bcss
Biodiversity Conservation Society Sarawak
Pertubuhan Konservasi Biodiversiti Sarawak
 
 

bcss wildlife stats trainers area


Trainers workshop, Kuala Lumpur, 10-13 January 2018

We will hold another "Elf Camp" in Jan 2018 with Ian Signer as the lead facilitator. This is immediately before the Boot Camp at Sunway University (15-26 January), and we hope some of the Elf Camp participants will come to the Boot Camp to practice their skills.

We have again booked facilities at Akademi Hevea, part of the Rubber Research Institute complex, Sungai Buloh, close to Subang airport. We encourage all participants to stay at the Lodge.

The workshop is by invitation only. It is free of charge, with the expectation that graduates will help with future BCSS workshops.


Who has helped with workshops since the last Elf Camp?

Codes are date (YYMM) and location, see Past Workshops for details.

  • Ngumbang: all of them! (That's his day job.)
  • Fen: 1601 RRI, 1603 GIS@PJ, 1607 GIS@KMUTT, 1609 Bhutan
  • Sylvia: 1604 Baluran, 1611 Kuching
  • Lukmann ("Bob"): 1607 NUS, 1707 JAGS@KMUTT, 1709 Bhutan
  • Dusit ("Soy"): 1607 Graphs@KMUTT, 1610 Jogjakarta, 1703 XTBG, 1707 JAGS@KMUTT
  • June: 1607 GIS@KMUTT
  • Win Sim: 1612 RRI, 1703 XTBG
  • Elan: 1612 RRI
  • Akchou: 1708 SUT

Strategies for the future

Towards the end of the 2015 Elf Camp we had an open discussion on future strategies of the Boot Camp, in the context of increased demand from potential participants. (We had done 5 Boot Camps in the previous 12 months.)

Some of the main points from the discussion were:

  • Aim to develop "satellite" teams of elves in each country in the region, eg, Lao-Thailand, Bangladesh, Indonesia.
  • Need for quality control: participation of "HQ" in all Boot Camps; exchange of elves between countries; centralised elf training (plus feedback from satellite teams); don't expand too quickly; limit participants to 20 to 24; provide more documentation (Green Cards are good prompt, but more background needed); elves meet for ~2 days before Boot Camp to review material.
  • Funding needed for regular elf training and support, cost should not be passed on to participants.
  • Filtering participants: Main target for Boot Camps is young researchers actually doing wildlife research; many postgrads fall into this category. Current methods seem to work pretty well, though we have less control over workshops outside Malaysia.
  • Credibility issues (ie, phasing out the white-beard): Profiles of elves on the web; emphasis on team rather than 'guru'; having a "driving licence" scheme [in Malaysia we have 'L' then 'P' then full licence, something like bronze-silver-gold may be better internationally].
  • Online versions of Boot Camp discussed briefly, but it's popular because it's hands on. Also lots of resources online already.

Trainers workshop, KL, 7-11 December 2015

We had 12 participants for the 5-day "Elf Camp" plus Ian Signer, Ngumbang and Mike as facilitators.

The first 2 days were devoted to learning principles, a framework for assessing lessons, and questioning techniques. Most of the time in the last 3 days was taken by practice lessons run by participants followed by feedback from the rest of the group. This brought up lots of suggestions for ways to improve the activities.

The 5-day format was much better than the 2-day sessions we have run for elf training in the past, and will be the format of choice for the future.


Why Bayes first?

We scrapped NHST in 2013 and confidence intervals were downgraded after Dec 2014. In Bangkok in July 2015 we ran a revamped Boot Camp introducing Bayesian analysis before frequentist and information theoretic (AIC) methods. All Boot Camps since then have used this "Bayes First" approach.

Frequentist methods seem simple and intuitive, but the intuitions are wrong: a p-value is not the probability that null hypothesis is true, a confidence interval is not a credible interval. For details see Gigerenzer, Hoekstra, Morey...

The Bayesian approach is intuitive and the intuitions are correct. It can be taught at elementary level, eg Albert & Rossman. The impression that Bayesian methods are difficult comes when people turn to Bayesian analysis when they encounter problems that cannot be solved with frequentist methods: the Bayesian approach works, but it's not going to be simple.

We no longer have to worry about Bayesian methods being criticized as "unscientific". A bigger worry is that a tradition of using uninformative priors seems to have taken hold in ecology.

 

Updated 4 August 2017 by Mike Meredith