Aston University Mumsnet Corpus FOI requests

[AnotherAngryAcademic]

I've been keeping an eye on the various FOI requests about the data scrape. One has been declined, with Aston saying they don't have the information requested. However

I'm not sure what the exact status of the data is at the moment (IIRC Aston have agreed to delete it) - but it is worth noting that analysis was being done on the student's own computer, which is not encrypted, so hopefully anything related to those analyses has also been deleted. There is a spreadsheet of usernames and the threads they post in, which will presumably also be deleted?

Snippets from the ethics form for the PhD project:

"There were two factors I needed to consider when measuring whether informed consent must be gathered for the purpose of this project: the first is how much contact I will have with the Mumsnet posters, and the second is the level of privacy afforded by Mumsnet fora (Eysenbach and Till, 2001). I will carry out a ‘passive’ (Eysenbach and Till, 2001:1103) analysis of the linguistic data, meaning that I will not be directly involved with the Mumsnet posters. Instead, I will analyse linguistic data that already exists on the website. According to Roberts (2015), this usually means that informed consent is not required."

"Raw data used in the study will only be analysed on my own password protected laptop, and stored on Box, the cloud device used by Aston University. During the data collection portion of the project (as the text files are being created), any identifying information pertaining to the posters will be redacted for anonymization purposes (for example, some usernames and locations).

A key will also be created of Mumsnet usernames to quantify the number of posters in each thread, which will give me a number of participants overall. This will be done by copying and pasting each username on each thread into a spreadsheet for each corpus, and cross-referencing how many threads feature posters with the same name. This will also allow me to assign pseudonyms to participants if needed, and to track which pseudonym belongs to each user. This key will be deleted after the redaction has taken place."

"There is a very small risk of potential loss of anonymity of posters."

"Overall, it will be unlikely that posters’ true identities will be accidentally revealed during the project. This has been discussed at length between me and my supervisors."

"Could participation cause discomfort (physical and/or psychological – e.g., distressing, sensitive or embarrassing topics), inconvenience and/or danger beyond the risks encountered in normal life? Please indicate the level of risk and plans to address these potential risks. N/A".

(Apparently a risk to the researchers was anticipated, but this has been redacted.)

Supervisor's comment:

"I can confirm that I have had extensive discussions with <student's name> about this application and the design of the project as a whole. We have talked at great length about the importance of redacting usernames and applying pseudonyms, and destroying the key to pseudonyms as soon as soon as this is completed. While strings of words will remain searchable and discoverable on Mumsnet.com, they will not point to an identifiable individual as this information is not shared on Mumsnet. We have also discussed the importance of researcher welfare and ensured adequate support is in place."

From correspondence about the ethics form:

"The second is just to flag the possibility that any direct quotes you may use (I wasn’t sure if this was a possibility) could be googled and lead back to the Mumsnet user which MAY including identifiable information (as we can’t guarantee pseudonymisation via username e.g. if I was xxxxxxxxxxxxx, someone could work out it was me!). It would be great if you could discuss how you’re mitigating this a little more to demonstrate that you’ve thought about the possibility (even if remote) and what steps you’re taking to ensure accidental identification doesn’t take place. This might be via paraphrasing quotes instead of direct ones, etc. If you’re not planning to quote at all, this is fine – I just wanted to flag that this is something the Chairs will be thinking about – it’ll just have to be explicitly stated in your form for better understanding, and our records."

(Given that this document is in the public domain I think it is reasonable to quote here. The whole thing is available at Mumsnet corpus FOI for anyone interested!)

Well they effed this bit of the ethics assessment up!

There are so many issues:

The data they scraped may include posts that the poster and / or MNHQ later decided was a safety risk and were since deleted. Aston will not have deleted these. This will be sensitive data which has extra protections under GDPR, Aston appear to have not even considered this (numpties).

There are identifiable named users on here - Glinner, Sarah Philimore, Maya F, Jo P., Dennis Noel K to name a few (Caroline Farrow?). So they are scraping posts from named individuals. Who have granted a license for copyright to MN but NOT Aston. Aston did not get permission from MN for this data scraping. In addition labelling these named individuals as 'transphobic' has already, in other circumstances, caused quantifiable harm - such as Glinners loss of work. It is not ethical to do research that libels named people in this way and causes them harm.

My opinion is they are breaking the law on GDPR / data protection and copyright and libel. At the very least. IANAL however, but I do think I'm right on this.

They might also be breaking the law on some commercial / business issues - they didn't adhere to MN terms, that's clear. If this becomes more widely known there will likely be many users who will not post any more or not post as freely which could impact MN's business.

Then there's the general issue of suppression or restriction of freedom of speech of women on one of the only online forums that welcomes women who are sex realists. Sex realism is a WORIAD. So this could constitute discrimination. No clue if it could be legally prosecutable, but ethically shady and I'd say poor choice of use of any public funding (which presumably Aston has some of).

...and this definition! "Gender critical" is literally what it says on the tin! Critical of gender. Even captured Wikipedia defines "gender critical" correctly!

