Average Rating: 5 vs. 7

I was just having a discussion between some friends of mine today about rankings from magazines or web site gives a game/movie/book/etc. a 7 rating, are they saying that it's average and using a school like grading or are they using 5 as average and 7 makes the thing great rating.

So my question to you all is what are your rankings average, 5 or 7? And if you think 5 is average, should reviewer on here, other sites, and other mediums change their ways to do the same and stop thinking 7 is average and use the full points range?

7 is Very Good imo. 5 Is quite neutral, decently done but doesn't impress you. 6 is in the middle. It's good.

For me, 7 would be a movie that I would watch again, but wasn't stellar. I save that for 8 through 10.

5 is for a movie that was watchable, but wasn't something I would watch again. And below that would be wincable movies.

So from five to seven is "Average," if that makes sense?

6 is average for me, I would almost never watch a movie generally rated a 5. To me anything rated 5 or below is bad.

If you look at the ratings distributions the majority of people rate above 5, here are the ratings distribution for Listal:

rated 1: 49k
rated 2: 36k
rated 3: 42k
rated 4: 80k
rated 5: 116k
rated 6: 227k
rated 7: 259k
rated 8: 295k
rated 9: 159k
rated 10: 235k

I find 5 to be average, I didn't LIKE it, but I didn't HATE it either

6 for me is I am not really sure I like it or not, 7 is I kinda liked it...

I generally rate above 4 just because I can normally find SOMETHING I might have found ok about the movie that'll bring the score up a point

I range 4-6 as average so I guess I'm in the 5 bracket. These are films that I couldn't find enough to love about them to warrant them 7 and up.

The way I rate it would seem I only rate five and up, but what's really going on is I try to not watch things that are less than a five, and if I don't watch it, then it doesn't get rated.

Ah, but then you aren't giving chance to the possible great movies and only get dissappointing movies that didn't meet your standard

I really don't understand the logic behind 7 being 'average'

After all, the way of finding the 'average' of something is to take all the numbers add em together and divide by the amount of numbers, and if there is only one number, cut it in half,

10 cut in half is 5, so it seems like the more logical choice,

unless you consider it based off of the grading scale, in which average would be a high 7(7.9) or a low 8(8.2), in which it makes a bit of sense, but not much...

When did 7 start becoming the 'average' for ratings, just curious?

Ah, but then you aren't giving chance to the possible great movies and only get dissappointing movies that didn't meet your standard

I'm aware of that and am 100% okay with it.

Thinking about this some more, a thought occurred to me. Most people don't tend to go to movies they're fairly sure that they won't enjoy. So the average rating for a person would most likely start a bit higher because you're tending to rate movies you thought you'd like.

That's not true of everyone, of course. But it would account for people's likelihood of rating things under 5. Which doesn't mean they won't, but that they're less likely to because they avoid those they think would fall into that category.

Does that make sense?

It does. I have a fairly big gap between movies I rate 1 or 2 and 5 or higher. I DO rate some as 3 or 4, but it is rare. A so-so movie typically gets a 6 or 7 from me.

How does this affect finding compatibility?
Ppl remember what they liked. They watch or play more of what they liked and feel more confident making ratings of it.
I often rate things higher not because i liked them but because i see merit and i think other people should see a better rating for this. That tends to cause an average at the 6-8 rating.

The 8 average rating started because people tend to read about what they like and reviewers wanted to please.

Compatibility is the most important thing in deciding ratings. Aggregate ratings don't mean much. At least when i look at any website's top list i get the feeling spam-bots have been all over it. When i am interested in checking something out its the contact's rating that means a lot more.

Here is an estimate of compatibility drop: If one person makes his average at 7 and another at 5 their comparability could drop about 10% if half of both their ratings are average.

If i have made a mistake about the comparability formula its because searching the forums every time is a pain and we really need a wiki. Would you like me to start one?

If i choose my average at 5 and i think a show deserves a 4 but im 200% certain i will rate the show a 2. Because ratings cant be Weighed.

One item with a 9 difference of rating should reduce comparability greater than two of 5. At least if the idea of the previous paragraph is used. Otherwise just Weighing a rating would be better.

Until then i will just prune 'followings' based on my contact's top ratings. No wait i will need to make dummy accounts with only the shows i feel strongly about so the compatibility works better.

Noo im completely wrong. The target rating is the rating ppl you want recommendations from make. As soon as you pass it in to the extreme direction every point in that direction subtracts the same amount of compatibility score for desired and undesired users.

