In a basic pricing scheme we can assume that most performers spend the minimal effort required in order for the task to be approved. It is certainly reasonable for performers to try to maximize earnings while minimizing effort and time spent. However, it is highly undesirable for requesters who want their data to be labeled with the best quality possible.
This controversy leads to the concept of performance-based payment. Instead of having a fixed price for each approved task, payment can be connected to the quality of the work done by the performer. There are various ways to calculate individual quality, but most of them rely on checks that are hidden from the performer. This means that to maximize their earnings they need to complete every assigned task carefully, thereby increasing the overall quality of data.
Here's an example of how a bonus of $0.50 can be awarded to performers with different quality. Note that in all three cases quality is reasonable, meaning that the performers have successfully passed through basic quality filters.