Thames Valley Speed Cameras

An independent report        




9.12 Acknowledgements, references and supplementary information (back to main report)




The author would like to thank everyone at TVSRP (Thames Valley Safer Roads Partnership) for checking, correcting and verifying the STATS19 data and for creating and supplying the database of collisions at all speed camera sites in Thames Valley. Errors in raw STATS19 data may not be confined just to Thames Valley therefore it is possible that the database TVSRP produced is now the most accurate database of collisions at speed camera sites that exists in Britain (reference 4 below).


The author would also like to thank everyone at TVSRP for supplying the supplementary information that was requested.




1: "Department for Transport: A cost recovery system for speed and red-light cameras ~ two year pilot evaluation" 11 February 2003. Adrian Gains and Richard Humble, PA Consulting Group, Professor Benjamin Heydecker and Dr Sandy Robertson, University College London. (report)


1a: p55/70 "This made Thames Valley a large pilot area that accounted for about one half of all the PIA data." (PIA is "Personnal Injury Accident", same as "all collisions")


1b: Large reductions in collisions of all severities were found to have occurred at speed camera sites:

p44/70 "there was a statistically significant reduction in PIAs at camera sites of 6% relative to the long-term trend"

p42/70 "there was a large and statistically significant reduction of 35% in killed and serious injuries at camera sites in six of the pilot areas. This equates to about 280 fewer people killed or seriously injured."


The report claims these results were:

p44/70 "the effectiveness of cameras in reducing the number of personal injury accidents occurring at camera sites. The approach undertaken was the same as that used to measure changes in KSI casualties."

The results, though, were not the effectiveness of speed cameras because the reductions were also influenced by the effects of RTM, co-intervention and diversion of traffic.


Note: Table 18 on p44/70 shows collisions increased (+13%,+13% and +14%) at speed camera sites in Thames Valley but the comparisons were made of the periods before and after the pilot study started, not necessarily before and after the speed cameras were installed. Because the majority of the fixed speed cameras in Thames Valley were installed long before the pilot started, the "before" period in table 18 for many sites may have been after fixed speed cameras were installed. The authors may have done this not to investigate the effect of installing speed cameras, but to investigate the effect of the increased enforcement expected as a result of the cost recovery system.


2: "The national safety camera programme Four-year evaluation report" Dec 2005: UCL, PA Consulting, Adrian Gains, Michael Nordstrom, Benjamin Heydecker, John Shrewsbury, Linda Mountain, Mike Maher. (Report)


RTM was estimated to be larger than all other factors combined at speed camera sites for KSI collisions:

p157/164 "RTM accounts for about three fifths of the observed reduction in FSCs with the effects of the cameras and trend each accounting for a fifth." (FSC is "Fatal or Serious Collision", same as "KSI collision")


3: As the accuracy of RTM analysis has improved, its effect has approached and exceeded the total collision reduction that has occurred at speed camera sites:


3a: RTM was 63% of KSI collision reduction (estimated using the EB [Empirical Bayes] method)

3b: RTM was 83% of casualty reduction (estimated using the full Bayes method, more accurate than Empirical Bayes)

3c: RTM was 99% of collision reduction (RTM excluded from results by partial FTP method, more accurate than estimates)

3d: RTM was >100% of KSI casualty reduction (measured using FTP method, greatest accuracy)


3a: "The national safety camera programme Four-year evaluation report" (see reference 2 above)

RTM and trend together were estimated to be responsible for 81% of the KSI collision reduction therefore the remaining 19% should be due to the speed cameras, co-intervention and diversion of traffic. The EB method, though, had been configured to produce an underestimate of the RTM effect, see 3.5 in government reports.


3b: "Linking road casualty and clinical data to assess the effectiveness of mobile safety enforcement cameras: a before and after study." 19 November 2012,  Neil Thorpe, Lee Fawcett

Final results: 2.6% reduction in casualties (-8/306) due to the speed cameras and diversion of traffic. (report)


3c: "Does reducing traffic speed using speed cameras reduce the number of collisions?" Feb 2012 Dave Finney.

Final results: 0.2% reduction in the number of collisions due to the speed cameras, co-intervention, diversion of traffic and some residual RTM. (report)


3d: "The effects of mobile speed cameras on road safety." 20/Nov/2012 Dave Finney.

Final results: a 19% increase in KSI collisions and an 11% increase in all collisions due to the speed cameras and diversion of traffic. (report)


4: During the first quarter of 2009, TVSRP undertook a "very extensive data cleansing exercise" on their collision records to produce a verified database of collisions at all speed camera sites in Thames Valley. (report)


5: The TVSRP database contains the number and severity of collisions and casualties within each speed camera site within each month from April 1993 to March 2009 inclusive, a total of 228 months (19 years). The full database contains a total of 517 speed camera sites. Of these 235 are fixed speed camera sites but 23 of these have been decommissioned. This report evaluates the effect of all of the 212 active fixed speed camera sites. The first of these was installed on 02/07/1993 (No603 in site TV4501) therefore the database contains 3.25 years (3 years and 3 months) of data prior to the installation of this speed camera. The most recent fixed speed camera was installed on 27/01/2003 (in site TV8016) and the database contains 6 years of data after the installation of this speed camera. Therefore every fixed speed camera site has at least 3.25 years of data before and 6 years of data after they were installed.


For the purposes of this report, the month the speed camera was installed is taken to be the first in the "after" period. TVSRP use the same method in their reports therefore collision numbers in this and TVSRP reports should match. Only data before Jan 2009 is used in this report because it was considered that the very latest police collision reports may not have been submitted or included during the final 3 months.


A sample extract from the database can be downloaded here: rbwm_raw_data.xls (contains data for all fixed speed camera sites in RBW&M).


