2017. Because there is an increase in the median HDI, the upper and lower
whisker and 1st and 3rd quartile have increased. The interquartile range has remained relatively
the same with 0.27 in 2000 to 0.26 in 2017. However, the range between the lower and upper
whisker in the year 2000 (0.57) and the year 2017 (0.52) had decreased slightly. This suggests
that the HDI of the 20 countries I have chosen have increased overall from 2000 to 2017. There
are no outliers in any of the box plots, which means that the HDI of the 20 countries I chose are
within either 1.5 times more or less of the median HDI.

With Outliers:

Without Outliers:
2000 2017

Upper whisker 103 122

3rd quartile 96.5 101.5

Median 93.8 96.35

1st quartile 66.55 80.4

Lower whisker 24.4 63

The median percentage of primary school completion of the 20 countries I chose has increased
from 93.8% in 2000 to 96.35% in 2017. Although the median only increased by 2.55%, the lower
whisker had greatly increased from 24.4% in 2000 to 63% in 2017. Along with the range in the
lower 25%, the overall range in 2000 has decreased from 78.6% to 59%, which shows that the
20 countries are starting to have a more similar percentage of primary school completion. The
range in the upper 25% (between the 3rd quartile and the upper whisker) has increased, which
suggests that more countries have a higher percentage of primary school completion than the
median. However, there is one outlier in each of the box plots, which means that one country
has an unusually low percentage of completion in primary school in both years compared to the
other 19 countries. It is very likely that the outliers in each box plot are the same country, and
although it has an unusually low percentage, the percentage in 2017 is closer to the lower
whisker than in 2000. Although 1 out of 20 countries have an unusually low percentage in both
years, the population of the other 19 countries have become more educated overall from 2000
to 2017.

2000 2017

Upper whisker 99.9 100

3rd quartile 96.9 96.9

Median 85.65 84.55

1st quartile 30.8 52.65

Lower whisker 8.56 16.1

The median overall access of at least basic sanitation has remained relatively the same, only
dropping by 1.1% from 2000 to 2017. Although the overall range decreased by 7.44% from
91.34% to 83.9%, the range in the middle 50% has decreased because the range in the lower
25% has increased. This suggests that the lower 50% are slowly catching up to the upper 50%.
There are no outliers, which means that the percentage of at least basic sanitation in all 20
countries are within 1.5 times of the median. The overall access of at least basic sanitation for
all 20 countries are relatively the same in 2000 and 2017, with visible change only in the range
of the lower 50%.

With Outliers:
Without Outliers:

2000 2017

Upper whisker 7.06 6.61

3rd quartile 3.56 3.48

Median 0.96 1.6

1st quartile 0.37 0.54


Lower whisker 0.07 0.2

The median CO2 emissions per person of the 20 countries I chose have increased by 0.64
tonnes per person from 0.96 tonnes in 2000 to 1.6 tonnes per person in 2017. Although this
shows that the overall CO2 emissions have increased, the upper whisker decreased from 1.06
tonnes in 2000 to 6.61 tonnes in 2017 by 0.45 tonnes per person. There is one outlier in the
2000 box plot, but the number increased in the 2017 box plot. This suggests that more countries
are beginning to emit 1.5 times more tonnes of CO2 per person than the median country.
Although the overall range remained relatively the same (decreasing slightly by 0.4 tonnes)
along with the interquartile range (decreasing by 0.25 tonnes), the range between the 1st
quartile and the median increased. There is a big range in the top 25% in 2000 and decreased
in 2017, albeit slightly by 0.37 tonnes. Although the increase of outliers and the value of the
median suggests that the 20 countries have emit more tonnes of CO2 per person overall, the
CO2 emissions per person remained relatively the same in both years in terms of the overall
range and the interquartile range.

Bivariate Statistics
In order to compare the data of my chosen index and the UN HDI, I have calculated the
correlation and will base the correlation strength on the correlation strength table below.

