Found the experiment really interesting and fun, and I love how well-written the entire piece is! It really got me thinking about a few things-
- I wanted to understand whether the distance was a fixed constant (like 3 ft, as shown in the diagram) or if it was always “roughly half the sidewalk,” as mentioned. I wonder how that affects the probability of people choosing to “pass through” rather than going around, because their “cost” to do so would vary in wider vs. narrower passages. For instance, if we fix the distance at 3 ft in a really wide passage, it might be much easier (lower cost) for someone to just walk around. But in a narrower passage, that same 3 ft might create a bigger obstruction, making it more likely they’ll just squeeze through. On the other hand, if you’re always blocking half the sidewalk, then in a narrow passage, the gap between the two persons might be so small that it feels more natural to go around. It’d be fun to see a robust method of controlling or varying this distance. It might also help boost that 7% rate of people passing through, giving more data points to analyse.
- Another point is how each location only had a constant 10 minutes of observation. I think that led to about 40-50% of the total data coming from the stations, possibly causing a bias by overrepresenting a “rush hour at a station” scenario vs a “lazy afternoon at a church” scenario. I was wondering how if instead of sticking to just time, trying aim for a set number of people at each location, or figure out some other way to balance the groups would differ the experiment. Even so, it’s really remarkable and surprising that the highest number of deviations happened at the church. I keep wondering if that was just an outlier or if its a general result as it’s non intuitive, which makes it even more fun.
- It would also be very interesting in seeing some sort of visualization showing a cross-sectional breakdown of ethnicities and age groups across all the different locations. Especially with the youth at the church deviating more. I was curious if these were mainly kids, or more like young adults?
- I also find the “sin theory” super interesting and honestly pretty realistic. I too feel that psychological factors can play such a big role given it is such an experiment. It made me think about another such factor- whether the consciousness of being perceived or judged in a structured, crowded, or familiar environment might push people to follow group norms more than in a setting where they don’t feel that same pressure. It’s just a hypothesis, but it would be cool to see if there’s a way to test this more formally or statistically in experiments.
- Lastly, I also think it would be pretty fun to run a logistic regression on all this data. You could treat features like gender, age category, ethnicity (I wonder if this is “AI ethical” xD), and location (maybe encoded in some clever way) as features, and then model the probability of “passing through” (1) vs. “passing around” (0). Would be fun to see its results.
Overall, I really enjoyed reading this, and hope you decide to expand on this one day! :)
Akhilesh, this is my favorite comment and might I just say it's a post in itself -- full of curiosity and thoughtfulness. I am sorry for not responding sooner!
So the distance was not fixed, but it wasn't random either. We kept it 'flex' as to almost always occupy half the sidewalk. I was pushing my colleagues for widening the distance between them to get more positive hits so to say, but that also has a negative flip side: it might lead to more false positives -- i.e. more people going between not because its rude but because its convenient. Tight rope to walk on here
Yeah the 10 minutes at each location was because one of us was recording discreetly and did not want to attract too much attention. Churches was so surprising, but to be fair, it was also the only closed door space in our sample. And yes, they were mainly young adults.
Yeah so many cultural norms and everything seep into the most inane decision making, but I wanted to fold all of those intricacies in the data and let that speak in the data. I guess I wasn't looking for robustness but just plain old patter, a refresher from all that stats courses.
You are absolutely right; we did not run any statistical method on this, it was just purely descriptive analysis. I'll try to do a follow-up post with more robust analysis, I am sure it could display some patterns that we don't realize here.
Thank you so much for your reading and deep thoughts on this :) My vague plan for expansion in this would be in India, and try and see if it can explain why some cities are more aggressive that others for ex: Mumbai vs Delhi (similar experiment in local metros)
This was such an interesting read! I loved the central idea of this experiment which even though seems trivial, says interesting things about human behavior. Great job on the blog post!
Thanks so much, Rutvi, yeah completely agree! We can all just tap into public places as research spaces and find all kinds of interesting things about behaviour if we are curious (and patient) enough to see them.
The data is interesting, the storytelling is captivating, Akshar! One can tell how much rigour has gone into this from how easy it is to read and feel.
Thanks for sharing this - the design and details of the social experiment as well as the dual, intertwined analysis of what you learnt from the results and what you learned for yourself during this process.
Thank you for this! I really enjoyed bringing out the personal here amongst all the data work — a lesson from you that I always try to embrace in my writing.
Really good job! Enjoyed reading the learnings you took away from our fieldwork project. I especially enjoyed how you were able to connect the importance of the data analysis process to the lessons learned from David Foster Wallace's timeless commencement speech.
