The rate of maternal
mortality, is extremely high in underdeveloped African countries, as well as
the population diagnosed with HIV. As HIV is the root of the deaths of many
residents, it would be interesting to investigate whether or not maternal
mortality rate is dependent on HIV. The correlation/relationship between
maternal mortality and HIV will be investigated throughout the investigation. The
independent variable is HIV whereas, the dependent variable is maternal
The data collected for the
investigation is secondary data. 48 African underdeveloped countries were used
for both HIV and maternal mortality. The first variable is HIV the age group is
15-49. The second variable is maternal mortality rate it is per 100.000. The
age group for HIV is appropriate, as people in most cases do not get pregnant
after the age of 49.
This investigation will
Scatterplot- a preview of
whether or not maternal mortality is affected by HIV
Box plots- for both HIV and
maternal mortality to show the distribution of data
test- Investigate whether or not maternal mortality in dependent on HIV.
The secondary data for the two variables: HIV and
maternal mortality was taken from globalis.dk.
This website shows/compares a country’s most important
info with statistics.
The African underdeveloped countries that contained
data for both HIV and maternal mortality were picked, and this included data
from 48 African underdeveloped countries. The reason for this is that the rate
of HIV and maternal mortality was higher in these African countries than any
other European, Asian etc. countries.
unit of measure for HIV positive population was in %, between the ages 15-49.
Whereas, for maternal mortality it was rate per 100.000.
Beneath, is a table showing the raw data of the two
variables: HIV and maternal mortality in 48 African underdeveloped countries.
However, only 12 countries are shown here. The full table is shown in appendix
Table 1: Maternal mortality rate per 100.000 and
HIV positive population (15-49%)
mortality per 100.000 (rate)
positive population 15-49 years (%)
The data will be used to create the boxplots for the
two variables: HIV and maternal mortality
Table 2: The distribution of data from the two variables: HIV
and Maternal mortality
Figure 1: The shape of the distribution of data for HIV. Orange
representing median and blue representing Q1
Figure 2: The shape of the distribution of data for maternal
mortality. Orange representing median and blue representing Q1
The boxplot in figure 1,
which represents HIV, is skewed in terms of the distribution of the data. 25%
of the data between upper quartile and upper value is between 2.2 and 12.1.
Whereas, 25% of the data between smallest value and lower quartile is between
0.1 and 0.2, this shows that the data is not well spread out, and is the cause
of the skewness of the data. However, in figure 2, which represents maternal
mortality, the data is more symmetrical. Between the smallest value and the
lower quartile, the data is between 0.3 and 0.6. Between the upper quartile and
highest value, the data points are between 1.2 and 1-5. Which shows a much more
symmetrical boxplot than the one in figure 1.
Figure 3: Relationship
between maternal mortality and HIV for 48 African underdeveloped countries.
Chi-Square Statistical Test
The scatterplot shown
earlier in figure 3, showed a weak correlation. There is a nonlinear relation
between the two sets of data: HIV and maternal mortality. Therefore, since the
results seem inconclusive according to the scatterplot, another test will be
conducted: Chi-Square Statistical Test.
Table 3, is a contingency
table showing the frequency of both maternal mortality and HIV positive
population. It is used to show the relationship between two categorical
The Chi-Square Statistic
Test, is used to show a categorical relationship between two categorical
variables. The test will be calculated manually as shown below. A conclusion
will be made, determining whether or not the null hypothesis (HO )
will be accepted or rejected. If the null hypothesis (HO) is
rejected, which states that there is no significant association between the two
variables, the alternative hypothesis (H1) will be accepted. The
alternative hypothesis, states that there is significant association between
the two variables.
H0= The maternal mortality is
independent on HIV positive population.
H1= The Maternal mortality is
dependent on HIV positive population.
