Thursday, August 05, 2010

Dodger Success vs Blog Traffic: A Graphical Dissolution

After a few email conversations with Orel and Sax regarding the impact, if any, of the Dodgers' on-field performance on SoSG readership, I decided to take a closer look. And since you readers have consistantly demonstrated your keen understanding of cause-and-effect relationships (case in point: blaming Leonard Nimoy for the Dodgers' July), I thought I would share my analysis for your consideration and critique.

Hypothesis: The better the Dodgers' on-field performance, the more interest the team generates and thus the greater the traffic to Dodger blogs such as SoSG.

To test this hypothesis, I plotted SoSG's weekly pageload traffic (dotted lines) against the number of games over 0.500 (solid lines) for the entire year - both in-season and off-season. And I've done this for both this year (in blue) and last year (in red). This is what it looked like (through Aug 2):

(click image to enlarge)

Based on this graph, I made some observations/conclusions. Let's start with the obvious ones:

  • Blog traffic is higher during the season vs the off-season. Duh.
  • Traffic is highest during the playoffs. Duh again.

Now, to truly evaluate the validity of the original hypothesis, one would have to isolate the impact on traffic, if any, of the Dodgers' success from that due to other factors, such as organic readership growth, off-field Dodger news, or a one-off popular post. To do this perfectly is of course impossible, but with the concept in mind, here are my next levels of observation with my corresponding conclusion:

  • Observation: There was a significant year-over-year traffic increase from January through March, before the season started. Comparing March 2010 YTD with the same period last year shows over an 80% increase. While some of this may be due to off-season news (McCourt divorce?), raised expectations, or other factors, we know it can't be due to actual team performance. And the magnitude of the increase suggests, to me, evidence of at least some organic readership growth*.
  • Observation: During the first two months of the season, 2010 tracked almost identically to 2009 (including a big spike in mid-May of both years). This in spite of the Dodgers' decent-but-far-from-as-good on-field start this year versus last year. To me, this could be interpreted as supporting the idea of organic growth. That is, if one believes that traffic has some dependency on team performance, the fact that the team performed far worse during the first two months of 2010 than the previous year yet still held steady in readership could be chalked up to organic growth. That's what I conclude.
  • Observation: Beginning in June, the 2010 Dodgers went into their tailspin, whereas the 2009 Dodgers continued to excel. During this period, 2010 readership fell below that of 2009 (incidentally, June 2010 was the first month that exhibited a year-over-year decrease in SoSG's brief history). I conclude that the Dodgers' June/July performance was so poor this year compared to 2009, that the difference overpowered any organic readership growth, causing a net readership decrease.
  • Observation: During perhaps the worst stretch of the Dodgers' 2010 season (the recent 6-game losing streak), traffic actually showed a notable uptick. Not sure what caused this...I'm thinking the trade deadline. Or perhaps misery has an inflection point beyond which we cease to separate and instead reach out for company?

That's all I've got for now, though I'm sure some of you readers have some thoughts. Let's hear 'em!

*I also think there's another factor at play here I'll call the "hard-core vs casual fan phenomena (HCVCFP)." I posit that the year-over-year growth in off-season readership reflects the organic growth in hard-core Dodger fans - i.e., those who stick with the blogs whether in-season or off-season. Once the season begins, those hard-core fans are of course still there, but the casual fans also join, clouding the YoY organic growth.

22 comments:

Greg Hao said...

I don't know about the rest of y'all but I'm just here for the snark and prizes. You could substitute the Dodgers for the Expos and it wouldn't matter.

Heh, jk. I think it's an interesting observation about HCVCFP, and it makes sense, as people blog and comment and interact with each other (such as the most recent SOSG Fest which I couldn't attend due to out of town obligations), what started out as a bunch of strangers snarking and shitty performances on the field has turned into a community of sorts. This is something I've observed on many forums/messageboards over the years.

Steve K said...

I (and I assume most of the people who read this blog) have given up on the Dodgers' 2010 season. I just don't feel the playoffs are realistic at this point.

But I'm a hardcore fan (although not as hardcore as some). I will still keep up with Dodger news and analysis and will continue to read this blog. I think most hardcore fans that read this blog will, too.

I think the drop-off is from casual fans who started giving up on the team in June. Attendance at Dodger games is down a little, which probably translates other areas of fandom, including blogs. I think it stems from an undercurrent of discontent with the McCourts. People started to realize that the Dodgers weren't good enough to win it all and the owners wouldn't be able to do anything about it.

Steve K said...

By the way, I recently joined Twitter, which means that I get your post links in your feed. As a result, I can just read individual posts when they pop up. Previously, I'd reload the page a bunch of times a day. If this is happening on a wide scale, it could also impact page views.

Steve K said...

One last note: It also seems to me like Off-Day Puzzle participation has decreased, which is most likely a byproduct of the same casual fan malaise.

Neeebs (The Original) said...

The Chart is nice. The conclusions are thoughtful.

