Around 6 years ago, while pitching for a digital marketing campaign, I decided to present the prospective client with data on Internet usage in Iran. No big deal I thought. I'll just put one or two stats on a PowerPoint presentation slide after googling the data. It didn't quite go according to plan.
It's not that I didn't manage to find any statistics, I found a bunch of figures but with either no source or with circular sources. The range of the numbers was quite varied, and some figures were suspiciously rounded such as 30 million as if plucked from mid-air.
What I found troubling was that there were numerous media outlets usually citing the highest figure with certainty, to back up a story. "Iran's internet users who now number over 30 million are a force to be reckoned with bla bla bla"...you get the picture.
This was in 2008. I finally turned to the ITU (International Telecommunications Union), a UN affiliated body, researching information communications technology on a global level. Their latest figures at the time were for the previous year, i.e. 2007 and at that time they stood at 11 million. The ITU has since revised the 2007 figure to under 8 million for individuals using the internet (making use of the internet at least once in the preceding 3 months when questioned). But where did their figures come from?
Well at the time from the 'Mokhaberat', which doesn't have the same sinister meaning it has in some other Middle Eastern countries, it simply refers to Iran's main (at that time state owned) telecommunications company. In fact I recently noticed that the ITU have mistakenly translated Iran's Mokhaberat company as the 'Ministry of Information' for their Iran data source, as this is the common definition of the term in the Middle East region, but crucially not in Iran!
But although these figures seemed more realistic, they were also a bit suspect. For example the following year the ITU figure jumped to over 20 million and then dropped back again to 13 million the consequent year. These figures have all now been revised by the ITU. At the time, I telephoned a close friend who worked as a top executive in one Iran's best knows ISPs (Internet Service Providers). He was able to tell me that the 2008 figure of 30 million was a wild fantasy. He indicated that the figure of 11 million may also be high, but he had no data to back up an alternative figure.
Fast forward to 2014
Ever since that time I have kept a close eye on Internet usage stats and can shed some light on them.
Ignore any stats on Iran's internet usage from the 'Internet World Stats' website www.internetworldstats.com
Now this source ranks very high on Google when you search for the number of internet users in Iran, hence it's picked up by many media outlets, who then go on to cite each other, despite the original source not being very transparent. You see the source of the Iran internet usage data on Internet World Stats is....wait for it....IWS, which as you may have guesses stands for Internet World Stats. Yes, the source for the data is the data itself! Okay I hear you say, but surely no one is going to take a site which has an ad of scantly clothed Filipino girls on its homepage seriously, are they? Hmmmm well you might be surprised.
For example, a Wikipedia entry (a source many journalists will turn to) on communications in Iran says:
"In 2012, there were 43 million internet users in Iran"
Where is the source? The source cited is an article on Peyvand News (payvand.com) which actually refers to 42 million internet users, so 1 million extra users seem to have appeared courtesy of an over enthusiastic contributor to Wikipedia. What is the source for the Payvand article? An article in Mehr News (mehrnews.com). What is the source for the article in Mehrnews? You guessed it it's Internet World Stats. IWS.
The Statistical centre of Iran
SCI sources of data
The statistical centre of Iran published a survey on Iran's internet usage in 2010. This was the last specific Internet usage survey. It is a bit ancient now in Internet terms, and hopefully they will do another one soon. At that time they found that around 14.7% of the total population were Internet users (had connected to the Internet at least once in the preceding 12 months - for the definition of their survey i.e. different from ITU's definition of once in the preceding 3 months). In absolute terms this equates to 11 million Internet users in 2010.
Apart from the 2010 Internet survey, the SCI, similar to most countries, carries out an extensive nationwide household income and expenditure survey every year. In that survey, there is a section on a number of assets held by the household. One of these questions in the past few years, has been whether the household make use of the Internet in the house. The specific question is this:
"Does the household make use of the following assets at home?" And then one of the checkboxes is the "Internet".
Using the latest SCI raw survey figures, which was for the year 2012, I weighted the raw figures using official SCI weights and calculated that nationwide only 12.7% of households said that they had made use of the Internet at home in 2012. I also broke down the overall figure by expenditure deciles and found that this figure jumped to 34.7% for the richest decile and was 0.4% for the poorest decile. In fact I found a strong correlation between Internet usage at home and household income (or strictly speaking aggregated expenditure).
But let's not get distracted into looking at characteristics of Internet users in Iran. With time and better infrastructure such characterisations will not be so apparent (I hope!). But it's a also a warning to journalists and Iranians who reside in North Tehran, who so often extrapolate data and opinion for the whole of the Iranian population based on their experiences in the richest part of Iran.
We must remember that the figure I have calculated refers to households and not individuals. Also it refers to the use of the Internet within the home and not outside the home, i.e. at work or Internet cafes. Most respondents probably did not even consider their mobiles as an Internet gateway in the questionnaire.
Okay so where does this leave us with the SCI, well basically apart from the 2010 survey, we don't have much to go on. The ITU use the SCI figures for their reporting of Internet usage figures in Iran.
For 2011, the ITU have done their own 'estimate' and state Iran's Internet penetration as 21% (up from 14.7% the year before). For 2012 the ITU cite their source as the SCI and give the figure of 27.5% and for 2013 similarly citing the SCI as their source they give the figure of 31.4%.
