1,000 Amazon Reviews

This weekend, Freakonomics received its 1,000th customer review on Amazon.com. Is this at all noteworthy?

Some of you may recall that we have previously posted here about Amazon reviews. Levitt wondered why people bother to write reviews at all. I wrote about one particular reviewer who somehow managed to always float his review to the top of the heap. (That reviewer, Loyd Eskildson, has disappeared, at least from the Freakonomics page; but he has been replaced by another top-lister who signs off as “Marilyn R. Barry, PhD in Sexuality.”)

The issue here is different: is 1,000 reviews a lot or a little? On the one hand, our book has sold roughly 1.5 million copies, which means that fewer than 1 in 1,500 buyers have reviewed the book on Amazon.com. (You don’t have to have bought the book on Amazon.com to review it there.) That’s pretty thin. On the other hand, Tom Friedman’s The World is Flat, which was published one week before our book and has consistently sold a bit more, has only 680 customer reviews as of this writing. Malcolm Gladwell’s runaway bestseller The Tipping Point, originally published in 2000 and still very strong in paperback, is one of the most talked-about books I’ve ever encountered; it has 632 customer reviews. Harry Potter and the Half-Blood Prince has 3,245 customer reviews, more than three times the number of Freakonomics reviews, but a whole lot more than three times as many copies of that book have been bought.

Could it be that people are more likely to review a book if they particularly like it, or particularly hate it? Hard to say from this group. The World Is Flat, The Tipping Point, and Freakonomics all average 4 out of 5 stars in customer rating; the latest Harry Potter averages 4.5.

While it is true that the Internet creates a public forum for a sort of carping and whining that generally isn’t tolerated elsewhere (witness some of the frequent commenters on this very blog, e.g.), that alone wouldn’t seem to explain why there has been so much Amazon commentary on our book.

So if you have any thoughts as to why we’ve passed the 1,000-review threshold, let us know. Carping and whining, as always, will be tolerated.


nphebel

Any information on the standard deviation of # stars in the reviews of the books you mention? If Freakonomics has a higher % of 1's and 5's compared to TWIF and TP (despite having the same mean), then you could assert that a book's devisiveness triggers more reviews.

Which brings up another question, who are these people who bother to write a 3 star review in the first place?

ftelegdy

I'm certainly no expert in the field of web site comments/ratings, but I do own a local restaurant dining guide that allows users to add ratings and comments for each restaurant. With that in mind, here are my thoughts...

Ratings and comments have to be separated to some extent. An entity can have 10 ratings and have an average rating of 4 and be the same rating as an entity with 1000 ratings with an average rating of 4. However, an entity with 1000 comments is vastly different from an entity with 10 comments. So, for the sake of discussion, I'd ignore the fact that all the books (The World is Flat, The Tipping Point, and Freakonomics) all have the same average rating other than to say that the same percentage of people who rated like the books.

Now, a bit of history about my own site to expand upon the comments portion of the discussion...

For an upcoming site redesign, I tried to transfer the old ratings into the new rating style: going from a 5-level rating to a 2-level rating. The idea for the transfer was to make ones and twos in the old system become a zero in the new system and fours and fives become a one. So, I had...

1/2 => 0
3 => (nothing)
4/5 => 1

I ended up not using this system for the new redesign because of something very interesting I found in the transfer. I found out that after the transfer was completed (in a test environment), I had over 4000 ones and only 800 zeros. In my rating system 1/2 was bad and 4/5 was good. So, after the transfer, I had over 4000 positive ratings and only 800 negative ratings. The threes, in the transfer, were considered "riding the fence" and accounted for about 2500 ratings.

It should seem obvious why I abandoned the new system, other than the fact that I was putting a new spin on data gathered by another means, given this.

