Some headlines grab your attention.

Sometimes, for the wrong reasons.

When MediaTel's newsletter landed in my inbox this afternoon, the number one story was "One in three Millennials watch no broadcast TV".

Obviously, I clicked straight through to see the article. Sure, its from the US, where the TV market is very different to the UK, but what does this massive statistic mean? Is it accurate? Who are these 33% of "Millennials" — are they living in accomodation where they can't get TV? Can they not afford subscription TV services? What is going on?

It would seem that, whatever the reason, online video is what is filling this hole in their lives that TV is leaving behind. So I had a look to see if there was any more information about the study out there.

The top result on Google (excluding "news") was a GigaOm post, echoing the same headline. So clearly, this isn't an issue with the MediaTel staffers misreporting a study.

Mashable, however, went with a slightly different headline; "34% of Millennials Watch More Online Video Than TV".

Well, thats quite a different story, isn't it. From "No broadcast TV" to "Less broadcast TV than online video."

Helpfully, they also provide the data behind the statistic;

So now we have 10% of Millennials watch no broadcast TV. OK - still a pretty big number, mind you. But quite different from a third. That is 1 in 10 of 18-34 year olds not watching TV. Based on a sample of 982.

Now, apparently 99% of US households have a television. So I doubt that it is lack of access to TV.

But there is another weaselly word in there — broadcast television. Thats a word that could have a number of different meanings, and without knowing the exact wording of the questionaire, its hard to say what it really means.

Maybe it means that 10% of millennials in the US say that they are only using their TV set for playing video games, watching DVDs and streaming online video.

But maybe it means that 10% of millennials in the US say that they only watch TV when its either pre-recorded on something like a DVR/TiVo — that they aren't watching "live" TV, but that they do watch (what I would call) "broadcast" TV content.

So, what does that mean?

There are a few ways that the research itself might be biased. I don't know how it was carried out, but if it was done online, then there is the risk that heavy internet users are over-represented in the sample. But lets assume that this hygeine factor has been accounted for by the research.

Lets also gloss over the fact that the way people want to represent themselves isn't always the truth. The average person probably doesn't want to admit to themselves that they watch more than 4 hours of TV every day — so they are unlikely to admit this to a faceless questionnaire.

Then there is the fact that saying and doing are quite different. If you're asking people what they do, then what you are really asking is what they remember doing.

You are more likely to remember something that you thought about doing than something you didn't think about doing. (eg. "What was the last film you saw at the cinema" vs "what was the last film you watched on TV" — chances are that the first question is much easier to answer, because its quite likely that you didn't make plans to watch a film on TV - you switched channel, it was on, you watched it. But for the cinema, you had to make plans, check listings, agree on a film to see — maybe where to see it, and at what time. Each step helping to reinforce the memory of what you were doing — before you had even watched a single frame of the film.)

An article over at The Economist ("The lazy medium") includes a quote from Sarah Pearson at ACB Research (who my employer has worked with on some TV viewing projects.)

This helps explain one of the oddest and most consistent findings of television research: that people seem unaware of their own behaviour. In surveys they almost always underestimate how much television they watch, and greatly overstate the extent to which they watch video in any other form (see chart 4). In particular, they underestimate their consumption of live television. One of Ms Pearson's subjects, a 27-year-old man, claimed to watch recorded television 90% of the time. In fact he watched live TV 69% of the time. He was probably not so much fibbing as misinterpreting the question. When asked how he watched television, he gave an answer that described his behaviour when he was alone, and thus did not have to compromise. But most of the time he watched with other people.

We have seen this effect time and time again — what people say and what they do is not the same. A few years ago, when the "TV revolution" was all about PVRs/DVRs/TVRs ("Sky plus" to most people), the fear of the advertising industry was ad-skipping. People claimed to only ever watch catch-up TV, and always skip the adverts. But the reality was (and still is) that 90% of TV viewing is "live", broadcast TV. (An interesting finding of our own research was that the amount of ad-skipping was outweighed by the increase in the amount of TV viewing that we saw in PVR owners.)

When you watch something you have recorded, you are choosing to watch it. That is easy to remember. When you watch something that just happens to be on TV that you have flicked over to, you haven't given the same amount of conscious thought to it. That makes it harder to remember.

So, my guess is that the truth behind the headline isn't anything to do with "new" behaviours around video-on-demand and streaming services, but an older, familar truth; people aren't very good at reporting on their own behaviour.

But thats an issue for researchers — or people who are digging for truths and insights from the statisticsand not taking the headlines at face value. The broader issue, as we turn to the internet for more and more of our "knowledge" is that for many publishers, the real friction is between viral and true.