ECON1193 – Business Statistics I

Asignment 2 - Inferential Statistics

Assigned topic: Mortality rate neonatal – data set 7

Word count: 2909

______________________________________________________________________________

Table of contents:

Part 1: Introduction

Part 2: Descriptive Statistics and Probability

Part 3: Confidence intervals (CI)

Part 4: Hypothesis Testing

Part 5: Conclusion

Part 6: References

Part 7: Appendices

1
Part 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . .>>>>> >>>..> >.>> >>>> >>> >>>> >>> >>>.>
Each year, 2.6 million babies die before they reach the age of one month. One million of them take
their first and last breaths on the day they are born. The neonatal phase- the first 28 days of life –
is the most dangerous time for a child’s survival (Unicef 2018). In fact, children face the highest
risk of deaths occuring in their first month of life at an global average rate of 17 deaths per 1,000
live births in 2019. Moreover, there was 2.4 million infants die in the first month of life, roughly
6,700 deaths a day, including over a third of all neonatal deaths happened within the first day and
nearly three-quarters happened during the first week of life (Unicef 2019).

Additionally, newborn survival is inextricably tied to a Gross National Income (GNI). In

comparison, high-income countries have an average newborn mortality rate of just 3.3 while low
income countries possess a rate of 27, which is considered as a substantial difference (Unicef
2018). Particularly, the main causes of newborn death are inextricably related to the mother's well
being and the treatment received during pregnancy and childbirth (Neal & Falkingham 2014).
Moreover, Unicef (2018) stated that increasing access to maternal and newborn health services is
a vital first step toward lowering infant mortality rates while high income levels indicate that
financial resources are available to invest in good health systems. This demonstrates that a
country's income level has a significant impact on its neonatal death rates.

The 3rd SDG of The UN is to ensure healthy lives and promote well-being for everybody at all
ages (UN 2021). The target 3.2 of the 3rd goal aimed to reduce neonatal mortality to at least 12 per
1,000 live births by 2030 worldwide. However, with 7,000 infants dying every day, progress in
reducing the death rate of newborn babies has been slow. It is partly due to the difficulty of
addressing neonatal fatalities with a single medicine or remedy — they necessitate a system-wide
approach. It is also because of a lack of momentum and international commitment to neonatal
survival (Unicef 2018). In other words, greater attention must be focused on newborn babies, the
neonatal mortality needs to be accelerated as its reduction can contributes greatly to the
achievement of this goal.

Overall, this paper will investigate the relationship between the GNI - the neonatal death rate and
the connect to SGD 3 by applying statistical analysis within a data set of 25 sample countries.

Descriptive Statistics and Probability

a. Probability
Low-Income countries (LI) GNI less than $1,000 per capita

2
Middle-Income countries (MI) GNI between $1,000 and $12,500 per capita

High-Income countries (HI) GNI greater than $12,500per capita

Figure 1: Table of categories of Low-income, Middle-income, and High-income counties
Mortality rate neonatal

High rate Low rate Total

Country Income Low-Income (LI) 4 0 4

Middle-Income (MI) 5 8 13

High-Income (HI) 0 8 8

Figure 2: Contingency table of country categories in terms of GNI per capita (current
$US) and mortality rate neonatal (per 1,000 people)

World Bank (n.d) stated that the mortality rate neonatal is the number of newborns that die before
they turn one year old per 1,000 live births in a particular year. The given dataset of 25 countries’
mortality rate neonatal is categorized as high and low rate segments as the rate exceeding 15 per
1,000 live births is defined as high rate. Further, the inspected nations are also classified into three
groups regarding their Gross National Income (GNI) per capita (current US$).

Test of statistical dependence:

To determine whether two events are statistically independent; theoretically, the probability of A,
P(A), must equal the probability of A given B, P(A|B). Mathematically, it can present as P(A) =
P(A|B) (RMIT University n.d). Applied to this case, the test compares a probability of countries
having a high mortality rate neonatal P(H) with a conditional probability of countries having high
rates given that they are high-income countries, P(H | HI).

9
��(��) =
25 = 0.36
��(��│����)
= = ��(����)
��(��
������
025
����)
825= 0

3
The result demonstrates that P (H | HI) ≠ P (H), high GNI and high
mortality rate neonatal are statistically dependent events. This
indicates that the probability of high newborn death rate is affected
by the event of the countries that have high GNI.