How does this relate? First, it goes back to the idea that ratings are just an average, so similar rated entities are necessarily the same. One other thing that I noticed in my data is that people tend to comment/rate the restaurants that are the most controversial or that really standout. For instance, the #1 restaurant on my site has all of 8 comments and 14 ratings. However, a restaurant not in the Top 10 that is looked upon favorably, but that I know receives a LOT of mixed reviews, has 34 comments and 66 ratings. Anoter, on the bad end of ratings, has 33 comments and 66 ratings.

In the end, I think it comes down to either a) Freakonomics being very controversial and/or b) People thinking Freakonomics is great/horrible. I'd personally lean towards A (How could you ever link abortion and crime rate drops?! Oh, the horror! ;), and that would include people defending the book through positive comments or condemning the book through negative comments, with a smattering of B.

By the way, if you're interested in the data from my web site, you can have it. If not, I hope the comment was still worth the read. :)

Read more...

divotjr

Tim Harford recently posted on his websited about this very topic. Check it out at

http://www.timharford.com/writing/2006/03/importance-of-being-negative.html

dd

...it makes people feel smart to review a book like Freakonomics. It means they got it.

Joubert

I love this question because in an alter-ego, I run a content-driven series of sites chock-full of reviews and because I have watched the rise of reviews from sites like Epinions to everyone having a review section.

Ranking items for the sake of relevancy is a great consumer tool (and even better tool for marketers). But as someone who hires contractors to write reviews, my marketing guru always begs me, "Who is your demographic? Why are they are going to do this for peanuts?"

In the case of Freakonomics, I think there are three elements at work (not including the obvious elements of literacy, access to the web, etc.)

The first is ego-boo, which I first saw in Wired but which Wikipedia sources to the 1950s. I believe the second portion is passion - for or against. I don't think we are wired to write neutral reviews. The third is the book's demographic, or rather, the demographic of the people who have actually read it all at least once.

Gladwell's books likely have the same demographic effect, as do those by celebrity historians and other "name" authors. People want to assoicate with them, people want to feel collaborative in some sense, and if the book has touched them positively or negatively, they're going to look for an outlet to share their thoughts.

I always love looking at Amazon's IMDb.com site for ratings. Check it out yourself. The passionate who want to be identified with a film or actor are the ones who write. My unscientific sample was to check the rating for the cheesiest holiday movie that makes me chuckle today (and yes, we own a copy): Santa Claus Conquers The Martians. The site has 2,688 ratings as of this writing. The lead comment is written by someone who has contributed 205 free reviews to this site.

Why?

My theory is the person wants to be viewed as an expert or otherwise receive a thrill for another web user finding his work helpful.

The seemingly rational thing to do would be to evaluate whether the time invested in reviewing the film would generate enough web traffic to warrant selling advertising or even the film itself through an affiliate program.

Freakonomics touched people that way because it made them think. That is where the passion comes in. I looked at some Stephen King titles to see if there was a trend. "The Stand", arguably his deepest work, has 838 reviews. "Cujo" has 238 and "Cell", the most recent, has 571.

Darn you guys. Now I'm going to end up plotting King titles vs other authors. I love a good puzzle that will take the night!

Read more...

Blandy

I haven't posted a review on Amazon, but I've recomended Freakonomics to lost of my friends, especially those studying economics. I do this because I think the book was a good read and that my friends might enjoy it too.

On further thought, my friends also recomend books to me. By sharing this information we are all able to increase the chances of us reading a good book. I guess I recomend books to friends in the hope that they will do the same for me.

Perhaps it's the same for reviews on the web

juliettelucie

If "so few" people have reviewed Harry Potter, it's probably due to the fact that most Harry Potter fans review it on the many fansites that exist. No offense, but there aren't as many fansites for Freakonomics as for Harry Potter...

CJ

Freakonomics has a larger demographic, in my opinion, than "The Tipping Point" or "The World is Flat." Think about the audience that any Harry Potter book reaches. Everyone from ten years old to eighty years old seems to enjoy Harry Potter, and the older the reader is, the more he tries to defend himself for reading a "kid book." There will be many more posts for a Harry Potter book because adults are attempting to convince other adults that it is intricate, captivating, creative, aka "more than a kid's book." It is also not surprising that Harry Potter would receive a 4.5 rating rather than a 4.0. If you're trying to convince another person that it is a "big boy" book why would you rate it a 4 (very good) instead of a 5 (excellent)?

