Estimated car cost as a predictor of driver yielding behaviors for pedestrians | Hacker News Reader

Estimated car cost as a predictor of driver yielding behaviors for pedestrians

sciencedirect.com

23 pointsbookofjoe12 comments

Open articleView on HN

I looked over the study to determine if there was a criteria for "yield" vs "non-yielding".

I think based on the study write-up under section "2.2. Pedestrian crossing", that yielding was based on:

- car in nearest lane is 200 yards away

- pedestrian placed one foot into crosswalk

- pedestrian attempted to make eye contact with driver

- driver reduces speed -> threshold for "safe to attempt"

- driver yields or does not -> fits criteria for "yields"

It isn't clear if they considered a failure by the driver to invigilate the entire crossing as a "failure to yield"; e.g. in my district, if a pedestrian is present in the cross walk at any part of the road, I cannot proceed full-stop, and that would be a "failure to yield" under my road laws act.

After thinking about this study, and attempting to not make myself look like a fool if my original issue was addressed, I noted a few more deficiencies in this study:

- road structure (other than mid-block crossing, and 35MPH; e.g. number of houses, other entry ways, number of business, two lanes or four lanes; traffic calming measures)

- road use (isolated neighborhood collector; retail zone; between two high-congestion areas; etc)

- volume of crossings by non-study participants

- recency of the crossing

- existing use of the crossing (inter-neighborhood? inter-shopping centre? school to park?)

- driver perceived age of pedestrians and pedestrian capability (infirm, underage, active)

This all leads into:

- driver awareness of existing pedestrian crossing

- driver familiarity with pedestrians at that crossing

- driver intention with the road

- brand new and underused crossing? old crossing with very low non-study volume?

- is 4x30 = 120 crossings at road 1 and road 2 high or low for that area?

I don't feel this study is useful in isolation, and I'd be irritated if someone used it as a basis for point-making.

> Of 461 cars, 27.98% yielded to pedestrians.

This is... pretty bad!

How do you estimate the cost of a car that you observe on the street? If someone drives a seven year old car that cost a lot new, how do you know if they bought it new or for a fraction of the price?

>The cost of each car was estimated using various pricing categories from the Kelly Blue Book (KBB). This consisted of using the video data to determine the vehicle make, model, year, and overall condition. The year of the vehicle was estimated by cross referencing each manufacturer's website to compare body styles. To estimate the overall cost of the vehicle, the low end and high end price estimates from private sale were averaged.

They appear to have used their estimation of current value, not their estimation of the original purchase price.

If you knew the cost of the car you wouldn't need to estimate it.

If it's uncorrelated to known characteristics, then arbitrarily choosing a number isn't really "estimating".

Good thing it is correlated to the characteristics in question.

Assuming it was just purchased. Which is my point. The average car was not just purchased, I feel confident in saying.

You don't have a point. A correlation doesn't have to apply for every single instance, not even for the majority.

My original post contained the suspicion that nobody was estimating cost. Although I never clicked the link, someone else ("def8cefe") posted as a response to me, that they indeed did not use cost but current value.

So, despite being downvoted, I think I made a substantive contribution in identifying the headline as misleading. This in turn would affect how a reasonable person interprets the meaning of the whole thing.

What is your point?

Cost correlates with current value.

>Estimated car cost as a predictor of driver yielding behaviors for pedestrians

Why not: > Correlation of vehicle price to the probability its driver yields to a pedestrian.

Still improve-able but “driver yielding behaviors” is too mafan.

Interesting work, But to be nit picky, the title is typical academese. I.e less readable than it needs to be even after taking the higher need of precise and accurate language.

Maybe:

Correlation of a driver yielding to a pedestrian with the vehicle price?

Reply on news.ycombinator.com