More fun with statistics: Do aggregate ratings matter?

I'm down into the meat of the stats now - well in between making a million meatballs (yes that is legitimate traditional christmas food in Sweden) and researching recipes for leftover ham, because the smallest one I could find in the store was 3.2kg. And we are two people. And there are also meatballs, sill (pickled herrings) and tons of other stuff to be eaten. We will be eating ham until easter I think. I would have skipped it but daughter got big sad-puppy-anime eyes at me, so I bought the damn thing. I had to move a shelf in the refrigerator to even fit it in there!


But anyway, I'm at the point in the christmas preparations where I am diving into the statistics for relief, but I'm also getting to the really interesting results too.


This is a little complicated, I hope you'll bear with my explanations, because it's a really meaty result for my paper, and I think one that will interest a lot of you.


I asked in the survey: Do high aggregate ratings make you more or less likely to want to purchase or read a book, than before you looked at the ratings? with the answers on a 1-5 scale from less likely through neutral to more likely. 


About half (56.8%) of people do think that high ratings influence them positively towards the book, however 27.2% think it doesn't sway them either way. Nothing groundshattering there. But fully 16% are negatively influenced towards a book with high ratings. 


Now here is where it gets really interesting. I also asked if high ratings make you want to read the reviews, again on a 1-5 scale. And there is almost a one for one correlation (the statistical likelihood of this being random is so small, I couldn't even calculate it, it's as solid a result as it's possible to get).


Which means: people who answered 1 on the first question, almost invariably answered 1 on the second. and people who answered 5 on one almost invariably answered 5 on both.


Finally I asked, which reviews people look at to find out more information: the high ones or the low ones. And here I got an inverse correlation. People who answered 5 (that is, yes I am quite likely to want to read the book MORE than before I looked at the ratings), and probably also answered 5 to (yes I'm MORE likely to look at reviews though) picked the low rated reviews to read.


In plain terms that means:

If a book has only high ratings, a good proportion of people (16% ish) are immediately put off it, and won't even read the reviews. 

Of the people who are positively influenced towards the books, the more positively the ratings sway them, the more likely they are to read the reviews. 

The more positively they feel towards the book after seeing the ratings, the more likely the reviews they will read to make up their minds are the lowest rated ones.


Furthermore, none of these results are affected in any fashion by any kind of demographic metric. They are stable for gender, books read, average rating, kind of reviewer (casual vs obsessive), primary book source (borrowers don't behave differently from purchasers).


I also asked the same questions, about negative reviews. Guess what folks (I'll just give the plain facts this time):

34% of people are still less likely to read a book after seeing a negative rating, and the number that feel more likely to read it drops to 25.6%

Once again, people who are negatively influenced by the ratings, won't even read the reviews, but people who are positively influenced will.

It's still the low rated reviews they read to make their minds up.


And once again, no amount of throwing this data up against the demographics in any combination made any kind of difference.


I'm not even close to writing up an analysis of what this means, because there's tons more statistics to play with, and also many more meatballs to fry for tomorrow. But here's my first thoughts:


- Pumping the reviews too high obviously can hurt perceptions of the book for a good number of people. And once you've lost those people, you've lost them (and the reviews won't get them back, because they don't read them)

- On the other hand, a very low rating has more effect on perceptions than a very high one does, but most readers will still make up their own minds either way.

- No matter what the ratings, it's the low/neutral rating reviews that readers turn to, to make their minds up.