Roger Watt
@rjwatt42.bsky.social
Emeritus Professor of Psych at
Stirling University
Trumpet player
Statistic-ist
Grandpa
Stirling University
Trumpet player
Statistic-ist
Grandpa
Yes. I imagine readers (and writers) of a popular science books will fail to see that nuance and will understand it to mean at a minimum stronger evidence for the hypothesis in question - which it doesn't.
November 14, 2025 at 5:22 PM
Yes. I imagine readers (and writers) of a popular science books will fail to see that nuance and will understand it to mean at a minimum stronger evidence for the hypothesis in question - which it doesn't.
Currently reading a popular book about nature and well-being. Riddled with statements about "highly significant" results as if that meant "large effect".
In some ways this usage is equally problematical.
In some ways this usage is equally problematical.
November 14, 2025 at 4:54 PM
Currently reading a popular book about nature and well-being. Riddled with statements about "highly significant" results as if that meant "large effect".
In some ways this usage is equally problematical.
In some ways this usage is equally problematical.
"a terrible line manager" puts you in the top 10% of managerial ability amongst academics, you know?
November 12, 2025 at 5:11 PM
"a terrible line manager" puts you in the top 10% of managerial ability amongst academics, you know?
Looking back, I think my own colour/saturation of ND was an enormous benefit to me.
November 12, 2025 at 10:42 AM
Looking back, I think my own colour/saturation of ND was an enormous benefit to me.
In the old days, I think that not going for promotion was sensible - pay and work satisfaction were both high without it. Not so sure today.
November 12, 2025 at 10:42 AM
In the old days, I think that not going for promotion was sensible - pay and work satisfaction were both high without it. Not so sure today.
Oh and also ffs, I just hijacked Katie and Olivia's very interesting conversation. Sincere apologies. I'm still hoping to learn how to be a good sm person...
November 12, 2025 at 10:33 AM
Oh and also ffs, I just hijacked Katie and Olivia's very interesting conversation. Sincere apologies. I'm still hoping to learn how to be a good sm person...
Oh and ffs, I have just been invited (=instructed) to do a cybersecurity online course. I did one 2 weeks ago and I passed but only by pretending that the words risk, likelihood and probability all mean the same thing: "bad".
November 12, 2025 at 10:33 AM
Oh and ffs, I have just been invited (=instructed) to do a cybersecurity online course. I did one 2 weeks ago and I passed but only by pretending that the words risk, likelihood and probability all mean the same thing: "bad".
And because you can't assess research quality, we are in a spiral of increasing costs trying to do so.
We are in effect doing incredibly costly research into what makes good research. But with very low power and theories/hypotheses that are too vague and timescales that are too short.
We are in effect doing incredibly costly research into what makes good research. But with very low power and theories/hypotheses that are too vague and timescales that are too short.
November 12, 2025 at 10:33 AM
And because you can't assess research quality, we are in a spiral of increasing costs trying to do so.
We are in effect doing incredibly costly research into what makes good research. But with very low power and theories/hypotheses that are too vague and timescales that are too short.
We are in effect doing incredibly costly research into what makes good research. But with very low power and theories/hypotheses that are too vague and timescales that are too short.
2. You can't assess research quality. Research just needs to be good enough be persuasive over the long run. Overall research is a tide; each of our own contributions is a wave on the shoreline. It's not possible to ask of each wave, did it make a difference to the tide?
November 12, 2025 at 10:33 AM
2. You can't assess research quality. Research just needs to be good enough be persuasive over the long run. Overall research is a tide; each of our own contributions is a wave on the shoreline. It's not possible to ask of each wave, did it make a difference to the tide?
Add up those hours per student and then take 10% of them and use those more sensibly to assess each student.
November 12, 2025 at 10:33 AM
Add up those hours per student and then take 10% of them and use those more sensibly to assess each student.
1. At Stirling, a student's outcome (degree class) is derived from maybe 100-200 individual assessments. And yet, the degree class is just a 3-way split, with proportions around 20-60-20. So an enormous amount of data to determine a largely non-informative outcome.
November 12, 2025 at 10:33 AM
1. At Stirling, a student's outcome (degree class) is derived from maybe 100-200 individual assessments. And yet, the degree class is just a 3-way split, with proportions around 20-60-20. So an enormous amount of data to determine a largely non-informative outcome.
My own, slowly forming conclusions are that everyone (staff, students, towns) can only now be improved by 2 major changes.
1. Stop continuous assessment.
2. Stop research quality assessment (REF).
1. Stop continuous assessment.
2. Stop research quality assessment (REF).
November 12, 2025 at 10:33 AM
My own, slowly forming conclusions are that everyone (staff, students, towns) can only now be improved by 2 major changes.
