Category Archives: Analytics

Need More Options

Yesterday I got conned into taking one of those phone call surveys. My phone is on donot call list and all but people seem to find ways to break that and just call. Usually I don’t entertain these calls, but now and then I just accept the calls to see where it goes. Yesterday’s call was about PG&E. Actually, it didn’t start off that way. I think the survey picks a few brands and eventually focuses on one and in my case that happened to be PG&E. Where I live, I don’t have choice of any other energy providers. Anyway, I am not talking about this when I mentioned about I need more options. I was able to patiently answer majority of the questions but at one point there was a question for which I have no idea how to answer as I wasn’t aware at all. But the guy who was taking the survey insisted that I give him an answer between 0 and 10. I almost lost my temper and told him I can’t. What I don’t understand is, I am doing him a favor, so why can’t there be just an option of “don’t know” so that it’s possible to move on? It’s not like that’s going to cause any statistical anomaly to the survey. Infact, while I gave very decent rating (between 7 to 10) for many questions, I did give 4 for a couple of them. He never bothered to ask me why I was giving such negative remarks for those questions. So, my answer is going to go into some standard deviation or mean with no meaning. But the real reason is their recent gas line explosion.

Now I learnt my lesson. Even for my amusement, I am not going to take any more survey calls. While I am at it, I want to talk about a few more lack of options on a few websites.

1) Netflix provides an option to rate each review. The options are “Helpful”, “Not Helpful” and “Inappropriate”. What does the last option mean? Is it a catch all? In my case, the problem is that for one movie that I just watched, the reviewer pretty much gave away the key surprise in the review and I wanted to rate it as “Spoiler”. But there is no such option. Should I use “Inappropriate”? Now, would that be appropriate? 🙂

2) Similarly, Netflix also provides suggestions based on past watched movies and the ratings. And they have an option to indicate “Not interested”. But what does “Not interested” really mean? That I don’t like that type of movies? Surprise, surprise, would that info go back as a feedback into some sophisticated neural network and adjust my “taste” weights? But the truth is, I am “not interested” to watch it again, as in I have already seen it else where.

3) On LinkedIn, there is an option to report about job positions of people. And the options are “has never worked at this company” and “no longer works at company”. But what about “did not work with this title”? I have seen a few ex-colleagues falsely claiming job titles that they never got while working at the company.

Life is full of options. Software can only do so much. Yes, companies are gathering more and more data. Terabytes, petabytes and want to use SQL, NoSQL, Hadoop and whatever you name it. But none of that matters if the model doesn’t consider critical pieces of information like the 3 examples I have given above.

Advertisements

Leave a comment

Filed under Analytics