Chiu wins Pierre Robillard award for best Ph.D. in Statistics in Canada in 2002

25/07/03

Congratulations to Dr, Grace Chiu for winning the Pierre Robillard Award for the best Ph.D. thesis in Statistics in Canada in 2002. She was recently recognized for this achievement at the 2003 Annual Meeting of the Statistical Society of Canada held in Halifax.

The Award is in honour of Pierre Robillard, a very talented and dynamic young statistician, whose untimely death in 1975 cut short what promised to be a brilliant and distinguished career in his chosen profession.

Grace's thesis was titled `Using the Bent Cable to Assess an Abrupt Change in Species Abundance and Other Phenomena' and was written under the supervision of Professors Richard Routledge and Richard Lockhart of our Department. She is currently a PIMS Postdoctoral fellow.

She previously won the International Biometric Society Western North-American Region (WNAR) Student Paper Award in 2002 for the same work.

Grace's thesis can be used in many practical situations. For example, one application of her work was in examining if the decline in sockeye abundance was gradual or abrupt.

"..The population [of sockeye salmon in River's Inlet, British Columbia] has declined from being one of the largest in Canada to an endangered remnant. The cause of the decline remains uncertain. Researchers are uncertain also over the timing and abruptness of its onset.

Did the decline being abruptly around 1993? Or was it more gradual, possibly starting earlier? To address this question, a researcher would normally fit a broken-stick model to the data estimating the unknown change point. In such a case, an a priori abrupt change in the abundance-year relationship is assumed. However the graph [shown in the thesis] seems also to suggest a range of years between 1985 and 2000 over which abundance gradually dropped. Instead of constraining the model to have a break point, we allow for a smooth change by "extending" the breakpoint to a curved "transition region" in the model - hence the bent cable.

.... The bent-cable fit allows the data to reveal whether an abrupt change or a smooth transition is more convincing ...

... The estimated bed ranges form 1986 to 1998. However, the data are consistent with a broad range of other values... The decline in sockeye abundance could have been accelerating steadily over most of the time range shown. Or it could equally well have begun abruptly around 1993. The data shown are consistent with both interpretations.

"... In the case of a declining fish population, knowledge of an abrupt change in, say, the abundance of an ocean predator to feed on sockeye, or a key aspect of habitat quality affected by logging activity could provide such evidence. In contrast, a bent-cable fit for these data would point to one or more source of influence took hold gradually or whose onset occurred at different instances over a range of years.

"... Thus interpreting the onset of the decline as abrupt without any solid evidence could lead to inappropriate conservation measures."

Please contact Grace Chiu (sgchiu@stat.sfu.ca), Rick Routledge (routledg@stat.sfu.ca), Richard Lockhart (lockhart@stat.sfu.ca), or the Statistical Society of Canada (info@ssc.ca)

The official press release from the SSC is here

A copy of her thesis can be accessed from our Alumni page