Damn Lies and Statistics

Damn Lies and Statistics or just fodder for rumination - I'm not sure. Recently I have been seeing several blog entries on numbers, traffic, blog size and / or type, etc. I became curious about how some available numbers stack up - and even more, how they stack up as a function of time. After all, I reasoned, on some level it is ultimately "time" that blog readers are investing in any particular blog.

Who is read "more", for example, Andrew Sullivan, Powerline, Protein Wisdom or Ace of Spades? Depending on how you want to interpret the numbers - and with the caveat that I'm no internet know it all - $20 says you probably guessed wrong. What if I told you that at least in one measure the numbers actually prove that both PW and ACE individually beat Powerline and Sullivan COMBINED!!

Can't be? OK, maybe I'm missing something but that is what the numbers seem to say. And if I am missing something, I apologize and I've no doubt I'll be corrected. But this isn't a ploy. It's just another way of looking at the available numbers that I could find through sitemeter.

That said, yes, there are different types of blogs. Clearly an Instapundit wouldn't anticipate people spending as much time on an entry as say a Powerline, right? That seemed to make sense to me, anyway. So, I was a bit surprised when that wasn't the case. And when you start looking at blog readership purely as a function of time - there may not be the gap between the large and medium blogs we'd expect from looking at NZ Bear's rankings.

It's important that I note here that I am in no way attempting to impugn the great work that Bear does. Nor am I claiming some great secret methodology that is somehow going to turn the blogosphere upside down. All I did was look at some very public numbers from a slightly different perspective. And I thought the results interesting enough to post.

Sampled blogs include many of the top twenty in addition to some I enjoy reading like Protein Wisdom and Ace of Spades. After all, it's my look at the numbers and the methodology is clear enough that anyone can crunch any available numbers they wish. I also won't be a bit surprised if someone wants to point out some shortcomings with the analysis. I'm a blogger not MIT. So, without further ado, here's a little different take on some numbers.

First the view to which we are accustomed, NZ Bear's current top twenty by traffic and then links:

TRAFFIC

1) Daily Kos :: Political Analysis and other daily rants on the state of the nation. 239440 visits/day (2)
2) Instapundit.com 137262 visits/day (1)
3) Gizmodo 117237 visits/day (116)
4) lgf: unh. can't help myself. 76206 visits/day (4)
5) Eschaton 63570 visits/day (5)
6) Wonkette 58641 visits/day (38)
7) Power Line 53456 visits/day (6)
8) www.AndrewSullivan.com - Daily Dish 52972 visits/day (8)
9) Gawker 52040 visits/day (189)
10) The Smirking Chimp 35306 visits/day (377)
11) Blog for America 33746 visits/day (308)
12) HughHewitt.com 29543 visits/day (24)
13) The Washington Monthly 29005 visits/day (12)
14) Defamer 28593 visits/day (625)
15) Belmont Club 22725 visits/day (25)
16) PoliPundit.com 17037 visits/day (141)
17) onegoodmove: I thought these things might be clues 15819 visits/day (1512)
18) Captain's Quarters 13933 visits/day (18)
19) Digital Photography Blog :: Digital Camera Reviews, News, Tips and Tutorials 13690 visits/day (3896)
20) Drudge Retort: Red Meat for Yellow Dogs 13402 visits/day (3034)

LINKS

Higher Beings
1.Instapundit.com (3984) details
2.Daily Kos :: Political Analysis and other daily rants on the state of the nation. (2303) details
3.lgf: unh. can't help myself. (2121) details
4.Talking Points Memo: by Joshua Micah Marshall (2099) details
5.Eschaton (1926) details
6.Power Line (1894) details
7.The Volokh Conspiracy - (1747) details
8.www.AndrewSullivan.com - Daily Dish (1653) details
9.DRUDGE REPORT 2005® (1570) details
10.Boing Boing: A Directory of Wonderful Things (1533) details

Mortal Humans
11.Blogs For Bush (1461) details
12.The Washington Monthly (1370) details
13.Redirect (1300) details
14.Dean's World - (1292) details
15.Wizbang (1262) details
16.Michelle Malkin (1251) details
17.LILEKS (James) The Bleat (1239) details
18.Captain's Quarters (1224) details
19.a small victory: Mark it 8, dude (1192) details
20.Slashdot: News for nerds, stuff that matters (1162) details

Now, before we get to pure time, which is the most interesting to me, here are many of the top twenty and some others of my choosing ranked in terms of page views per visit. There are many factors which might influence this but it at least begins to get to the question of how much time do people actually invest in a given blog on a particular day - or how engaged do they get on a single visit. This will be lower for a more frequently updated blog. But that is addressed in the last piece measuring overall reading time. Blogs linked above are not re-linked below.

Blog                             Page Views per Visit

protein wisdom	2.3
Drudge Retort	2.0
Winds of Change	1.6
ACE	1.5
Wizbang	1.4
KOS	1.3
Captains Quarters	1.3
Volokh	1.3
AsmallVictory	1.3
LGF	1.2
INDC	1.2
Wonkett	1.2
Belmont Club	1.2
Instapundit	1.1
Eschaton	1.1
Hewitt	1.1
powerline	1.0
Andrew Sullivan	1.0

Now to actual time. Not the average time any one individual spends on a blog for each visit. But on any given day how much "reading time" of the total blog reading population's time does a particular blog pull or consume.An arguable point for several reasons, it is one measure of "traffic" you don't usually see and the results are, if nothing else, interesting I believe. I took total number of average daily visits for a blog multiplied by the average visit length and did the math to get the result in hours per day "readers" spend on each blog studied. Some of the top twenty, Malkin for instance, don't appear to have public access to sitemeter. And any blog included or left out is not meant to imply any bias. I included what I could of the top twenty supplemented with some of my favorites. I apologize to some favorites I just didn't have the time to do. Maybe you'll find it interesting, but then, as when you read any blog, there are no guarantees that the particular writer isn't half nuts and wasting his, her or your time! ymmv

