Subman,
No, you do not understand what I am after. You have given two variables. Number of executions, and number of murderings per head of population. All nice and well. But every scientist who would conclude by that that the one variable influences the other, would be laughed about, because for such a conclusion the type of data you quote simply is not good enough. You conclude on a causal connection, that is not backed by that data. The link between both variables that you conclude by the look of the two graphs - is in your eye only. There is no causal explanation that these two variables would support. Maybe it is there in reality, but the graphs and numbers do not allow you to take that as a given fact. So far, you just believe it. If that causal context is given, it would be needed to prooven by according statistical data you have won in experiment or by research, and even additonal variables, that the graphics simply does not contain.
Honestly, not kidding you, but there is not that conclusion in that graph that you want to see in it. The graphs only describe the up and down of two variables over time. It is tempting to see them interacting, for it matches your hypothesis, but the type of data does not support that. They do not say the slightest thing about wether both variables are related to each other, or not. They are purely descriptive, they describe something like a correlative context only,
not a causal one. You may think this is something minor, or just a cheat, but it is not, not by logic, and not in science and statistics. Graphs like the one you have given we had been warned about time and again in statistic classes. If during the statistic exam I would have made a causal conclusion on the basis of that low-quality data that effectively describes only a correlation, it would have been game over for me.
A correlation, even a highly significant one that is close to 1 or -1, never means a causal connection by itself, you need to do different statistical work if you want to proove that causal connection. You need additonal processing of the raw data if you have a high correlation that makes you believe that eventually this might be a hint that there is a causal connection, and the more variables are involved, the more work it becomes.
Such statistics and graphs like this one are given because the author does not think about what he is doing or actually does not know it (the trap you just fell for yourself is very easy and tempting to step into), or he knows it but wants to fool the reader. The data only hint at that there
might be a connection between variables, but not of what kind that connection is, if it is a mutual influence or not, if a third or even more variables are involved that mediate between the primary two.
I hated statistics back then, and I still hate it today, and now I hate you becasue you made me going back to it all!!!
No more comment on statistics from me. The more sophisticated stuff I already have forgotten anyway.