Pearson’s correlation Correlation strength

Very weak 0.00-0.19

Weak 0.20-0.39

Moderate 0.40-0.59

Strong 0.60-0.79

Very strong 0.80-1.0

The table below shows the correlation between each of my chosen index and the UN HDI in the
years I have chosen and the correlation strength.
Correlation 2000 Correlation strength 2017 Correlation strength

Primary school
completion (% of
relevant age group) vs
HDI 0.92 Very strong 0.58 Moderate

At least basic
sanitation, overall
access (%) vs HDI 0.76 Strong 0.84 Very strong

CO2 Emissions
(tonnes per person) vs
HDI 0.74 Strong 0.73 Strong

The table below shows the data of all the 20 countries and their GNI group, HDI, and each of
my chosen indices.

Year: 2000
Primary school At least basic CO2 Emissions
GNI completion (% of sanitation, overall (tonnes per
Country Group HDI relevant age group) access (%) person)

Burundi L 0.293 24.4 45.2 0.43

Mozambique L 0.301 16.6 10.5 0.07

Nepal L 0.446 66.5 15.1 0.13

Togo L 0.426 66.6 99.9 0.27

Burkina Faso L 0.286 25.4 11.3 0.09

Djibouti LM 0.361 26.7 47.2 0.51

India LM 0.497 71.5 16.4 0.98

Kyrgyz Republic LM 0.594 93.6 92.5 0.94

Moldova LM 0.609 97.8 74.3 0.85


Ghana LM 0.483 69.1 8.56 0.32

Mauritius UM 0.674 103 90.1 3.75

Cuba UM 0.686 95.5 87 2.33

Jordan UM 0.702 94.5 98.5 2.97

Samoa UM 0.638 94 97.6 0.82

Peru UM 0.679 96.8 64.2 1.14

Croatia H 0.749 90.9 96.2 4.45

Finland H 0.858 96.2 99.4 11.00

Estonia H 0.78 95.7 99.7 11.00

Seychelles H 0.712 98.9 94.1 7.06

Lithuania H 0.755 102 84.3 3.37

Year: 2017

Primary school At least basic CO2 Emissions


GNI completion (% of sanitation, overall (tonnes per
Country Group HDI relevant age group) access (%) person)