Unless of course you are looking for ppl who rate to the extremes. In other words it works just like the 5v7 issue. For those who you have the same opinions with choosing different rating strategies will reduce compatibility by : 1/5 ratings are extreme, with a difference of 3 pts makes 3/5 or about 7% difference.
However, unlike in the Average issue, the 7% drop is probably for everyone EXCEPT those who you are compatible with already.

The real reason not to do this is the recommendations you effectively give others and your ability to show you liked something slightly less or more.

I use a rating of 5 as an average. If 7 is average, what's the point of having a full 10 point scale? Using the full scale helps to really separate out the merely "better than average" films from the "great" films.

If 7 is the average compatibility becomes based much more on things ppl dislike. That is bad.

Anni's ratings A:10 B:1
Ben's A:7 B:1
Ryz's A:10 B:7

A-B 3 or 85%
A-R 6 or 70%

And then there's the people who rate 7s based on what they perceive to be what others would like even if rater doesn't like it. To put another way the rating 5 needs to be "i liked it a little and i think it deserves a good rating even if i only liked it a little."

To me 5 is just ok, it wasnt great but at the sametime wasnt terrible, 7 to me is good id watch it again, some of my fav movies are 7s

1 2 3 I could not watch this/possibly like it i don't know how anyone can and i don't want ppl who could to recommend me things.
4 5 6 7 I almost certainly didn't watch the whole thing. I can understand why other people would and enjoyed it somewhat.
8 9 10 I watched the whole or large part of the series. I looked forward strongly to watching each episode. I may have put the rest of my activities on hold to finish this.

6 7 I watched and enjoyed and may have watched all of but i can understand why others would not.

7 I didn't enjoy it much but i do see the value innovation or worth to others.

I put 6 or 7 if I want to rate something as average.
I don't know if it's my subconscient who speaks louder because I'm used to the college rating, but I always did that.

But I think 5 would be most clever.

Thinking of a 7 as average must have originated from the educational system where the 70-79 range on tests is called a C and thought of as "average," probably because the vast majority of scores fall somewhere in the 50-100 range. I don't think its helpful to think in these terms when rating movies. Not many test-takers score a zero, but there's plenty of movies I would rate a zero. 5 should be called average on a 10-point scale.

I think of average in move ratings primarily in terms of how it compares to all the other movies I've seen simply in terms of how much I enjoyed it. I analyzed my ratings recently and found that 47% of my 862 ratings were 5 or lower, so it looks like I'm pretty accurate when I assess how a movie I just watched compares to all the other movies I've seen.

If I think a movie I didn't like has some redeeming value for others and is just not my type of movie, this seldom has any impact on my rating. I'm not estimating predicted ratings for other people; I'm trying to make as accurate an assessment of my own preferences as I can.

5 is average. I neither strongly liked or disliked a 5 movie. I watched the whole thing without feeling motivated to shut it off is about all I can think to say about it. But I have no interest in seeing it again.

A 6 movie I feel afterward I've gotten something out of other than just a way to take my mind off things--generally not enough that I'd see it again, but occasionally I'll rate 6 for movies that have parts I both strongly like and dislike.

Movies below 5 I either didn't finish or go away feeling bored, traumatized, depressed, or tricked by. I try to be as generous as possible and not just rate things 1/10, but if there are parts of a movie bad enough, I'll sometimes rate it 1/10 to convey just how bad they are, even if I could think of some redeeming aspects.

When you have to use a 5-point scale, such as at Netflix, you're forced to choose between either 2/5, which is 4/10, or 3/5, which is 6/10. I tend to choose 2/5 so that Netflix's algorithm won't think that I "Liked" the movie when factoring it in to its recommendation calculations.

Contrary to my earlier posts i don't advocate the 5 average anymore.
My opinion is divided in to two parts; my enjoyment of the show and my opinion of ppl who enjoy this show. 7 is a show i mildly enjoy and would watch if forced by circumstances. 6 is not. 4 is a show who's fans are normal forgivable people. 3 are trolls spambots and ppl i don't want recommendations from. You could say 5 is the Median of shows and that would mean i enjoy less than 1/2 of the shows available. Way less since 5 is far from 7. The main reason i use this system is that it makes voting easier.