Found the experiment really interesting and fun, and I love how well-written the entire piece is! It really got me thinking about a few things-
- I wanted to understand whether the distance was a fixed constant (like 3 ft, as shown in the diagram) or if it was always “roughly half the sidewalk,” as mentioned. I wonder how that affects the probability of people choosing to “pass through” rather than going around, because their “cost” to do so would vary in wider vs. narrower passages. For instance, if we fix the distance at 3 ft in a really wide passage, it might be much easier (lower cost) for someone to just walk around. But in a narrower passage, that same 3 ft might create a bigger obstruction, making it more likely they’ll just squeeze through. On the other hand, if you’re always blocking half the sidewalk, then in a narrow passage, the gap between the two persons might be so small that it feels more natural to go around. It’d be fun to see a robust method of controlling or varying this distance. It might also help boost that 7% rate of people passing through, giving more data points to analyse.
- Another point is how each location only had a constant 10 minutes of observation. I think that led to about 40-50% of the total data coming from the stations, possibly causing a bias by overrepresenting a “rush hour at a station” scenario vs a “lazy afternoon at a church” scenario. I was wondering how if instead of sticking to just time, trying aim for a set number of people at each location, or figure out some other way to balance the groups would differ the experiment. Even so, it’s really remarkable and surprising that the highest number of deviations happened at the church. I keep wondering if that was just an outlier or if its a general result as it’s non intuitive, which makes it even more fun.
- It would also be very interesting in seeing some sort of visualization showing a cross-sectional breakdown of ethnicities and age groups across all the different locations. Especially with the youth at the church deviating more. I was curious if these were mainly kids, or more like young adults?
- I also find the “sin theory” super interesting and honestly pretty realistic. I too feel that psychological factors can play such a big role given it is such an experiment. It made me think about another such factor- whether the consciousness of being perceived or judged in a structured, crowded, or familiar environment might push people to follow group norms more than in a setting where they don’t feel that same pressure. It’s just a hypothesis, but it would be cool to see if there’s a way to test this more formally or statistically in experiments.
- Lastly, I also think it would be pretty fun to run a logistic regression on all this data. You could treat features like gender, age category, ethnicity (I wonder if this is “AI ethical” xD), and location (maybe encoded in some clever way) as features, and then model the probability of “passing through” (1) vs. “passing around” (0). Would be fun to see its results.
Overall, I really enjoyed reading this, and hope you decide to expand on this one day! :)
Akhilesh, this is my favorite comment and might I just say it's a post in itself -- full of curiosity and thoughtfulness. I am sorry for not responding sooner!
So the distance was not fixed, but it wasn't random either. We kept it 'flex' as to almost always occupy half the sidewalk. I was pushing my colleagues for widening the distance between them to get more positive hits so to say, but that also has a negative flip side: it might lead to more false positives -- i.e. more people going between not because its rude but because its convenient. Tight rope to walk on here
Yeah the 10 minutes at each location was because one of us was recording discreetly and did not want to attract too much attention. Churches was so surprising, but to be fair, it was also the only closed door space in our sample. And yes, they were mainly young adults.
Yeah so many cultural norms and everything seep into the most inane decision making, but I wanted to fold all of those intricacies in the data and let that speak in the data. I guess I wasn't looking for robustness but just plain old patter, a refresher from all that stats courses.
You are absolutely right; we did not run any statistical method on this, it was just purely descriptive analysis. I'll try to do a follow-up post with more robust analysis, I am sure it could display some patterns that we don't realize here.
Thank you so much for your reading and deep thoughts on this :) My vague plan for expansion in this would be in India, and try and see if it can explain why some cities are more aggressive that others for ex: Mumbai vs Delhi (similar experiment in local metros)
This was such an interesting read! I loved the central idea of this experiment which even though seems trivial, says interesting things about human behavior. Great job on the blog post!
Thanks so much, Rutvi, yeah completely agree! We can all just tap into public places as research spaces and find all kinds of interesting things about behaviour if we are curious (and patient) enough to see them.
The data is interesting, the storytelling is captivating, Akshar! One can tell how much rigour has gone into this from how easy it is to read and feel.
Thanks for sharing this - the design and details of the social experiment as well as the dual, intertwined analysis of what you learnt from the results and what you learned for yourself during this process.
Thank you for this! I really enjoyed bringing out the personal here amongst all the data work — a lesson from you that I always try to embrace in my writing.
Also, am I your 300th subscriber? :)
I’ve always wanted to be that subscriber/follower who rounds it off to a milestone number xD
Really good job! Enjoyed reading the learnings you took away from our fieldwork project. I especially enjoyed how you were able to connect the importance of the data analysis process to the lessons learned from David Foster Wallace's timeless commencement speech.
Thanks Rhodwin, I really enjoyed working with you on this. Your work ethic and attention to detail inspires me to write better.