Assume you've some data gathered from a repeated experiment. If you want to make inferences about the data collected, and knowing that it is highly unlikely that a density function will provide a perfect representation of the data, how important is it to choose a correct probability density model? Should you be concerned about the likelihood that it is highly unlikely that a density function will provide a perfect representation of the data?