"Freakonomics" reaches a large audience, and an especially large college audience. I attend Fordham University and "Freakonomics" is on the shelves of our bookstore, along with Fairfield University's bookstore, and the Westchester Community College bookstore. These are the schools that I personally know of. There's probably more factors than I can imagine that would result in 1,001 reviews, as of this post, but I think the answer is much simpler than everyone thinks. Freakonomics is a very well-written, innovative book, and in the hands of liberal college students, it spreads like wild fire. I feel like adults and students alike genuinely want to spread the word that this book is a teaching tool and a new way to think and draw correlations. Readers want to inform others that it is not just another boring economy book jam-packed with statistics and figures. It gives real examples, stories, and testimonies that prove new points about topics that most would be hesitant to discuss.

Simply, it reached the 1,000 review mark because it deserved to. "Freakonomics" has changed my perspective on things. Since I have finished reading it, I convinced three of my roommates to read it. The results were predictable; they were all hooked, and how could you not be? There all also so many topics to discuss because as you say, it "explores the hidden side of everything." There are easily 1,000 different points that I could bring up throughout "Freakonomics", so I see it fitting that there are over a 1,000 reviews.

Read more...

pkimelma

I think it is simpler. I do agree that people very happy or upset with a product will review it. In particular, people review products they had a problem with, especially if they couldn't get help - it is a kind of revenge I think.

But, I also think that many reviews I see in Amazon and elsewhere are the see-saw style. Someone sees a really bad review and feels compelled to write a good one in defense. Obviously the opposite as well.

I read a study last year, where the author intentionally wrote carefully worded reviews that were quite extreme (some good and some bad) as well as some which were very neutral or balanced (pro/con together). They then measured the kinds of follow-on reviews. What they found was probably "overcompensation". That is, the follow on review was stronger in the opposite direction. When their review was neutral, it took much longer for a follow-on review and it tended to be less extreme (on average).

This was tested on Amazon, Netflix, Rotten Tomatoes, and some other shopping site.

It is interesting that this would tend to continue on, as the effect is to more extreme and not less.

Read more...

christopher

Comment #9 sounds plausible to me, but here's another possibility:

Within Freakonomics, a single book, the authors provide provocative and/or very uncommon analyses of MULTIPLE and VARIED subject matters. Meaning that they're more likely to interest/intrigue/anger/delight more people than the average book that comments only on a single or relatively fewer discrete topics.

In other words, the more and varied the topics discussed -- particularly when what is said about each of them is so provocative -- the greater the likelihood of trespassing on the particular sensibilities and interests of MORE and different people.

A bestselling book on real estate, or one on immigration, or one on the life of Da Vinci -- all will generate x number of reviews on Amazon from the particular audience of each.

But a book like Freakonomics, which deals with many different topics, with such compelling and thought-provoking conclusions, seems likely to stimulate a whole population of reviewers for each topic covered. So not only do you get all the readers were particularly intrigued by the particular ideas about real estate agents; you also get all the readers who have a deeply felt opinion on abortion or the crime situation; and all the readers who are concerned about the job they're doing as parents; etc. As well as all the readers who are just plain shocked -- and very happily so -- to find a pair of independent minds with the honesty and intelligence to seek the real story behind things and the courage to print exactly what they find, popular or not. Something definitely worth five stars.

Read more...

JGUNS

The link to the story mentioned before states that he found it was more likely that the smart reader will discount the positive reviews and instead place more emphasis on the negative reviews. They found this to be true in their analysis.