Likewise, the same idea is shared with the categories of Middle and Low-Income countries, given
that the death rate of infants is high and the death rate of infants is low. In other words, all possible
outcomes are tested.
��
���� 5
��(� )
� 5
���� 25
��(��│����) = = =
��(� 13
25
���)
13

→ ( ) ��(��│����) ≠ ��(��) 1 ≠ 0.36

����)
��(�� 4
���� 25
��
��(��│����) = = =
��(�� 4
��) 251
→ ( ) ��(��│����) ≠ ��(��) 1 ≠ 0.36

16
��(��) =
25 = 0,64
����)
��(�� 8
���� 25
��
��(��│����) = = =
��(�� 8
��) 251
→ ( ) ��(��│����) ≠ ��(��) 1 ≠ 0,64
�� 8
��(� ���� 25
� )
���� 8
��(��│����) = = =
��(� 13
25
���)
13

→ ( ) ��(��│����) ≠ ��(��)813 ≠ 0,64

����)
��(��
0
���� 25
��
��(��│����) = = =
��(�� 4
��) 250
→ ( ) ��(��│����) ≠ ��(��) 0 ≠ 0,64

4
The calculation denotes that the probability of countries having high mortality rate neonatal is
different from the probability of countries having high mortality rate neonatal given that they have
high, middle, or low income. Similarly, the same results are witnessed in the chance of low
mortality rate neonatal compare to this but given that the GNI is high, middle, or low. Thus, it leads
to the conclusion that the Gross National Income and newborn death rate are statistically dependent
events. Expressly, the infant death rate closely relates to country income.
)==
��(����)
025
��(������ 8 = 0
25
������)
��(��│����
����
���) 5
��(�
5
���� 25
��(��│����) = = =
��(�
���)
 ��(� ���) 425
���� 13
25 13
����

��(��│����) = = =
��(�� 4
��) 251

→ �� (�� | ����) < �� (�� | ����) < �� (�� | ����) ( 0

5
< 13 < 1 )

The comparison reveals that high-income nations have the likelihood to obtain a lower newborn
death rate as 0% compare to middle and low-income nations, which are 38,46% and 100%,
respectively.

b. Descriptive Statistics:
Min >,<,= Lower Max >,<,= Upper Outlier
limit limit

HI 0.9 > 0.8875 2.6 < 3.1875 no outlier

MI 6.1 > -3.45 24.6 < 32.15 no outlier

LI 26.5 > 17.425 37.4 < 42.425 no outlier

Figure 3: Comparison between upper threshold, lower threshold and outliers of neonatal
mortality (per 1,000 live birth).

The table above conducted an outlier test, showing that the dataset does not have any outliers
over 25 observations within 3 categories of high, middle, and low-income.

5
High income Middle Low income
income

Mean 1.93 14.29 30.6

Median 2.1 12.7 29.25

Mode 2 9.9 -
 Figure 4: Table for central tendency of neonatal mortality (per 1,000 live birth).

Since there is no extreme value in the dataset; the Mean is not drawn further away from the center
therefore the interpretations of the Mean remain accurate. Further, it includes every value in your
data set as part of the calculation, making it the most represetative out of three central tendency
values. Therefore, Mean is the most appropriate measurement in the case.

The Means of three categories are remarkably different from the others. In particular, the Average
point of middle-income is nearly 7.5 times higher than the high-income corresponding value.
However, it just account for nearly a half of low-income Average value. It can be interpreted that
there is a reduction of the average newborn death rate in wealthier countries group, especially a
high-income segment notices an extreme lower value than others 2 groups.
High income Middle Low income
income

Standard 0.63 6.04 5.09

Deviation

Sample 0.40 36.54 25.91

Variance

Coefficient 32.94 42.29 16.64

of
Variation
(%)

Figure 5: Table for measures of variation of neonatal mortality (per 1,000 live birth).

According to Figure 2, the Means of 3 groups are considerably different from each other, which
may cause the misleading interpretation of Standard Deviation. On the other hand, the Coefficient
of Variation (CV) is effectively utilized to compare the relative dispersion between two datasets as
it is the ratio of the Standard Deviation to the Mean. Therefore, the use of CV can eliminate the
obstacle of SD in this case, meaning that this is the most preferable measurement.

6
In comparison, the group of middle-income countries has the highest ratio of CV (42.29%),
followed by high and low-income countries, which are 32.94% and 16.64%, respectively. The
percentages indicate that the middle and high-income nations witnessed a significantly wider
fluctuation around the Average values, compared to the low-income ones. The rate in these 2 two
segments are more variable than the low-income one.