1. Stop continuous assessment.
2. Stop research quality assessment (REF).
1. Stop continuous assessment.
2. Stop research quality assessment (REF).
At the minute, I think that Scotland is more like the UK than elsewhere. Plus we have weather...
November 12, 2025 at 10:12 AM
At the minute, I think that Scotland is more like the UK than elsewhere. Plus we have weather...
I think I have PDA, therefore I'm not going to ignore you. 😀
November 12, 2025 at 10:05 AM
I think I have PDA, therefore I'm not going to ignore you. 😀
Mackerras and the SCO for me. A distinctive sound and played with great swagger.
November 11, 2025 at 6:47 PM
Mackerras and the SCO for me. A distinctive sound and played with great swagger.
At the end of weeks of evidence from all sorts, the judge concluded that my explanation was the most likely. On my train home, along the same route, I found that RailTrack had replaced the signal in question with a fibre optic one, eliminating the potential problem in one go. Felt quite vindicated.
November 11, 2025 at 4:30 PM
At the end of weeks of evidence from all sorts, the judge concluded that my explanation was the most likely. On my train home, along the same route, I found that RailTrack had replaced the signal in question with a fibre optic one, eliminating the potential problem in one go. Felt quite vindicated.
Cue giving them an impromptu tutorial on reflections/refractions with curved windscreens because the rail users group had an idea. Then,
lawyer:"Did I think apparent motion might have been involved?".
me: What??
lawyer:"Just wondered whether you knew about apparent motion and had considered it."
lawyer:"Did I think apparent motion might have been involved?".
me: What??
lawyer:"Just wondered whether you knew about apparent motion and had considered it."
November 11, 2025 at 4:30 PM
Cue giving them an impromptu tutorial on reflections/refractions with curved windscreens because the rail users group had an idea. Then,
lawyer:"Did I think apparent motion might have been involved?".
me: What??
lawyer:"Just wondered whether you knew about apparent motion and had considered it."
lawyer:"Did I think apparent motion might have been involved?".
me: What??
lawyer:"Just wondered whether you knew about apparent motion and had considered it."
Giving evidence to the Inquiry itself was very tough. In front of survivors who had been badly burned, I was pushed as far as the lawyers could.
"You aren't going to speak beyond your professorial expertise, are you?"
"No, but I am going to use all that expertise and we can see whether it helps".
"You aren't going to speak beyond your professorial expertise, are you?"
"No, but I am going to use all that expertise and we can see whether it helps".
November 11, 2025 at 4:30 PM
Giving evidence to the Inquiry itself was very tough. In front of survivors who had been badly burned, I was pushed as far as the lawyers could.
"You aren't going to speak beyond your professorial expertise, are you?"
"No, but I am going to use all that expertise and we can see whether it helps".
"You aren't going to speak beyond your professorial expertise, are you?"
"No, but I am going to use all that expertise and we can see whether it helps".
Jumped off the tube at the next stop to get a copy for my mum, but the edition had changed and although the story was still there, the photo had gone. Talk about 10 minutes of glory.
November 11, 2025 at 4:30 PM
Jumped off the tube at the next stop to get a copy for my mum, but the edition had changed and although the story was still there, the photo had gone. Talk about 10 minutes of glory.
To my surprise I did figure it out. On the way back to my hotel after giving my evidence to the Cullen Inquiry, I was sitting on the tube and the person opposite me folded their copy of the Evening Standard - letting me see my mugshot on the page they had just read. "Stirling professor explains..."
November 11, 2025 at 4:30 PM
To my surprise I did figure it out. On the way back to my hotel after giving my evidence to the Cullen Inquiry, I was sitting on the tube and the person opposite me folded their copy of the Evening Standard - letting me see my mugshot on the page they had just read. "Stirling professor explains..."
There were curious moments. I needed to know exactly where the sun was in the sky at the moment the driver would have sighted the signal. To do this, you need to use Julian dates (so seconds elapsed since noon UT on January 1, 4713 BC) a big number and so symbolic maths software to avoid overflows.
November 11, 2025 at 4:30 PM
There were curious moments. I needed to know exactly where the sun was in the sky at the moment the driver would have sighted the signal. To do this, you need to use Julian dates (so seconds elapsed since noon UT on January 1, 4713 BC) a big number and so symbolic maths software to avoid overflows.
A train had passed a red signal, carried on and hit a train coming the other way. I was asked to try and find out why. The driver was dead along with 30 others. So began a year of fascinating work. I said at the time that they needed an explanation for two things: why then and why not before.
November 11, 2025 at 4:30 PM
A train had passed a red signal, carried on and hit a train coming the other way. I was asked to try and find out why. The driver was dead along with 30 others. So began a year of fascinating work. I said at the time that they needed an explanation for two things: why then and why not before.