In trying to give incentives to get what you want, how you give the incentive may actually matter. I mean, for example, if you want to hire a kid to do some menial tasks for you for a week, you can give the kid $5.00 a day, or you can simply and randomly give the kid a one-time payment, like $35.00, at any one day for that week. The fact that it's random means the kid won't know when he's going to be paid, only that he will be paid. Robin Hogarth and Marie Claire Villeval actually have shown an economic validity to this.
In their latest paper from the Institute for the Study of Labor entitled, "Intermittent Reinforcement and the Persistence of Behavior: Experimental Evidence," Hogarth and Villeval points out that attention should also be given to the effects of regularity and frequency of giving incentives:
"We contrasted three ways of rewarding participants in a real-effort experiment in which individuals had to decide when to exit the situation: a continuous reinforcement schedule (all periods paid); a fixed intermittent reinforcement schedule (one out of three periods paid); and a random intermittent reinforcement schedule (one out of three periods paid on a random basis)."
Their findings? Intermittent reinforcement led to higher total effort, while continuous reinforcement led to decrease in efforts dramatically:
"Overall, intermittent reinforcement leads to more persistence and higher total effort, while participants in the continuous condition exit as soon as payment stops or decrease effort dramatically. Randomness increases the dispersion of effort, inducing both early exiting and persistence in behavior; overall, it reduces agents’ payoffs."
It actually justifies one of Machiavelli's famous quotes, "Benefits should be conferred gradually; and in that way they will taste better." Benefits conferred gradually means you don't shower your subjects with benefits all the time. Much like the opposite of a continuous payment scheme where incentives are given periodically. Conferring benefits gradually is similar to not letting know your subjects when you're going to give them benefits.
The authors also made one application that makes sense: the experiment may prove why agents persist in activities although they lose money, such as excess trading in stock markets. But it actually has many applications. Bottom line is, if the kid expects to be paid frequently, the kid'll get sloppy. If you withold information about the payment method, the kid'll straighten up otherwise the kid might not get paid.
So the carrot can also be used as a stick. Go figure.