Numbers can lie

“Without data, you’re just another person with an opinion,” said W Edwards Deming. Which is why extreme care is required about what kind of number one decides to throw out there. Even more important is how one decides to interpret that particular figure as it can have multiple interpretations.

Dissecting a fact or figure needs one to ensure that they keep the context in mind. The context is important because a story can be twisted if the context is changed slightly.

For example, figures collected by the Pakistan Telecom Authority (PTA) which are routinely quoted in technology- related articles. The PTA claims that the cellular teledensity in Pakistan is roughly 73% — the number of cellular subscribers divided by the total population. However, the figure is over-reported and the interpretation is misleading. Firstly, does this figure count active users? Or does it also have connections that are dormant? What exactly is a subscriber-- is it a person or a SIM/connection?

Secondly, does this figure also take into account multi-SIM usage? GSMA Mobile intelligence claims that Pakistan has roughly 80 million unique cell phone users, where a unique user may use one or more SIMs.

If you follow the GSMA route the teledensity would decrease to roughly 40%! Hence, the PTA numbers are not incorrect, they are simply misleading. The reality is that less than 50% of Pakistan owns a cell phone connection.

Similarly, the PTA potentially under-reports the number of internet users in the country. It tells us we have roughly 54 million users of broadband, out of which a whopping 51 million use mobile internet while three million use other connections. The issue is that these other connections are categorised as either fixed line internet connections (such as DSL or Fibre Optic) or wireless broadband (such as Wi-Tribe).

A DSL subscription is utilised by more than one person; usually an entire household. The same could be said for USB dongles and pocket Mi-Fi devices. If a household size is considered to be 4-5 the number of users would suddenly swell to over 10 million!

As corrective measures perhaps the PTA should include definitions on their website or add caveats that caution the reader about what the numbers represent. For instance, they could define what they mean by a ‘subscriber’ — is it one person or one connection. Similarly, they should define what they are taking into account when they report the ‘teledenisty’ of Pakistan.

Ideally a metric like ‘teledensity’ should be the number of unique users divided by the number of people who are older than 15 in Pakistan. The rationale behind the age limit is to take into account that it is less likely for individuals who are aged 14 or less to have a cell phone connection. This is subjective and you could argue to increase the limit to 18 or even drop it to 10; the standards used here is the one set by ESOMAR, which is an international organisation of the research industry.

In the case of internet usage things get a little tricky. The PTA could multiply the average household size with the number of fixed connections to arrive to a more well-rounded figure for users. Preferably they should conduct independent research that deduces the overlap of internet technologies (such as mobile internet, DSL, etc) to arrive at the ‘unique user’ figure.

Hence the current PTA numbers should be used as a proxy to understand the telecom trends and not be considered as the bible of telecom numbers. It is also important to note that no number is perfect and that researchers and statisticians always keep a margin of error.

Understanding the context behind any number can help dismantle a lot of mistruths and half-truths out there. In a world where we are constantly bombarded with information, we must realise that sometimes numbers too can lie.

Published in The Express Tribune, August 11^th, 2018.

Like Opinion & Editorial on Facebook, follow @ETOpEd on Twitter to receive all updates on all our daily pieces.