How much damage would disagreeing make? For the following examples these people have perfect intended correlation. Idk what the addition of random(non-correlated) votes would do to these guesses.
I said before: "Here is an estimate of compatibility drop: If one person makes his average at 7 and another at 5 their comparability could drop about 10% if half of both their ratings are average."
In the worst case of both parties voting only their respective average they get -20% or an 80% compatibility.
If the person who votes 5 has one item for every rating or 10 items, 7's appropriate effective distribution can be approximated.
1 2 3 4 5 6 7 8 9 10
1 3 4 5 7 7 8 9 9 10
0 1 1 1 2 1 1 1 0 0
8/100 or 8%
If the intended distribution is a bell curve or something similar the % lost increases. In my experience that is straight from my top 30 into average so we need to agree.

The best "average" for everyone to use is the one that gives the most space to place votes. Actually let me get back to you on this.

Hyomnil thats a Median.
Edit: and i forgot, (10+1)/2=5.5 the average.

Since there's no rating of zero, 5.5 would technically be the average:

(0+1+2+3+4+5+6+7+8+9+10)/11=5
(1+2+3+4+5+6+7+8+9+10)/10=5.5

But since you can't rate on a 20-point scale, you have to either round that down to 5 or up to 6 when picking a rating you think of as average; I think it makes more sense to round down to 5 because that gives you an equal number of ratings choices for below-average and above-average.

I don't see any use for thinking of 7 as average unless you need a greater range of choices to distinguish between below-average movies than you do above-average movies. If 7 is average, then you have 3 choices for above-average (8, 9, 10) and 6 choices for below-average (1, 2, 3, 4, 5, 6). Why would you want twice as many options for distinguishing how bad a movie is vs. how good it is? Tom posted the listal ratings distribution, which has 7 as the most frequently occuring rating (the "mode"), but that could be because people are simply more likely to bother entering a rating at all for movies they liked vs. those they didn't.

If you're just choosing whether you hated, were so-so on, or loved a movie right after you've seen it, it doesn't make much difference, but when you're making Top 10 lists, you have to go back and think critically "Among these movies, which would I really consider the best ones now that I've seen other better movies since." If all your above-average ratings are either 8, 9, or 10, that's going to make it more difficult than if you have 6, 7, 8, 9, and 10. At criticker.com and inveni.com, they use a 100-point scale. You take all your, 8-rated, movies, say, and sort them by which ones you like better now: 80, 81, 82, ... , 89.

The "median" rating is found by arranging the values in order and then selecting the one in the middle.

If you have ratings (1, 1, 10, 10, 10), mean=6.4 and median=10.

Median is useful in cases where the distribution has very large extreme values which would otherwise skew the data. But on the 10-point scale, there's an equal distance between each possible value, so I don't think its of much help here other than its easy to calculate.

The median rating is the rating that 50% of a user's total ratings are less/greater than or equal to. If my median rating were greater than 5, I would think of that as meaning I was particularly good at choosing movies I enjoy; if it were lower than 5, I would think I was particularly bad at choosing movies I enjoy.

Please define average. First, is the average based solely on your enjoyment, your opinion of its worth including others or your opinion of the people who enjoy this movie?
Second, is your definition of average "So-so", like watched whole enjoyed a bit would not watch it again and equivalents, OR the Median, like half of the movies I watched are worse and half are better.
Third, is this average of your current collection OR the entire world collection.
A total of 3X2X2 or 12 possible "averages" arise. I use two at the same time. Mine are personal joy and so-so at 7 with opinion of people , Median and world collection at 5.5.

If the decision on what the average should be will be based on spreading out distribution then the MEDIAN should be around 5.5. Otherwise, if you are the average person, you will see a lot of stuff outside the genera you enjoy and the low scores will pile up. I mean the "So-so" can't be at 5.5 because there is more to dislike. Well that could depend on the individual.

"20-point scale" You mean an odd point scale?
"round down to 5" Um 4 and 5.
Your last paragraph Hyomil + condition of keeping 50% of votes in 1-5 and the other 6-10 means use MEDIAN. Well the median of your current collection or the world's but median.

The 'most space to place votes" is probably not the best criteria for choosing the average.

Please define average. First, is the average based solely on your enjoyment, your opinion of its worth including others or your opinion of the people who enjoy this movie?

There's the mathematical definitions, the "mean" or the "median," depending on which type of average you want to use, and the common-use definition, "so-so" or "mediocre." After seeing a movie I found so-so, I rate it a 5 because that's the middle value (median) on the 10-point scale I have to choose from at listal. If I did an analysis of all my ratings and found that my median rating was 4 or 6, or even 7, I don't see why that should influence which rating I use to convey that I found the movie so-so.

