Okay. Thanks for your comment. This discussion we’re having here is one of the few threads that hasn’t devolved into name-calling so I appreciate that.

I have two responses to this.

The first is that I’m still not so sure I agree with the framing here regarding cherry-picking or bias. Your concern seems to be (and correct me if I’m wrong) something like this: in most cases nothing out of the ordinary happens, so if we only focus on the few cases where something paranormal seems to have happened then we disregard the vast majority of the data and are only focusing on anecdotes. It’s more scientific to focus on the bulk of the data, where nothing interesting happens. (Again, please correct me if this is a misrepresentation.)

I don’t agree with this characterization because important data is often few and far between. But we shouldn’t discount it simply because it is rare. For example, consider Hawking radiation. From what I hear it’s an important concept in theoretical physics. But Hawking radiation is very hard to observe. In fact, it’s only been observed once, and the observation wasn’t even in the wild; it was in a lab. This was an important observation; it provided experimental support for an important concept. Say I was a physicist and I was sceptical of Hawking radiation. What should I do with this information? Should I say “well, this data doesn’t matter, most of the time we can’t observe Hawking radiation anyway so this data is just anecdotal”? No, that would be an improper response. Sure, this data is rare, but that doesn’t mean I can just label it as anecdotal and reject it on that basis. Because the data, though rare, still is very hard to explain without the concept of Hawking radiation. Similarly: it is possible that interesting data regarding near-death experiences are rare. Does that mean that this data is anecdotal and should be ignored? No. So long as we have cases that are genuinely hard to explain without supernatural explanations (and, I believe, we do) then that data will be very important. Because we still have to explain what was going on in those cases.

Another example of this is in the Earth sciences, where the large portion of the field is literally trying to create theories to explain one-time events. For example, the extinction of the dinosaurs. Should we reject the theory that they were killed by an asteroid or meteor or whatever simply because that event only happened once, and the event is therefore merely anecdotal? No. Even events that only occur once may require us to construct novel theories. So long as we cannot explain the event with current theoretical frameworks then it is our duty to invoke a new framework. As it is with dinosaurs so too with NDEs: even if there was just one, spectacular event that was difficult to explain with current frameworks, then it is our duty to invoke novel theoretical frameworks (so long as we actually want to know what’s going on). If the data leads us to theories that are paranormal in their character then, oh well, that’s just where we’ll have to go. If we want to follow the data to where it leads, then we cannot rule out certain destinations ahead of time.

It’s also worth pointing out that focusing on single cases is common practice in psychology and medicine. Sure, it’s not a replacement for theoretical understanding or large-scale studies, but it is still informative (for the reasons mentioned above). When researchers document and discuss a single interesting case it is known as a case study.

The second thing I wanted to say was regarding your estimates of total NDEs versus potentially paranormal NDEs. You seemed to be trying to aggregate over all the NDEs that have every happened and tried to find the ratio between the NDEs that are interesting versus the one that are amenable to mundane explanations. But I don’t know if this is super helpful. Because, for one thing, we’re largely left guessing at the numbers (how do we how many were interesting? how do we know how many were mundane? there’s literally no way to know). Even if we only look at all the data that we do have then we have to do that in a controlled manner, otherwise we’ll run into issues. If we only run thing haphazardly, back of the envelope style then we don’t know our scope (how many cases are we dealing with?) and we cannot control for any confounding variables (is this data interoperable?) or trace the data chain-of-custody (how did we even get this data to begin with, and how did that colour its presentation?). In short, it’s too messy.

So what we need, instead, is something more controlled. Ideally for something like this we’d want to look at a meta-analysis. But unfortunately I don’t know if anything with the required scope exists (if you can find one though, let me know). So the second-best thing to look at is an individual study. You mentioned earlier that you were looking at some studies. If you found any that you thought were interesting then it would probably be more productive to poke holes in that study specifically. I would be happy to discuss the merits of any study of your choosing and then take things from there.

But if we do that then I think the ratios you were discussing in your message would dissolve. This is because its practically impossible to conduct an NDE study with a large sample size (it’s hard to predict if/when/how/where someone will die, and the vast majority of those that do die don’t come back to talk about it). And with small studies even a single hard-to-explain NDE would be a relatively large percentage of the total sample (which should, I think, mitigate the concerns you expressed in your earlier message; but correct me if I’m wrong on that).