Thursday, April 19, 2007

WAAL 2007: DataFerrett!

Virtual Processing of Govt Data Files
Mike Watkins

Mike, the govdocs guy at UW-Oshkosh, gave us a whirlwind tour of
DataFerrett ( , a very cool tool for
dealing with some government statistical datasets. This would be very
powerful for students and faculty working in a variety of fields,
including political science, public policy, health care, economics,

Below are my "scribbled" notes from Mike's PPT. They in no way reflect
the depth of Mike's knowledge about DataFerrett...

FERRETT: Federated Electronic Research, Review, Extraction and
Tabulation Tool

  • extracts and tabulates data from "small" data sets
  • first release was a collaboration between BLS and Census Bureau; now
    Census and CDC
  • applet and/or application which must be downloaded -- can be a problem
    on restricted computers


  • microdata
  • aggregate data
  • time series data
  • longitudinal data


  • small samples taken from surveys
  • scientific sampling techniques
  • decennial census 1% and 5% samples
  • small sample means large geographic areas: smallest is the MSA

Public Use Microdata Raw Data

  • Presents students and researchers with data to actually manipulate -->
    you may be aggregating a unique dataset

American Community Survey

  • Most current information -- may eventually replace the decennial census

Many datasets available, from ACS to FHWAR... Some of it is getting
kinda old, but some is pretty current.

Demo: veterans educational attainment
select census dataset
select variables --> Education
select required (geographic) variable --> Wisconsin --> Appleton-Oshkosh
MSA, etc.
select variables --> Military Service

Some tips:

* keep tabulations simple by limiting geog. area and overall number
of variables
* you can download the file if online can't process it -- in a
variety of formats: SPSS, xls, etc.
* fewer categories can make the data much easier to read and
manipulate -- recode variables
* can save final tables in a variety of formats too, including PDF...

Tutorials and user guide are pretty good and take about 45 minutes to

Other tips for finding stats:

* find "like" stats in StatAb and go to source...
* USAGov
* Google Gov

No comments: