Blair's Book Blog

Book ratings aren't helpful

People love rating things. Food, people, movies, and books. Let’s look at book ratings.

Ostensibly ratings convey how someone feels about a book. Did they love it? 5 stars. Thumbs up. Did they hate it. Zero or thumbs down. But what about that murky middle? They sorta liked it, but that once scene, where the dragon kissed a donkey, that wasn’t realistic. Everyone knows dragons eat donkeys! 4 stars! 4.5 stars!

Wait. 4.5 stars? What does that even mean? Goodreads and Amazon reviews are rife with reviews that start “3.5 stars, but I have to pick 3” or “4.5 stars, but I have to pick 4”. As a reader looking for a book to read, a fractional rating doesn’t convey any additional helpful information. I find that fractional ratings tend to raise more questions than they answer.

Years back, I interviewed and did annual performance reviews with a system that allowed fractional reviews beyond half-steps. I could assign a ratings such as 2.2 (considered fairly bad) or 4.6 (considered very good). Oh, but the arguments folks could have when selecting a rating between 3.2 and 3.8. A 3.0, on a 5-point scale, isn’t usually a bad rating. Consider it middle of the pack. That rating would imply the person or thing rated was exceptional or poor. In theory, someone would do quite well with a 3.0. But what does 3.2 mean? What does it mean in context with someone with a 3.5? Why did they not get a 3.5, then? And the same questions arise from a 3.8 rating. Why didn’t this person rate a 4.0? Is someone with a 3.8 rating distinctly better than someone rated 3.5?

These fractional ratings caused a lot of wasted time as people argued their particular ratings, strongly believing that they held solid truths of the person that they rated. But, often, they didn’t. The ratings often meant something to the individual who selected a particular rating, but to other folks the fractional ratings failed to convey that someone performed well versus someone who performed poorly. Good managers would instruct folks to stick to N.0 if at all possible, and N.5 ratings if someone just had to use a fractional rating. And avoid, at all costs, any other fractional values. Be decisive!

Thankfully, fractional ratings don’t exist any longer (to the relief or complaint of many). But I think the point illustrates the problem with rating a book: fractional ratings reflect that folks have a hard time being decisive with their ratings. Perhaps they enjoyed a book, as a whole, but that dragon/donkey scene really took them out of the story. A 4.5 absolves someone of choosing between loving a book with a 5.0, or liking it strongly, with a 4.0. It is an avoidance maneuver to own up to a rating.

I suspect that books with harder material, diction, or thematic elements routinely receive lower ratings. They may not be bad books given time and energy, but they may make the collective readership uncomfortable. Or they may be inaccessible. And therefore the lower ratings indicate less a book’s intrinsic “goodness” as the ratings reflect how accessible the material in a book is to the wider audience.

A further problem with ratings is that folks often disagree what they mean. I think a 3.0 is a solid, good rating. I’d recommend a 3.0 to a friend. But I have friends that feel that a 3.0 implies a book is awful, that I hated the book. If we can’t agree on what a particular rating means, can we derive any value from aggregate ratings on book sites? Can you reliably use aggregate ratings to pick a good book? I don’t think so.

An aggregate rating of 4.1 would typically indicate that the median rating is fairly high. If we’re lucky, we have some bar graphs that indicate the spread of reviews, so we can determine if the ratings are clustered around 4, or scattered wide between 1 and 5, with the bulk of ratings closer to 4. But that numerical rating doesn’t amount to much if each person rating a book has a personal meaning attached to each rating. What does 4.1 mean if each person has their own meaning for each numerical rating? If we can’t agree that 3.0 is a good book? At best, it means that the mathematical mean of all ratings is 4.1. That’s it. That’s all the information that number could provide. I don’t think that it can provide any deeper information on how well a work is liked, if we can’t agree to a scale.

At the end, the problem isn’t so much that ratings exist. The problems is that ratings exist as different concepts for each person reviewing the item. Without a consistent guide to ratings, without context or meaning, numerical ratings aren’t helpful. Perhaps if folks could agree on a consistent guide, then personal ratings would be helpful. In the mean time, I think the best that each person can do is to find and cultivate reviewers they like or trust for solid ratings and reviews.