MOOSE JAW, Sask. — A University of Saskatchewan agricultural economist is examining Canadian Transportation Agency rail decisions to find out if they are unbiased.
James Nolan told the recent Farming For Profit conference that it should be possible to see random decisions in the data.
For example, he said decisions shouldn’t be a balanced yes, no, yes, no pattern, but would have runs of a few ‘no’ decisions, followed by a few ‘yes’ decisions. He likened it to watching roulette in a Las Vegas casino.
Read Also
Alberta eases water access for riparian restoration
Alberta government removes requirement for temporary diversion licence to water plants up to 100 cubic metres per day for smaller riparian restoration projects
“Sometimes there are runs in the numbers,” he said.
“We’re going to look at these runs (in data) to see whether or not the CTA are doing their job in a random fashion.”
He lacks enough information yet to be sure, but he said bias by decision-makers is not new and somewhat logical.
The theory is that regulatory agencies become biased because the people working for them come from industries they are regulating.
He said statistical evidence would prove or disprove the idea.
The CTA lists all its decision on its website but doesn’t maintain a database, which Nolan said indicates they don’t want to be analyzed or overseen.
Alexandre Robertson, media relations officer for the CTA said in an email that CTA members are required to follow a code of conduct, which spells out that they must comply “with the rules of natural justice and procedural fairness, approach each case with an open mind and are impartial and remain objective at all times, ensuring that proceedings are conducted in a manner that is transparent and fair.”
But Nolan said his observations, which he based on about 35 cases, found anomalies one would not expect in completely random data.
“The casual observation of several high profile rail decisions appeared to be a balancing act, with decisions, wins or losses, simply alternating through time,” he said.
“That kind of run sequence is not random. That indicates something is amiss. That indicates the agency is not making decisions in an unbiased manner.”
The other extreme would be all wins or all losses, he said.
There are tests to determine how many runs should be present in a random binary sequence, but at this point Nolan doesn’t have enough data to decide.
“What was clear in the data (is) there was more losses for plaintiffs than victories,” he said.
Sixty percent of the decisions went in favour of the carrier, but Nolan said it doesn’t necessarily mean a bias is present and more study might produce different results.
He intends to get all the rail data from about 2,000 decisions to see if he can pick up runs in the data.
For now, he said, the CTA appears to be doing OK, he said.
“There is no significant evidence of capture or bias based on the runs, but these are very small samples and pure count evidence hints at bias toward carriers,” he said.
