The US Census Bureau was supposed to finish counting households by October 31. But yesterday, the Bureau announced that it would end field data collection one month earlier. This change gives them six or seven weeks to finish a count that was supposed to take 10 weeks. With only about 60% of households counted so far, many people worry that they’ll undercount the population.
The Bureau issued a press release stating (in part):
So, the Bureau’s new political leadership wants to improve the speed of the count by (1) conducting additional training sessions, and (2) providing awards to field staff who work more hours.
However, there’s no indication that workers would benefit from additional training. As I’ve written about in my new book, The Conclusion Trap, when people frame their problems poorly, their countermeasures fail because they don’t address the underlying issues. In this case, additional training doesn’t address the real problem of a too-short data collection period.
Moreover, while spending more hours in the field might lead to a (more) complete count, providing “awards” is a recipe for mischief. As Mark Graban has written about before, when workers are asked to do a better job, they have three options:
They can improve the system
They can distort the system
They can distort the data
Since the Bureau isn’t improving the system, and the field staff don’t have the ability to easily distort the system, it leaves workers with the temptation to distort the data so that they can collect awards. To be clear, I’m not suggesting that census workers will distort the data. But the Bureau leadership is creating badly aligned incentives rather than taking the time to improve the quality of the survey.
The Census Bureau—under pressure from the Trump administration—seems determined to follow the lead of traditionally managed (i.e., top-down, command and control management) organizations. That’s not going to end well for the Bureau. Or the American people.