I recently had the pleasure of speaking with Statistical Analyst, Social Media Man and Principal at Anderson Analytics, Tom H.C. Anderson. Many of you may also be familiar with Tom through his online community for market researchers: Next Gen Market Research. Tom is very open about sharing ideas regarding social media and how it relates to the market research industry. I was curious to know what his thoughts were specifically about how Social Media and Text Analytics fit into the (online) qualitative realm.
Q
There is a lot of chatter among qualitative researchers looking to public blogs and forums to gain insights. How representative do you think these mediums are and should researchers be using them as a credible source for data?
A
Depends... If there is a discussion board dedicated to a specific brand or category, then this is where you will find their most loyal and knowledgeable customers as well as your clients’ competitors’ customers. While it may not represent every single type of customer, these customers are far more important than any other type. A great resource for quant and especially for qual research.
Blogs and other social media and social networks are a different story and usefulness of these needs to be evaluated based on specific project objectives.
Q
Are social networks, in your opinion, aware of market research and how market researchers are using them.
A
Sure, certainly they are; I know market researchers at Facebook, LinkedIn and MySpace. I do not think Twitter has anyone in MR. None of them are prioritizing MR as a revenue stream currently other than supporting ad sales. This is partly because of lower margins, but also because they are considerate of user experience and they somehow view MR as more intrusive than advertising.
None of them had any notes in their TOS about screen-scraping a few years ago. Now they do. But I don't think this means you can't do screen-scraping, just that if you do, that account might get banned. I do know companies that still try to do this on all of Facebook, or at least to parts they can get to and try hard to get the stuff they technically shouldn't be able to get to. Not something we try to do. If a client’s brand is being discussed on their Fanpage on Facebook say, then we definitely feel they have a right to get that data by screen scraping. But we're not hopping over walled gardens; we're expecting them to come down soon. Instead we prefer to think intelligently about areas where we find the most interesting discussion and go there.
Q
Which social network is most accommodating to market research? Why?
A
Depends on what you mean by accommodating. Do you consider Twitter a Social network? If you do, as it works on blog/RSS technology, arguably it's the easiest to work with since most of it is publicly available. But it is very different from FB & LI in many ways. The people on it are very different, and there is relatively little info available on them. 140 character chucks intersperses with links and tags; I'm not sure how useful it is for quant, never mind qual, but I suppose qual researchers should be able to answer that better than I can.
I think best practices for qual researchers in this area may actually be more important than for quant.
Facebook and LinkedIn, respectively look like they will both be very interesting in the future for both quant and qual. While much of the data has been behind ’walled gardens’, these walls seem to be coming down, and they are easier for Qual researchers to 'hop over' than for quant in any case.
I was sad to see LinkedIn leave the MR sample provider service, but for qual I don't think that matters as much. Qual researchers can probably still use LinkedIn to recruit B-B sample as they need smaller groups.
Currently though recruitment is still the best way to leverage SNS. I hope it will be easier to leverage in other ways in the near future as the walled gardens come down.
Q
What do you think about online communities being used for market research?
A
I assume you mean MROC's. Well it's not very different from convenience panels in many cases. I have a bit of a problem with the wording. Why should we call panels communities all of a sudden? If you recruit and pay people to be in a ’community’, it's not quite the same thing as people interacting freely in social media/social networks elsewhere. I find it hard to accept that some companies think they have a social network or are active in social media by having a small MROC.
Don't get me wrong. I think MROCs can provide valuable feedback. But I've always been more interested in the biggest focus group in the world, the one Al Gore invented.
Q
Do you think the qualitative research industry is doing a good enough job in keeping up with advancements in online technology?
A
It's a lot easier to adapt if you are in qual than quant. Like I said, I think qual needs some standards on what constitutes or at least how to define a qual sample on social media. How to interact or not interact.
i.e. should we be anthropologists or get involved, or both.
But again, hard for me to say too much about qual yet as I'm really a quant researcher being pulled towards qual, because of text analytics and social media. I think the risk for both qual researchers and quant researchers is being left behind as we move towards quali-quant hybrid thanks to NLP (natural language processing)/Text Analytics. I'm learning more about qual now so we can make better use of it in these hybrid cases.
Q
Is there a place for text analytics in qualitative research? If so, how have you seen it used?
A
Yes, in more than one way. On small samples you can still use TA. But more importantly, as mentioned earlier, I think Quals need to become more quant and vice versa.
I think basic text analytics (i.e. word count), as well as those programs that focus on emotions (often based on a-priori word choice models) can be very useful in qual. These were used historically to do at-a-distance psychological analysis, for instance, in speeches given by political leaders. Even if a speech is written for you, the words you choose to use when you deliver it, say something about your current emotional state. Text analytics is ideal for picking this up and comparing it to normative data and control groups. We have a white paper or two about these on our site.
-End Interview-
Author – Derek Sawchuk
Marketing Manager at Itracks
@SawchukMR
