IPCC’s 5th Assessment Report: Interpreting 95% certainty in science, law, and from a climate change denier’s perspective.
In the time since the first section of the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5) was released a few weeks ago, one number from the document has received substantially more attention than the others. Surprisingly enough, this number doesn’t relate to projected future carbon dioxide emissions, sea level rise, or even increases in temperature. The number is 95% - how certain we are that humans are the driving force of climate change.
Climate deniers, of course, wasted no time – they grabbed this number and ran.
Scientists (and those who understand basic statistics) recognize the frustration experienced by the IPCC, as well as other organizations and individuals who are also attempting to communicate this finding, in their endless attempts to defend the proper translation of this value from the report to the real world.
I have personally attempted numerous times to explain ‘95% certainty’ within the context of this report to deniers and near-deniers (i.e., those who are skeptical but aren’t sure why), and I’ve yet to find a straightforward means of explaining this concept. Climate deniers jump at the opportunity to focus on the ‘unaccounted for’ 5%. In their arguments, 95% certainty translates into the presence of some sort of gap or a lack of understanding. I picture climate deniers as Lloyd from Dumb and Dumber, all with bowl cuts, saying, “So you’re saying there’s a chance!” (Note: I genuinely quoted this line before discovering that I am not the first one to do so in reference to the IPCC Reports and climate change deniers, though it is not clear if the author of this blog was thinking of Dumb and Dumber as well…). It’s relatively simple to convince others to adopt your point of view on this figure if they fail to understand ‘95% certainty’ from a statistical standpoint. ‘95% certainty’ within the context of the Assessment Report should not necessarily be analyzed as something simply occurring 95 of out 100 times. For instance, it is not meant to be compared to, say, a 95% certainty that your favourite beer will be on tap over the course of 100 visits at your local restaurant, while the keg will unfortunately be empty during the remaining 5 trips. It is much more complex than ‘beer’ or ‘no beer’.
95% Certainty in Science.
In science, at least within the context of climate science, a range of percentages correspond to a particular level of certainty. An event is ‘likely’ at 66% certainty, ‘very likely’ at 90% certainty, ‘extremely likely’ at 95% certainty, and ‘virtually certain’ at 99-100% certainty.
In hypothesis testing - one of the most common statistical methods used within scientific research - 95% is an important value here as well. Though argued as ultimately arbitrary in nature, the results of a study are accepted as going against the null hypothesis when the p-value, “the probability that the observed data under the assumption of no effect of obtaining a result equal to or more extreme than what was actually observed”, is less than 0.05 (which translates to 5%). In other words, it is the confidence that your null hypothesis (which states that there is no relationship between the variables being measured) is right – though a scientist would be shot if they “accepted the alternative hypothesis” instead of “rejecting the null hypothesis”. Essentially, if the p-value is less than 0.05, this means that the null hypothesis has a less than 5% chance of being right. So, if p<0.05, it is reasonable to reject the null hypothesis (which, in the world of research, is another way of saying, “Yay! Our results mean something!”). When p<0.05, you also get the right to use the term ‘significant’ as it pertains to your results, which would not be accepted if you described your results as such with a p-value larger than 0.05. The convention of using 95% as a basis for hypothesis testing in science states that you (that’s a collective ‘you’, as in the entire scientific community) are willing to accept being wrong 5% of the time, yet is still considered significantly different from the null hypothesis.
