Life as a Stochastic Modeler
Alan Gelfand, Duke University
F
irst, let me say I am flattered to be invited to write about my
career in statistics. I do not know if many readers will find it
interesting, but I do know it has evolved along a path that
may be viewed as a reflection of the evolution of our field.
What I have become is a stochastic modeler. In particular,
a modeler whose primary focus is in the area of environmental
processes, including ecological systems, exposure assessment, and
climate processes. A common ingredient of this analysis is the col-
lection of data across space and, often, across time. I am usually
studying a complex process with different types of information—
theoretical results drawn from physical principles, mechanistic
insights based on knowledge of aspects of process function, and
empirical knowledge as a result of previous data collection and
relevant laboratory and field experiments.
I imagine the process is described at multiple levels, with the
foregoing information entering in at different places and lev-
els. I typically represent the process through an acyclic-directed
graph with some nodes observed and others unknown, and then
I infer about the unknown nodes given the observed nodes. I
formulate the joint model in a hierarchical fashion driven by
the graph and fill in the stochastic details needed to complete
the model specification. In essence, a stochastic modeler seeks
the posterior distribution of what we don’t know given what we
have observed, so we usually fit these graphical models within
the Bayesian framework.
It has emerged, more clearly than ever, that such modeling is
my greatest strength as a statistician. Moreover, being part of a
team of researchers assembled to ‘brainstorm’ a complex prob-
lem is an exceptionally stimulating and rewarding activity. In
this setting, the modeler becomes a central player in synthesizing
inputs from team members, shaping progress on the problem,
and becoming a richer scientist as a result.
I think this research view serves as a contemporary perspective
of our field. The team research concept, which it presumes, dra-
matically revises the role of the statistician from someone brought
in at the end to carry out data analysis and create ‘pretty’ pictures.
Rather, the statistician is able to illuminate what we can learn with
what we have, as well as what we need to collect to learn about
what we want. In the midst of all this, it is essential that the mod-
eler retain technical rigor, attention to detail, and appreciation of
the properties and features encompassed by the modeling.
SEPTEMBER 2008 AMSTAT NEWS 21
SEPTEMBER AMSTAT FINAL.indd 21 8/20/08 2:26:57 PM
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