original and reinterview survey under Daniel Kasprzyk on many
data. Traditionally, the reinter- important government policy
view program measured these issues, including the free and
rates with a response variance reduced-price lunch program.
study under the assumption that These projects made me feel I
the original survey had the same had a role in making a real differ-
level of misclassification as the ence in people’s lives.
reinterview survey by means of I decided in 2003 to return to
replication. The response-bias federal service in the Department
procedures relied on additional of Labor at the Office of Federal
probes or reconciliation to gener- Contract Compliance Programs
ate a presumably negligible level (OFCCP). OFCCP is charged
of error. By using various alge- with the difficult task of moni-
braic manipulations of the prob- toring the personnel practices of
ability structure and examining federal contractors to ensure their
different assumptions, we were actions are nondiscriminatory.
able to gauge the level of bias My role was to help the agency
in both reinterview techniques. understand the utility of various
This, with the work of many types of data and how to best use
other talented researchers, led to it statistically to judge the actions
These groups are supported by a new reinterview program. This of the audited companies. The
our information technology and example highlighted the impor- instructive nature of this position
publications units, and we pro- tance of the mathematical rigor required me to revisit some of the
vide critical grants to improve in the student’s classes and how foundational theories and concepts
state record-keeping systems to their experiences from working I learned in academia so I could
enhance criminal history records in different subject areas could explain statistics and defend the
for conducting background be applied in practice. results to nonstatisticians. As many
checks and supporting state-level After leaving the U.S. Census have told me, you don’t really learn
“
… the best statistical analysis centers (SACs).
Bureau in 1992, I began a peri- statistics until you have to teach it
advice I can
I began my statistical career in od of employment in private to someone else.
1986 at the U.S. Census Bureau industry focusing on marketing I now find myself in a new
offer to
with a new master’s degree from research and social programs. subject area, with an exciting new
Oklahoma State University. While I often hear statisticians position at BJS. However, through
students is
During my time at the bureau, I downplay the importance of all these changes, my statistical
to conduct as
took advantage of an opportunity what might be referred to as less- training has been constant and
to work under John Thompson, rigorous statistical work in mar- continues to serve me well. Every
much hands-on Henry Woltman, and Jon Clark
keting or similar areas, I found position has unique strengths and
data analysis as
in the then Statistical Studies this work could be quite chal- weakness, but I have found both
Division (STSD) and to pursue lenging and meaningful. In par- government service and private
possible …
”
and complete a PhD in statistics ticular, I helped a company assess industry projects to be fulfilling.
at George Washington University. the future market for beverage What makes sense for an indi-
I also had the fortune of select- medium, namely sport drinks vidual will depend on what it is
ing Joseph Gastwirth to be my and flavored water beverages, the agency or employer does and
adviser, which led to a long-term that were at that time only con- how that relates to one’s personal
relationship and the opportunity sumed by a few. In this project, I style and interests. As practitioners
to learn from a master about the had the chance to apply numer- and artists of our work, we should
measurement of nonsampling ous statistical methods, including demonstrate the great value of the
errors and the use of statistics multistage and double-sampling theoretical and applied academic
in examining discrimination in procedures; factor, cluster, and courses we rely on and recognize
personnel decisions. regression analysis; complex vari- those who teach these courses
As part of my early career, I ance estimation techniques; and to meet our needs. Probably the
conducted research to improve survey nonresponse adjustments. best advice I can offer to students
the U.S. Census Bureau’s rein- Though no one can say what is to conduct as much hands-on
terview program. Gastwirth had effect the study had, these prod- data analysis as possible and take
put forth the idea of building ucts are now quite prominent in advantage of any opportunities to
on an approach from medical the market today. improve your writing and presen-
diagnostic testing to measure During this period, I also tation skills. This will serve you
the rate of false positive and worked at Mathematical Policy well, wherever you find yourself in
false negative responses in the Research Incorporated (MPR) the future. ■
46 AMSTAT NEWS JULY 2009
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84