Final report vuhoangdung 17110077

Final report Quantitative and Qualitative AnalysisName: Vu Hoang Dung Student code: 17110077 Program: Public Policy – 2 nd intake – Vietnam Japan University • Variables: Dependent variab

Final report Quantitative and Qualitative Analysis

Name: Vu Hoang Dung Student code: 17110077

Program: Public Policy – 2 nd intake – Vietnam Japan University

• Variables:

Dependent variable: Internet users (per 100 people)- Y

Independent variables:

GDP_PPP - GDP per capita, PPP (constant 2000 international $)- X1

Ser_import - Computer, communications and other services (% of commercial service

imports)-X2

Ur_pop - Urban population (% of total)- X3

Sample size: 19 countries

• Model 1:

• Model 2: Deleting Japan

• Model 3: Log GDP_PPP

* Method: Number of variables is 3, size 19 countries, not overfitting and underfitting Use

stepwise to determine the appropriate model, which will then determine the statistically significant variable

> vif = diag(solve(cor(x))) ; vif

Service import GDP_PPP Ur_pop

1.010008 2.267654 2.259349

Model 1 is a normal stepwise method after checking VIF, there is no multi-co linearity between these variables and no need to delete any variable From the regression results table, we can see that R2 value is 0.7899, which is quite high, and the significance level is quite good and AIC value is 142.7617 However, in the graph country number 9 (Japan) are near the line value 1, which can have effect on regression results

Model 2 is obtained by deleting Japan

> vif = diag(solve(cor(x))) ; vif

GDP_PPP Ser.import Ur_pop

4.202452 1.000479 4.202732

Model 2, after deleting Japan, checking the multiple co linearity and stepwise method, model 2 has R2 = 0.795 and AIC = 135.4048 Comparing model 1 and model 2, we can see that, model 2

can be better than model 1 However, the significant level of variables higher than 0.05, so we can use logarithm function with GDP_PPP and stepwise method to find out model 3

Model 3 is created by stepwise method and use logarithm with GDP per capita

Coefficients:

Estimate Std Error t value Pr(>|t|)

(Intercept) -61.786 9.169 -6.739 4.76e-06 ***

GDP_PPP 9.915 1.193 8.309 3.38e-07 ***

-Signif codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 8.009 on 16 degrees of freedom

Multiple R-squared: 0.8119, Adjusted R-squared: 0.8001

F-statistic: 69.04 on 1 and 16 DF, p-value: 3.38e-07

After checking 3 models, we can see model 3 is the best one, with t-value has statistical significance, R-square = 0.8119 and AIC is the smallest Although model 3 has the fewest variables, it can explain the relationship between GDP and internet penetration (per 100 people)

• Conclusion:

Internet users are defined as individuals who have access to the Internet at home, through computers or mobile devices As a result, the number of computers and mobile devices will affect the number of internet users in each country

Moreover, in the 2000s, internet access was mostly via computers This has led to an increase in the import of computers and other communication devices, which has been linked to the proportion of Internet users in this period (model 1)

In addition to importing, the percentage of computer users also depends on the proportion of urban areas in the country In countries with large urban areas, the proportion of internet users will increase due to higher demand for work and living conditions

Today the use of the internet for everyday use and application is indispensable A country that successfully applies technology applications to life will make the country more prosperous People will have higher living standards and income According to statistics, countries with high numbers of internet users such as Australia, Hong Kong, Japan and Singapore are all high income countries So GDP and the number of internet users are closely related (model 3)