The internet has made going to restaurants a much more efficient process. Now instead of relying on what friends suggest, you can quickly find out what hundreds of people think about any given restaurant. Even things not possible in the previous generation such as finding the 5 best pho restaurants in your city for under $20, only takes 30 seconds and one Google search. Simply put, we have more options today thanks to the internet.
I think Yelp is a truly amazing service. If someone wants to go out for [insert criteria of food here], they have no excuse not to go yelp and find the highest reviewed restaurants in the city for that. That being said, Yelp has some deficiencies. After all, Yelp itself only has a 2.5 rating on Yelp.http://www.yelp.ca/biz/yelp-san-francisco
Yelp ratings are susceptible to bias for several reasons including:
- Yelp reviewers might not be representative of the cities food interests as a whole and might favour certain types of restaurants
- Some restaurants encourage their customers to post on yelp
- High yelp scores create a self fulfilling prophecy based on an anchoring effect where individuals will be more likely to give a high scoring review to a restaurant that they think is better regarded
- Yelp users might select their restaurant choices based on yelp reviews, leading to high ranking restaurants receiving more customers and thus more yelp reviews compounding their score
In order to evaluate how reliable yelp is, I needed some data to compare it to. I decided to look at Chowhound.com which is the pre-eminent foodie forum. Despite being a “foodie” community, it really is a non-pretentious place. What separates chowhound users is not the sophistication of their palate, but their desire to learn about and try all of the best restaurants in their respective cities. They love cheeseburgers equally as much as they love foie gras.
Every year since 2011, Chowhound Toronto compiles a list of the best restaurants in the city as voted by its users. Each year, there are 100’s of submissions for each posters favourite restaurants that they ate at in the last 12 months. I took the results from the past four years to create my master list of the best restaurants in Toronto.
(10$ = Cheap, 10-30$ = Affordable, 30-60$ = Nice, 60$+= Fine Dining)
I then compared this list to the megalist list of Toronto’s most prominent food-critic, Joanne Kates’ of the Globe and Mail, as found here: http://www.postcity.com/Kates-Top-100/2014/ .
Seeing such a similar list gives me confidence in the methodology and credibility of Chowhound’s users. Some interesting findings are:
- The best restaurants in Toronto are not the most expensive. In fact, there are more affordable restaurants in this list than fine dining.
- Bar Isabel has only been around for two years. If it opened earlier, it would have by far the highest amount of votes from Chowhound. Not only is it likely that Bar Isabel is the best restaurant in Toronto, it probably is by a large gap.
- There is only one steakhouse on the list. I doubt this has anything to do with Jacob and Co being vastly superior to other steakhouses, but merely for the fact that Jacob and Co is ever so slightly better than the other steakhouses. If people are going to splurge on an expensive steak dinner, they want to make sure they get to the best one. This would lead to Jacob and Co being the only steak house that Chowhound users would dine at, skewing all the steak votes in Jacob and Co’s favour.
I then wanted to see how this data compared to Yelp rankings in Toronto. I planned on doing this by comparing the amount of votes a restaurant received on chowhound to the amount of 5 star reviews it has received on Yelp, using this as a proxy for the restaurant’s popularity. One immediate discovery I had is that Yelp has a huge preference for cheaper restaurants. This makes sense as its not really fair to compare the amount of 5 star reviews for a 10 dollar sandwich and a 100 steak simply based on the amount of people that will eat that sandwhich. In order to mitigate this problem, I had to separate restaurants into groups by price. I settled on using the Yelp price ranges for three groups: 0-30, 30-60 and 60+ dollars per meal.
The horizontal axis in these graphs show the chowhound ranking, displaying a restaurants quality and the vertical axis displays what I consider to be the restaurants popularity, the amount of support it receives from Yelp. The distance below the line represents how “underrated” the restaurant is by being undervalued by Yelp. Conversely, the further a restaurant is above the line, the more Yelp has overrated it.
Based on the results, my takeaways are the following:
- Chantecler is by far the most underrated restaurant in Toronto.
- There are a few restaurants that have created a massive hype on Yelp. Other than that, Yelp seems to be very accurate for restaurants below $60 dollars per meal.
- Yelp has no clue about fine dining – this intuitively makes sense as only a small subset of the population can afford to eat at these restaurants, and the typical clientele is unlikely to be a Yelp user. Still, having a negative correlation with chowhound’s favourites is pretty surprising.
- Most of it stems from the incredibly high support on Yelp for Jacob and Co and Canoe. This is not entirely surprising as Canoe is hugely popular due to its Canadian authenticity while Jacob and Co is a steakhouse, and steakhouses are the most likely choice of an expensive restaurant for a non-routine fine diner to go to.