Thread by @DrBrocktagon

Remember that study claiming that gut bacteria from autistic kids cause "autism-like" behaviour in mice?

I had some fun with @f_l_o_u_r_i_s_h #dataviz and #opendata. It's pretty illuminating...
medium.com/metascience/ca…

OK. So you're claiming that mice show "autism-like" behaviours. You'd better have strong evidence of social impairment, right?

This is the data from the 3-chambers test - ubiquitous in mouse studies of autism.

Each box is a kid. Each dot is a mouse who received their gut bacteria from that kid's faeces sample.

If you're thinking there's no difference between the groups, you're right.

NB This sentence is the only mention of this null result in the whole paper.

Here is the *only* evidence of social impairment - from the direct social interaction test.

Note that there are only 5 autistic donors and 3 controls. And 2 of the controls are indistinguishable from the autistic kids.

TBC...

Next, communication impairment, measured in terms of ultrasonic (high frequency) vocalizations.

Here's all the data provided. Authors report significant difference between the 4 controls and the 5 ASD kids (excluding the 3 Mild ASD cases)

And now marble burying which is considered analogous to repetitive and restricted behaviours in autism.

I reckon this is the most compelling evidence of group differences. But again the significant group difference depends on (presumably post hoc) exclusion of Mild ASD cases.

Marble burying was correlated with ADOS scores of the donors.

"Children who displayed more autistic behaviours donated their faeces to mice who buried more marbles," is not a sentence I ever imagined myself writing.

Finally, mice with ASD donors showed reduced locomotion.

1. Really hard to believe there is any group difference in this data.

2. This was one of 9 measures extracted from this test with apparently no corrections

3. What has this got to do with autism?

And that's it! Don't know about you but I need more convincing.

Lots of other critiques in this earlier thread including @WiringTheBrain @Dereklowe @LukeHolman_Evo @surt_lab

https://twitter.com/DrBrocktagon/status/1134339288907206656

Also hat-tip to @alchemytoday who gave me the idea to visualise the data like this

https://twitter.com/alchemytoday/status/1135154680827469824

Final thoughts - I really sympathise with reviewers. The behavioural experiments were only a small part of the paper.

To review it competently you'd need to be an expert in animal behaviour, clinical psychology, genetics, microbiology, neurobiology, stats...

If nothing else, it's a GREAT advert for crowd-sourcing peer review.