OPINION24 May 2010

The social media lies we tell

Tech blogger Robert Scoble has an interesting post today about the lies people tell on social networks. To be more precise, these are ‘lies by omission’ – that is, important things about ourselves that we choose not to share in order to maintain a certain identity online.

Scoble focuses on what he calls “the like economy”, powered by websites such as Yelp.com and Facebook, that allow people to display their liking for brands, shops, restaurants, products, etc.

While merrily ‘liking’ hip, high-end restaurants, Scoble noticed he wasn’t showing his liking some of his other favourite food stops.

“I like McDonalds and Subway. But I wasn’t clicking like on those. Why not? Because we want to present ourselves to other people the way we would like to have other people perceive us as. Translation: I’d rather be seen as someone who eats salad at Pasta Moon than someone who eats a Big Mac at McDonalds.”

Are there implications here for social media research and marketing? Is the data that advertisers are using – to profile users, identify potential customers and market to them – fundamentally flawed?

On the other hand, researchers have long known that what people say and what they actually do are often two entirely different things – making this a web 2.0 variant of a perenial problem. Can the same fixes used offline be applied online?

I’d love to hear your thoughts.

@RESEARCH LIVE

7 Comments

13 years ago

This is what separate data collectors from researchers. Researchers know exactly how to create valid and reliable research results from what appears to be random collections of highly skewed, vague, and horrible quality data. We have learned the appropriate techniques to apply to the data to tease out real results. Data aggregators may have trouble but researchers have trusted techniques.

Like Report

13 years ago

Thanks for your comment, Annie. Would you be able to share some of these techniques in the interests of furthering the discussion (and my own knowledge)?

Like Report

13 years ago

Well, there's the "technique" of looking someone in the eye and round the tummy AS they tell you their eating habits! And then raising ONE eyebrow. That generally works quite well :-) We know there can be pressure to 'impression manage' online. I don't think anyone can pretend otherwise anymore. The AQR/QRCA conference delegates last week in Prague debated some of these issues. If you have to use an online methodology, I think I could sum up their conclusions simply as "MROCs yes, twitter-trawling no".

Like Report

13 years ago

Great point, Brian. If we think social desirability bias is bad when a respondent is trying to impress an interviewer, imagine how extreme it can be when they are trying to impress their relatives, friends and business contacts. If we think extreme response style is bad in surveys, imagine what it must be like on social networks when people primarily share the good, the bad and the ugly.

Like Report

13 years ago

The techniques I speak of are well known by all researchers. Intense data quality process, precise sampling and weighting techniques, precise scaling and standardization techniques. These are the same procedures used for all research whether the data is sourced from focus groups, surveys, or anywhere else. Data is data. It's the skills and expertise that researchers have that turn it into research results.

Like Report

13 years ago

I think we will always have face to face qualitative research for the reasons just given. However, encouraged by Ray Pointer, I think that much of our present survey techniques will disappear. My current thinking is that the pressures we are facing in online quantitative research (poor data quality, respondent bias, speed over accuracy) will become much more apparent, and will morph into pressures for the use of social media research instead. After all, the data is already there, and it reflects what people want their friends to hear, rather than what they think a researcher wants to hear. We have always had data quality issues and respondent bias, and to my mind we always will. The key is using traditional qualitative techniques to understand and document this bias, so that we can account for it in our quantitative models. And by this I mean thinking about the world, and the quality of the data that we have, as much as using depth interviews and focus groups to probe, listen and respond. What do others think?

Like Report

13 years ago

It's a true observation OK! You do think about your 'audience' a little before you post stuff, unless you're a sociopath. Online behaviour is like anything else in that respect, we wear different masks at different times and searching for a 'core truth' about a person is often beside the point. We all complex strands to ourselves that interweave, coalesce and separate. Understanding what these strands are, how they interact at different times and how it affects the way people think and act is what research is all about. How to deal with it? There's no silver bullet of course, but good old multi-methodologies go a long way. Online gets you part of the picture, a group gets another, face to face interviews another, quant numbers will give you a bit of the picture and ethnography something else again. Each part is just one bit of evidence - it's the same danger you have with relying on any single methodology. But obviously the more time you spend with people and the more varied the circumstances in which you see them, the more you have to go on. So offline methodologies can definitely play a strong role in projects about online behaviour.

Like Report