Using the past to predict the future

PredictionsPhoto

Successful decision-making needs to predict the future. How can we do this best?

In most cases, this is not easy: in biodiversity conservation, predicting future outcomes is complicated by complex ecological, economic, and social contexts.
But these challenges do not negate the need for us to make these predictions when deciding what conservation interventions to implement, and where.

In late 2015, we invited causal inference expert, Professor Paul Ferraro from John Hopkins University, and a team of international researchers to join us at the University of Queensland, to really nut out what it takes to translate some of the latest causal inference thinking and apply it to causal predictions. Considerable discussion and debate later, we emerged with three key nuggets of advice:

1. Clarify your causal assumptions. To predict the future, and evaluate decision alternatives, we need a model of how the world works, and some alternative scenarios to compare. These are really important to clarify, and justify, especially for interpreting the predictions, but these details can easily get glossed over, forgotten, or altered as the analysis progresses. Clarifying these details to yourself and others is essential for informed decision-making.

2. Use better data. Data is rarely perfect, but if we know what the biases are, or could be, then we are able to build this knowledge into the interpretation of predictions. Knowing where and how biases may occur, and how to mitigate biases in experiments, when collating data from the literature, and when eliciting information from experts, is a crucial skill to learn and practice.

3. Use data better: Under what conditions do the predictions hold? What do our assumptions mean for our interpretations? Conducting sensitivity and uncertainty analyses, including using techniques to examine the impacts of key assumptions, is due diligence in predicting the future.

In the paper we present a number of techniques that can be used to clarify causal assumptions, identify and develop better data, and use data better in predicting the future outcomes of conservation interventions. This confidence in the robustness of the science is, of course, only one element contributing to the wider salience, legitimacy, and other forms of credibility of policy advice and of policies themselves, but it is an important element to maintain public trust in science. Broader recognition and uptake of these tools and approaches will help to develop more scientifically credible projections of impacts, and thereby, if heeded in policy development, better outcomes for conservation.

 

Citation:

Elizabeth A. Law, Paul J. Ferraro, Peter Arcese, Brett A. Bryan, Katrina Davis, Ascelin Gordon, Matthew H. Holden, Gwenllian Iacona, Raymundo Marcos Martinez, Clive A. McAlpine, Jonathan R. Rhodes, Jocelyne S. Sze, Kerrie A. Wilson, Projecting the performance of conservation interventions, Biological Conservation, Volume 215, November 2017, Pages 142-151, https://doi.org/10.1016/j.biocon.2017.08.029.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s