Of burdens and black dogs

oCaptainMany years ago, I had a student that was really, really difficult.

He was unpleasant. Disruptive. Challenging. And he openly said that he did not like me. I am not going to lie: I was very happy when the semester was over!

 

One year later, he showed up at my class door. He wanted to tell me that, in the previous year, he had experienced a very traumatic episode. That episode left him scared and angry. He felt confused. He was hurt. And he explained that, in the process of dealing with all that pain, he had been really nasty to those that were trying to help him.

 

I don’t know what is made of him, now. You see, this happened long before LinkedIn or Twitter, and I completely lost contact with him. But I never forgot him, for he taught me a very important lesson: that the person standing in front of me – student, colleague, random stranger in the street… – may be carrying a very heavy burden, or struggling with the black dog named depression.

 

Most of the time, I will not be aware of that burden, and I won’t be able to see the black dog. Or I will be unable to help despite my desperation to do so. Most of the time, the best I can do is to be compassionate.

 

In honour of all of those fighting their demons, and in memory of those that lost the battle, I climb on a table and say: O Captain! My Captain!

 

This is not a post about marketing. But it’s an important one.

Have a great day. And be compassionate.

Analysing photographs and other visual input

20140810-141554.jpgWith photos and videos representing an increasing proportion of the content shared online, I am very interested in their potential for my own research. However, I struggle to incorporate visual data in my work because qualitative analysis software (at least the ones that I am familiar with) can only process alpha-numerical data. This means that I have to analyse visual data manually, which is a slow process.

 

The paper by Hu et al (2014) that I mentioned in my last post (you know, the one about cats not ruling the Internet) describes a computer-assisted approach to analysing photographs. This is what the authors did, as described in section 3.2 of the paper:

Coming up with good meaningful content categories is known to be challenging, especially for images since they contain much richer features than text. Therefore, as an initial pass, we sought help from computer vision techniques to get an overview of what categories exist in an efficient manner. Specifically, we first used the classical Scale Invariant Feature Transform (SIFT) algorithm (Lowe 1999) to detect and extract local discriminative features from photos in the sample. The feature vectors for photos are of 128 dimensions. Following the standard image vector quantization approach (i.e., SIFT feature clustering (Szeliski 2011)), we obtained the codebook vectors for each photo. Finally, we used k-means clustering to obtain 15 clusters of photos where the similarity between two photos are calculated in terms of Euclidean distance between their codebook vectors. These clusters served as an initial set of our coding categories, where each photo belongs to only one category.

 To further improve the quality of this automated categorization, we asked two human coders who are regular users of Instagram to independently examine photos in each one of the 15 categories. They analyzed the affinity of the themes within the category and across categories, and manually adjusted categories if necessary (i.e., move photos to a more appropriate category or merge two categories if their themes are overlapped). Finally, through a discussion session where the two coders exchanged their coding results, discussed their categories and resolved their conflicts, we concluded with 8-category coding scheme of photos (see Table 1) where both coders agreed on, i.e., the Fleiss’ kappa is κ = 1. It is important to note that the stated goal of our coding was to manually provide a descriptive evaluation of photo content, not to hypothesize on the motivation of the user who is posting the photos.

Based on our 8-category coding scheme, the two coders independently categorized the rest of the 800 photos based on their main themes and their descriptions and hashtags if any (e.g., if a photo has a girl with her dog, and the description of this photo is “look at my cute dog”, then this photo is categorized into “Pet” category). The coders were asked to assign a single category to each photo (i.e., we avoid dual assignment). The initial Fleiss’ kappa is κ = 0.75. To re- solve discrepancies between coders, we asked a third-party judge to view the unresolved photos and assign them to the most appropriate categories.

 

Hum… sounds like I a need to get a Computer Sciences degree to be able to do this :-( Plus, the process described still relies heavily on manual analysis to refine the coding scheme and to do the actual categorisation. Still, it sounds like a promising ‘starting point’ to develop an inductive coding scheme. So, I am adding a note to my diary to look into this during my sabbatical (one can always dream big!)

 

Do you use visuals in your research or work? How do you analyse them?

 

References

Lowe, D. G. 1999. Object recognition from local scale-invariant features. In CVPR.

Szeliski, R. 2011. Computer vision: algorithms and applications. Springer.

It’s official: cats don’t rule the Internet

Quick. Answer this question: What is the most popular category of photos on Instagram?

 

I thought it was food, but I was wrong. And if you thought that it was cats you were wrong, too.

 

According to this paper by Hu, Manikonda and Kambhampati, nearly a quarter of content posted on Instagram are selfies. This is closely followed by pictures with / of friends. Pictures of pets are the smallest category!

