A Look at Twitter in Iran

By Mark Evans - Sunday, June 21st, 2009 at 8:25 am  

Over the past two weeks since the Iranian presidential elections and the subsequent protests about the results that saw incumbent Mahmoud Ahmadinejad attract 66% of the votes cast while rival Mir-Hossein Mousavi received 33%, Twitter has emerged as a key tool used by many Iranians to talk about what’s happening.

Twitter’s value as a communications tool was highlighted when the U.S. State Department asked Twitter to postpone scheduled maintenance so Iranians would have access to the service at a time when thousands were taking to the streets to protest.

To get some insight into how many Iranians are using Twitter, and how Twitter is being used to talk about the presidential election, Sysomos analyzed its database to pull some of the more interesting facts.

As a starting point, there are now 19,235 Twitter users in Iran, compared with 8,654 in mid-May. (Note: We determined the number of users by reindexing over 13 million Twitter accounts. Location is based on the information provided in a user’s profile. Update: Locations are based on information as disclosed in May for users who joined before June to avoid counting those who changed it later to Tehran).

Next, we looked to see when Twitter accounts were created over the past 15 months. As you can see from the chart below, the number of Twitter users in Iran has grown strongly in 2009 with the most active months being March and June when 9.81% and 9.93% of all Twitter accounts were created respectively.twitteriranchart1

Here’s a table that shows when Twitter accounts were created in Iran.

Picture 3-51

Of all Iranian Twitter users, 93% are located in Tehran, while 0.94% are in Shiraz and 0.83% in Mashhad.

Iran Chart-1

To get a sense of what Twitter users in Iran were talking about before and after the presidential election, we created two buzz graphs – one on June 11 (the day before the election) and another done on June 19. (A buzz chart shows the major discussions taking place, and the associations between these topics.)

On June 11, there was a lot of conversation about presidential candidate Mir-Hossein Mousavi with the strongest association to “Iran”, “freedom”, “Iran” and “vote”.

Buzzgraphjune11

On June 19, the major conversations coming from Iranian Twitter users involved the keywords “Iran”, which had the strongest links to “Mousavi”, “Tehran” and “Protest”. This reflects the protests taking place in Tehran by Mousavi’s supporters.

Buzzgraphjune19

We also looked into the Tweets using the query “Iran Election”. On June 11, 51.3% of all these Tweets came from Iran, while 27% came outside the country, and 21.6% of Tweets did not include a location.

Picture 3-52

On June 19, 40.3% Tweets about the election came from outside Iran as the media and blog coverage about the protests in Tehran attracted global attention. Meanwhile, the percentage of Tweets from Iran fell to 23.8%, while 35.7% of users did not provide a location. The lower percentage of Tweets from Iran could also could be due to reports the Iranian government is blocking access to the Internet and Twitter.

Picture 2-83

More: The New York Times has a story looking at how Twitter is being used to organize political protests in Iran, and six lessons to be taken into consideration, while techPresident has a post looking at how reTweeting is a form of reporting.

Tags: , , ,

15 Responses to “A Look at Twitter in Iran”

  1. Pablo says:

    It should be noted that many Twitter users are changing their location to Tehran or creating accounts with Tehran as the location in order to complicate the process of tracking Iranian users. Is it possible to quantify this?

  2. Green_Rev says:

    Nice work, but what you are not taking into consideration is the effort being made by supporters of the Iranian protesters to get non-Iranians to change their twitter location and time zone to Tehran in order to prevent Iranian authorities from locating and arresting real protesters in Tehran.

  3. Karen says:

    I think at this point it would be impossible to tell how many of the post-election-created accounts are truly from Iran. But I do believe you can use some creative math to at least identify a certain subset of users who would most likely be from outside the United States.

