Words are coming out of my mouth, but they seem detached. The effluent flows uncontrollably into the swarming background fuzz. I could, like Deepak Chopra or someone whose diet consisted solely of Rocky Mountain oysters, be spouting any old bollocks.

My attention starts to peel away from the corporeal grit of the experience. Instead, time slows as I painstakingly abstract significance from her words and gestures. I think we’re talking about music. My eyes saccade all over her face, and my awareness of this only augments the countervailing pressure to also hold steady eye contact.

Be normal, don’t fuck this up.

As it usually does in even the most perfunctory or romantically uncharged of conversations, the eye contact becomes uncomfortable, so I look around the room. A mutual friend is watching the discourse like a patient butler, half bored, half awaiting her cue for interjection.  There is nothing in her expression to gauge my performance in the main conversation.

Alas, too much time elapses; so I am forced to resume contact with the original face:  It lets out a short globular laugh. Have I just made a joke? Is it possible that the effluent contained some witticism, some enamouring clamour? I see myself as a sexual gold-panner, feverishly panning her communicative river in the vain hope of some nugget of interest.

Aaargh, I need some sign, some confirmation that I might be in here.

The analysis continues. My eyes slowly drift to her right, when one gesture protrudes in my consciousness. Infinite curls of blonde hair erupt laterally over her shoulder like overheated porridge brimming over the edge of an excessively microwaved bowl. Within these sinuous curls her right index finger worms around, creating hirsute ripples that glisten in the light.

She is playing with her hair.

Yes, I’m in here surely!

A presumably natural break occurs in the conversation and so I retire to the kitchen with the previously on-looking, mutual friend.

“I think Felicity likes me!”  I proclaim.

“Really, what makes you think that?”  replies Kate.

“We had a pretty long conversation and she was playing with her hair.”

Kate drolly responds, “Ha! Girls play with their hair all the time, you can’t take that as a sign that she’s interested in you!”

Unfortunately for me, Kate is right.

Cupid and Chlamydia

A girl playing with their hair is a poor indicator of her having a romantic interest in you. To borrow a term from epidemiology, the hair-playing test has a low positive-predictive value.

But what does that even mean? Perhaps some clarification is in order. Let us briefly depart from the triviality of annoyingly prolonged celibacy to the hilarity, sorry graveness, of a medical disease, say, chlamydia.

With regards to the disease, people may either have chlamydia or not have chlamydia. This is known as the disease status.

In order to diagnose chlamydia, a physician will want to undertake some tests. Let’s take one of these, say a urine test.  Unfortunately urine tests are not always correct, so physicians need some way to objectively assess how reliable they are at picking up certain diseases.

Now, imagine one person with chlamydia. He goes to his doctor and gets a urine test, which, lo and behold, indicates the presence of the C. Trachomatis bacterium causing chlamydia. In epidemiological terms, this person represents a true positive – someone who has the disease and is correctly identified by the test as having the disease.

In order to ensure that as many of the right people as possible get appropriate treatment, a good test ought to maximise the amount of true positive cases.

Sometimes, however, the test can give the wrong result. Another person may indeed have chlamydia (he is positive in disease status), but the test incorrectly indicates that he doesn’t have chlamydia (he is negative in test status). This person is known as a false negative.

Clearly, someone who actually has a disease but is wrongly given the all-clear is going to miss-out on potentially life-prolonging treatment. With regards to chlamydia, he or she may continue subversively mocking my 27 year-long dry spell by shagging lots of other people and, in the process, passing on the disease to others. In addition to maximising the amount of true positives then, a good test ought to minimise the amount of false negative cases.

In other words, when someone actually has a disease, we want the test to have a high probability of correctly identifying them as having the disease. This is known as the sensitivity of a test.

Sensitivity is the probability that a test will be positive when the disease is present.

Looking at the table below, we measure sensitivity by calculating the proportion of those correctly identified by the test as having chlamydia (true positives) to all those who in fact have chlamydia (the sum of true positives and false negatives).

Lots of people are fortunate enough not to have chlamydia. It’s actually one of the few perks of adult virginity; almost enough to mitigate the adult virgin’s pariah status, stigmatisation, and incessant, unwarranted media stereotyping as overweight, chronic masturbators harbouring a child-like fascination with My Little Pony. The stereotype is not true; we’re not all overweight.

Anyhow, a urine test of these people ought to show up negative. These cases are true negatives – someone who doesn’t have the disease and is correctly identified as not having the disease. A good test would maximise the amount of true negatives.

Unfortunately, some people will test positive but in fact not have the disease at all. These are known as false positives.

Informing someone that they have chlamydia when in fact they don’t is a practical joke even Harold Shipman may have thought twice about. Wrongly diagnosing someone with a disease also has repercussions in terms of exposing that person to unnecessary, possibly damaging, treatment. Therefore, a good test, in addition to maximising the amount of true negatives, ought to minimise the number of false positives.