However, I submit that it depends on the TYPE of book and that one would find a varying degrees of statistical trends depending on the book. For example, political books often get reviews even before they are released to the general public. If one was to look up a book by say, Ann Coulter, one would see a huge number of negative reviews given either before or on the day the book was released. Clearly these people haven't read the book and they are more interested in "bringing down" the rating of the book. In fact, nearly any book that challenges ideology will have the same problem.

1000 plus reviews in the case of this book does not necessarily mean a lot of people have actually read it (although we know that they have), but that there are strong feelings that people have about the book.

I think the motivation for people to right reviews is more about ideological motivations or passion for a particular genre of books in the case of fiction books. For instance I read a lot of "fantasy" type books, and I don't discount the negative reviews as much because they can often be done by people that don't "get it." Instead I look for well done reviews that can discuss the content and also show that the author of the review understands the genre and can put it into context with similar books of the genre.

A smart buyer doesn't necessarily give more credit to negative reviews, instead they are looking at the content of the review and basing their opinions on their judgement of the author writing it.

Read more...

SteveSailer

Freakonomics' scattershot, quick-and-dirty approach appeals to people with short attention spans, and there are a lot of them.

It's like a blog in book form!

Unfortunately, Dr. Levitt saves his more interesting ideas for his paying outlets, because his natural talent is for blogging -- run an idea up a flagpole and see whether people who know more about the subject than he does salute or not. If he'd put his ideas here on his blog first and see whether they fly, he could save himself a lot of embarrassment and the world a lot of misinformation.

Similarly, Malcolm Gladwell is putting the cart before the horse. On his blog, his recent New Yorker review that credulously fell for everything in a book about sports statistics by three economists entitled The Wages of Wins is currently getting ripped to shreds by readers who know a lot more about sports and statistics than Gladwell does. Unfortunately, most people who read the review will never see the debate over it.

http://gladwell.typepad.com/gladwellcom/2006/05/the_case_for_ke.html#comments

Read more...

meteorplum

Like "nphebel". I didn't have enough data to make an informed opinion, so I decided to get some data. Here are the distributions of Amazon reviews for the books mentioned in the blog entry, as well as The Da Vinci Code as a second "control":

Freakonomics (out of 1003 reviews)
5 stars - (37.9%)
4 stars - (29.4%)
3 stars - (15.7%)
2 stars - (9.7%)
1 star - (7.4%)

The Tipping Point (out of 635 reviews)
5 stars - (51.0%)
4 stars - (27.9%)
3 stars - (8.8%)
2 stars - (5.0%)
1 star - (7.2%)

The World Is Flat (out of 682 reviews)
5 stars - (45.2%)
4 stars - (22.3%)
3 stars - (11.9%)
2 stars - (8.8%)
1 star - (11.9%)

Harry Potter and the Half Blood Prince (out of 3248 reviews)
5 stars - (60.2%)
4 stars - (19.7%)
3 stars - (10.1%)
2 stars - (5.7%)
1 star - (4.3%)

The Da Vinci Code (out of 3288 reviews)
5 stars - (39.8%)
4 stars - (15.8%)
3 stars - (13.9%)
2 stars - (11.1%)
1 star - (19.5%)

The data point tantalizingly to the idea that controversial books get more reviews because they get more "bad" reviews, but the numbers are not very conclusive. And though Harry Potter 6 seems to be a counter-example to my almost theory, it *is* both the sixth book in a series and, as most HP fans I know agree, a better one than #5. So I decided to get the same data for #1:

Harry Potter and the Sorcerer's Stone (out of 5217 reviews)
5 stars - (85.1%)
4 stars - (9.3%)
3 stars - (3.0%)
2 stars - (1.4%)
1 star - (1.3%)

Given all the negative press around the original release of Harry Potter 1, I was not expecting that the book would have only garnered 1.3% (66) of one-star reviews. However, a quick glance at the one-star reviews for Harry Potter 1 showed at least two anti-Amazon comments (didn't receive the book) and one review saying that it is "a good book [and] people should read the Harry Potter series."

