It's only in statistics that y is the independent variable & x the dependent variable. In economics, when we draw demand & supply curves, by convention, y is price and x is quantity.

Dear Ted,

Thank you very much for clarifying that convention and context. Very useful.

I guess 'quantity' and 'effectiveness' are often related/synonymous, which then explains the cost-effectiveness planes.

Best,

Mikkel

Hi Mikkel,

Yes, it is a bit of a trivial question because in the end it does not lead to different conclusions.

I tend to agree with you that it looks odd to put effectiveness on the x-axis and cost on the y-axis. I have asked why the convention in cost-effectiveness of health care interventions is like that. Someone reasoned that in healthcare, costs may be seen as something resulting from our efforts to try make people healthier. (Didn't convince me... :-)

The only thing I can come up with in favor of the convention is:

The cost-effectiveness plane is not a function (unlike supply and demand curves) but cloud of points relating the total or incremental costs and effectiveness. The costs considered are not just the input costs but include downstream costs, some of which may by incurred well after the intervention under consideration has stopped. That makes it a lot more arbitrary which to put on the x- or y-axis. With that, it may be best to order the decision alternatives according to increasing health, i.e. effectiveness on the x-axis.

I agree with Robert and the cost-effectiveness plane is just a cloud of uncertainty. The cost should include both intervention costs and cost offsets.

Long Khanh-Dao Le: Mikkel Godsk was not talking about the results of an uncertainty analysis / PSA but about plotting the results of different scenarios. He may want to use that plot to draw an efficient frontier, which can be regarded as a function where all plotted dots are either above or below the frontier, depending on what is on which axis...

Hi Robert

Really good points. Thanks.

I agree, I don't think the healthcare example is fully convincing (especially due to the reverse chronology of getting well and the costs/efforts for the treatment). However, your point about not being a function sounds like a good, plausible explanation. So, by having the effects on the x axis I guess you stress that there is not (necessarily) a mathematical/statistical relationship between the costs and effets and that you are focused on effects and, secondly, what it takes to obtain that effect. That could also explain the healthcare example.

Best, Mikkel

Thanks!