[naasking]

The problem is not just with "belief", but in the process itself. Non-replicable studies are actually cited more than replicable studies:

https://advances.sciencemag.org/content/7/21/eabd1705

Science journalists probably are vulnerable to the same thing influences that lead scientists to do this, except they have even less review on their claims and so they become pop culture sound bites.

As for why non-replicable results are cited more, I'd speculate that non-replicable results are often more unintuitive and surprising, and per the above link, reviewers apply lower standards on these papers in the hopes of finding something truly interesting and/or exciting. Not just in the results mind you, sometimes papers also apply a novel methodology that might be worth wider discussion. I'm not sure that's worth the reduction in credibility though.

[theptip]

> As for why non-replicable results are cited more, I'd speculate that non-replicable results are often more unintuitive and surprising

This seems the likely explanation; I saw a paper recently that showed that lay people can predict what will replicate with above-chance accuracy[1]. I imagine scientists are even better than lay people at this.

So non-replicable results are almost by definition surprising (i.e. they are hypotheses that don't match our current model of how the world works), and surprising results are definitely better news than unsurprising results.

[1](https://journals.sagepub.com/doi/full/10.1177/25152459209196...)

[edanm]

> As for why non-replicable results are cited more, I'd speculate that non-replicable results are often more unintuitive and surprising

I mean, a kind of well known thing is that in general, something false can be more interesting than something true, since it has more degrees of freedom. You can make up anything you want - the truth has to conform to what's actually real.