Skip to main content

It's just a Review, Right?

Finding a restaurant has never been so easy—open up Google Maps and look up what you want to eat, scan reviews, and then go there, but is there actually a methodology to a review?

In America, 98% of Americans look at the reviews of a restaurant before they will decide to go there to eat.1 As college students, we excel at eating out often and eating out at new restaurants. Many times, our friends or roommates take us to a new place to eat, but often we may want to try something new together for fun. In this case, one of the easiest ways to find a good restaurant is to look up restaurants and search menus and reviews to find a restaurant our entire group will like.

Pexels Photo by Tahir Osman

Restaurant owners know most customers look at their reviews before they decide to look inside the restaurant, so to restaurant owners, customer reviews are crucial. Restaurant owners can check their reviews for effective feedback and comments on how they can improve as a restaurant, but this can be time consuming and overwhelming.

Dr. Matt Baker and Dr. Brett Hashimoto conducted research on how to make restaurant reviews more beneficial and less time consuming for restaurant owners. In their article “Expression of Customer (Dis)satisfaction in Online Restaurant Reviews: The Relationship Between Adversative Connective Constructions and Star Ratings” they researched how the use of adversative connective constructions like but, however, in addition to, etc. affect the restaurant’s star ratings and how they do.1


Dr. Baker and Dr. Hashimoto focused on Yelp restaurant reviews from the years 2018-2019. They sampled non-chain restaurants across 1-5 star ratings and across prices ($,$$, $$$). The reviews also included at least one of 17 adversative connectives (ACs): although, at the same time, but, conversely, however, in comparison, in contrast, instead, nevertheless, on the contrary, on the other hand, rather, still, though, whereas, while, yet.

ACs connect two linguistic units, whether those be words or groups of words that function similarly. Because ACs are adversative, reviewers use them to contrast or contradict the ideas expressed in the units. For example, in the review “I liked the food, but the service left a lot to be desired,” the reviewer sets up an expectation of a positive review about the food, but then that expectation is defied by a negative review about the service. Linguistic theory suggests that people shift the most important information in their communication to the end of sentences (a concept called end weight), so the negative second unit (“the service left a lot to be desired”) communicates what the reviewer wanted to emphasize in the review—or how satisfied the customer was with the experience.

Using the concepts of ACs and end weight, Dr. Baker and Dr. Hashimoto generated concordance lines of all the ACs in their sample. Then two BYU students served as coders who categorized the ACs by whether the second linguistic unit of each AC ended negatively or positively. They called these positive or negative AC constructions. Ultimately, Dr. Baker and Dr. Hashimoto wanted to see if the AC constructions could be used to unlock insight into how satisfied the reviewers were.

With the data coded, Dr. Baker and Dr. Hashimoto conducted a mixed-effects ordinal regression to see if an increased frequency of positive (or negative) AC constructions in reviews correlated with higher (or lower) star ratings. They found that more positive AC constructions did, indeed, increase the likelihood of a higher star rating; in addition, they found that increased negative AC constructions associated with lower star ratings.

Through additional data coding, they also found that if the restaurant had a high rating (5 stars), the comments focused on what reviewers liked about the food, service, or environment. If the restaurant was a medium rating (3-4 stars), the reviewers would typically be commenting on the bad quality of the food. The lowest ratings (1-2 stars) focused on the bad service at the restaurant.

Pexels Photo by Gustavo Fring


Through analyzing restaurant reviews, Dr. Baker and Dr. Hashimoto found that when restaurants went above and beyond their customers’ expectations, they tended to get higher ratings, but if the restaurant lacked in food quality or customer service, they would receive lower ratings which would affect their restaurant image. Specifically, using the results of the regression, Dr. Baker and Dr. Hashimoto found that for restaurant owners to increase a reviewer’s star rating by 1 star, they must meet or exceed the reviewer’s expectations twice (i.e., double the frequency of positive ACs in the review). To decrease the reviewer’s star rating by 1 star, restaurant owners must fail to meet the reviewer’s expectations by eight times (i.e., greatly increase the frequency of negative ACs in the review).

Although there is still more research that can be done with restaurant reviews, Dr. Baker and Dr. Hashimoto’s research helps restaurant owners focus on some of the most important information in reviews. Using this information, restaurant owners can focus on ACs in reviews and seek to capitalize on the positive things and remedy the negative things reviewers include in the second units in AC constructions.

The next time you are at a restaurant or are giving a review, start to note if you tend to use ACs in your review and which star you would rate the restaurant at. You may find that there is more to a simple review than you originally thought.

Read the Original Article:

1Baker, Matt, Hashimoto, Brett. 2023. Expression of Customer (Dis)satisfaction in Online Restaurant Reviews: The Relationship Between Adversative Connective Constructions and Star Ratings. Sage Journals 61(1). 148-180.