July | August 2017


 

 

 

 

 

 

The Causes, Costs and Consequences of Government Bad Data

By Lisa McKinney, CSG communications associate
Data is essential for lawmakers to make evidence-based decisions about how to allocate limited resources—but not all data is created equal. CSG Senior Fellows Katherine Barrett and Richard Greene discussed what makes data reliable and what steps can be taken to improve the quality and availability of data at The Causes, Costs and Consequences of Bad Government Data policy workshop Dec. 11 at the 2015 CSG National Conference in Nashville, Tenn.
“The problems [with data] are becoming increasingly transparent while states are increasingly relying on data to make decisions,” said Greene. “The benefits of good data are huge and the risks of bad data are enormous.”
During their research into the prevalence of bad data in government, Barrett and Greene surveyed 75 officials in 46 states; about 7 out of 10 said data problems were frequently or often an impediment to doing their business effectively. None of the officials surveyed that work with program data said this was rarely the case.
Barrett and Greene found that much of the faulty data stemmed from poor study design by government employees collecting data who may not be trained in reliable research techniques. The Pennsylvania Legislative Budget and Finance Committee surveyed businesses to determine how many jobs were created through the state’s Keystone Opportunity Zones program. However, the survey did not define the word “job” or the time period to be considered. Some respondents only counted full-time jobs in their answers; others included part-time and contract jobs. Some reported the number of jobs created since the program’s inception; others reported numbers from the previous year.
“If data is not collected in a consistent way, it becomes meaningless,” said Barrett. 
Another key issue with government data is that it is often siloed, said Greene. “There is a desire on the part of agencies not to share data with other agencies,” he said. “There is a question around whose data it is—the state’s or the agency’s.”
There is also widespread belief on the part of agencies that there are statutes in place that prevent them from sharing the data they collect, particularly in agencies that deal with sensitive materials such as mental health data, he said. Often, that is not the case and there are ways to share data while protecting privacy.
Even when data is collected using reliable techniques and shared with the appropriate parties, it is sometimes the wrong kind of data. For example, police officers in New York City were being evaluated primarily on the number of arrests they made, said Greene.  “What you really want is less crime not more arrests,” he said. “It can cause officers to be more inclined to make an arrest rather than to settle an issue.”
Collecting good data and making it available to the people who need it requires investing in technology and people. There is a need for more data scientists and analysts in government, said Greene. Creating inventories that track all of the data the state is collecting and considering all the agencies that could use the data can help make agencies’ data collection efforts more worthwhile.   
“Everyone needs to be a lot more concerned than 15 years ago about the data they are using because it is much more pervasive,” said Barrett.
 

 

 

< Prev 1 | 2 | 3 | 4 | 5 | 6 | 7 Next >