Ofer Malamud and Cristian Pop-Eleches look at the effects of a program that gave poor households a voucher to purchase a computer. ... Households with incomes directly below a cutoff level were given a voucher while households with incomes directly above the cutoff were not. Thus, households which were very similar were treated differently and this lets the authors use a regression discontinuity design that makes their results credible as representing a causal effect. The results of a regression discontinuity design are also very easy to explain with figures.
The income cutoff is shown by the red line. Beginning at the top left we see that households with incomes just below the cutoff were much more likely to have a computer than households with incomes just above the cutoff - thus the voucher program has a big effect on computer ownership. The top right figure shows similarly that the voucher program increased computer usage since computers were used much more often in households with incomes just below the cutoff than in the non-eligible for voucher households with incomes just above the cutoff.
(Originally from David Youngberg at See the invisible hand, which looks terrifyingly addictive in its own right.)