I'm fairly sure in at least one, possibly more of the recent court cases won by GC / sex realist women the court was extremely unimpressed at the deliberate smearing of women as 'transphobic' if they held sex realist views.

Tsk Tsk Aston. Not ethical.

One has been declined, with Aston saying they don't have the information requested

I don't believe them. Hasty deletions all round, perhaps. Very much 'not in the spirit'.

Also this... "Though non-linguistic studies relating to a potentially transphobic culture on Mumsnet (e.g., Pedersen, 2022; Thurlow, 2022), and discourse analyses of other radical online communities such as incels or Islamist extremists (e.g., Koller, 2021; Krendel, 2020) have been undertaken, the language of transphobia on Mumsnet has not been specifically examined. This study aims to address this gap."

Incels and radical Islamists..?! Those people (mostly men) going around fantasizing about killing people (in the case of Incels, mostly women), or literally killing people?!

Has she ever actually been on the feminism board and read anything we post?!

Sorry to ask on this thread, but I was around when all this started to break (and Justine was about to meet with the VC) but haven't caught up since then - is there a summary thread anywhere on what exactly has happened between Aston, the researchers, and MN?

but it is worth noting that analysis was being done on the student's own computer, which is not encrypted

At my university, doing this, storing it unencrypted on a personal computer, would immediately invalidate the student using ANY data or conclusions drawn from such data in their dissertation.

Such shoddy shoddy ethics and research. But then, it is Aston, not exactly renowned as a research institution. The best comp linguistics research unit in the Midlands is Aston's far superior neighbour in Edgbaston ...

I'm part of a university ethics community and you are required to store information on their secure system. On your own computer is not allowed.

Surely it's been a month now since this was first brought to light and Aston contacted?

A key will also be created of Mumsnet usernames to quantify the number of posters in each thread, which will give me a number of participants overall.

Hur hur. Ahem. No it won't. Doesn't s/he know there are only six of us, with a lot of usernames?

😎😎😎

Shhhh don't let on @CatonmyKeyboard

Squeek fur yurself, Ize defnnitly no sok poppet!

I do so love my multiple identities....

I tried to find the reference given in the student's proposal to Roberts (2015) - as in 'According to Roberts (2015), this usually means that informed consent is not required.' Bear in mind that 2015 was a decade ago which in terms of online forums (research and development thereof) is longer than 10 human years ...

I found Roberts but also found something even more interesting during my search. LSE's guidance on Using data from the internet and social media in research: Ethics and Consent. https://info.lse.ac.uk/staff/divisions/research-and-innovation/research/Assets/Documents/PDF/ethics-Using-internet-and-Social-media-data-v8.pdf Which contains this.

'Even where data is publicly available, it is usually ethically advisable to anonymise any information that could identify the poster, including by not using direct quotes, where this is compatible with the integrity of the research. Cases where this is not, in general, necessary include: if the person who has posted the information is a public figure (e.g. a politician, a celebrity, etc.); if the content of the postings are, in the main, non-contentious, or by contrast, if the posted content contains dehumanising, hateful and/or other material that it is in the public interest to draw attention to.'

Note my bold bits. So LSE advice is that its permissible to not anonymise if the content is dehumanising or hateful. Drawing parallels to communities of incels or islamist extremists (as pointed out by @WitchyWitchersonabove) could mean NOT having to ask for consent, ethically and make a researcher's task easier or less bothersome re, anonymisation. (I assume the logic is that people who engage in this sort of talk do not deserve consideration). Back we go to how this student and supervisor decided the content was transphobic before engaging with it.

Or indeed how they decided it was transphobic if they did engage with it.

Excellent point by all pps and you, @Gettingmadderallthetime.

And 'Roberts (2025)' obviously preceded GDPR (it became enforcable on the UK of 25th May 2018, after passing through the European Parliament in 2016).

Roberts (2015) unless he has travelled to the future!

I do not think it is the case that Aston have agreed to delete the PhD data set from earlier this year; they agreed to delete the dataset which was the scrape of the whole website (the ‘sandbox to play in’) from 2018.
I posted on the site stuff thread asking MN to clarify if Aston were using the data for the PhD with their permission, and await a response.

I also came to the conclusion that Aston could argue public interest for content which was criminal (eg dark web or similar), but failed to see how this applied to either the LGBTQ+ children board or the relationship board, even if one accepted the premise that the content on the FWR board was or is or criminal (I do not accept this premise and think the research question is problematic).

Place marking to keep on top of this…

@Talulahalula I wondered whether the 2018 data was not sufficiently fresh to be of great use so fine to get rid of. I imagined that given the emphasis in this PhD that its the FWR board that has been scraped this time. I agree its not hateful or dangerous of threatening. By any normal standards.

There have been several court cases now that have reinforced that GC beliefs are WORIADs so I think Aston are on shaky ground. Obviously they think that they get to decide what 'extremism' is but that's not how the law works.

This is infuriating.

There are plenty of posters who pop in who don't understand what gender critical means.

But I always assume they're either American, extremely young, a bit thick, or being deliberately goady.

I expect better from an academic.

Imagine, your whole enterprise is pointless because you couldn't be bothered to learn a crucial definition. So fucking arrogant.