 

IG paper

 

 

The proportion of selfies and friends photos was pretty much stable across different type of users and levels of engagement (defined as the number of photos posted by a user). In contrast, there was high variance for pets and fashion postings – some people posted lots of photos in these categories, most posted none at all.

IG paper2

 

Source of picture: Hu et al, 2014

The study’s authors conclude that Instagram is mostly used for self-promotion and for connecting friends. I agree with this, but would take the analysis further and suggest that:

  • These findings show that Instagram has broad appeal and is well past its niche stage
  • The data and, particularly, the variance for categories other than selfies and friends, also shows that we can segment and target Instagram users based on their revealed interests
  • Cats don’t rule the Internet :-)

 

Surprised with these findings?

Home or away – deciding where to publish your blog post

Well, call me absent minded, but I only just realised that LinkedIn has a new(ish) feature: blogging. Well, actually, the company calls it ‘Long-Form Posts’ so let’s refer to this feature as LFPs, for short.

 

Your LFPs are displayed as part of your profile and are accessible to LinkedIn users not on your network. According to this overview on the company’s website, LFPs are a way for:

members to contribute and share professional insights on LinkedIn. We’re expanding LinkedIn’s publishing platform, by allowing members, in addition to Influencers, to publish long-form posts about their expertise and professional interests.’

 

Professor Gary wrote a short but very interesting analysis of his early experiences with LFPs here. And, at face value, it seems that blogging on LinkedIn is the way to go.  Do I hear you say ‘Oh, no, not another platform to learn about / maintain / monitor’? My sentiment, exactly.

 

To decide whether to add LFPs, or Medium, or any other platform to our social media presence, we need to think about what we are trying to achieve. Blogging is not an end in itself. It is just a means to an end. So, what is your end goal?

 

Sepp et al (2011) classified individual bloggers’ motivations into three broad categories: process, content and social. Based on their classification, I suggest the following approach to decide where to publish your blog posts.

 

Type of gratification: Process

The first type of gratification concerns the direct and immediate benefits for the blogger.

 

The drivers could be a desire to improve skills such as writing or reflexivity, to release negative emotions or to engage in an enjoyable past-time. In this case, you are unlikely to benefit from blogging in a third party platform, in addition to your own blog, unless you want to experiment with different types of writing in the various platforms.

 

Type of gratification: Content

The second type concerns the subject matter of the blogs, and could deliver benefits long after the content has been created.

 

The drivers could be: a desire to keep a journal of your activities; to express your opinions on various matters (e.g., a news article); to promote ideas, ideals or products; or to attract advertising revenues. All of these goals would be best served by focusing on your own blog, and attracting readers there.

 

However, this type of blog could also be associated with a desire to entertain or enlighten others. In that case, you want to make sure that your message reaches the targeted audience. In addition to using your own blog, it makes sense to place content in other platforms to reach new readers and build awareness.

 

Type of gratification: Social

The third type concerns the interactions with others.

 

In this case, blogging is driven by the desire to obtain different perspectives on a topic of interest, or to engage in dialogue with others. It is also a way of meeting interesting people and keeping in touch with friends, letting others know about what you do or what you have achieved, or even to obtain tangible or intangible support. To maximise these benefits, you need an audience and third-party platforms may be the best way to achieve that.

 

Here is a summary of the classification and types of goals, as per Sepp at al (2011)  and my suggested approach:

Slide1

 

Example

I created this blog as a repository of content for my classes. I wanted to capture examples of marketing principles working in everyday life, so that I could refer to them promptly in my lectures or to provide links as and when needed. This is a journaling-type of goal best served by focusing on my own blog.

 

Some time later, the blog evolved as a mechanism to bridge marketing theory and practice, and I started summarising academic papers and capturing what the findings meant for practitioner. This is an educational-type of goal that would benefit from placing content in other platforms.

 

Finally, I like to discuss half-baked ideas with others. It helps me get fresh perspectives on teaching or research matters. I am fortunate to have a small number of friends that really extend my thinking and that challenge what I write – be it through comments here on the blog, or via e-mail. I often extend this conversation by commenting and asking questions on other blogs that discuss matters that interest me. Occasionally, I also post on discussion forums or write for other blogs to tap into the brains of people that do not read my blog.

 

Your turn: why and where do you blog?

July round-up

July was going to be the month I was going to get on top of my to do list. It didn’t happen. It was nonetheless a good month, which included the following highlights.

 

Researching

At the beginning of the month, I attended the Academy of Marketing conference, in Bournemouth. I presented 2 papers, including one I co-authored with Yuvraj Padmanabhan, and which won two awards: Best Paper in Track and Best Paper at the Conference. So happy.

