A Standard Experiment: How to determine present day efficacy
A Standard Experiment: How to determine present day efficacy
Author: Bernd Wechner
Published on: February 1, 2003
Having reviewed the state of the art in hitch-hiking experiments last month, we saw that of eight experimenters each approached the question independently, designing very different experiments along the way. Comparing them is a questionable task with a broad range of conditions, technique and result. All the same, they produced results, the only results we have - the only objective historical record we have of how efficient hitch-hiking is or was.
I’ve often been faced with the suggestion that hitch-hiking is not nearly as efficient now as it was then. It derives I suspect, from the generally sound observation that there are far fewer hitch-hikers now than then. Yet, as I noted last year, there is not a soul on record that has, since the 1970’s set out to measure the truth of such a conclusion - how efficient it actually is. It is by no means sound logic to conclude from the decline in popularity of hitch-hiking, that the efficacy has similarly declined. All the same, perhaps it has, who knows?
Robert Prins kept exquisite notes over 18 years and 1072 rides. He recorded an average wait of just under an hour, but didn’t spend any time counting passing cars, so we’re not easily able to compare his results with those of objective studies. All the same with such a lengthy wait, it would seem he's waiting much longer than your average experimenter from the '70s did. Robert's the first to admit that his wait times are a little inflated because of his inclination to solicit rides directly at services and wait around until the right, comfortable car comes along ... not an objective technique for assessing efficacy so much as a clever technique for getting around comfortably!
Over a couple of year I kept some notes on my experiences, I too recorded my wait time but not a count of passing cars. On some roads I’m sure 100 cars passed my in a few minutes, on others, one car every few hours, so wait times are not easily comparable with the results of the 1970s experiments (which all counted the portion of passing drivers that pulled over). I can say with confidence though that around Europe and Australia after 704 recorded waits my average wait was 34 minutes. It is somewhat distorted by outback Australian waits of several days in length though and neglecting these a more modest 18 minutes emerges.
To cut through this speculation and promote some uniformity, I’d like to lay out the bare necessities of a standard experiment that can conceivably be picked up by college students everywhere, or active thumbers, as a guideline for experimental hitch-hiking. It would aim, unlike my notes of years gone by, or even those of Robert Prins, to produce results that are comparable with those of the 1970s. Experiments conducted today, following this guideline should reveal not only useful information about the here and now, modern motoring attitudes and driver psychology but also conceivably lend some insight into that dogged question “how have things changed over the last few decades?”
With that in mind, and eight role models from the 1970s, I present:
The Suite 101 Standard Hitch-hiking Experiment
An experimenter attempts to solicit a ride from passing traffic.
A trial is a single effort by an experimenter (or experimenters) to solicit a ride from passing motorists, lasting as long as it takes to win the offer of a ride.
The success rate is one divided by the number of cars passing the hitch-hiker before the offer eventuates. Only cars in a position to offer a ride are counted, meaning busses, taxis, police cars, ambulances and fire engines are not counted1.
On multi-lane roads, if traffic is dense, only cars in the outside lane are counted, if traffic is sparse, all lanes are counted2.
The precise conditions of experiment and rules followed in counting are to be clearly documented.
A variable is an experimental condition that is altered between trials in an effort to determine if changes in that condition have a statistically significant effect upon the success rate.
An experiment is a series of one or more trials and may legitimately have any number of variables (none, one or more) or trials. The more variables chosen, the more combinations there are to explore. Aim for 10 or more trials for each combination of variables being explored.
Standard variables and suggested values are:
- Sex: male, female, male+female, male+male, female+female
- Dress: sloppy, casual, formal
- Eye contact: yes, no
- Time: day, night
- Location: as desired3
The following standard observations for each trial are suggested:
- Count of passing vehicles
- Time of wait
- Driver characteristics4: apparent age, apparent sex, apparent ethnicity, age, sex, ethnicity, hitched before?
- Vehicle characteristics5: manufacturer, model, apparent age, type, age
- Number of occupants
Further, the specific appearance of the experimenters and their practices are to be recorded.
Recommended for such an experiment is an independent observer who can count passing traffic while the experimenter hitch-hikes from a discrete position, and a standard tally counter to aid in reliable traffic counting.
I hope one summer soon to run just such an experiment in Tasmania. I would like to encourage others in other places to do the same. It would involve three people ideally, a male and female thumber and a traffic counter, time keeper, coordinator. A perfect college experiment producing a nice paper. Let's see it happen.
1. Occasionally a bus, taxi, police car, ambulance or fire engine may offer a ride (I have hitched rides with all these bar the fire engine myself), and this will be counted. On the whole it is safe to say that they are not in a position to offer rides to hitch-hikers and various experimenters made similar assumptions in their counting.
2. Dense and sparse are subjective decisions based upon the assessment of on-site of traffic’s ability to cross the outside lane(s) to pull over on the curb.
3. But well documented in the experimental report.
4. Of the driver offering the ride, not all passing drivers.
5. Of the vehicle which pulled over, not all passing vehicles.