6: At the end of this page on the TVSRP website, "There was a change in the guidelines for classifying serious collisions on 1st January 1999. This resulted in an increase in the number of collisions reported as serious".


7: Data and analysis to identify the SSP at fixed speed camera sites: Fixed speed camera SSP identification.xls


8: The 74 most recent fixed speed camera sites were analysed using the full FTP method because these sites had the greatest number of siteyears of data during the PreSSP and ASBiC periods, see "siteyears" page of spreadsheet in reference 7 above.


9: Results (including data and calculations) are in this spreadsheet: Fixed speed camera results.xls


10: The British government's largest report on speed cameras is "The national safety camera programme Four-year evaluation report" (reference 2 above).


10a: There was a 42% reduction in KSI at speed camera sites across Britain p4/164: "after allowing for the long-term trend, but without allowing for selection effects (such as regression-to-mean) ... 42% fewer people were killed or seriously injured."


10b: RTM was estimated to have caused a 34% reduction in KSI. Table H9 on p158/164 shows the observed annual FSCs/site for urban sites before speed cameras were deployed was 1.05, and the RTM effect was estimated to be -0.36. This is therefore a 34% reduction due to RTM (-0.36/1.05).

11: There are only two speed camera reports in which the final results do not include any RTM effects. The first was "The effects of mobile speed cameras on road safety" (reference 3d above), and the second is this report.


The FTP method is the only method that can fully exclude RTM effects from the results but, in order to achieve this, the SSP must be identified. The only reports to date in which the SSP has been identified are this and reference 3d.


Both reports found increases in fatal and KSI collisions at speed camera sites. The first report found there was a 25% increase in fatal collisions and a 19% increase in KSI collisions after mobile speed cameras were deployed. This report finds there was a 38% increase in fatal collisions and a 16% increase in KSI collisions after fixed speed cameras were installed.


12: The FTP method has been used in three reports so far:


Report 1: "The effects of mobile speed cameras on road safety" (reference 3d above)

Report 2: "Guidance on Use of Speed Camera Transparency Data" Richard Allsop University College London May 2013

Report 3 "The effect of fixed speed cameras on road safety" (this report)


Reports 1 and 3 identified the SSPs for their included groups of sites but report 2 assumed that the SSP might have been a duration of 3 years. The analysis in report 2 included fixed speed camera sites in the following nine areas:


• Cambridgeshire and Peterborough (47)

• Leicester, Leicestershire and Rutland (15)

• Lincolnshire (50)

• Merseyside (33)

• South Yorkshire (56)

• Staffordshire and Stoke on Trent (68)

• Sussex (55)

• Thames Valley (203)

• Warwickshire (24)


Report 2 included 203 fixed speed camera sites in Thames Valley. The Thames Valley sites were over a third of the total number of fixed speed camera sites in report 2, and around three times the number within the next highest area.


Report 3 (this one) has identified the SSP at fixed speed camera sites in Thames Valley to have been a duration of 5.5 years therefore the actual SSP for the group of sites evaluated in report 2 was significantly longer than the 3 years that had been assumed. Report 2 may therefore have significant levels of RTM in the results.


Supplementary information:


Why was there such a difference in the duration of SSP for fixed and mobile speed camera sites in Thames Valley?


In Thames Valley, the SSP duration for the group of fixed speed camera sites was 5.5 years, and for the group of mobile speed camera sites was 2.5 years. This marked difference was not due to the type of speed camera deployed, it was due to when the sites were selected. Most of the fixed speed camera sites were selected by the various Local Authorities before 1999, at a time when it was common practice to use 5 year SSPs. After 1999, national guidelines were drawn up and these encouraged the use of much shorter SSPs. In 2000, speed camera enforcement was taken over by the newly formed partnership (TVSRP) and they selected most of the mobile speed camera sites in Thames Valley.


In summary, the fixed speed camera sites were selected by Local Authorities who traditionally used 5 year SSPs, and the mobile speed camera sites were selected by the partnership in compliance with the new national guidelines. Hence the very different SSP durations for fixed and mobile speed camera sites in Thames Valley.


Why might the number of serious collisions increase following the deployment of speed cameras?


Speed cameras, like most safety devices, can have positive benefits and negative side-effects. The initial problem is that the positive benefits may be very small. Collision investigations find that fewer than 8% of KSI collisions involve a vehicle that was speeding and all the other KSI collisions (over 92%) therefore occurred when no-one was speeding.


Further analysis of collision investigations suggest that fewer than 0.5% of motorists will cause a KSI collision while speeding in their lifetime and, if the behavior of this 0.5% can be changed, collisions might be prevented. To achieve this benefit, though, speed cameras would have to successfully target just the 0.5%.


The other 99.5% of motorists, though, would never have caused a KSI collision while speeding and this is where a negative side effect may exist. Speed cameras may change the behaviour of many, if not all, of the 99.5% therefore, if motorists who would never have caused a KSI collision while speeding then change their behaviour as a result of the speed cameras, this might explain the increase in KSI collisions at sites after speed cameras were installed.


The proportion of motorists that might cause a KSI collision while speeding in their lifetime is estimated using 2010 figures on spreadsheet motorists causing KSI in lifetime.xls.


Summary of spreadsheet: In 2010, there were around 33,600 KSI collisions and Police found that 7.1% of those that were investigated involved a vehicle exceeding a speed limit (speeding) (table 1.2). This suggests that around 2,400 KSI collisions involved speeding in 2010 and therefore nearly 140,000 KSI collisions might involve speeding during the 58 years a person might be a motorist. As there are around 30 million motorists, these collisions could be caused by up to 0.5% of them. Therefore, the other 99.5% of motorists would not cause a KSI collision while speeding in their lifetime.


Potential positive benefits and negative side-effects of speed cameras are examined here: effects of cameras.