But....I still blame Leonard Nimoy.

Viva La Count Chocula!

Eric Karros said...

Yeah, J Steve you bring up a good point about pageloads being an imperfect metric for traffic. It's also worth noting that a load of the 'post a comment' page of a post (as opposed to the main page) doesn't register.

Greg Hao said...

look at j. steve trying to pump up his comment stats by posting not once, not twice, but thrice!

on the effects of the attendance drop off, I think part of it has to do with the mccourt divorce, but I think that's really only the hard core fans like us who're paying it so much attention. For the casual fans, I think the drop off is mostly having to do with the fact that the economy is so shite.

Steve K said...

If I can milk out a few more, I might crack the top 10 commenters!

Steve Sax said...

J steve, when you pull up the posts through Twitter, does the counter still load in the right hand side sidebar?

Steve K said...

They do, but when I come directly to the page when Twitter notifies me, you only get one hit. Before, I might have refreshed five times between posts, resulting in six hits if the sixth had a post.

Steve Sax said...

Aha. Well, feel free to refresh anyway. Often. Please.

Steve K said...

I also realized I haven't clicked on any ads in a while... cha-ching!

Steve Dittmore said...

Another thought-provoking opportunity to bring academic theory to the Sons of Steve Garvey. I think EK's initial hypothesis is absolutely true, and it is recognized in the literature as BIRGing (basking in reflected glory) and CORFing (cutting off reflective failure).

A good, albeit outdated, review of the two constructs can be found at the following link:
http://www.athleticinsight.com/Vol5Iss2/FanPDF.pdf

I think it is impossible to isolate a cause and effect here. If we view page traffic as a dependent variable, the number of independent variables which may influence page traffic can be numerous, most of which were identified (winning, trade deadline, etc).

If the Sons were willing to do this (and the readers were willing to participate seriously), I would be happy to assemble a survey which would measure a number of constructs (fan identification, motivation, etc.) and run a regression with traffic as the DV to find out the degree to which these IVs impact blog traffic.

Sons - are you game? I'll give you the results if I can retain the right to turn them into a research article.

Steve K said...

Steve D: I like the line of thinking, but I wouldn't get started putting together a survey of this type. I doubt there would be much validity to the results.

The main issue is with the amount of information that I believe the Sons can provide. Your DV is time-series data on an aggregate level (all visitors per week to the blog). From what I understand, you'd need individual-level page visits as a whole (i.e., how many times do DB vs. MLASF visit per week) to match to the individual-level survey data. You could theoretically do this by IP address, but then you'd need every respondent to figure out the IP address for every device (home, work, iPhone, library, etc.) they visit the Sons with.

At that point, you'd also realize that there are a lot of other IP addresses that are unaccounted for because the sampling distribution would be out of whack. The only people who would complete the survey are the hard-core, regular visitors. I think you would get a lot of motivated fans who identify with the team. You'd need a much more random sample, including many casual fans who only stop by once or twice, to make any broad conclusions.

Hope I made sense...

Fred's Brim said...

I think I cracked my brain reading the last two comments. You guys are SMRT!

Eric Karros said...

Hey Steve D, thanks for the offer. It opens up a few potentially interesting possibilities.

I think, to one of J Steve's points, a balance would have to be considered and struck between, on one side, making the survey short and simple to gather a representative response from both casual and hard-core readers, vs detailed enough to actually produce worthwhile results. And I'd think we'd have to settle for analyzing both survey responses and blog behavior in aggregate, which though suboptimal I think could still be meaningful.

Let me think this through though...now that I know (or more accurately, now that I am reminded) you are qualified to create surveys and run regression analyses, I may use that knowledge.

QuadSevens said...

Reading all these comments makes me feel like I'm working.

karina said...

I want a shirt that says "I love SoSG's Eric Karros graphs".

Where were you when I took Statistics in college?

Steve Dittmore said...

@J Steve - You are correct, there is no perfect way to operationalize the DV. As I was typing the initial comment I thought we could measure web site hits on the day the survey was posted. This would provide interval level value. It would also mean the study would be a case study, and thus subject to difficulty with generalizations. We could ask an open-ended question like "how many times a day do you visit the Sons site?" But this is problematic because, using me as example, I mainly read it through RSS and don't actually view the page. These can be overcome by acknowledging several study limitations.

Steve Dittmore said...

@EK - let me know. There is nothing in the academic literature right now which discusses what you hypothesize. It could be a continuation of a study I published in 2008 using Josh Rawitch's Inside the Dodgers site as a way to demonstrate the importance of relationship building using an official organizational web site. I'll send a PDF of it to the Sons Gmail address.

We (I have a doctoral student at my disposal) could massage several previously validated scales to fit the Sons site. Sample size might be an issue, but again that is not insurmountable.

Steve Dittmore said...

@Fred's Brim - If I am SMRT, how come the Puzzles are so damn hard? I give up on those things.

Eric Karros said...

Cool, we'll be in touch professor!