Measure of Development:

Education Index

Submitted To: Dr. Javed Iqbal

Name ERP ID Class ERP Percent Signature

Contribution
Contents
1. Introduction ................................................................................................................................. 2
1.1 Education Index Defined ......................................................................................................... 2
1.2 Research Report Outline ......................................................................................................... 2
2. Comparative Analysis .................................................................................................................. 3
2.1 Descriptive Statistics ............................................................................................................... 3
2.1.1 Mean values of the data sets ............................................................................................. 3
2.1.2 Median values of the data sets ......................................................................................... 3
2.1.3 Standard Deviation values of the data sets ...................................................................... 4
2.1.4 Skewness values of the data sets ..................................................................................... 4
2.1.5 Coefficient of Variation (CV) values of the data sets ...................................................... 4
2.2 Histograms .............................................................................................................................. 5
2.2.1 Americas ........................................................................................................................... 5
2.2.2 Africa ............................................................................................................................... 6
2.2.3 Asia .................................................................................................................................. 7
2.2.4 Europe ............................................................................................................................. 8
2.3 Box and Whisker Plot .............................................................................................................. 9
2.4 Remarks on Descriptive Statistics and Possible Reasons ......................................................... 9
3. Regression Analysis ................................................................................................................... 11
3.1 HDI and Education Index ....................................................................................................... 11
3.1.1 Statistical Analysis .......................................................................................................... 11
3.1.2 Interpretation ................................................................................................................. 12
3.1.3 Test of Model (Predictive Performance of the Model) ................................................. 13
3.2 GDP Per Capita and Education Index ..................................................................................... 13
3.2.1 Statistical Analysis ......................................................................................................... 13
3.2.2 Interpretation ................................................................................................................ 14
3.2.3 Test of Model (Predictive Performance of the Model)................................................. 15
3.3 Comparing the Models .......................................................................................................... 15
4. Conclusion ................................................................................................................................. 15
5. Data Sources and Software Used ............................................................................................... 16

1|P a ge
1. Introduction
1.1 Education Index Defined
An Education Index is a component of the Human Development Index published every year
by the United Nations Development Programme. The United Nations Development
Programme obtains Education Index metrics by measuring the mean of years of schooling for
adults aged twenty five and over and expected years of schooling for children of school entering
age. The reader must keep in mind that the lowest Education Index score that can be received
by a country is 0 and the highest is 1.

1.2 Research Report Outline

This Research Report is focused on comparing the Education Indices in 2019 of the following
Major Regional Groups:
1. Americas (North and South)
2. Africa
3. Asia
4. Europe
Countries in Oceania have been excluded because Oceania has fewer than fifteen countries.
Two methodologies have been employed to carry out these comparisons:
1. Comparative Analysis: In this section, the use of tables, box plots and histograms is
employed to ascertain and understand the quantitative and visual differences between
education quality in each of these regions.
2. Regression Analysis: In this section, the correlation between Education Index and
another factor such as Human Development Index ranking is obtained and a theoretical
model is built to relate the variables. The effectiveness of the model is then put to the
test by obtaining expected data sets and comparing them to real life data sets.

2|P a ge
2. Comparative Analysis
2.1 Descriptive Statistics

Americas Africa Asia Europe

Mean 0.6933 0.4842 0.6564 0.8363
Standard Error 0.0167 0.0172 0.0209 0.0107
Median 0.695 0.486 0.678 0.8315
Mode 0.74 0.557 0.678 0.711
Standard Deviation 0.0991 0.1257 0.1403 0.0712
Sample Variance 0.0098 0.0158 0.0197 0.0051
Kurtosis 0.6129 -0.6861 -0.6876 -0.9860
Skewness -0.2006 0.0841 -0.3359 -0.3100
Range 0.444 0.487 0.533 0.239
Minimum 0.456 0.249 0.35 0.704
Maximum 0.9 0.736 0.883 0.943
Sum 24.268 25.667 29.54 36.799
Count 35 53 45 44
Coefficient of Variation 0.1430 0.2596 0.2137 0.0851
Table 1: Descriptive Statistics

The table above does an excellent job of comparing the Education Index values of the four
different geographic regions. Of all the measures presented above, we are particularly
interested in comparing the values of the:
1. Mean
2. Median
3. Standard Deviation
4. Skewness
5. Coefficient of Variation

2.1.1 Mean values of the data sets

Mean is the average of any data set and is one of the most reliable measures of comparing data
sets. From Table 1 given in Section 2.1, it can be concluded that Africa has the lowest education
quality (with a mean Education Index of 0.4842) whereas Europe has the highest Education
Quality (with a mean Education index of 0.8363). The mean Education Index of the Americas
(0.6933) and Asia (0.6564) is quite close.