Part 2: Confidence Intervals

a. Calculation

The confidence level is chosen at 95%; therefore, the corresponding level of confidence (α) is
0.05
Significance level �� 0.05

Confidence level (1- ) �� × 100% 95 %

Population standard deviation �� unknown

Sample standard deviation S 10.73 per 1,000 live births

Sample mean X 12.94 per 1,000 live births

Sample size n 25 countries

Degree of freedom d.f. 24

t-critical value t ± 2.0639

Figure 6: Statistics summary table for mortality neonatal rate

�� = �� ± ��(����)

⇒�� = 12.94 ± 2.0639 ×10.73

25

⇒ 8.51 ≤ �� ≤ 17.37

Conclusion: We are 95% confident that the world mortality rate neonatal falls between 8.51 and
17.37 deaths per 1,000 live births

b. Discussion on assumptions:

7
Due to the fact that the standard deviation of population is unknown, we need to check for the
Central Limit Theorem (CLT). However, the sample size (n) is 25, which is less than 30
observations within the dataset, the CLT is not applicable. Thus, we need to assume that the
population is normally distributed.

c. The possible impact on the confidence interval results

Supposing that the standard deviation of the global newborn death rate is known, the Z-value table
can be applied instead of the T-value table. In comparison, a T-distribution is located more toward
the tail than Z for a normal distribution, resulting in “fatter" and shorter bell-shaped of T (Appendix
1). To clarify, the Z-score is measured in terms of Standard Deviations from the Mean of
population, T-score is used as same as Z-score but for a sample. However, the Standard Deviation
of the sample is random variables while population parameters remain constant (Naghshpour &
Shahdad 2012). This indicates that sample SD have a greater uncertainty compared to population
SD.

On the other hand, the width of the confidence interval will be smaller when the underlying
population has a smaller standard deviation since more variability makes sample statistics less
reliable. The wider the confidence level, the less precise the estimate (Panik & Michael 2012).
Therefore, as the world standard deviation of Mortality rate neonatal is identified, the result
presents more accurately.

Part 3: Hypothesis Testing

In accordance with the World Health Organization report, the world average mortality rate
neonatal is 18.6 deaths per 1,000 live births in 2016. Nonetheless, the calculation of Confidence
Intervals above forecasted that we are 95% confident that the true world average death rate of 2017
locates somewhere between the range of 8.51 and 17.37. To put these statistics in comparison, it
is exposed that the previous year data (18.6) is higher than the interval scale, denoting a great
chance that the mean in 2017 will decrease.
Significance level �� 0.05

Confidence level (1- )*100% 95 %

Population standard deviation �� unknown

 Sample standard deviation S 10.73 per 1,000 live births

Population mean 18.6 per 1,000 live births

8
Sample mean X 12.94 per 1,000 live births

Sample size n 25 countries

a. Hypothesis Testing (Critical Value Approach)

Step 1: Check for Central Limit Theorem (CLT)

Since the sample size (n) is 25, lowering than 30, the CLT can not be applicable, we must
assume that the sampling distribution is normally distributed.

Step 2: State null and alternative hypotheses:

o The null hypothesis Ho: �� ≥ 18.6

o The alternative hypothesis H1: �� < 18.6

Step 3: Choose the level of significance and determine test type

In corresponding with the chosen confidence level (0.95), the level of significance is computed
as 0.05. Given the sign that H1: �� < 18.6, the test type is a one-tailed test - lower tail.

Step 4: Determine table type

As the population standard deviation (��) is unknown and the sampling distribution is
normally distributed, the t-table is used.

Step 5: Determine critical value (CV):

o Degree of freedom: ��.�� = �� ― 1 = 25-1 = 24

o Level of significance: �� = 0.05
o Lower-tailed test: t = - 0.711
Step 6: Calculate test statistics
18.6
�� ― ��
=12.94 ―
��′ = -2.64
��
��
25
10.73 =

Step 7: Make a statistical decision:

9
t-critical value > t’ (- 0.711 >-2.64) => t' falls into the Rejection Region, we reject H0

. Step 8: Explaination

As the null hypothesis H0 is rejected, thus we are 95% confident that the world mean of death
rate will possibly decline in the future.

Step 9: Determine the type of error

The hypothesis test has the probability to commit a type I error when H 0 is rejected. This type of
error mistake to ignore to the true result (RMIT University n.d). If type I error occur, we reject
that the Mean value will increase in the future, but in fact, it is correct.

b. Possible impacts of decrease a half of sample size on the hypothesis testing results:

Supposing that the number of countries of your dataset will become a half, there will be some
impacts as some change are made. Initially, as the sample size n decreases, the margin error in the
CI will increase (Appendix 2), resulting in the less accuracy of estimation (Naghshpour & Shahdad
2012). Further, looking at the equation of the CI:

�� = �� ± ��(����)

It demonstrate that the sample size (n) has an inverse ratio with the CI ( ), the contracting in n will
�� naturally rise . Notably, in the hypothesis test, we based on the CI to estimate the true Mean
of �� newborn death rate in 2017, and compare with the given value of 2016 to make an
assumption. Hence, the change of CI range can seriously lead to the wrong assumption as presented
in these two cases:
+ If the CI range increase and including the value 18.6 (2016 Average rate), the logical assumption
should adjust to a great chance that the mean in 2017 will remain unchanged as the uncertainty is
build up, we can not know axaclty which Average points is the higher. Therefore, the old
assumption need to be replaced.

