Correlation - scatter plot

[Anne-Claire Jo]

Dear stata users,

I have generated a scatter plot in order to study the correlation between variables related to political stability and education.
However, different from general (or most usual) graph which x-axis starts from 0, my graph x-axis starts from negative value to positive value (attached is the graph).
I am focusing on Africa case, and I wanted to verify if this makes sense to interpret for negative values in x-axis.
I would say that for Africa, it has much negative correlation between the two variables.
Could someone here please help me with this ?
(I apologize to ask here if it is not relevant but I had no one to ask.)


Thanks,
AC

[Fei Wang]

Anne-Claire, the x-axis starts with negative values because the political stability index can be negative. If it's correct (the PS index can be negative), then there is nothing wrong with the graph, and the overall correlation of education and PS is positive. But I would not say the correlation in Africa is more negative as there are only two points from Africa. I would at most say that, the points from Africa are not clearly following the overall positive correlation.

[Anne-Claire Jo]

Fei Wang Thanks for your reply ! Then the graph implies that in Africa, when PS increases, education decreases right ? Since they have negative correlation..
And if the points from Africa are not clearly in the confidence interval, then it could say that they are not significantly correlated ?

[Fei Wang]

Anne-Claire, my point in #2 is that you can hardly draw clear conclusions with only two points from Africa. If you have more points from Africa, then you may directly calculate the correlation among the points and compare it with the correlation of points from other regions -- the comparison of correlations has nothing to do with the confidence interval in #1.