2.1.2 Median values of the data sets

Median is another metric used to compare data sets. It is obtained by arranging the data in
ascending order and then picking the “middle value”. In other words, half the data is above the
median value and the other half is below. Moreover, median is resistant to presence of outliers.
3|P a ge
From Table 1, it can be observed that Europe has the highest median Education Index (0.8315),
while Africa has the lowest median Education Index (0.486). Asia (median Education Index:
0.678) and the Americas (median Education Index: 0.695) receive a close score.

2.1.3 Standard Deviation values of the data sets

This is a measure used to illustrate the spread of the data around the mean value. From Table
1, one can see that the Education Index of European Countries is closely clustered around the
mean Education Index for Europe, owing to the low standard deviation of 0.0712. This implies
that there is less Education Inequality between European Countries than countries in the
Americas (standard deviation: 0.0991), African countries (standard deviation: 0.1257) and
Asian countries (standard deviation: 0.1403).

2.1.4 Skewness values of the data sets

Skewness is a metric that helps in visualizing the shape of the distribution. A distribution is
negatively skewed if the skewness is less than -0.5. On the other hand, a positively skewed
distribution has skewness greater than +0.5. If the skewness value lies between -0.5 and +0.5
(both values inclusive), the distribution is said to be approximately symmetric. Based on this
criteria, the distribution of Education Index scores of all regions is approximately symmetric.

2.1.5 Coefficient of Variation (CV) values of the data sets

CV can be calculated by dividing the standard deviation by the mean. A lower CV indicates a
more consistent data set and is a better means of comparing variation when mean is different.
Therefore, the data set for Europe is most consistent and the data set for Africa is least
consistent.

4|P a ge
2.2 Histograms

2.2.1 Americas

Education Index

Figure 1

The distribution of Education Indices in the Americas seems to be approximately symmetric.

Majority of the countries had an Education Index between 0.6 and 0.8 (26 out of 35). Only one
country had an exceptional education system, with an Educational Index of more than or equal
to 0.9.

5|P a ge
2.2.2 Africa

Education Index

Figure 2

The Histogram shown above indicates that the distribution of the value of Education Index in
African countries is approximately symmetric or very slightly positively skewed. More than
half of the countries have values of Education Index below 0.5, which indicates that most of
Africa has a poor learning infrastructure in place. No country in Africa has an Education Index
of more than 0.8. Generally, this means that even the best education systems in Africa are
nowhere near as good as the average standard for Europe.

6|P a ge
2.2.3 Asia

Education Index

Figure 3

The distribution above, at first glance looks heavily negatively skewed. But, the mean value
lies at the center of the 0.6-0.7 class interval and so distribution is approximately symmetric,
but less symmetric than that of the Americas, Africa and Europe. Most (12) countries in Asia
have an Education Index in the class of 0.6-0.7, but there is no country with excellent Education
Index value greater than 0.9. There are however, many countries that delivered subpar
education. 14 out of 45 countries have an Education Index of less than 0.5, which is a poor
score.

7|P a ge
2.2.4 Europe

Education Index

Figure 4

The x-axis of the Histogram for European countries shows that Europe has the best Education
System in place out of all four regions analyzed in this report. The distribution in seen to be
approximately symmetric. No country in Europe has an Education Index lower than 0.7. A
staggering 12 countries have excellent education systems as shown by them having an
Education Index score of more than 0.9.

8|P a ge
2.3 Box and Whisker Plot

Figure 5

A box and whisker plot is one of the best ways to compare data sets qualitatively in a
standardized manner. The diagram above shows Europe leading the line in terms of quality of
Education, followed by the Americas, Asia and then finally Africa. The box and whisker plot
also helps showcase measures like the mean, median and Interquartile range of the data sets.
Asia has the greatest Interquartile range, which reflects it having the highest standard deviation
within its data set for Educational Index. Africa follows Asia on second place in this regard.
Data for Asia is also the least symmetric out of all four geographic regions. The Americas and
Africa have a symmetric relation as shown by the differences between the median value and
the first and third quartile on the box and whisker plot. It is interesting to note that the American
geographic region also has outliers. This is because the range of this data is much larger than
the Interquartile range and it also has a large and sharp peak in its distribution at its center.

