In this section, we aim to examine the factors that affect the transmission and death rates of COVID-19 across 95 top-affected countries. We used life expectancy, smoking level, population density, and outsiders’ effect (tourism and global export indicators) as the independent variables and total cases, total cases per million, death rate, and death rate per million as the dependent variables. We used t-tests to compare the means of the dependent variables between two groups of countries based on the median values of the independent variables. We also used fuzzy logic to model the chance of transmitting COVID-19 based on the weights of the independent variables derived from the statistical results.
2.2.2. Data Analysis
We performed t-tests to test the null hypothesis that there are no significant differences in the means of the dependent variables between the two groups of countries for each independent variable. We used a significance level of 0.05. We reported the mean, variance, observations, pooled variance, degrees of freedom, t-statistic, p-value (one-tail and two-tail), and t-critical value (one-tail and two-tail) for each t-test. We also presented the results in tables for each independent variable. We rejected the null hypotheses if the p-value (two-tail) was less than 0.05 and accepted them otherwise.
We found that life expectancy and outsiders’ effect had significant effects on the death rate and the total cases of COVID-19, respectively. We also found that smoking level, population density, and temperature had no significant effects on any of the dependent variables. We concluded that life expectancy and outsiders’ effect were the most important factors for the transmission and mortality of COVID-19, while smoking level, population density, and temperature were less relevant.
We also used fuzzy logic to create a model that can estimate the chance of transmitting COVID-19 in a region/state based on the four independent variables. We used the following steps to create the model:
Step 1: We selected the independent variables that had significant effects on the dependent variables based on the t-tests. We chose life expectancy and outsiders’ effect as the input variables for the fuzzy logic model.
Step 2: We calculated the weights of the input variables based on the p-values of the t-tests. We used the following formula: Weight percentage = (1 − p-value) * 100/2.15788, where 2.15788 is the sum of the 1 − p-values of all the independent variables. We rounded the weight percentages to the nearest integer. We obtained the following weights: outsiders’ effect: 46%, life expectancy: 21%, temperature: 10%, and others: 23%.
Step 3: We defined the fuzzy sets and the membership functions for the input and output variables. We used three fuzzy sets for each variable: low, medium, and high. We used triangular membership functions for the input variables and trapezoidal membership functions for the output variable. We used the median values of the input variables to define the breakpoints of the membership functions. We used the following ranges for the output variable: low: [10, 12.5], medium: (12.5, 16.5), and high: [16.5, 20].
Step 4: We defined the fuzzy rules for the inference process by a specific fuzzy rule base.
Step 5: We applied the fuzzy inference system to the input data and obtained the output values. We used the Mamdani method for the inference process and the centroid method for the defuzzification process. We presented the output values in graphs and tables.
Case 1 (Life expectancy): Life expectancy is a statistical indicator of how long an individual is predicted to live on average based on their birth year, current age, and other demographic variables such as biological sex. It is different in different regions and periods. The death rate due to COVID-19 is higher in countries where life expectancy is higher. In this section, the analysis is performed.
H0-Null Hypothesis. There are no significant differences in death rate when comparing low (less than or equal to 76.65 years) to high values (greater than 76.65 years) of life expectancy of COVID-19-affected countries.
H1-Alternative Hypothesis. There are significant differences in death rate when comparing low (less than or equal to 76.65 years) to high values (greater than 76.65 years) of life expectancy in COVID-19-affected countries.
First, data from the countries were sorted as per life expectancy in increasing order. Then, we divided the list into two groups: a country list whose life expectancy is low, less than or equal to 76.65 years, and a list whose life expectancy is high, greater than 76.65 years. Then, we compared the death rates between the two groups. As per the results of the
t-test for equal variances, the mean was 0.023584419 for low values and 0.050352366 for high values of life expectancy, and variances are shown in
Table 1. The
p-value (two tails) was 0.000284975, which was less than 0.05. Thus, the null hypothesis can not be accepted. Hence, there are significant differences in death rate when comparing low to high values of life expectancy of COVID-19-affected countries. This concludes that life expectancy has a high impact on death chances. Older persons having COVID-19 have death chances almost double those compared to younger people.
Case 2 (Smoking):
H0-Null Hypothesis. There are no significant differences in death rate while comparing low (less than or equal to 0.226 indexes) to high values (greater than 0.226 indexes) of smoking levels of COVID-19-affected countries.
