The April 18, 2024 column by Professor Sylvain Charlebois, entitled “Can we really trust Statistics Canada? », arouses severe reservations on our part. She suggests that the data collected by Statistics Canada (SC) to develop price measures in the food sector is neither precise nor reliable.
The author, however, does not explain how the methodology developed by himself and his team is superior to that of SC, which constitutes a significant gap that significantly weakens his claim.
However, we were able to obtain more details on Mr. Charlebois’ methodology by communicating directly with him, which allowed us to draw the following conclusions:
1) The technical terms used by Mr. Charlebois to qualify his price values are statistically inappropriate. The latter speaks of “real observed values”, whereas with a price sample, we cannot make such an assertion. To do this, we would need to have all possible observations, which is an elementary notion in statistics.
2) Further in the article, the columnist states that “47% of food items (16 out of 34 items listed) are underestimated by Statistics Canada”. The inflation rates for these items estimated by SC are indeed lower than those estimated by Mr. Charlebois, but it is incorrect to say that they are “underestimated”, because they are averages from samples .
3) In addition, the columnist repeatedly attacks the statistical agency in an unjustified manner, without ever presenting the methodology used by the latter. Let us look at two examples:
a) “(…) when assessing the accuracy of our lead federal agency in measuring changes in food prices, it is clear that much work remains to be done”;
b) “(…) this suggests that the agency’s reports do not always accurately reflect food inflation.”
These judgments are based solely on differences in average prices of food items. To issue them, it would be necessary to compare the methodologies and indicate what is wrong with that of SC. Furthermore, in the table presented in his column, Mr. Charlebois nowhere mentions over what period the price variations are observed.
4) SC’s methodology is available on its website1. We note that “the average prices included in (its) table are calculated from optical reader data”. Additionally, this data includes “prices from thousands of stores across the country, which represent millions of weekly food prices and cover the majority of the Canadian food market.”
1. Consult the Statistics Canada website
In an email exchange with Mr. Charlebois, he mentions that his team collected “approximately 13,780 data points in total”, but without specifying how this data was weighted to construct price indices. The large difference in sample size between that of SC and that of Mr. Charlebois strongly calls into question the quality of the statistics collected by the latter.
This column joins many others with approximate rigor in the media where Mr. Charlebois continues to operate. In the past, other academics in Quebec (Daniel Mercier Gouin of Laval University2) or elsewhere in Canada (Andrew Leach of the University of Alberta3), severely criticized Mr. Charlebois.
2. Read the article from News writer
3. Read the line about X (in English)
Hippocrates, the father of medicine, said: “First, do no harm. » Members of academia are supposed to enlighten us and contribute to public knowledge, not undermine it.
Mr. Charlebois seems to deliberately do the opposite.
reference: www.lapresse.ca