Blog                 Total Daily  Hours of Readership

KOS	399.07
Drudge Retort	335.05
Instapundit	228.77
LGF	211.68
protein wisdom	169.39
Wonkett	162.89
Winds of Change	155.02
Eschaton	141.27
Volokh	135.02
Captains Quarters	127.72
Wizbang	116.26
Belmont Club	113.63
ACE	112.97
INDC	58.00
Hewitt	57.44
AsmallVictory	56.49
powerline	44.55
Andrew Sullivan	44.14

To replicate the above numbers you need only take the average number of daily visits times the number of seconds per visit. That gives you total viewing time in seconds. Dividing that by 60 yields minutes and dividing by 60 again gives you hours. For example Powerline gets over 50,000 hits a day - but their average visit length is in the 3-4 second range. Contrast that with Protein Wisdom that might only get about 3,500 hits, but PW has an average visit time of 2:51 or 171 seconds. When you do the math - PW has more than twice the "read" or "viewing" time of Powerline. Just do the Math.

Lastly, if there are unavailable statistics or multi-site combinations I'm unaware of, I apologize. I was only working with what I saw and thought was correct. And as a conservative, either Kos spends all day clicking and reading his own site, or the blogosphere is full of some real left wing loons. Forgive the editorializing - it's late!

UPDATE: Initially in the example above I had PW at 2:11 or 171 seconds. Checking my numbers which I saved from sampling on Tuesday night the PW avg. visit length at that time was in fact 2:51 and the 171 seconds is correct. The general calculation example stands.

Comments

[chrys]

What ever you say - you lost me in the . . . .

[David Mobley]

Have you looked to see how many of those blogs use expanded entries? In my experience, one big thing that runs up the sitemeter "time" stuff on my own blog is when I write posts which have continuations. This forces my readers to view a second page to read what I'm actually saying, and reduces the number of 0 second visits.

Instapundit almost never uses continuations -- so the only time he would see visits longer than 0 seconds, I think, is if someone clicks a link away from his page and then returns within several minutes (In my experience, sitemeter seems to recognize the same ip address coming back if it happens within a few minutes as a single visit of a larger amount of time; if it's hours later, it's a distinct visit), or clicks another link to elsewhere on his site. In contrast, Evangelical Outpost (for example) uses lots of continuations.

What I'm saying is that I suspect that blogs that use lots of extended entries, or spend lots of time linking to themselves, would probably see higher average time per visit than those which put the entire entry for on the initial page. Have you thought of looking at whether this is correct?

Another factor also may be those blogs which use a lot of links elsewhere. They probably would see higher time per visit than those that don't (due to people going away on a link, and then coming back, which I think sitemeter counts as a single longer visit).

In summary, if I'm right, I would expect those blogs with the longest average time per viewer to be those which use lots of outgoing links, or links to elsewhere on the blog, and use extended entries. Those with the shortest would be the ones that don't link out much, and put everything on the main page.

[Jeremy]

I think David has a point...

But I think the answer lies with the "comments" as well. Neither Powerline nor Sullivan use comments on their sites--thus you never redirect to another page within their site. You just scroll. Perhaps that has an affect. That might explain (aside from shear readership) why LGF and KOS etc. have high numbers. Drudge Retort is a different type of site all together. Volokh does a lot of self linking--even though it doesn't have comments.

Possibly? Maybe?

[greg]

As per Sitemeter's FAQ:

Why do some of my visitors have visit lengths of 0:00?

That means the visitors are only staying to view a single page and then leaving. The only way that Site Meter knows how long someone is on a site is by the times of each page view. If they only look at a single page and then leave, we don't know how long they looked at the page. If they looked at two pages and left we would know they at least were on the site during the time of the first page view and the second page view. The difference between those two times would be the length of the visit.

So yes, any blog that is formatted so that visitors don't click around very much, if at all, are much more likely to have vastly lower reading times. To be able to compare reading times, a counter would have to keep track of the total time between the first instance of the domain being accessed and another domain being accessed (either through an outside link or through bookmark/favorite lists).

"Site Meter defines a 'visit' as a series of page views by one person with no more than 30 minutes in between page views."

and

"When you are browsing a site, every time you follow a link, it is treated as a single 'page view'."

I don't have the time to do the analysis, but if you compared the average amount of time spent at a site with the average number of page views, you'll probably get a pretty direct correlation.

As to why they don't measure the way I mentioned? Probably because keeping count for as many sites as sitemeter does would so overload the system that it would be ineffective. And so we are stuck with the 0:00 visitors, who could have spent 3 days sitting on the site, and never clicked a link or refreshed, so we don't know how long they accessed the site.

[The Commissar]

Jeremy nails it. He made the same point over at TTLB.

Comments increase page views. In Goldstein's case, his entries spark a LOT of engaged discussion.

Subject to the 'zero-second visit' arithmetic, your analysis makes good sense. And Jeremy hit on WHY certain sites have high views-per-visitor.

[MyManMisterC]

Hey thanks for adding to me the Carnival. My memories of carnival are fond, except for when I was in the 6th grade and I yakked on the tilt a whirl. If you saw the load I was sitting next to you would have understood that my stomach never stood a chance.

Anyway, hope you enjoy the Doormatt, I have been getting good numbers so far.