Put simply, when someone is disease-free, we want the test to have a high chance of correctly showing that they do not have the disease. This is known as specificity.

Specificity is the probability a test will be negative when the disease is absent.

Specificity is measured by calculating the proportion of those correctly identified by the test as not having chlamydia (true negatives) to all those who actually do not have the disease (true negatives plus false positives).

In addition to values for sensitivity and specificity, we may also work back from the test results to see whether someone has the disease or not. In this vein, the positive predictive value measures the probability that a disease is present when the test is positive. Logically then, negative predictive value is the probability the disease is not present when the test is negative.

The differences between sensitivity, specificity, positive and negative predictive values are subtle, but hopefully the table and equations below will elucidate things.

Sensitivity  = True Positives / (True Positives + False negatives)

Specificity  = True Negatives / (True Negatives + False Positives)

Positive predictive value = True Positives / (True Positives + False Positives)

Negative predictive value = True Negatives / (True Negatives + False Negatives)

In Hirsute of Love

Right, so why have I wasted 719 words being so painfully didactic? Perhaps it’s an uncomfortably apt analogy, but a girl being romantically interested in me may be compared to having a disease. She’ll either consider going out with me or not. If she is indeed interested, we can say that she is positive in “disease status.”

Being British, directly asking a girl whether or not she likes me is tantamount to social Harakiri.  It would be less embarrassing and more self-effacing to pull down my trousers and ask her what she thought of my withered genitalia. Naturally, I’m prone to use other diagnostic tests to assess her potential for reciprocating romantic interest. The diagnostic test in question is the hair-playing test. A positive result on the test equates to her playing with her hair, a negative result is her not playing with her hair. Simple.

Now, let’s make a new table for the hair playing test. I reckon I have probably conversed meaningfully with around 100 girls. Of course, there is no way of knowing whether all of these girls liked me or not. Instead, I will use an average of my success rates at eliciting a response or match on OKCupid and Tinder respectively. That’s approximately 15%, so lets pretend that of the 100 girls with whom I conversed, 15 of them were willing to go out with me. In reality, that’s probably a massive overestimate, but never mind.

I went on a few dates using OKCupid and the overwhelming majority of girls played with their hair during conversation. So let’s say that of those 15 girls who were actually interested in me, 14 of them played with their hair. These 14 girls tested positive on the hair test and so represent true positives.

Positive predictive value               = 14/84         = 16.67%

Negative predictive value              = 15/16         = 93.75%

Sensitivity                                         = 14/15         = 93.33%

                Specificity                                         = 15/70         = 21.42%                 

As you can see, this gives a remarkably high sensitivity of 14/15 or 93%. From this we can surmise that if a girl actually likes you, there is a high chance she will play with her hair. Only 1 girl (the false negative) didn’t play with her hair, yet still liked me.

Excellent, so Felicity must have been interested in me, right? Not necessarily. As previously mentioned by Kate, girls always play with their hair: when they’re bored, when they’re stressed, or when it’s cold and they forgot their gloves so needed another means of warming their hands. Of the 100 people I met, lots of them, about 84, played with their hair. Yet, only 14 of these people were genuinely interested in me. The remaining 70 played with their hair, thus testing positive, but were actually not interested in me at all: they were false positives.

The positive predictive value of the hair-playing test is then 14/84 or 16.7%, which is quite low. This suggests that just because a girl plays with their hair, there is a low chance she actually likes you: the vast majority are just playing with their hair for whatever other reason.

But what if a girl doesn’t play with her hair, what does that mean? I have met 15 people who didn’t play with their hair, and it as fairly obvious from subsequent conversation, that they weren’t interested in me. They either already had boyfriends, or, more likely, hated me for reducing female behaviour into 2×2 tables. As mentioned, only the 1 girl didn’t play with her yet was interested in me. Maybe she was bald, I can’t remember.

The negative predictive value is high, at 93.75%. This suggests that, if a girl doesn’t play with her hair (thus testing negative), she is unlikely to be interested in you.

Given that there of lots of girls that play with their hair, but are not interested in me, it should be no surprise that such girls have a fairly low chance of not playing with their hair. This is reflected in a low specificity of 21.42%.

To summarise; a girl playing with their hair as a diagnostic test of romantic interest has a high sensitivity and a high negative predictive value. If a girl is truly interested in you, she is likely to play with her hair. If a girl doesn’t play with her locks, then it’s highly likely you should stop fantasising about her and move onto your next infatuation. But, when this next girl plays with her hair, a low positive predictive value would suggest that, rather than subliminally communicating romantic interest, she is probably removing the drool of saliva that you have inadvertently precipitated upon her through your awkward, Hugh Grant-style stuttering.