Visible Certainty - The Verifiable.com Blog

May 15 2009

“IT MAKES SENSE TO BUY GOLD !”

Something as simple as visualizing the value of gold over the past two centuries highlights two important and frequently overlooked problems when charting price data.

Here’s a typical historical price chart, from an informational website advocating gold as an investment:

gold price graph

The first problem with this chart is inflation.  Changes in a commodity’s price do not necessarily correspond to changes in its value over time, because of price inflation — so if you don’t convert into a constant unit of value, it makes no sense to chart prices over time (unless your goal is actually just to show inflation, using the commodity as an example).

If we take the same source data and adjust for inflation (how?), the graph looks like this:

The second problem, even with this second chart, is that we’re visualizing absolute changes in price, rather than relative changes in value.  It doesn’t really matter to anyone whether gold jumps from $5 to $10/ounce, or from $500 to $1000/ounce (even if you’re using the same 2007 dollars in all cases): either way, gold doubled in value.  But if you chart gold prices on a linear scale, the former doubling is virtually invisible and the latter looks gargantuan.  You’re not visualizing what you’re trying to visualize, which is the relative change in the value of a single ounce of gold.

But if we take the inflation-adjusted chart, and switch to a logarithmic scale, the graph looks like this:

As you can see, it’s shows a totally different story than the original chart we found.  (To see the original and final versions together, see here.)

Why doesn’t everyone do this when they chart this kind of thing?  Assuming no malice, one issue is obviously ignorance.  They don’t teach this stuff in schools (usually), and it requires a comfort level with simple math.  But another reason is the sheer difficulty of making the necessary adjustments, in practice.  You’ve got to find the inflation data set, hope it covers the entire time period you need (and uses the same units of time), manually combine it with your data (often by hand, by reformatting or retyping it), and then you’ve got to do the math to actually convert everything into inflation-adjusted figures.

Excel doesn’t just have a “adjust this line for inflation” feature, and it certainly doesn’t provide the inflation data you need to do it.

We do.

And we’re planning to make it even easier in the future.  Let us know your thoughts.

- Peter C., Verifiable.com

Comments (View)
Apr 30 2009
Comments (View)
Apr 21 2009

Tax Freedom Day

We’ve posted a new featured chart this morning, showing the power of data set joins on Verifiable.com. After one of our users uploaded a great data set listing Tax Freedom Day for each state, I wondered what the implications of low taxes were for a state’s population. I searched for other data sets to join it with (by state), and in just a few minutes came up with this.

The Tax Foundation website appears to be down, which is too bad as I’d love to get the 2006 data to see if an apples-to-apples chart looks much different.

I’d love to see what else people can do with this data.

- Peter C., Verifiable.com

Comments (View)
Page 4 of 9 Newer Entries →
I want my StackExchange