In science, 95% certainty is accepted. The reviewers of an article that is submitted to a scientific journal for publication (and other scientists in the field, for that matter), do not criticize the use of p<0.05 (unless, of course, the research itself is flawed, and in that case, wouldn’t make it into the journal). It is the accepted value at which the null hypothesis is said to be ‘wrong’. Generally, statistical analyses at this level generally stay within the journals and within the scientific community. That’s not to say, though, that significant results from interesting studies or studies of public interest will not be reported within the news. They occasionally are, but it is most likely that only the conclusion will be broadcast. For instance, if the 6 o’clock news reports the findings from a study by a group of researchers who study some peculiar combination of invertebrates and ethanol, they may say: “Drunk earthworms are more likely to wiggle in a circle,” (completely hypothetical research, obviously), but withhold ‘p = 0.032’, because really, the pubic doesn’t care so much about the stats. People just want to know what it means. With the IPCC reports, however, the results are being scrutinized from all possible angles. So, even though the vast majority of scientists do not disagree with the significance of the relevant research published (since reviewers of such research would not allow faulty science to be published), 95% certainty, at least in statistical terms, is accepted and rarely debated. However, once an issue surrounded by controversy (i.e., climate science) comes into the public’s viewfinder, 95% is now open to interpretation.
(I am aware that the context of ‘95% certainty’ as relevant to AR5 does not stem from a single scientific study or p-value, but rather from a general consensus as to the likelihood that the reasons for our changing climate is anthropogenic in nature. My main point here is that, within the disciplines of science and statistics, 95% confidence is sufficient to conclude that the results are significantly different from a null hypothesis).
I recently went to a briefing for the public and media in Vancouver regarding the outcomes of the IPCC AR5 Working Group I report, with presentations by members of Working Group I itself (who investigate the physical sciences behind climate change). I very much enjoyed this experience because, even though aspects of the report were challenged and questioned, the questions did not address the usual items addressed by climate deniers (i.e., trivial items, typically). The presentation remained focused on the science, was unbiased, and did not even address the policy implications or potential adaptation and mitigation measures from this section of AR5. (In fact, it was politely asked that the audience refrain from asking questions regarding policy or future mitigation efforts, because these issues do not fall within their area of expertise).
95% Certainty in the Law.
Andrew Gage, a lawyer with West Coast Environmental Law, also explains various issues with climate deniers’ interpretation of 95% certainty. He addresses how certainty is interpreted within the context of the law to demonstrate just how much proof 95% certainty carries. This level of certainty is substantial enough to assign charges not only within civil law, but is also considered ‘beyond a reasonable doubt’ – the standard of legal proof used in criminal law.
Gage explains how different percentages of likelihood relate to both civil and criminal law. Within the context of civil lawsuits, “a balance of probabilities” is used as legal proof, meaning that something is ‘more likely than not’ (i.e., when certainty exceeds 50%). Historically, a balance of probabilities has been the type of legal proof used in assessing whether or not to act on climate change. Recalling that the term ‘likely’ corresponds to a 66% certainty, any statements out of the IPCC reports that include the terms ‘likely’, ‘very likely’, and ‘extremely likely’ when referring to anthropogenic drivers of climate change carry enough certainty to be proven at (above) a balance of probabilities. To put it simply, 95% far exceeds 50%, and so, within civil law, it could be proven that, yes, humans are the driving force behind climate change.
If we were evaluating this statement within the realm of criminal law, however, one may not expect such a bold statement to be proven as true. Shockingly, however, the report’s declaration that climate change is ‘extremely likely’ to be caused by humans can very well be proven “beyond a reasonable doubt”. Yes – that’s the same “beyond a reasonable doubt” required to convict someone of murder. Though the courts have refrained from assigning a specific number to describe what exactly constitutes “beyond a reasonable doubt”, some say that this corresponds to approximately 90% certainty. Considering both AR4 and AR5 reported the likelihood that humans are responsible for climate change as 90% and higher, proof that anthropogenic activity is to blame for climate change stands from both a civil and criminal law perspective. With this level of certainty, you could convict someone with murder. Just sayin’. Gage sums it up well: “Once a murderer is convicted, we stop referring to them as ‘alleged murderers’. I think that the same can be said for the fact that humans are contributing to climate change.”
Rather than poking holes in the Assessment Reports for the purpose of sparking debate and initiating controversy to further their agenda, it is time for deniers to accept the truth – it is extremely likely that humans are the cause of climate change. No matter how you interpret ‘95% certainty’, let’s face it – it’s pretty darn certain we’re to blame.