A scientific formula for a successful relationship?

Wouldn’t it be dandy if there was a scientific formula which predicted how long a relationship would last? We wouldn’t have to rely on the advice of buddies, astrologers, and agony aunts. Most importantly we would have a leg up before investing time, money, and energy into one of the most important decisions of our lives.

Recently, a high-profile research paper published in Psychological Science suggested that the best predictor of how long a relationships lasts is a parameter known is Language Style Matching or LSM. USA Today published a short description of this innovative approach:

The kind of language style the researchers focused on was the use of such words as personal pronouns (I, his, their); articles (a, the); prepositions (in, under), and adverbs (very, rather) — the types of words most people don’t give much thought to.

“You are four times more likely to match and probably go on a date if your language style matching is even just above average”…

Earlier in the year, a blogger at The Economist commented on LSM too:

A drop in the LSM score can mean a relationship is going down the tubes, though not necessarily; for instance, one year Freud and Jung’s LSM score dropped when they were still on good terms, which the researchers think may have been because Jung was ill and stressed that year.

Okay, so this is heavy-duty stuff. So I decided to investigate a bit more.

It turns out that there are two research papers which describe LSM  in writing and how the algorithm “predicts relationship initiation and stability“. What I really wanted was to put it to the test before reading more about the algorithm or the claims made in the media. Fortunately, the authors created an online application which you use to check the Language Style Matching of two people in order to predict how successful their relationship will be.

Once you enter your words in the interaction and then the other person’s words, you will get a number back that assesses the degree to which the two of you match. This number, called a language style matching, or LSM score, ranges from about .50 to 1.00. The closer you are to 1.00, the more in synch the two of you are.

Other studies have found that the LSM score is associated with how long a relationship lasts and its overall quality.

It sounded good to me. In theory, it is plausible that two people engaged in a conversation who use a similar structure in their language might be more likely to have a successful relationship. Or even the opposite: two people in a long-term relationship might be more likely to share similar language structure. Or whatever. Psychology isn’t a discipline always known for giving two hoots about trifling matters such as correlation not being equal to causation.

But I digress. What better way to test the predictive power of LSM than to examine the writing of two people in a relationship who very much love each other?

In other words, me and me.

Or put another way, I decided to test a random sample of my own writing to see if my relationship with myself might be predicted to last my own lifetime. For the examination, I took the text from the most popular post on this blog and compared it to the one I wrote less than a week later. I then compared the popular post to the first chapter of my PhD thesis. Next, I compared two posts which I had deliberately written in the same flippant style here. Finally, I compared the text of an editorial I wrote on the lung cancer genome last year to a business task force recommendation I wrote the same week.

And what did I find?

When I checked how my relationship with myself was going based on my different writing styles I got back Language Style Matching numbers for each pair which always varied from 0.69 to 0.71. So what does this LSM mean?

According to the website:

Compared to other general writing samples that we have analyzed, your LSM score is within the average range. To give you an idea, most LSM scores for general writing samples range between .60 and .90, with the average being around .78. The more the authors of the two samples are thinking in similar ways, the higher the LSM.

So, in other words, the matching between two different samples of my writing is less than the average for two different random people. But in my case, for the single test I ran with writing in what I thought was in the same style, I got the same result which was also lower than for two random people.

Apparently, I don’t even think like myself.

Now, I understand the problem in examining different types of writing for different purposes. And I see a broader use for this application which might even result in better informed outcomes for hapless folks such as mangliks which does  not involve tree marriage.

But excuse me if I’m not totally sold on this yet.

Will you get the same results that I did? Will I get the same results again with other samples of writing? Or do I suffer from multiple personality disorder in which one me absolutely hates another me?

I don’t know. But for now, I’m sticking to reading coffee grinds. They may not predict how much I love myself, but I certainly feel better after drinking the Turkish brew at the top of the cup.