Perhaps the only conclusions I can reach from this paltry research are a) more research is necessary, and b) this research would be easier if Amazon made it easier to scrape the actual reviews from their site with fewer page views (you can only see 10 reviews at a time).

Read more...

danny

Penguin Australia promised to send me a review copy of Freakonomics, but it never turned up. But if there are 1000+ reviews at Amazon there's probably no need for more of them - most of the books I review have no reviews online anywhere, at Amazon or otherwise.

ermagetton

How about:
The World is Flat looks good on a shelf, while Freakonomics is something people actually read and discuss. Maybe more copies of the World is Flat are sold because some people have extra shelf space (whereas Freakonomics would be on the bedside table with a bookmark in it).

jonksbargains

Freakonomics is a great read. But everyone already knows that. I don't see any reason to post a review - I just tell my friends all about it instead.

Brendan Moore

Diminishing returns.

As you approach 1000 reviews, the true evaulation of the book is better reflected in the aggregated reader reviews. Therefore, the marginal benefit I get from my review contribution (of bringing the evaluation of the book closer to what I perceive to be its true value) isn't worth the marginal cost.

nphebel

Any information on the standard deviation of # stars in the reviews of the books you mention? If Freakonomics has a higher % of 1's and 5's compared to TWIF and TP (despite having the same mean), then you could assert that a book's devisiveness triggers more reviews.

Which brings up another question, who are these people who bother to write a 3 star review in the first place?

ftelegdy

I'm certainly no expert in the field of web site comments/ratings, but I do own a local restaurant dining guide that allows users to add ratings and comments for each restaurant. With that in mind, here are my thoughts...

Ratings and comments have to be separated to some extent. An entity can have 10 ratings and have an average rating of 4 and be the same rating as an entity with 1000 ratings with an average rating of 4. However, an entity with 1000 comments is vastly different from an entity with 10 comments. So, for the sake of discussion, I'd ignore the fact that all the books (The World is Flat, The Tipping Point, and Freakonomics) all have the same average rating other than to say that the same percentage of people who rated like the books.

Now, a bit of history about my own site to expand upon the comments portion of the discussion...

For an upcoming site redesign, I tried to transfer the old ratings into the new rating style: going from a 5-level rating to a 2-level rating. The idea for the transfer was to make ones and twos in the old system become a zero in the new system and fours and fives become a one. So, I had...

1/2 => 0
3 => (nothing)
4/5 => 1

I ended up not using this system for the new redesign because of something very interesting I found in the transfer. I found out that after the transfer was completed (in a test environment), I had over 4000 ones and only 800 zeros. In my rating system 1/2 was bad and 4/5 was good. So, after the transfer, I had over 4000 positive ratings and only 800 negative ratings. The threes, in the transfer, were considered "riding the fence" and accounted for about 2500 ratings.

It should seem obvious why I abandoned the new system, other than the fact that I was putting a new spin on data gathered by another means, given this.

How does this relate? First, it goes back to the idea that ratings are just an average, so similar rated entities are necessarily the same. One other thing that I noticed in my data is that people tend to comment/rate the restaurants that are the most controversial or that really standout. For instance, the #1 restaurant on my site has all of 8 comments and 14 ratings. However, a restaurant not in the Top 10 that is looked upon favorably, but that I know receives a LOT of mixed reviews, has 34 comments and 66 ratings. Anoter, on the bad end of ratings, has 33 comments and 66 ratings.

In the end, I think it comes down to either a) Freakonomics being very controversial and/or b) People thinking Freakonomics is great/horrible. I'd personally lean towards A (How could you ever link abortion and crime rate drops?! Oh, the horror! ;), and that would include people defending the book through positive comments or condemning the book through negative comments, with a smattering of B.

By the way, if you're interested in the data from my web site, you can have it. If not, I hope the comment was still worth the read. :)

Read more...

divotjr

Tim Harford recently posted on his websited about this very topic. Check it out at

http://www.timharford.com/writing/2006/03/importance-of-being-negative.html