Help Interpreting a log-log scale scatter plot

[Jason Browen]

Dear all,


I am using Stata 16, on mac. I used the command twoway scatter lnGDP lnpopulation. GDP is in millions. I have included an image of my graph. Is there anything else to mention besides the fact that the graph illustrates a positive trend?

Thank you in advance for your help


Jason Browen

[Jason Browen]

I am sorry please ignore the bar graph I accidentally attached it.

[Clyde Schechter]

I would also mention that it is a pretty strong relationship--the slope appears to be around 1. And I would mention that the relationship is pretty tight: most of the points lie very close to the trend line.

I would also look into those two outlying points to see if there is some reason for their departure from an otherwise very strong connection.

[Jason Browen]

Thank you so much Clyde for your help!

[Nick Cox]

It's easy enough to get the same graph, but with more intelligible axis labels.

twoway scatter GDP population, ysc(log) yla(20e6 "20" 10e6 "10" 5e6 "5") ytitle(Population (m)) 

with similar detail for the x axis. Note that the same basic idea was covered just yesterday in your previous thread

https://www.statalist.org/forums/for...ummy-variables

See also niceloglabels (SSC, SJ):

https://www.statalist.org/forums/for...lable-from-ssc

https://journals.sagepub.com/doi/abs...urnalCode=stja

Tufte's 1974 book Data Analysis for Politics and Policy (Englewood Cliffs, NJ: Prentice-Hall) has a really good expository chapter on these matters. It is out-of-print but completely up-to-date on this point!

https://www.edwardtufte.com/tufte/ebooks gives details of purchase of a .pdf at USD 9. As you are asking this question, it may be the best 9 dollars you spend all week.