I have yet to come across any published report by the SCI confirming the figures for 2012 and 2013 and more importantly how they calculated these percentages. i.e. this may have been communicated to the ITU but as far as I can tell these percentage have not yet been published along with the methodology for the public to see. The World Bank figures for Iran are also using these figures from the ITU (or should I says SCI). If someone from the SCI is reading this, I would be most interested in knowing where the figures for 2012 and 2013 come from.
Iran's ICT Ministry
Iran's Information and Communication Technology Ministry (ICT) is in charge of a state organisation confusingly for our purposes called the ITC (Information Technology Organisation - Yes I know the acronym no longer matches the name). This organisation is in charge of planning and developing Iran's IT network. It has an affiliated division responsible for reporting Internet usage statistics. This division is called MATMA. Okay, with the introductions out of the way, let's get to the point.
MATMA's ( http://www.iriu.ir/matma/ ) approach to reporting Internet usage is to simply sum up all the Internet subscribers. i.e. all Internet enabled mobile subscribers + all dial up subscribers + all fibre optic subscribers + all ADSL subscribers + Wimax subscribers etc.
It then uses an algorithm to multiply the number of subscribers to arrive at a total number of Internet users (adopting the logic that one subscriber is equivalent to more than one internet user). It may be that they only multiply certain subscription types such as ADSL but not mobile subscriptions for example. But I have seen no documentation on this, so it's only a guess. Their publicly available methodology document does not state the algorithm for multiplying their number of subscribers into internet users. I have found no explanation of the multiplier whatsoever.
A simple division shows that it currently stands at 1.74 for the national Internet usage figure. i.e. currently the number of subscribers nationally x 1.74 = Total number of internet users. But of course an algorithm is being used to arrive at the Internet users figure and is not a simple multiplier.
MATMA acknowledges that this methodology is flawed but argues that it's simple to understand and is superior to the SCI method which relies on surveys which are complicated and very hard to carry out regularly.
The flaws are obvious. One person who has an ADSL subscription, a mobile subscription and a WIMAX subscription will be counted as 3 subscribers. Furthermore, the multiplier will then increase the Internet usage figure further. So 3 subscriptions from one person may conceivably be transformed into 5 Internet users.
MATMA argues that this doesn't matter, as the Internet penetration rate should not be limited to being under 100% and by adopting this approach they acknowledge that one day their penetration % will be over 100%. Their philosophy is to keep the the figure simple to compile and simple to understand.
This is fine, but I think most observers are after a figure which gives an idea of how many 'people' make 'regular' use of the Internet. As mentioned before, regular according to the ITU is at making use of the Internet at least once in the past 3 months.
The problem with the MATMA figure is that the media often cite this figure as Iran's Internet penetration rate without caveats. Most readers then interpret this as the number of people who use the Internet. It is also not comparable to the methodology adopted by other countries.
MATMA's latest figure for Iran's penetration rate as this post is published was 49% for 2013 which according to their own figures is down from 61% the previous year of 2012. This seems a bit strange, either they have made modifications to their methodology or subscriptions have really gone down. But in their time series data on their homepage these are the stats which they display.
MATMA justify the global validity of their methodology by pointing to the highest international internet penetration rates, in countries such as Norway, with a 94% Internet penetration rate in 2013.
They point out that as almost 19% of Norway's population is under 14 years of age and almost 16% over 65, then their penetration rates must also include duplicates and therefore in order to make Iran's internet penetration rates comparable, Internet penetration rates should be reported using the methodology MATMA have adopted. This reasoning involves a few leaps of logic.
A look at the ITU figure for Norway shows the source as Eurostat, and in turn their source is "Statistics Norway". A quick look at their Internet survey on the Statistics Norway website, and it becomes clear that their penetration rate is reported for the ages of 16 to 79. They also go on to report that for the near future, Internet penetration is not going to change. This makes it clear that they are not going to report a figure of above 100%.
It is also worth noting that young toddlers in high Internet penetration countries are now making use of the Internet. My 4 year old son has been using apps connected to the Internet since the age of 1. Similarly the vast majority of people aged over 65 years are also using the Internet, as subsets of Norway's survey makes clear.
In short, MATMA has adopted a methodology which is simple but ad hoc and not suited to arriving at an accurate % of Internet usage.
So what figure do we go with?
With the SCI not carrying out Internet surveys regularly and MATMA's methodology questionable, what figure do we go with? The answer is that until a new reliable survey is carried out by the SCI, there is no reliable figure to go with. Looking at the figures by MATMA and SCI we can say that the Internet penetration rate in Iran is somewhere between 25% to 50%, and probably more likely to be between 30 to 40%.
These wide ranging figures may not help a journalist or an academic who just wants to cite one number, but it is important to acknowledge the reality that a single reliable figure does not exist at the moment. If you are really keen to report one figure, then it's probably best to report the ITU (sourced from the SCI) figure 31.4%. Although as stated before I have seen no publication or documentation reporting or showing the methodology for arriving at this measure by the SCI.
What you shouldn't do is to cite figures from publications whose sources are not clear, which is what's happened in the case of the Wikipedia entry on the subject.