common UI complaint about frontpage catalog graph

[alee]

When showing this graph a common UI complaint / point of confusion was about the y-axis values on the right hand side of the graph being scaled between 0 and 0.2. Without careful inspection it makes it seem like code available is much more common than it actually is.

Other thoughts from @MarcoAJanssen @seanbergin ?

[seanbergin]

The obvious change would be for the x axis of the red line to 100% Yes you wouldn't see the increase well, but I think you would see the change from 5 - 18% ?

Can we fill the bars based on % of code available?

Alternatively, you could compromise and just do the x axis as 50 or 75%

[MarcoAJanssen]

In fact the graph is incorrect since 0.2% would be a fraction 0.002 of the whole population. Indeed the availability of code has increased quite a bit. Not only is the percentage increased to 20% also the absolute number of publications has increased. It is true that one could use a scale to 100% but than we see a relatively flat line. It is also helpful to zoom in the changes over time.
So I would make sure it is 20% instead of 0.2% and then it is also up to the reader to check the axis

[seanbergin]

We used the graph at the science gateways meeting in Chicago and the audience found it very confusing. The graph seems to suggest that most model code is published - unless you notice the scale.

If the point of the graph is that the proportion is going up then maybe that graph shouldn't be overlaid over the total number of publications? Or number of publications could be a line graph as well so that it doesn't look like there is a proportional relationship?

[MarcoAJanssen]

So an alternative will be a stacked graph where we have a stacked bars of code available and code unavailable publications. People will not be able to read the percentage, but it shows the increasing share.