2.4 Remarks on Descriptive Statistics and Possible

Reasons
From the Comparative Analysis conducted so far, it is abundantly clear that Europe has the
greatest education system followed by Americas, Asia and then Africa. Mean and median are
two excellent measures that highlight these differences. The mode metric is found to not be
very suitable. The data used in this analysis consists of numerical values of 3 decimal places.
It is clear that the mode therefore may coincidentally appear at the “ends” of the data
distribution and may give a misleading view about the data. Europe has the lowest standard
deviation of Educational Index values whereas Asia has the highest. Africa comes a close
9|P a ge
second in this regard. All distributions are, according to the Skewness coefficients,
approximately symmetric.
There are multiple reasons for the data to look like the way it does. Firstly, Africa is the most
politically unstable geographical region, while Europe is the most stable. European countries
are richer and have a higher budget for Education than countries in other geographic regions.
Moreover, diseases and food shortages are very common in Africa and cripple the growth of
the education sector there. Asia is a very diverse region, with systems of education and country
HDI (Human Development Index) varying greatly between countries. This gives rise to the
large standard deviation and large range of Education Index data. As shown on the box and
whisker plot, the American geographic region has outliers and may be due to the diversity of
education standards, GDP (Gross Domestic Product), etc. between North and South America.

10 | P a g e
3. Regression Analysis
In this section of the report, the real life relationship between Education index for a country
and two other variables (each one at a time) will be investigated. Then, two separate models
will be formulated and tested. The two variables being taken are HDI (Human Development
Index) and GDP (Gross Domestic Product) per capita. It is important to note that Education
Index values, HDI ranks and GDP per Capita numbers are all taken from the year 2019.

3.1 HDI and Education Index

HDI is a “summary measure of human development” and can help people judge a particular
country’s standard of living. In this regression analysis, a theoretical relationship between HDI
and Education index will be established.

3.1.1 Statistical Analysis

Here are the estimates of the regression model:

Regression Statistics
Multiple R 0.952703903
R Square 0.907644727
Adjusted R Square 0.907116983
Standard Error 0.052802642
Observations 177

ANOVA
df SS MS F Significance F
Regression 1 4.795164905 4.795164905 1719.85662 1.89885E-92
Residual 175 0.487920824 0.002788119
Total 176 5.283085729

Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 0.939498046 0.007885565 119.1414981 1.8689E-169 0.923934996 0.955061096 0.923934996 0.955061096
HDI Rank -0.00299486 7.22155E-05 -41.47115407 1.89885E-92 -0.003137385 -0.002852334 -0.003137385 -0.002852334

11 | P a g e
The Regression Line:

Average of all Education Indices = 0.656915254

Average of all HDI Ranks = 94.3559322

3.1.2 Interpretation
The R-squared value is 0.907644727. Hence, approximately 90.76% of variation in Education
Index is explained by HDI rank in 2019. Square root of R-squared = 0.9527. The sign of the
gradient of the regression line is negative. Hence, correlation coefficient = -0.9527. Therefore,
this is a very strong, negative linear correlation. Hence, HDI rank is a very good predictor of
Education Index.
The equation of the regression line is:
y = -0.003x + 0.9395
This implies that for every one unit increase (this corresponds to one step down on the HDI
list) for a country in 2019, the Education Index of that country falls by 0.003 units. A country
with HDI rank of 0 in 2019 (which implies that country is highest rated on HDI list) will have
an Education Index of 0.9395. This is not practically possible since the ranking starts at position
1.
Elasticity of Education Index with respect to HDI rank
= Gradient x Average of all HDI Ranks /Average of all Education Indices

12 | P a g e
 = -0.003 x 94.3559322/0.656915254
= -0.431
Thus a one percent increase in HDI rank of a country is associated with a decrease of Education
Index value by 0.431% on average.

3.1.3 Test of Model (Predictive Performance of the Model)

Country HDI (Human Predicted Actual Residual

Development Education Education
Index) in 2019 Index (2019) Index (2019)
Brazil 84 0.6875 0.694 0.0065
Egypt 116 0.5915 0.618 0.0265
Pakistan 154 0.4775 0.402 -0.0755
United 13 0.9005 0.928 0.0275
Kingdom
United States 17 0.8885 0.9 0.0115

3.2 GDP Per Capita and Education Index

GDP per capita is a measure of the productivity of a country per person. According to the
World Bank, “GDP per capita is the sum of gross value added by all resident producers in the
economy plus any product taxes (less subsidies) not included in the valuation of output, divided
by mid-year population”.