Table 3: Maternal Mortality
and HIV Positive Frequency Contingency Table (Observed frequencies)
x<0.4 0.4? x<0.8 x ?0.8 TOTAL x< 0.4% 5 10 8 23 x ?0.4 8 5 12 25 TOTAL 13 15 20 48 The values in table 2, in terms of observed frequencies are named as seen below: 11 12 13 21 22 23 Using the following formula: Fe =Column x row total total sum the calculation of expected frequencies based on observed frequencies will be calculated. "Expected frequency is defined as the number of times that we predict an event will occur based on a calculation using theoretical probabilities" (Glover, 2003) This means, these would be the values expected if the two variables were independent. The formula will be used to calculate the expected frequencies, using the observed numbers from table 2 Fe11=Fe21= 23 x 13/48=6.22 Fe12 =Fe22 =23x15/48=7.19 Fe13=Fe23 = 23x20/48=9.58 Fe11=Fe21 = 25x13/48=6.77 Fe12 =Fe22= 25x15/48= 7.81 Fe13=Fe23 =25x20/48=10.4 Table 4: Observed and Expected Frequencies Cell FO FE FO-FE (FO-FE)2 (FO?FE)2 FE 11 5 6.2291667 -1.229167 1.5108508 0.2425446 12 10 7.1875 2.8125 7.9101563 1.1005435 13 8 9.5833333 -1.583333 2.5069443 -0.1652174 21 8 6.7708333 1.2291667 1.5108508 0.22314104 22 5 7.8125 -2.8125 7.9101563 1.0125 23 12 10.416667 1.583333 2.5069434 0.24066656 SUM 2.65417831 Using the formula also seen in table 3: X2 Calc = ? (Fo – Fe2) Fe the chi-square statistical test will be calculated. The sum therefore is: 2.65. The alpha level shows the probability that the null hypothesis might be rejected, when in reality the null hypothesis is true. (Paret, 2012) The alpha level chosen is at 1%, and the degrees of freedom is 2. The degrees of freedom is the calculation done to show how many variables within the observed frequencies, can exist independently. The formula used to find this is: DF = (number of rows - 1) x (number of columns - 1). Meaning: DF= (2-1) x (3-1) = 1x2= 2. At an alpha level of 1% and degrees of freedom at 2, a critical value (CV) of 4.605. The CV of 4.605, will be compared to X2calc of 2.65. The chi-square statistic test states: X2calc > CV = H0 is rejected?
X2calc < CV= H0 is not rejected.? Heading back to the H0 (Null hypothesis), that states: The maternal mortality is independent on HIV positive population. Applying this to the results of this investigation: 2.65<4.605à H0 is not rejected. This means maternal mortality is independent on HIV positive population. Interpretation of results and conclusion The aim of carrying out this investigation, was to determine whether or not there was a relationship between HIV and maternal mortality. The Chi-square statistical test was carried out because the scatterplot showed no correlation between the variables whatsoever, the correlation was weak, and no linear regression could be drawn. However, since the critical value: 4.605 was higher than the calculated value: 2.65, the null hypothesis could not be rejected, and this conclusively meant there was no dependency of maternal mortality on HIV. A significance level of 1% was chosen, as this gave a more accurate conclusion. A level of 5%, could have meant concluding a dependency of maternal mortality on HIV, even if there was no dependency. Even though, there could have been a possibility that there was a relationship between HIV and maternal mortality in underdeveloped African countries, as the rate was high in these countries, some of the countries only had a high rate of one of the variables. Validity A bigger sample could have been considered, as there were 48 countries. However, it also has to be kept in mind that it was specifically aimed at African developed countries, which made the sample much more limited. There was enough data to perform a chi square statistical test, as all the expected frequency numbers were above 5. As shown in appendix 2. It needs to be considered that maternal mortality was between 15-49 years, however, this makes sense since women between these ages most likely sexually produce. An error from the secondary data collected, is that the population affected by HIV, counted for both women and men, whereas the maternal mortality rate counted for women. This could have been a reason for the skewed boxplot seen in figure 1. The data was way more distributed than the data collected for maternal mortality.