It's been suggested that purchasing a miner (e.g. Block Erupter) is futile because ROI should be calculated in terms of mined coins and not fiat.  This line of reasoning is based upon the idea that it is better to purchase coins and sit on them rather than buying a miner.

Contrary to this belief, calculating ROI of a miner in terms of fiat has its merits.

I purchased a few dozen block erupters a while back with BTC, but then immediately repurchased the BTC spent on the miners.  In other words, I essentially purchased my miners at a set fiat price since my total BTC holdings didn't vary.  

For kicks, lets say I purchase these miners with 15 BTC at a $100 USD conversion rate.  Is it likely that I will mine 15 BTC with these erupters?  Not likely.  But, since I effectively purchased them with fiat, recouping 15 btc is no longer a concern.  I simply need to recoup $1500 in mined BTC to achieve 100% ROI in the scenario.  If the value of BTC increases and continues to do so significantly, achieving 100% ROI becomes more and more likely.

This method protects against a drop in BTC value because the hardware itself serves to hedge against the initial investment.  If I spend $1500 to purchase 15 Btc and the price per btc declines to $70, then I'm down $450 (I.e. 30 x 15 = 450).  But, if I purchase the miner instead and the price drops to $70, I always retain the option to sell the miner.  For example, selling the miner purchased originally at $1500 for $1000 at a $70 conversion rate, I can receive 14.28 btc for it, and that's on top of the few btc I would've mined already.

In other words, purchasing a miner with fiat is better if the price of btc drops, but it's also better than doing nothing at all if the price of btc goes up.  Buying btc outright may be better if you assume that btc value will only continue to increase, but this is a riskier approach.