Home > Science Journalism > Does predictive texting make children do more mistakes?

Does predictive texting make children do more mistakes?

The news

Last week, there was this article in the Hindu, which talked about how the predictive texting feature of mobile phones leads to errors that affect relationships. I do not know if that is true, because I would assume that any such error can be easily explained to the other party in the relationship. But there was something else more curious about the article. What was it doing in the Science section of the newspaper? And then I noticed that it had the following sentence.

A study in 2009 found predictive text messaging changes the way children’s brains work and makes them more likely to make mistakes generally.

This sentence made me curious. Is it really possible to tease out such a connection from the complicate causes and effects of human behaviour? Did the study really find such a result? I tried googling a bit and I found some articles from 2009 which covered this study. Let me quote one of them.

In a groundbreaking study, Professor Michael Abramson analysed the mobile phone use of children aged between 11 and 14 and their ability to carry out a number of computer tests.

A quarter of the children made more than 15 voice calls a week and a quarter of them wrote more than 20 text messages a week.

When researchers studied the way in which the children handled IQ-type tests they found that increased mobile phone use appears to change the way their brains work.

Prof Abramson, an epidemiologist at Monash University, Melbourne, Australia, said: “The kids who used their phones a lot were faster on some of the tests, but were less accurate.

“We suspect that using mobile phones a lot, particularly tools like predictive texts for SMS, is training them to be fast but inaccurate.

“Their brains are still developing so if there are effects then potentially it could have effects down the line, especially given that the exposure is now almost universal.

Some questions

The other two articles did not give any more information. None of them told me the number of children tested. I tried to get the original paper from a couple of my friends, but guess they don’t have access. But that need not prevent us from having a discussion on it. The article says, a quarter of the children wrote more than 20 messages a week. Frankly, 20 messages a week is not much. That is less than 3 messages a day. Some of the questions that immediately popped up in my head were these (I can keep adding to the list, but consider this be a sample).

  1. How many children were part of the study?
  2. Were the children monitored for the length of the messages they send? Were they just sending messages like “I am at school”, “I am on my way home”, “Happy Birthday” etc or were they sending out long messages?
  3. Are the kids really using “predictive texting” or do they prefer to key in each letter of the word individually by tapping the keys multiple times?
  4. What kind of IQ tests were done?
  5. Most importantly, how did the researches control for other factors to conclude that it is indeed texting that is causing the increased errors?

The first few questions are obvious, but let me explain the last point a bit. It is a common error to confuse correlation with causation. Just because two things always happen together, does not mean one causes the other. For example, just because two of your friends come to college daily at the same time, does not mean, one talks to the other to ensure that their arrival times are synchronised. It could just be that they both live in the same area, from where there are no buses to college, and so both wait for the same train every day to board. Thus their arrival at the same time (correlation) is not caused by either of them, but by a third factor of their using the same train.

Similarly, just because kids who are using texting on their mobile phones do not do well on IQ tests, does not mean texting causes them to do badly on IQ tests. An immediate thing off the top of my head is that this could be because those who are texting are not attentive in class (they are obsessed with the mobile phones) and thus do not give enough attention to their academics. This leads to lower IQ.

I will be the first one to admit that the previous conjecture is riddled with holes (after all sending just 3 messages a day cannot by any means, be called “being obsessed”). I do not suggest it seriously, but am just trying to point out what other factors could be possible. There are numerous other factors that could show such a correlation. A proper study would control these factors. In case of my specific theory about lack of attentiveness in school because they are on their phones, it must be ensured that the study takes into account the times when the child uses the phone.

But all this this assumes that there is indeed a correlation. There are too many pitfalls in the way the study could conducted that a false correlation can appear, especially when the plausibility of the mechanism is so tenuous. But to see all that, I need to see the original paper and I could not.

Somebody saw the paper

But one person has seen it and written about it. He makes a good point about accuracy and speed. But he also points out that the way the texting of the children was analysed was with a questionnaire. I am not kidding. They just asked children an estimate of the number of messages they were sending and did their study. It was not even asked whether they use that predictive texting or not? Is that not weird? But there is a final death blow to the whole paper.

He also points out something else. Though it appears on first impressions, that predictive texting leads to errors, does it really do that? Imagine you dont know the spelling of a word, and key it in wrongly on the mobile, you will end up with a junk word. So you have to do a few backspaces and then type it out correctly again. Such a mechanism only penalises an error and does not encourage it? So if at all anything, such texting should encourage children to get their spellings right. There goes the fundamental premise of the whole study.

Go read that whole post. After destroying any case for using “predictive texting” instead of “word completion”, he also questions the statistics involved. They don’t look good.


It is important to learn the lesson that whatever appears in a newspaper is not true (even if it is The Hindu). As with any field, mediocrity is rife in Science too. It is the newspaper’s job to filter out such stuff. But then again, there is mediocrity in journalism too. So it is upto all of us to be aware of such problems (Oh yes, there is mediocrity in blogging too, you are probably looking at one such example :-)). The subject of the study and the results are not very consequential (except for the money spent). But it provides a good case study to understand what could go wrong, and what we need to look out for. If we learn that, then the paper, even though faulty, would have served some tiny purpose.

  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: