Forum Topic

It is an interesting study but with flaws.The main flaw is that the data needs to be re-analysed based on relative risk taking into account changes in number of journeys.  In fairness the authors acknowledge that flaw.  "The above analyses do not consider changes in travel patterns, and therefore do not estimate risk to an individual, e.g. risk per trip"This is important because the data could be under or over-estimating the effect of the LTNs.  Take for example the data for cyclist injuries: the data shows a negligible change in injuries pre and post the LTN both for within the LTN boundaries and just outside.  You could look at the data and state that the LTN made no difference to cyclist safety, but if the LTN encouraged more cycling then the relatively small change in absolute numbers could mask a much bigger change in relative risk. (in the interests of completeness I consider comparing the data to the rest of London to be a inadequate method of assessment because the whole point of LTNs is to change the volume and mode of traffic compared to the rest of London).  Again this is something which the authors of the study fully accept - "It is therefore likely that the effects shown in Table 1 [the table with the data] somewhat underestimate improvements in risk per trip for walking and cycling inside an LTN but overestimate improvement for driving, which may show a more modest decrease in risk per trip."The final significant flaw, again fully acknowledged by the authors of the study, is that the volume of the data is relatively small which leads to limited statistical power.  I have no issue with that, all it means is that we need more data over a longer period (12 months is the absolute minimum, I think many people more expert than I would say 3-5 years).  As the report explicitly states "Future follow-up assessment of injury impacts is warranted, once schemes have had more time to ‘bed in’, and once the impacts of the Covid-19 pandemic on travel behaviours subside" and I completely agree with that statement

Justin Stephenson ● 1921d

Whilst I agree that the data shows that LTNs significantly reduce all injuries in absolute numbers within the LTNs that is cherry picking one conclusion and ignoring other possible conclusions from the data.So I will also cherry pick the data.  According to the data there is a 77% that LTNs made no difference to cyclist injuries inside the LTN and a 48% chance they made no difference to cyclist injuries on the boundaries of LTNs (a bit of a surprise). When you look at injuries to cyclists  (not just death or serious injury) it is factually the case that the volume of data is so small that there is limited statistical power. But as you referred to all injuries by any mode the data shows that there is a 35% chance that LTNs made no difference to injuries at the immediate boundaries of the LTNs - that chance is 7x the normally accepted threshold for statistical significance.  For me that is a significant surprise.However, it should also be noted that the data only covers a period relatively shortly after the introduction of the LTNs.  As research going back decades shows, this can be highly misleading as over time traffic adapts to road management schemes in different ways.  That is why until we have much longer term data any conclusions can only be labelled tentative.IMO the research is an interesting starting point, but only a starting point.  We definitely need longer term data and I would also like to see the data being analysed against traffic volumes. We know that LTNs are intended to reduce traffic volumes and previous analysis suggests that this true not just within the LTNs but also on the immediate boundaries (and some level of traffic evaporation but the amount is highly variable). So I am very surprised about the data for injuries at the LTN boundaries.  The lack of any statistical significant change in absolute numbers of injuries at the boundaries may mask a significant change in relative risk or it may just be a quirk of the short term data.

Justin Stephenson ● 1920d

> Paul, I think some people have had enough of experts. These Culture War threads are always amusing, especially when the expert question rears its fuzzy head. What people forget is that there’s two types of expertise, that or open and of closed systems. Of which does this topic fall into?“ However, it is worth drawing a distinction between these two types of expertise—the kind people question, and the kind people don’t. In short, people value expertise in closed systems, but are distrustful of expertise in open systems. A typical example of a closed system would be a car engine or a knee joint. These are semi-complex systems with ‘walls’—that is to say, they are self-contained and are relatively incubated from the chaos of the outside world. As such, human beings are generally capable of wrapping their heads around the possible variables within them, and can therefore control them to a largely predictable degree. Engineers, surgeons, pilots, all these kinds of ‘trusted’ experts operate in closed systems.Open systems, on the other hand, are those that are ‘exposed to the elements,’ so to speak. They have no walls and are therefore essentially chaotic, with far more variables than any person could ever hope to grasp. The economy is an open system. So is climate. So are politics. No matter how much you know about these things, there is not only always more to know, but there is also an utterly unpredictable slide towards chaos as these things interact.”https://quillette.com/2018/06/04/the-limits-of-expertise/

Russell Pearson ● 1920d

The last few tradesmen who visited our place all came via non-car means.The plumber had a scooter for instance which he parked in the hallway.Given that it took 3 separate visits and a trip to B&Q to get parts, I suspect he was able to save a fair amount of time over using a car or van.  There is this assumption that all tradesmen require vans or cars in order to do their jobs and that is not correct.  I don't live in an LTN myself, but that doesn't mean we don't get traffic or that there's always a parking space outside.  While Brian Sargent, 43, the plumber from East London quoted in the article is hiking his prices from £45 to £70 to offset his travelling time, his competitors aren't because they're using public transport, bikes and scooters to visit customers and saving time.Of course, the blame on LTNs causing congestion is debateable in itself - they lower traffic inside the LTN and on the boundaries and higher congestion on main roads is also attributable to the collapse in public transport use during the pandemic - still at around 50%. Brian again from the article:"To most people who live in areas with LTNs they seem to just be taking traffic away from the nice, posher areas and making everything longer for us working people." That's incorrect - in Hounslow for instance it is roughly equitable.https://osf.io/preprints/socarxiv/q87fu/If your tradesman is hiking his or her prices by 55% and saying LTNs are the cause then consider trying another tradesman.

Ed Saper ● 1918d

The article picks on some interesting statistics, but then fails to properly understand the implications. And the author's approach to data from the Netherlands has some glaring inconsistencies.The drop in car driver and passenger injuries between the two periods in London outside the LTNs, some 25%, is indeed very likely to be due to the reduced number of motor vehicles as the article states, but given that most of these injuries are caused by collisions with other motor vehicles, the reduction in vehicles outside LTNs would be 11%, not 25%. The 11% increase in cyclists injures can similarly be put down to an increased number of cyclists. However, almost all injuries to cyclists in London come from collisions with motor vehicles, not from each other as the author wrongly claims, so the increase in cycling has to be larger than that proportion to account for the lower number of vehicles, assuming the risks otherwise are unchanged. This implies an increase in cyclists of 25%, assuming no change otherwise to the risks. Similarly, pedestrians numbers would have reduced by 22% over the same period, which could be explained by lockdown, fewer shopping on the street and much less commuting.The numbers are rather small to do the same analysis inside the LTNs, and the assumption that otherwise the risks are unchanged is certainly less justified. The author makes a glaring error in using data from the Netherlands to imply that cyclist in London are mostly colliding into each other. Given the extensive and heavily used cycling infrastructure in the Netherlands, it is no surprise that cyclists hit other cyclists more often than in London, but the number of cyclists killed in collisions with other cyclists in the Netherlands is still less than 3% of the total.In fact he's even contradicting himself. In May he wrote "It turns out Dutch cyclists aren’t dying on their famous segregated cycling paths where cars are prevented from mixing with cyclists. To the contrary,  it’s on the roads where cars and bicycles are NOT segregated that they keep getting killed." He was actually right then: collisions with motor vehicles are 30 times more likely than with another bike to lead to a cyclist death. It's not that cyclists only hear what they want to hear, but rather the author makes a series of false and inconsistent assumptions to attempt to make a spurious case that is then lapped up by those who want their prejudices confirmed.

Tom Pike ● 1918d

p values are a very poor way of assessing whether the data is significant where there are multiple alternative explanations for data changes. Effectively the researchers are saying this data either shows that LTN reduce injuries or the results are by chance - that is overly simplistic.  Any decent statistician can provide a better explanation of the problems of p values than I can.Lets look at the definitions from the research because this highlights the problemsInjuries within a LTN in the data are in fact injuries that are both within the LTN AND more than 25m from the boundary road.  So they were not measuring injuries within the LTN merely those that were an arbitrary distance inside the LTNInjuries on a boundary are injuries within 25m of the actual LTN (so effectively a 50m wide strip, 25m inside the LTN and 25m immediately outside the LTN) and so inevitably will include some injuries that actually occurred with the LTN.Without knowing how many injuries happened on a boundary but that were also within the LTN, you cannot say that LTNs halved the injuries.What you could say is that the p- value, flawed as it is, does indicate that if you only stay more than 25m (effectively a full street) inside the actual LTN boundary, then the LTN has a significant effect on reducing injuries by all modes of transport  - up to a 50% reduction - but not for cyclists (the LTN makes no difference for cyclists which is a surprise) AND that if you live inside the LTN but close to the boundary the chances are the LTN makes no difference to the injury rates

Justin Stephenson ● 1917d