As I said, if my median was 7, that would indicate I was good at choosing movies I would like. Should I then start rating all my so-so movies 7 just so that I can have a nice even distribution of my ratings with half on either side of 7? What would I gain from that?

I also don't see how my opinion of others who enjoy movies I don't rate highly factors in. Its the extremes that matter there. If rate a movie a 1 and someone else rates it a 10, that strongly affects my opinion of our compatibility. If I rate a movie as so-so and someone else rates it as 1 or 10, that doesn't have much effect on my opinion of our compatibility.

Third, is this average of your current collection OR the entire world collection. A total of 3X2X2 or 12 possible "averages" arise. I use two at the same time. Mine are personal joy and so-so at 7 with opinion of people , Median and world collection at 5.5.

In what way do you use two at the same time? 12 possible averages? Are you saying you choose your ratings with 7 as so-so but assume that others will choose 5 as their so-so rating and weigh that when deciding whether to accept their recommendations?

"20-point scale" You mean an odd point scale?

5-point scale: 1, 2, 3, 4, 5
10-point scale: 1, 2, 3, ... , 8, 9, 10
20-point scale: 1.0, 1.5, 2.0, 2.5, ... , 8.5, 9, 9.5, 10

meaning 20 possible ratings you have to choose from.

"round down to 5" Um 4 and 5.

You pointed out that, mathematically, 5.5 is the mean on a 10-point scale. You can't rate anything 5.5, though, so you have to pick a mean rating. That can be 5 or 6. I prefer 5. I don't know where you get 4 unless you've just decided arbitrarily to think of 4 as being so-so.

Your last paragraph Hyomil + condition of keeping 50% of votes in 1-5 and the other 6-10 means use MEDIAN. Well the median of your current collection or the world's but median.

That's why I began that paragraph with "The median rating is..." :-)

Right, question number 2 is Mean vs Median. "Enjoyed a bit" does not mean Median or Mean. For all future discussion "so-so" will mean "enjoyed a tiny bit." "enjoyed a bit" is not the mean or median of the Enjoyment value.
I must correct my 3x2x2 idea because "Enjoyed a bit" is separate and has no "current library vs world" or "Mean vs Median" context. You partly corrected me Hyomil. Effectively that means 3x2x2 + 1 = 13 possible "averages".
Also "so-so" is not an Average because some ppl just enjoy every thing more and others guess better shows to watch or forget worse ones.

"If I did an analysis of all my ratings and found that my median rating was 4 or 6, or even 7, I don't see why that should influence which rating I use to convey that I found the movie so-so. As I said, if my median was 7, that would indicate I was good at choosing movies I would like. Should I then start rating all my so-so movies 7 just so that I can have a nice even distribution of my ratings with half on either side of 7? What would I gain from that?"
Most of 3x2x2 is for consistency. I want to be sure we are all on the same page and haven't missed an issue to discuss. You would need to move your "so-so" down to 4 or 3 to make the median sit at 5. One CRITERIA (gain) for picking Listal's Averages is "most space to place votes." That probably is the same as "equal weight in correlation for liked and disliked". The last criteria is almost the best one. My personal reason for the 7 "liked a bit" is to leave space for "the fans are idiots" 1-3. Or is it "leave a space for average?"


"I also don't see how my opinion of others who enjoy movies I don't rate highly factors in. Its the extremes that matter there. If rate a movie a 1 and someone else rates it a 10, that strongly affects my opinion of our compatibility. If I rate a movie as so-so and someone else rates it as 1 or 10, that doesn't have much effect on my opinion of our compatibility."
That literal means you will get more "so-so" in the top of your comparability top list. So yeah it matters. Unless you are talking about my first question of "Enjoyment vs Value vs Idiot/Troll avoidance." The third option may be moot, insignificant and complicated in which case i would like to leave it for later.

Following Paragraph Needs A Separate Thread! And i will start or point to one if anyone replies:
Afaik Listal's internal compatibility calculations makes 2 5 vs extreme votes worth the same as one 1 vs 10 vote. Listal is Linear not quadratic and the most common "correlation" calculation method uses quadratic. I mean "quadratic" as squaring the distance between two votes like 9^2 vs 4^2 + 5^2 which means 81 vs 25 + 16.


"In what way do you use two at the same time? 12 possible averages? Are you saying you choose your ratings with 7 as so-so but assume that others will choose 5 as their so-so rating and weigh that when deciding whether to accept their recommendations?"
Read my previous posts.