    It wasn’t suggested (nor believed necessary) that anyone outside of location change their local to Tehran until after the election fallout. You give the following two numbers:

    Current Twitter users in Iran: 19,235
    Twitter users in Iran in mid-May: 8,654

    I would suggest going back through the index, if you have the data, and see how many of the 19,235 purported Iranian Twitter users have accounts that were created prior to June 12. I would theorize, with confidence, that it’s highly unlikely anyone outside of Iran felt motivated or inspired to change their location to Iran prior to June 12. Therefore, if you look at the number of accounts located in Iran created prior to June 12, presumably, that total will be greater than 8,654 (the total number of *actual* users in Iran prior to June 12.

    In conclusion, the difference between those two numbers would be the minimum number of users who are likely *not* actually located in Iran. I think it is impossible to tell how many of accounts created post-election are actually in Iran without painstakingly analyzing their updates and trying to glean from the content whether they are tweeting from the midst of the crisis or from afar. But at least it would give a greater depth of perspective to have those additional numbers.

    And another interesting anaylsis would be to look at these numbers again, say, a month from now or a year from now, and see how many accounts created from, say, mid-May to the end of June are from Iran versus somewhere else, as I am guessing at some point, most people will want to modify their bio information to more accurately reflect where they are. That would also give yet another layer of perspective to the numbers.

    Just my opinion!
    Karen – @LegalTranscript

  4. RevengaBE says:

    ridiculous!

    from the start of the protests non-iranians have been changing their settings to Iran-teheran +3h30 GMT… and that’s the data this whole study is based on.
    ‘Location is based on the information provided in a user’s profile’

    This has no value whatsoever.

  5. nilesh says:

    Pablo, Green_Rev and RevengaBE: We actually looked at the user profile data as disclosed in both the months of May and June. Hence we were able to distinguish between those who were always from Iran and those who changed their locations later.

  6. steveolson says:

    nilesh:

    a) Thanks for the amazing post!
    b) I am still confused about factoring out the locations. If you are looking at new accounts in June, does this take into account that non-iranians were likely opening new twitter accounts and setting their locations to iran? Or is the only data way you’d catch location-switchers would be if their accounts existed in May as well?.

  7. evano says:

    Do you have some method of accounting for all the people in Iran who, motivated by safety concerns, claim that they are not in Iran? The prevalence of proxies, Tor, and other means of evading the state’s control over the Internet in Iran even before the current unrest argues that there are significant numbers of Iranians with incentive not to disclose their true locations.

    Then there are the people — in Iran and out of Iran — who don’t put any location or time zone at all. Unless you have some sort of metric which allows you to determine the portion of the Iranian Twitter population who don’t reveal a true location, then your numbers are valueless, despite your attractive infographics.

  8. nilesh says:

    @steveolson: We looked at data from may where available. While this does not account for new users joining in June and changing their location, the number of such users is small. Most users who changes their locations were more savvy Twitter users as opposed to new ones.

    @evano: This analysis is based only on self-disclosed information. There might be more Twitter users from Iran who did not disclose their locations, but there is no way for anyone to know that. But there are other independent studies to suggest that Twitter was not used that much in Iran.

  9. Scott Watlsh says:

    Thank you, I would now go on this blog every day and check for new updates!

  10. Jenny says:

    I think you also have to consider those people who are in Iran and cancelled their location trying to hide.

  11. Amin says:

    HI all,
    does any body Know what the percentage of Iranian users are in jun 2010?

  12. vpn service says:

    The only reason you would need to use a VPN connection to your work is if you need to be able to access resources at your work like files or databases. A VPN basically gives you secure network access to your work without being on your work computer. A VPN is not something you will just be able to setup at home on your own and connect to your work. You will have to coordinate with your IT department to set this up.

  13. Tim Jennings says:

    @nilesh Thanks for some interesting data. Do you have a view on the number of people outside of Iran who did change their location to Tehran?

  14. Mark Evans says:

    Tim,
    I don’t think we have that kind of data but it’s an interesting question.

    Mark

  15. Normally reasonable? Aside from the Barak people here have long halted to have a discussion – rather they harass other posters in lieu of staying with the challenges. And if you are reading this, they have created a screen name on here similar mine to minimic & harass. That isn’t dialogue either.