H1-Alternative Hypothesis. There are significant differences in death rate while comparing low (less than or equal to 0.226 indexes) to high values (greater than 0.226 indexes) of smoking levels of COVID-19-affected countries.
Data were sorted as per smoking levels in increasing order. Then, we divided the list into two groups: a country list whose smoking level was low (less than or equal to 0.226 indexes) and a list whose smoking level was high (greater than 0.226 indexes). Then, we compared the two groups. As per the results of the
t-test for equal variances, the mean was 0.031749961 for low values and 0.042013 for high values of smoking level, and variances are shown in
Table 2. The
p-value (two tails) was 0.177094358, which was higher than 0.05. Thus, the null hypothesis is accepted. Hence, there are no significant differences in death rates when comparing low (less than or equal to 0.226 indexes) to high values (greater than 0.226 indexes) of smoking levels in COVID-19-affected countries. This concludes that the smoking level of a person has a minor impact on death. Highly smoking persons having COVID-19 have death chances larger (but not significantly) compared to non-smoking people.
Case 3 (Population density):
H0-Null Hypothesis. There are no significant differences in affected cases (affected cases/million) while comparing low (less than or equal to 92/sqkm) to high values (higher than 92/sqkm) of population density of COVID-19-affected countries.
H1-Alternative Hypothesis. There are significant differences in affected cases (affected cases/million) while comparing low (less than or equal to 92/sqkm) to high values (higher than 92/sqkm) of population density of COVID-19-affected countries.
Data were sorted as per population density in increasing order. Then, we divided the list into two groups: a country list whose population density was low (less than or equal to 92/sqkm) and a list whose population density was high (higher than 92/sqkm). Then, we separately compared the two groups for affected cases and affected cases/million. In
Table 3, the results of affected cases are shown, and in
Table 4, affected cases/million are shown.
p-value (two tails) was 0.186759117 for
Table 3 and 0.933317888 for
Table 4; these were large values (greater than 0.05). Thus, the null hypothesis is accepted for both cases. Hence, there are no significant differences in affected cases while comparing low to high values of population density of COVID-19-affected countries. This concludes that the population density of a country has no impact on COVID-19 transmissions.
Case 4 (Outsiders’ effect):
H0-Null Hypothesis. There are no significant differences in affected cases (or affected cases/million) while comparing low (less than or equal to 0.2 global sharing) to high values (greater than 0.2 global sharing) of percentages of outsiders (tourism and global export indicators) of COVID-19-affected countries.
H1-Alternative Hypothesis. There are significant differences in affected cases (or affected cases/million) while comparing low (less than or equal to 0.2 global sharing) to high values (greater than 0.2 global sharing) of percentages of outsiders (tourism and global export indicators) of COVID-19-affected countries.
Data were sorted as per sharing percentages of outsiders (tourism/global export indicators) in increasing order. Then, we divided the list into two groups: a country list whose sharing percentages of outsiders (tourism/global export indicators) was low (less than or equal to 0.2 sharing) and a list whose sharing percentages of outsiders (tourism/global export indicators) was high (greater than 0.2 sharing). Then, we compared the two groups for affected cases and affected cases/million separately. In
Table 5, the results of affected cases are shown, and in
Table 6, affected cases/million are shown.
p-value (two tails) was 0.012169463 for
Table 5, which was less than 0.05, and was 0.249465449 (greater than 0.05) for
Table 6. Thus, the null hypothesis is not accepted for
Table 5 and accepted for
Table 6. Hence, there are significant differences in affected cases while comparing low to high values of sharing percentages of outsiders (tourism/global export indicators) of COVID-19-affected countries. This concludes that sharing percentages of outsiders (tourism/global export indicators) of a country has a significant impact on the transmission of COVID-19.
Other cases:
A few other cases were also analysed in this section. In
Table 7, it is observed that life expectancy has no significant effect on transmission of COVID-19 as the
p-value is higher than the 0.05 significance level. In
Table 8, we find that temperature is also a non-significant factor for the transfer of COVID-19.
Table 9 shows that the smoking level of countries has some impact on the transmission of COVID-19. However, these factors are not significant to spread such a deadly disease.