I think i have been going about this the wrong way. The point of this "average" discussion is to pick the "average" and in fact the scale that listal users should to most adjust for 1. a person's enthusiasm for that media or understanding of "little" and 2. the effects of remembering better shows better or just never watching bad ones. JUST HAVING ANY "Enjoyed a bit" agreed upon number solves #2. Having a few more such very concrete meanings like "couldn't stand to watch 30 min" or "could not stop watching" would help with issue #2. We can rely on people to put their votes in the right order or compatibles having the same order.


DON'T READ THIS PARAGRAPH. ITS INCOMPLETE. SUMMARY IS STANDARDIZATION IS ONLY PART OF AN ANSWER!
Issue #1 is harder. We need to help Mary find Pete. They are actually very compatible, or at least have the same opinions, but Pete is depressive and Mary loves movies alot more than him. They are both using 6 as "liked it a tiny amount". Since they are compatible they may have watched more things in common (as a % of their respective totals) and definitely have almost the same order to their votes. The same things goes for Appelton. He is also depressive like Pete. Unlike pete, Appys determined and has a huge number of votes on the low side.
The only idea i have aside from using uber-complex statistical math is to work with the idea that Order of the items is what matters. Btw this is a solution for #2 as well. Take all the items the two have in common. Number all of these items in the order of their rating by one of these people and again for the other. If multiple items had the exact same rating give them all the same number equal to the average these items would have taken. Also the items after these start with the number after the last of the multiple would have taken.
Take the differences of the numbers of each item and average these differences.
Here is the example with the faults already approaching.
Mary Pete Ruk Lit M P R L Diff P R L
a 10 9 4 10 1 1 1.5 1.5 0 .5 .5
b 8 8 4 10 2 2.5 1.5 1.5 .5 .5 .5
c 7 8 3 6 3 2.5 3 3 .5
d 5 3 2 2 4 4 4 4 0
e 3 1 1 1 5 5 5 5 0
1/5 1/5 1/5
Obviously that has got problems. In fact i think Lit may be a problem that would get fixed if Standardization instead of the numbering was used. Infact the numbering may be a method of STD but not the better, more common one. R vs M is still a huge problem.


No wait, this is just "most standardized." Lets be sure and start from the top: Find people who when i add as "follows" will only improve my Followeds Top Tv list. By improve I mean bring that list closest to what I would have if I had watched everything. Actually past the "liked it a little" point it wont matter. That means heavy Normalization to find compatibles would require applying the same normalization on the "followed's" ratings before adding them to the Toplist. Tom has work to do. Omg headache will be back.

And then i realize: Q 1's answer is obviously "that which i wish i would be recommended if i haven't already watched before." The issue of Value vs Enjoyment is thus resolved. More on this later.

10/10: Excellent
9/10: Wonderful
8/10: Great
7/10: Good
6/10: Nice
5/10: Average
4/10: Mediocre
3/10: Bad
2/10: Worse
1/10: Horrible

5 points are the middle of the 10 points scale. But in terms of quality rating, I think that 50% of perfect is not average. It's like in school or university, when you get only half of the maximum points you are not average - you're at least below average if not bad. The most movies that I watch get 7 points, so this is the average rating for me, but it doesn't mean that the movie is average - actually these are good movies. I rarely watch movies that I don't like, so I don't give 6 points or less that often. But that doesn't mean that these bad movies aren't there, I just don't watch them. That's my philosophy... ;-)

the recommendation system needs ratings. its also what makes listal unique. Lots of sites do it your way or can be wrangled to.

Anautix; its 'median' not 'average'. The average of 10 9 8 8 7 7 7 4 3 1 is 6. The median is 7.

Absolute earmarks are the best system. They are easy to rate with. Let tom add some standardization maths to the find similar people system.

I must say I like the words used by another site to "describe" movie and TV ratings. Of course, you might want to use any other choice of words of your liking.

10: Totally Ninja
9: Superb
8: Great
7: Good
6: Fair
5: Meh
4: Poor
3: Bad
2: Terrible
1: Weak sauce :(


@Gary
You don't have to rate anything if you don't want to. You can ignore the Rating system completely, if you wish.
As to me, I think of ratings as a reminder to myself, or a personal journal if you will. I don't mind sharing my ratings with others, but sharing is not my primary goal.

If you read one review it´s implicit that you trust the author opinion or have some kind of similar opinion, but I cannot see any good reason for rating.