Chao Wang
@excel-wang.bsky.social
Associate Professor in health and social care statistics at Kingston University. PhD in econometrics.
P-value has been severely criticised for 7 decades (the paper in the screenshot was published in 1994). I guess it's not going to go away anytime soon...
November 18, 2025 at 10:24 AM
P-value has been severely criticised for 7 decades (the paper in the screenshot was published in 1994). I guess it's not going to go away anytime soon...
One can certainly view it as an intercept-only linear regression model. One can also view it as a nonparametric moving-average smoother with the purpose of smoothing noise and giving a baseline, which does not attempt to model the data-generating process.
November 17, 2025 at 12:24 PM
One can certainly view it as an intercept-only linear regression model. One can also view it as a nonparametric moving-average smoother with the purpose of smoothing noise and giving a baseline, which does not attempt to model the data-generating process.
Yes but this could either mean "the deaths are expected to continue to decrease" (without pandemic) or "the deaths are expected to increase due to 'dry tinder effect'" (www.sciencedirect.com/science/arti...). A simple average is not perfect but it's more model-free.
November 17, 2025 at 1:09 AM
Yes but this could either mean "the deaths are expected to continue to decrease" (without pandemic) or "the deaths are expected to increase due to 'dry tinder effect'" (www.sciencedirect.com/science/arti...). A simple average is not perfect but it's more model-free.
Strange to cherry pick two states. How about a comparison of death rates across all blue and red states? Note red states generally had less stringent policies. www.bsg.ox.ac.uk/sites/defaul...
November 16, 2025 at 4:15 PM
Strange to cherry pick two states. How about a comparison of death rates across all blue and red states? Note red states generally had less stringent policies. www.bsg.ox.ac.uk/sites/defaul...
I think assuming the number of deaths being roughly stationary is more reasonable than assuming the deaths would always increase or decrease at the same rate.
November 16, 2025 at 2:51 PM
I think assuming the number of deaths being roughly stationary is more reasonable than assuming the deaths would always increase or decrease at the same rate.
Too early to tell anything meaningful from just two years' data. "Machine learning engineer" saw the biggest increase (+39.62%), and other data jobs seem healthy: research/applied scientists (+11%), Data Analyst (+0.5%), Data Management Specialist (+1.1%), against the "baseline" -8%.
November 14, 2025 at 4:03 PM
Too early to tell anything meaningful from just two years' data. "Machine learning engineer" saw the biggest increase (+39.62%), and other data jobs seem healthy: research/applied scientists (+11%), Data Analyst (+0.5%), Data Management Specialist (+1.1%), against the "baseline" -8%.
Yes it adds to the evidence base, albeit not a very strong one.
November 10, 2025 at 12:14 PM
Yes it adds to the evidence base, albeit not a very strong one.
Yes but they are all methods trying to match the UK with other countries, which is unnecessary. A more flexible approach is synthetic difference-in-difference, which only creates "parallel trend". See the comparision between SC and SDID www.nber.org/papers/w25532. The true effect may be smaller.
November 10, 2025 at 11:59 AM
Yes but they are all methods trying to match the UK with other countries, which is unnecessary. A more flexible approach is synthetic difference-in-difference, which only creates "parallel trend". See the comparision between SC and SDID www.nber.org/papers/w25532. The true effect may be smaller.
Synthetic control method is a bit outdated tbh.
November 10, 2025 at 10:54 AM
Synthetic control method is a bit outdated tbh.
Also: AI peer-reviews papers (it's "peer"-reviewed because papers were written by AI)
November 7, 2025 at 11:15 AM
Also: AI peer-reviews papers (it's "peer"-reviewed because papers were written by AI)
Looking at the first plot, the UK's downward trend began well before 2016. It seems quite difficult to notice any significant changes around 2016 in the plot.
November 1, 2025 at 3:30 PM
Looking at the first plot, the UK's downward trend began well before 2016. It seems quite difficult to notice any significant changes around 2016 in the plot.
The correct word is "he" dictionary.cambridge.org/dictionary/e...
October 30, 2025 at 2:13 PM
The correct word is "he" dictionary.cambridge.org/dictionary/e...
The impact of smoking on Covid? thorax.bmj.com/content/77/1...