Burundi L 0.421 68.4 45.8 0.47

Mozambique L 0.442 47.1 29.4 0.28

Nepal L 0.574 122 62.1 0.33

Togo L 0.51 99.9 16.1 0.43

Burkina Faso L 0.429 63.5 19.4 0.20

Djibouti LM 0.492 63 63.6 0.67

India LM 0.643 94.4 59.5 1.84

Kyrgyz Republic LM 0.671 102 96.5 1.53

Moldova LM 0.709 89.9 76.3 1.18

Ghana LM 0.591 94.7 18.5 0.62


Mauritius UM 0.793 98.2 95.5 2.27

Cuba UM 0.777 89.5 92.8 2.48

Jordan UM 0.722 71.3 97.3 2.47

Samoa UM 0.706 107 98.2 1.29

Peru UM 0.756 97.4 74.3 1.68

Croatia H 0.835 95.5 96.5 4.48

Finland H 0.942 101 99.4 8.12

Estonia H 0.879 97.2 99.1 14.20

Seychelles H 0.8 102 100 6.61

Lithuania H 0.866 102 93.4 4.72

I will be explaining the results of the graphs below using the tables above.
As you can see from the graph above, the correlation between the percentage of the relevant
age group that completed primary school and the UN HDI in 2000 was stronger than that of the
correlation in 2017. The Pearson's correlation coefficient in 2000 is 0.92, and has decreased by
0.34 to 0.58 in 2017. Although both graphs are monotonic, the graph in 2000 is a curve that
levels off as the percentage of primary school completion and HDI increase. Countries with a
higher HDI and GNI have relatively the same percentage of primary school completion in 2000
and in 2017 because this index is measured in percentage, which means that those countries
have already reached the maximum percentage. There is one outlier that is significantly lower in
the year 2000, which may be because the country, Mozambique, was in the same GNI group in
2000 and in 2017. This means that countries with a lower GNI are less able to provide access to
basic education as they don’t have the money to do so. There is also an outlier that is
significantly higher than the trendline in 2017, and it may be because of a migration of another
country’s population into the country, therefore raising the percentage of primary school
completion to above 100%. The decrease in correlation between the two indices shows that the
percentage of primary school completion is becoming less relevant to human development as
basic education is becoming more and more accessible.
According to the graph above, the correlation between the percentage overall access of at least
basic sanitation and HDI have increased slightly from 2000 to 2017. The Pearson’s correlation
coefficient increased by 0.8 from 0.76 in 2000 to 0.86 in 2017. Although the slope of the
trendline both graphs are relatively the same, the data in 2017 are closer to the trendline than
the data in 2000. Both graphs are linear, and therefore monotonic. There are several outliers in
the 2000 graph, one of which is significantly higher than the trendline, and the others are
significantly lower. This could be due to the low GNI group the country belongs to, therefore
there are anaccuracies in measurement of the population due to lack of equipment and/or
underdeveloped healthcare technology and systems, leading to lack of access to information
about sanitation. The outliers in the 2017 graph are significantly lower than the trendline, which
could also be due to inaccurate measuremtn of the population. The correlation between the
percentage of overall access to at least basic sanitation has increased from 2000 to 2017 most
likely due to fewer outliers as a result of more accurate measurement methods.
The Pearson’s correlation coefficient between the tonnes of CO2 emissions per person and HDI
stayred relatively the same, only decreasing by 0.01, with 0.74 in 2000 and 0.73. Both of the
graphs above are monotonic and are curves that curve upward, which shows that there is not
much change in the data in 2000 and 2017. This could be due to wars that were happening in
less developed countries or due to their low income (and therefore low GNI group), therefore
limiting their access to energy sources that would emit carbon dioxide. Some countries that
have a higher HDI have a relatively consistent tonnes of emissions per person on average while
the carbon dioxide emissions have dropped for others, because they are countries belonging to
a higher GNI group that have started to look for more renewable energy sources, which could
also be seen as an improvement in human development. However, since the UN HDI doesn’t
include an index that focuses on technological advancement, there is no increase in the UN HDI
even though the slight drop in the average carbon emissions per person suggests that countries
are starting to look for renewable energy sources that would leave a smaller negative impact on
the environment.

Conclusion
From the univariate and bivariate statistics above, it can be concluded that my prediction that
the percentage of relevant age group that completed primary school and the percentage of
overall access to at least basic sanitation and the UN HDI have a positive correlation in both
years. All the 20 countries have an increase in the UN HDI and my chosen indexes overall.
Although there is a positive correlation between the percentage of relevant age group that
completed primary school and the HDI, the correlation has decreased over time, which means
that this index is becoming less relevant to human development. This may be because more
countries are giving easier access to basic education, therefore an index that focuses on
achievements in higher education may have a higher correlation to the HDI. On the other hand,
my prediction that the tonnes of carbon dioxide emissions per person and the UN HDI have a
negative correlation was proven to be incorrect. This could be because the carbon dioxide
emitted as a result of producing energy was used for human development purposes such as
building schools. This suggests that even though an increase in carbon dioxide emissions is
harmful for the environment, it is good for human development.

Spearman's Correlation between My HDI and UN HDI


I will be using the most recent year, 2017, to compare my HDI and the UN HDI. The graph
below shows the correlation between the ranks of countries in my HDI and in the UN HDI, which
is also the Spearman’s correlation.
The Spearman’s correlation between the UN HDI rank and my HDI in 2017 is 0.766, which is a
strong positive correlation. The correlation between my HDI and the UN HDI is only ‘strong’ and
not ‘very strong’ because the correlation strength between the percentage of relevant age group
that completed primary school and the HDI is only moderate. This is due to the fact that more
countries are able to give more acess to basic education, and that this index doesn’t consider
the achievements and completion of a higher education. Thus, the education index of my HDI
should also consider the percentage of population that completed an education higher than
primary school in order to portray the human development of countries more accurately. This is
because completion of a higher education means a higher percentage of population are more
educated in specific areas of work, and are able to improve society with their knowledge.

You might also like