3.2.1 Statistical Analysis

Here are the estimates of the regression model:
Regression Statistics
Multiple R 0.588525713
R Square 0.346362515
Adjusted R Square 0.342627444
Standard Error 0.140473137
Observations 177

ANOVA
df SS MS F Significance F
Regression 1 1.829862862 1.829862862 92.73250325 7.03888E-18
Residual 175 3.453222867 0.019732702
Total 176 5.283085729

Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 0.589326297 0.012678606 46.48194885 2.1278E-100 0.564303643 0.614348951 0.564303643 0.614348951
GDP per Capita/ $1000 0.004417328 0.000458716 9.629771713 7.03888E-18 0.003512001 0.005322655 0.003512001 0.005322655

13 | P a g e
The Regression Line:

Average of all Education Indices = 0.656915254

Average of GDP per Capita/ $1000 = 15.3008692

3.2.2 Interpretation
The R-squared value is 0.3464. Hence, approximately 34.64% of variation in Education Index
data is explained by GDP per Capita in 2019. Square root of R-squared = 0.589. The sign of
the gradient of the regression line is positive. Hence, correlation coefficient = +0.589.
Therefore, this is a moderate, positive linear correlation.
The equation of the regression line is:
y = 0.0044x + 0.5893
This implies that for every $1000 increase in GDP per capita for a country in 2019, the
Education Index of that country increases by 0.0044. A country with GDP per Capita of $0 in
2019 (which implies that that country has the lowest possible GDP per Capita) will have an
Education Index of 0.5893. This is practically not possible as any country must have some
Gross Domestic Product if it exists.
Elasticity of Education Index with respect to GDP per Capita in Thousands of Dollars
= Gradient x Average of GDP per Capita in Thousands of Dollars /Average of all Education
Indices
= 0.0044 x 15.3008692/0.656915254
= 0.102

14 | P a g e
Thus a one percent increase in GDP per Capita of a country is associated with an increase of
Education Index value by 0.102% on average.

3.2.3 Test of Model (Predictive Performance of the Model)

Country GDP per Predicted Actual Residual

Capita/ $1000 Education Education
(2019) Index (2019) Index (2019)
Brazil 8.936 0.6286 0.694 0.0654

Egypt 3.161 0.6032 0.618 0.0148

Pakistan 1.169 0.5944 0.402 -0.1924
United 42.628 0.7769 0.928 0.1511
Kingdom
United States 64.949 0.8751 0.9 0.0249

3.3 Comparing the Models

The Education Index vs HDI rank model has a coefficient of determination (R-squared) value
of 0.9076 in 2019. The Education Index vs GDP per Capita model has a coefficient of
determination (R-squared) value of 0.3464 in 2019. Also, the average value of residuals (all
taken to be positive) for Education Index vs HDI rank model is lesser (value: 0.0295) than the
same value obtained for Education Index vs GDP per Capita model (value: 0.08972) in 2019.
Therefore, for the year 2019, Education Index vs HDI rank model is a better model and
predictor for Education Index as compared to The Education Index vs GDP per Capita model.

4. Conclusion
From the analysis conducted in this report, it is abundantly clear that Europe has the best
education systems out of all the geographic regions analyzed with not only the greatest average
Education Index, but also the least variation and most consistency (data-wise). Africa, on the
other hand, is a long way from establishing good education systems for all the opposite reasons.
Americas and Asia were about average in their performance and that may be due to the diversity
present in those regions.
The regression analysis clearly showed that the Education Index and the Human Development
Index were very strongly positively correlated. One might expect GDP per Capita to greatly
affect Education Index, but it does not. Clearly, income inequality is a big reason why this is
so. Other factors such as political stability, investment and health systems also play a big role
in determining how educated a country is.

15 | P a g e
5. Data Sources and Software Used
HDI rank and education Index Data for 2019 has been gathered from the United Nations:
http://hdr.undp.org/en/data#
List of Countries and corresponding continents has been taken from:
https://github.com/dbouquin/IS_608/blob/master/NanosatDB_munging/Countries-
Continents.csv
GDP per Capita data for 2019 has been obtained from:
https://unstats.un.org/unsd/snaama/Downloads
HDI (Human Development Index) Definition:
https://databank.worldbank.org/metadataglossary/africa-development-
indicators/series/UNDP.HDI.XD#:~:text=The%20Human%20Development%20Index%20(H
DI,a%20decent%20standard%20of%20living.
GDP per Capita Definition:
https://databank.worldbank.org/metadataglossary/statistical-capacity-
indicators/series/5.51.01.10.gdp#:~:text=Metadata%20Glossary&text=Long%20definition-
,GDP%20per%20capita%20is%20the%20sum%20of%20gross%20value%20added,GDP%20
data%20in%20local%20currency.
Data Analysis has been carried out using Microsoft Excel

16 | P a g e