If the overall price weren’t changing much, then maybe the relative contributions of different components would be an OK story, maybe less misleading, but even there I’d say chart (b) is going to tell that story as well as chart (a), and provide some additional information to boot.

And in that case, shouldn’t it be a stacked bar, since the components add to 100% and that should be called out?

For purposes of political discussions, the price of crude may not be not independent, but that is a different argument. Also notice that I said “cost” of gasoline, not “price”. Production and distribution variations, as well as market and supplier induced supply and demand fluctuations, can also impact day-to-day gasoline prices. Ever notice that the price of gas increases right before Memorial Day (end of May in the US)? This is because the weather is nice and people start taking vacations and driving more.

It’s possible the choice of graphic was influenced to minimize the visual impact of crude price. I couldn’t tell by reading he chain of references which eventually led me to the CIO article Gas Prices: How Oil Companies Use Business Intelligence To Maximize Profits. The article was a bit of fluff extolling the virtues of all the wonderful BI software that’s available. The featured graphic, by the way, is NOT at al related to BI.

Good points. I think a good composite would be dividing the chart space into four quarters, two-by-two, putting the cost contribution chart across the top two, to provide adequate resolution in the value axis, and putting the other two into the bottom left and right quadrants, perhaps sharing the category axis.