Smoking and COVID-19 outcomes: an observational and Mendelian randomisation study using the UK Biobank cohort
Background Conflicting evidence has emerged regarding the relevance of smoking on risk of COVID-19 and its severity.
Methods We undertook large-scale observational and Mendelian randomisation (MR) an...
thorax.bmj.com
October 29, 2025 at 9:20 AM
The impact of smoking on Covid? thorax.bmj.com/content/77/1...
Maarten has a paper on that diagnprognres.biomedcentral.com/articles/10....
October 9, 2025 at 9:22 AM
Maarten has a paper on that diagnprognres.biomedcentral.com/articles/10....
I think it's partially due to funding (is social science famous for ample research funding as in medicine?), partially culture/tradition. A few years ago, Elvik (well known in road safety) asked why so few RCTs in road safety research. Seems an interesting read. www.sciencedirect.com/science/arti...
Why are there so few experimental road safety evaluation studies: Could their findings explain it?
Randomised controlled trials (also known as experiments) are widely regarded as the best design of studies that aim to estimate the effects of a treat…
www.sciencedirect.com
October 5, 2025 at 7:43 PM
I think it's partially due to funding (is social science famous for ample research funding as in medicine?), partially culture/tradition. A few years ago, Elvik (well known in road safety) asked why so few RCTs in road safety research. Seems an interesting read. www.sciencedirect.com/science/arti...
There are actually lots of RCTs in behavioural research within "medical" research, for example efforts to increase cancer screening uptake (www.cochrane.org/evidence/CD0..., www.cochrane.org/evidence/CD0...). I only knew this b/c I used to work in an institute of "preventive medicine".
Do invitations, lay health worker interventions and educational interventions increase the uptake of cervical screening? | Cochrane
www.cochrane.org
October 5, 2025 at 7:43 PM
There are actually lots of RCTs in behavioural research within "medical" research, for example efforts to increase cancer screening uptake (www.cochrane.org/evidence/CD0..., www.cochrane.org/evidence/CD0...). I only knew this b/c I used to work in an institute of "preventive medicine".
He basically says all methods have limitations, which is of course true. It's like saying gold could have less value than silver depending on how a gold product was made, e.g. whether the manufacturing protocol was well followed.
October 5, 2025 at 2:25 PM
He basically says all methods have limitations, which is of course true. It's like saying gold could have less value than silver depending on how a gold product was made, e.g. whether the manufacturing protocol was well followed.
Their relatives may already been in UK or British citizens.
October 2, 2025 at 4:50 PM
Their relatives may already been in UK or British citizens.
Even that’s true, who said anything about sick relatives’ entry???
October 2, 2025 at 3:31 PM
Even that’s true, who said anything about sick relatives’ entry???
Reposted by Chao Wang
The future is bleak:
AI "researchers" will send questionaires to AI "respondents" to collect data on what they "think" about Topic, poorly summarise the results and generate a paper for a mill.
AI peers will review and leave comments.
Other AI will cite the paper (and nonexistent ones).
Wooooo.
AI "researchers" will send questionaires to AI "respondents" to collect data on what they "think" about Topic, poorly summarise the results and generate a paper for a mill.
AI peers will review and leave comments.
Other AI will cite the paper (and nonexistent ones).
Wooooo.
October 1, 2025 at 4:47 AM
The future is bleak:
AI "researchers" will send questionaires to AI "respondents" to collect data on what they "think" about Topic, poorly summarise the results and generate a paper for a mill.
AI peers will review and leave comments.
Other AI will cite the paper (and nonexistent ones).
Wooooo.
AI "researchers" will send questionaires to AI "respondents" to collect data on what they "think" about Topic, poorly summarise the results and generate a paper for a mill.
AI peers will review and leave comments.
Other AI will cite the paper (and nonexistent ones).
Wooooo.
I think she just means unpaid work (e.g. looking after sick relatives) is also valuable (don't you agree?) and should be considered in terms of ILR application. I doubt anyone would do unpaid work for 10+ years just to get an ILR.
September 30, 2025 at 10:09 AM
I think she just means unpaid work (e.g. looking after sick relatives) is also valuable (don't you agree?) and should be considered in terms of ILR application. I doubt anyone would do unpaid work for 10+ years just to get an ILR.
And sometimes coauthors.
September 28, 2025 at 1:32 PM
And sometimes coauthors.