+ If the CI range increase but not including the value 18.6, the assumption is not affected since the
Mean of 2016 still seems to be higher than 2017, as compare to 2017 CI. However, as mentioned
above, wider range can result in the less accuracy.

Part 4: Conclusion

10
The calculation and analysis of a sample of 25 countries can tell a lot and give an idea about the
whole population parameters. Initially, the figures of probability have proved that GNI of a
country and their mortality rate neonatal are statistically dependent events. It means there are
certain linkages between these two events. Specifically, the level of GNI, which changes from
high to low, has a direct impact on the newborn death rate. There is a tendency that the higher
the income, the lower the death rate, completely aligning with the findings in the beginning. In
detail, an exact ratio of probabilities gives a transparent view about the newborn death rate in
these countries. While 100% of low-income regions obtain a high risk of death in the infant
phase ( as 15 per 1,000 live births), these places are all lacking in medical facilities for mothers
and children. Further, urgent support should be placed there, in order to achieve the UN third
goal of reducing newborn mortality rate globally by 2030.

Secondly, in the descriptive statistics part, the interpretation of Mean has shown that the average
death rate decrease from the high to low-income groups, which strengthens the findings from the
probability calculations. Also, it demonstrates that the high-income nations have a very low rate
(1.93), compared to two other segments (14.29 & 30.6). As the average rate of the middle group
nearly reaches 15, which is a point to be considered as high rate, showing the fact that the surge
in the income of this class is still not enough to obtain an impressively positive impact on
newborn mortality. Moreover, the rate of the poorest group (30.6) is even more than a double of
the point 15 per 1,000 live births. The number could raises an awareness for the alarming reality
in these regions.

Lastly, the confidence intervals suggest that 95% the Mean is within the range between 8.51 and
17.37 birth lives per 1000 babies. In comparison with the average death rate of 2016 (18.6), 2017
has downward trend, as reinforced in the hypothesis part. Linking to the third goal of UN, the
calls for reducing newborn mortality rate to 12 per 1,000 live births by 2030 (UN 2021), the
statistic has shown that the world is on pace to achieve this goal. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . .
 11
References

Damasceno B. (2020) Hypothesis Testing. In: Research on Cognition Disorders. Springer,

Cham. https://doi.org/10.1007/978-3-030-57267-9_16 (11)

Damasceno, B 2020, ‘Hypothesis Testing’, in: Research on Cognition Disorders (eds), State
University of Campinas, Campinas, pp. 157-166

Naghshpour & Shahdad 2012, Statistics for Economics, Business Expert Press, ProQuest Ebook
Central.

Neal, S & Falkingham, J 2014, ‘Neonatal Death and National Income in Developing Countries:
Will Economic Growth Reduce Deaths in the First Month of Life?’, International Journal of
Population Research, vol. 2014, https://doi.org/10.1155/2014/989485

Panik & Michael, J 2012, Statistical Inference : A Short Course, John Wiley & Sons,
Incorporated, ProQuest Ebook Central.

Sample Size Determination and Power, First Edition. Thomas P. Ryan. © 2013 John Wiley &
Sons, Inc. Published 2013 by John Wiley & Sons, Inc.

The World Bank 2019, Mortality rate, neonatal (per 1,000 live births), The World Bank, viewed
19 August 2021, < https://data.worldbank.org/indicator/SH.DYN.NMRT> .

Thomas, P 2013, Sample Size Determination and Power, in Thomas P(ed.), John Wiley & Sons,
Inc., viewed 20 August 2021, < https://onlinelibrary-wiley
com.ezproxy.lib.rmit.edu.au/doi/pdfdirect/10.1002/9781118439241.ch1>.

UN 2021, #Envision2030 Goal 3: Good Health and Well-being, UN, viewed 20 August 2021,
https://www.un.org/development/desa/disabilities/envision2030-goal3.html.

UN 2021, Goal 3: Ensure healthy lives and promote well-being for all at all ages, UN, viewed
20 August 2021, < https://unric.org/en/sdg-3/>.

Unicef 2018, Every child alive - The urgent need to end newborn deaths, Unicef, viewed 19
August 2021, < https://data.unicef.org/resources/every-child-alive-urgent-need-end-newborn
deaths/>.

Unicef 2019, Goal 3: Ensure healthy lives and promote well-being for all at all ages, Unicef,
viewed 19 August 2021, < https://unstats.un.org/sdgs/metadata/files/Metadata-03-02-02.pdf>.
Volkmar, F 2021, ‘Type I Error’, in Volkmar F.R. (eds), Encyclopedia of Autism Spectrum
Disorders, Springer.
 12
Appendices

Appendix 1: Three t-distributions and a standard normal (z-) distribution (JMP 2021)

Appendix 2: Margin for Error (Lumen n.d)

13