15 Comments

I challenge Gareth to do some science.

Gareth Morgan wants to help New Zealand wildlife by restricting cat numbers. Last night on Campbell Live, Gareth Morgan was brought back on the show for a second run to make an announcement. As I understood it he is offering to pay the SPCA $5 for every cat that is not taken up by a family and micro-chipped, to euthanatise it rather than than, as he implied, have them release it.

If he doesn’t plan the experiment, measure the outcomes and test the results, no-one will have any idea if it helped or not.

I’m not going to argue his final point. Others say it’s true that in some locations cats are released. Something for investigative journalists to look into. (Putting it as the SPCA “has been taken over” isn’t helpful to my mind.)

I’m also not going to argue if these cats not taken up by families should be killed or not. Others can debate the science and morality of that.

But I will say that if you are going to do this, do it properly and treat it as an experiment. Hire ecologists. Plan it. Measure the current situation. Measure the resulting situations. Compare them.

If you just stick the money in without measuring the impact, you’ll have no idea how much effect it had and what effect it had.

Gareth is entitled to put his money where he wants. It’s his money, after all. But it’s evidence that will show if it was to good effect or not.

Anecdotal claims after the fact won’t cut it.

An important element when determining if one thing caused a particular outcome is to know of or rule out the role of any other things that might have caused the outcome. You want to deal with the different factors (variables) that might affect what you’re measuring.

I’m not a field ecologist. The details should be dealt with by people who are. Ecologists will know what previous studies have shown and have a better idea of what is important or not, but I imagine factors like location,[1] type of wildlife, predator competition and environmental conditions[2] all play a role.

You’d have to measure the outcome on each of these.

Don’t say obvious “it’ll be obvious.” That’s just one giant anecdote and will count for diddly-squat.

Don’t say you can work it out afterwards. That just doesn’t work. You have to plan data analysis before you do the experiment.[3] In addition to the usual making sure you have a procedure that will yield useful results, you’ll be wanting to measure how things are ‘before’ doing anything to compare with later.

If you want to tackle this, measure the effect of what you’re doing. Perhaps run a pilot programme or three or whatever it takes. In different settings, perhaps. Fund ecological studies.

Show it works.

Footnotes

Facebook users can see the Campbell Live page for early commentary, e.g. these posts.

* I know my title is just a little unfair. But I’ve got your attention now?

1. I’m thinking of within a large town or city, rural outskirts or fringe of small town, rural farming location, out in the (remote) bush, that sort of thing. (There’s a related thought: what locations matter most in preserving bird life?)

2. The weather, fire, etc. All the things that affect wildlife numbers.

3. I’m a computational biologist. Data analysis is a large part of what I do. For one post lamenting this issue, see External (bioinformatics) specialists: best on the grant from the onset.


Other articles in Code for life:

Thieves in gold-mining era campsites

Vaccination – why learn the hard way?

Mad on Radium

When things grow wild – post-earthquake natural succession in Christchurch gardens

Sea stars and mosaics

 Haemophilia – towards a cure using genetic engineering