Since the first cases of COVID-19 appeared in Wuhan in December 2019, the disease has spread worldwide and become a global pandemic. More than 400 million infected people have been registered globally, including more than 5.8 million fatalities with a presumably high number of unreported cases [1]. To counteract the outbreak of the disease, regulations and lockdowns were imposed by many governments, which in turn imposed many restrictions on daily life. Despite this, the pandemic has not yet been stopped and the hopes of many people lie in vaccines, which were first developed at the end of 2020. Vaccination could slow down the infection, or at least weaken the symptoms of the disease, leading to fewer hospitalizations and thus keeping the health- care system intact. Many countries have launched their vaccination campaigns during the spring of 2021. At first, only vulnerable groups were able to get vaccinated, but toward the middle of the year, the vaccine was released to the entire population in most countries. Although vaccines are promising, vaccination coverage varies widely across countries. The reasons for this have not yet been conclusively clarified. In many countries, vaccination is not mandatory, but the government encourages people to get vaccinated. Therefore, it could be that trust in the government has an impact on vaccination rates, along with other socio-economic factors. The current study thus aims to investigate which socio-economic factors have an influence on vaccination rates for COVID-19.
In the second part we visualized and analyzed that data. This needed to be done in such a way that the underlying relationships between the variables would be observable. For that we plot each of our measures with our target variable, which is the vaccination rate, on a two dimensional scatter plot that has the respective measure on the x-axis and the vaccination rate on the y-axis. This lets us visualize the relationship of each measure with the target variable. One important indicator of that relationship is the correlation coefficient. We expect some degree of linearity between a factor and the vaccination rate if that factor has an influence on the vaccination process. For that reason the correlation coefficient is a good measure to quantify the relationship. Additionally we visualize the correlation by fitting a linear regression line on the points of each plot. That way a high degree of correlation will be easily observable by a high concentration of points close to the regression line. In the last step we calculated a p-value for each of the plots. The p-value indicates the probability of two random variables producing a dataset with the given correlation value. Given a dataset it gives us the probability of that data being produced by two independent random variables. In general a p-value of 0.05 and below is considered to be statistically significant. For that reason we only considered measures that had a value in that range to be influencing the vaccination rates. We performed the analysis using a Python based stack including the packages Jupyter-Lab, Matplotlib, Sklearn, Scipy, Pandas and Seaborn. Please have a look at the pipeline for the code.
Figure 1 reveals a high positive linear correlation between the share of people fully vaccinated and the gross domestic product per capita with a correlation coefficient of 0.701 and p-value of 7.54x10^−15 , meaning it is very likely that these two variables are correlated. The same is shown by Figure 2 for the relationship between the share of people fully vaccinated and the life expectancy. These two variables have an even higher correlation coefficient of 0.782 and also a small p-value of 1.29x10^−07 . Figure 3 shows a negative linear correlation between the share of people fully vaccinated and the working hours of employees with a correlation coefficient of -0.48. The correlation is slightly weaker than for the previous two factors, but the small p-value of 0.006 indicates that correlation is very likely. For the other factors the p-value was always higher than 5%, which makes a correlation between the share of people fully vaccinated and the respective factors very unlikely.
Figure 1: Correlation between share of population vaccinated against COVID-19 and GDP per capita
Correlation Coefficient and p-Value: 0.7006793183652773, 7.542031256159133e-15
Figure 2: Correlation between share of population vaccinated against COVID- 19 and life expectancy
Correlation Coefficient and p-Value: 0.7815731285293231, 1.288725407220812e-07
Figure 3: Correlation between share of population vaccinated against COVID- 19 and working long hours
Correlation Coefficient and p-Value: -0.4823137806158691, 0.005999820173321409
The ongoing COVID-19 pandemic has cost the lives of many people and also damaged the economies of many countries. In addition to strict regulations by governments, which in turn restrict daily life, vaccination is one way to mitigate the effects of the pandemic, making a high vaccination rate desirable. The current investigation therefore examined which factors have an influ- ence on vaccination rates. The results show that there are three factors that have a relevant influence on the vaccination rate: the GDP per capita, the life expectancy as well as the percentage of people working more than 45hours per week. The higher the GDP per capita and the life expectancy the higher is the vaccination rate. However, the more people work long working hours in a country the lower is the vaccination rate of a country. The other factors did not have a statistically relevant influence. The limitations of this study were a limited number of data points due to a limited number of countries. Moreover, data on socio-economic factors is hard to quantify, because some of the indices were collected by asking people about their attitudes and these findings are not entirely reliable. It is important to mention that the GDP per capita and the life expectancy have a high relation among each other, they represent wealth. This is reasonable because a lot of poor countries have limited access to vaccines and therefore not even the possibility to get vaccinated even if they would want to. The more money a country has, the more it can spend on its health system. It was also shown that countries in which employees work longer have a lower vaccination rate. This could be because in poorer countries people earn less money and therefore have to work more to make a living. That’s why another valuable investigation would be to look into the influences of vaccination rates for groups of countries separately, for example, to examine which socio-economic factors are of relevance within a group of richer countries with a high GDP versus within a group of poorer countries with a lower GDP.
https://www.kaggle.com/nitishabharathi/gdp-per-capita-all-countries
https://www.kaggle.com/rsrishav/covid-vaccination-dataset
https://www.kaggle.com/uniquekale/covid-19-data-analysis-2021/data
https://sites.psu.edu/dictators/
pip install -r requirements.txtjupyter-notebook pipeline.ipynb