AMaward2AMaward

 

I also presented at the MOPAN conference, with my colleague Sarah Quinton.

MOPAN

Even better, the team is making good progress on the project looking at Digital Marketing in SMEs. And I am setting up a new project looking at customer profiling in the age of cheap, abundant data.

But the best of the best had to be the kick off meeting of the digital citizenship project, with our partners in local councils. This is a project that I am working on with Sarah Quinton and Thom Oliver.

 

Writing

I worked a bit more on a paper about the role of CRM systems in surveillance :-) However, the paper on crisis management in social media was rejected :-(

And the Management Research book is out. Yay.

 

Teaching

I have been working on the website materials for the Management Research book, which includes slide sets for instructors, exercises, reading lists and short videos. It’s been a very interesting experience, despite the technical problems.

 

Learning
bournemouthThis month, learning came mostly in two forms. First, I attended several interesting presentations at the Academy of Marketing conference and caught up on the latest thinking about segmentation, digital and research methods.

 

Second, I have been reading about the use of blogs and social media in course work, as I am trying to jazz up the assessment in my Consumer Behaviour module. If you have experience of this (either assessing or being assessed via social media), do share it with me, please!

 

 

At a personal level, I am loving the sunny weather this month, and seeing my kids so enthused about coding as a result of joining the YRS’s Festival of Code.

YRS

What were July’s highlights for you?

The performative power of the score

There are two interesting articles in the news, today. They are about two very different companies but, essentially, the same issue: the performative power of the score. Or, in others words, about how much a simple number can influence our life.

 
The first article is about passenger transportation company, Uber. It was revealed that Uber drivers rate their passengers, and that this information is made available to other drivers, to help them decide whether or not to take up a fare. This means that if you have a low rating, you may find it difficult to book one of Uber’s cabs. The concerns are that the ratings are subjective and may result from factors beyond the passenger’s control. For instance:

You hear stories from people who missed a pickup because of buggy notifications, for example, and those people all of a sudden just can’t catch a cab. Any kind of technical error can skew the ratings, but because they’re invisible they’re also treated as infallible.” (Source: Guardian)

 

 

The second article is about dating website, OKCupid. This company conducted an experiment whereby it manipulated the ratings that signal the extent to which your profile matches that of other users of the website – a higher score signals a better match between 2 persons (and, presumably, a higher chance of developing a successful relationship). During the experiment, users still had access to unaldetred profile information. The experiment revaled that users’ behaviour (i.e., whether they contacted the potential match or not) was highly influenced by the compatibility score provided by the company (as opposed to the qualitative information provided by the users in their profiles). This means that people were making important decisions regarding a potential life partner based on a simple score:

“In one experiment, the site took pairs of “bad” matches between two people – about 30% – and told them they were “exceptionally good” for each other, or 90% matches. (…) Further experiments suggested that “when we tell people they are a good match, they act as if they are. Even when they should be wrong for each other”.” (Source: BBC News)

 

 

Scores are not new. For decades, credit decisions have been made based on whether someone’s credit score falls below or above a particular threshold. Over time, the credit screening process has become largely automated as machines are deemed to be more accurate than staff. You see, humans make mistakes, are inconsistent, and are expensive to train and employ. And machines don’t.

However, evidence from default rates in the subprime loans market (yep, the one that triggered the credit crunch we are all recovering from) indicates that customer screening that relies solely, or predominantly, on quantitative data tends to systematically under-predict the default rate. In contrast, screening that incorporates soft data leads to better lending decisions. In other words, decisions that incorporate soft data outperform those that rely only on a quantitative score. See this paper for further details.

 

 

With digital technology making it so easy to collect, store, analyse and access data, quantitative scoring is spreading to more and more areas of our lives. As with credit scoring, numbers are easy to understand. They are clear cut. But they provide an incomplete picture of the person behind that score. And, yet, we are making important commercial and personal decisions, like whether to take on a client or contact a potential partner, based on those fallible scores.

 

 

I am really interested in this topic and I am trying to set up a project looking at customer profiling in the age of cheap, abundant data. If you want to join or, even better, if you want to fund this project, get in contact :-)

 

In the meantime, let me know: Can you remember a situation when you relied solely (or mostly) on a quantitative score to make a decision?

 

Filming – take 2

The book is now available for purchase! Yay.

 

And Susan, Nigel and I are recording materials for the book’s companion website. Again.

 

We met 10 days ago to film the chapter introduction. It was scorching hot – 87F / 30.5C inside the room – but we persevered. All day. However, there were some technical issues with the videos and we need to re-do them.

book videos

I thought that writing the book was the hardest part ;-)

 

What exciting things are you doing this week?