Well, all of last week and the first part of this week have been dedicated to "cleaning" and formatting the data submitted for my two survey areas: Police Services and Purchasing. The data cleaning process is made up of "logic checks" (i.e. if x equals y, then it can't be Z kind of stuff as well as detecting outliers that are more than 2 times the standard deviation) and after that I eyeball the data to look for outliers that the computer does not detect, which at times are quite egregious. Once the problem data points have been identified, I email the jurisdictions and ask for them to verify their data, sometimes they correct the error, sometimes they verify that it's correct, and sometimes they simply ignore me (those data points get removed).
After I receive the responses from the jurisdictions, I then have to "extract" the data from our online survey tool. It extracts in a very raw form that is simply unreadable in that state, so I am required to format it, and transform some data into percents and the like. Once the data is legible it is posted online for our participants (city managers, analysts, etc.) to use in their annual reports and other performance management tools.
Sounds pretty basic, but it can be cumbersome at times as getting 220+ cities, counties, and towns to submit comparable data can be like herding cats.
Subscribe to:
Post Comments (Atom)
1 comment:
If the research has been run formerly, it is essential to keep uniformity in regards to example sources, our group at Data cleaning could manage the complete process for you throughout (questionnaire design, dealing with fieldwork suppliers, information cleansing, coding and the last analysis, discussion), we can make your lives less complicated by leaving everything approximately us. We will certainly additionally be happy to utilize the very same fieldwork suppliers you have previously used or conversely we will certainly locate the ideal ones for you.
Post a Comment