SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey

Page created by Nicole Griffith
 
CONTINUE READING
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
SWOV

  Monitoring the drink-driving trend
  Tips and tricks on how to design a roadside survey

                  Dr. Sjoerd Houwing
         SWOV Institute for Road Safety Research
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Background

•   Drink driving is an important factor in road
    safety crashes

•   20%-28% of all road fatalities in Europe are
    alcohol related

•   Drink drivers use their seat belts less and are
    more often speeding than sober drivers

                    IBSR, April 23rd 2015
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Background

Combating drink driving is one of the pillars
of a succesful road safety policy!

                 IBSR, April 23rd 2015
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Who is this drink driving offender?

No typical drink driver exists, but relatively often:

•   Male (18-35),

•   Unemployed or labour worker,

•   Single or divorced/separated,

•   Prior driving offences or other criminal records.

                    IBSR, April 23rd 2015
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Why do we monitor drink-driving?

•    To get information on the prevalence of drink driving

•    To monitor the development of specific driver groups

•    To evaluate the effectiveness of measures
        – before vs after
        – study group vs reference group
        – study area vs reference area

                     IBSR, April 23rd 2015
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Best indicators in theory

         •   Direct crash-based indicators

         •   Observed safety performance indicators

         •   Self reported safety performance indicators

         •   Indirect crash-based surrogate indicators

         •   Non-crash-based surrogate indicators

      IBSR, April 23rd 2015
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Best indicators in theory

         •   Direct crash-based indicators

         •   Observed safety performance indicators

         •   Self reported safety performance indicators

         •   Indirect crash-based surrogate indicators

         •   Non-crash-based surrogate indicators

      IBSR, April 23rd 2015
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Roadside surveys ‘gold standard’

•   Test all active road users
•   For all psychoactive substances
•   During all times of the day and all days of the week
•   Collect as much additional information as possible.

                            STOP, have you
                            been drinking?

                  IBSR, April 23rd 2015
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Roadside surveys in practice

           IBSR, April 23rd 2015
SWOV Monitoring the drink-driving trend - Tips and tricks on how to design a roadside survey
Some examples of limitations
Practical limitations
•  Restrictions in budget, time, and manpower
•  Limited cooperation police

Legal limitations
•  No random breath testing allowed
•  No interference of standard police procedure

                 IBSR, April 23rd 2015
Step 1: check legal issues and willingness
of the police to cooperate
 •   Study design should be according to the legislation
     and ethical approved

 •   Participation police is very important

 •   Alternatively, breath testing on a voluntary base
     (e.g. at parking lots and gasoline stations).
            -   Participation rates
            -   Safe working conditions
            -   Ethical issue: How to respond to drivers
                with BACs over the legal limit?
                   IBSR, April 23rd 2015
Step 2: Conduct a power study

Conduct a power study to determine how many
samples you will be needing to answer your research
questions.

You will probably need information from previous
studies or expert opinions to fill in the expected
values that are required for the power analyses.

                 IBSR, April 23rd 2015
Step 3: Design your study in full
detail
Selection bias is most important issue in roadside
   surveys on drink-driving

•   Representative selection of research sites
•   Random testing
•   Minimize non-response, especially selective non-
    response
•   Be aware of possible confounding (stratified
    sampling, adjustment in analysis)

                  IBSR, April 23rd 2015
Taking a representative sample
Design a grid of locations that is representative for
traffic in these time periods (e.g. main roads in urban
and rural areas)

Locations should have save working conditions and
limited possibilities for drivers to avoid the
police
                  IBSR, April 23rd 2015
Taking a representative sample
Avoid survey sessions at locations or days that may
not be representative for normal traffic conditions
during the study period (e.g. festival or holidays)

Avoid selective testing based on type of car or type of
driver (e.g. use random sampling techniques)

                  IBSR, April 23rd 2015
Selection of additional information
 •   Gender and age
 •   Type of vehicle                      Either all drivers that are
 •   Passengers                           tested or those drivers
 •   Nationality                          who have used alcohol
 •   Novice driver

 •   Location of drinking
 •   Recidivism                           Drink driving offenders

                  IBSR, April 23rd 2015
Step 4: Conduct a pilot study

Conduct a pilot study to check if all works out in
practice like you have planned.

                  IBSR, April 23rd 2015
Step 5: Start your study

Adjust the study protocol if necessary or, if all went
well, start directly with the main study.

                  IBSR, April 23rd 2015
Case studies: Canada
Roadside survey to measure effect of new measures implemented in
September 2010

Design: Before-after study, 21:00-03:00 in Wednesday through Saturday
nights

Target N= 2,500 per year, based on a power study (95% C.I. of ± 1.1%).

Police officer directs next driver to research location, but researcher tests
for psychoactive substances (alcohol and drugs).

Incentive for drug test: 10 dollar coupon for gasoline, taxi or designated
driver service was provided for drivers over the legal limit

Response rate 90% alcohol breath tests and 70% drug tests in oral fluid

                            IBSR, April 23rd 2015
Case studies: Netherlands
national study
Roadside survey to measure the development of drink driving
and the effect of new measures (1970-present; annual or each
2 years)

Design: Friday and Saturday nights, 22:00-04:00.

Target N= 25,000 per year, 2 or 3 sessions per region
(sufficient number for each region, no power study).

Data drivers recorded by researchers for all drivers (age,
gender, years of possession driver license, location of drinking,
passengers).

                       IBSR, April 23rd 2015
Case studies: Netherlands regional
study
Measure the development of drink driving on a regional level
(Province of Zeeland since 2003, other regions may follow)

Design: 21:00-04:00 in Saturday nights

Target N= 5,000 per year, to collect sufficient data (power study).

Survey sessions spread over year: One session per subregion (3)
in each quarter of the year.

Police officer collects additional information from offenders only
(age, gender, passengers, novice driver, location of drinking,
recidivism).
                       IBSR, April 23rd 2015
Case studies: Belgian national
study
Roadside survey to measure the development of drink driving and
the effect of new measures (2003-present; each 2 or 3 years)

Design: 4 time periodes: week and weekend days 06.00-22.00,
week and weekend nights, 22.00-0200.

Target N= 10,000 -12,000 per year, 2 or 3 sessions per zone
(sufficient number for each time period, power study).

Logistic regression used to e.g. estimate influence of variables on
alcohol use. Traffic counts to determine share of drink driving.

Data car drivers recorded on (age, gender, driver license,
location of drinking, passengers (+age and gender).
                       IBSR, April 23rd 2015
Case studies: DRUID-study
Roadside survey to measure the share of drink
driving per country

Design: 13 EU countries, 23 psychoactive substances, all days
of the week and times of the day

Target N= 1,000-5,000 per country, based on a power study, in
total 50,000 drivers.

Procedure of stopping and testing varied between countries
due to legal and medical ethical limitations

Incentives varied per country, Large variation in response rate,
some countries suspected of selective testing
                      IBSR, April 23rd 2015
Conclusions (1)

 •   Participation of police and legal limitations
     very important factors when designing your
     study

 •   Find a balance between practical and
     theoretical optimal design, use weight factors
     and power studies for assistance

 •   Conduct a pilot study

                   IBSR, April 23rd 2015
Conclusions (2)

•   Don’t reinvent the wheel: use the experience from
    previous studies (both national and international)

                   IBSR, April 23rd 2015
Thank you for your attention!!

         IBSR, April 23rd 2015
You can also read