The vocabulary we currently use to describe sexual orientation is hopelessly inadequate, with labels like 'gay', 'straight' and 'bi' falling far short of the complex reality, a large long-term study suggests.

Far from being being a fixed preference, the findings suggest that sexual identity and attraction undergo extensive and often subtle changes throughout a person's life, continuing long past adolescence and into adulthood, with women showing slightly more fluidity than men.

"Sexual orientation involves many aspects of life, such as who we feel attracted to, who we have sex with, and how we self-identify," explains the lead author Christine Kaestle, a developmental health scientist at Virginia Tech.

"Until recently, researchers have tended to focus on just one of these aspects, or dimensions, to measure and categorise people. However, that may oversimplify the situation."

Kaestle's research is different in that it takes all the dimensions of sexuality into account, and not just on one occasion. Using data from a national survey of American students, her research tracks the sexual identity, sexual behaviour and romantic experiences of over 6,000 students from the age of 16 to 32.

At four separate points during these years, participants were asked about their romantic attractions, their recent sexual relationships, and whether they self-identify as 'straight', 'gay' or 'bisexual'.

Combing through the results, it quickly became apparent to Kaestle that these three words needed company. Both male and female participants were found to have fluid sexual orientations, not only in their teens, but right through their 20s.

For the male participants, Kaestle identified four categories of orientation, including 'straight', 'mostly straight or bi', 'emerging gay', and 'minimal sexual expression'; while for the female participants she identified five: 'straight', 'mostly straight discontinuous', 'emerging bi', 'emerging lesbian', and 'minimal sexual expression'.

Not only were women more difficult to pin down and categorise, they also showed greater fluidity within these groups, taking up more space in the middle of the spectrum.

For instance, women who fell in the 'mostly straight' category were attracted to both sexes in their early 20s, but by the time they reached their late 20s, almost all of them were interested in just men.

In contrast, male participants tended to fall more on the extremes of the spectrum, as either 'straight' or 'emerging gay'. Yet even though women were more likely to explore the full length of the spectrum, those men who identified as straight in their teens were more than twice as likely to be attracted to both sexes, compared to women at the same age.

"In the emerging groups, those who have sex in their teens mostly start with other-sex partners and many report other-sex attractions during their teens," explains Kaestle.

"Then they gradually develop and progress through adjacent categories on the continuum through the early 20s to ultimately reach the point in the late 20s when almost all Emerging Bi females report both-sex attractions, almost all Emerging Gay males report male-only attractions, and almost all Emerging Lesbian females report female-only attractions."

Kaestle thinks this is probably because a human's early 20s are a time of increased independence, when people begin to accept, explore, question and acknowledge same-sex attractions, without their preferences being obscured by a long term partner.

"At the same time," Kaestle explains, "as more people pair up in longer term committed relationships as young adulthood progresses—this could lead to fewer identities and attractions being expressed that do not match the sex of the long-term partner, leading to a kind of bi-invisibility."

It's subtleties like this that make research on sexual orientation so difficult. The broad labels we currently use often mean that those in the LGBT community are lumped together, with some individuals slipping through the cracks of our poorly defined parameters. 

Figuring out a way to accurately define specific sexual minorities is a challenge that may be impossible, but it is also one of the most important missions in health research.

Today, LGBT individuals face a huge disparity in health and well-being, and their presence is completely hidden in the national census. Not only do individuals in this group suffer from unusually high rates of psychiatric disorders, substance abuse and suicide, they are also less likely to receive routine and reliable health care, including cancer screenings and STI checks.

Part of the problem is when health experts make population estimates and comparisons by defining sexual orientation solely in terms of behaviour at one specific point in time. The new research shows just how exclusionary those tactics can be.

"We will always struggle with imposing categories onto sexual orientation," Kaestle admits.

"Because sexual orientation involves a set of various life experiences over time, categories will always feel artificial and static."

The goal, however, should not be perfection. Instead, we need to focus on creating nuanced, person-centred, multidimensional, longitudinal studies that encompass as many sexual minorities as possible.

This study has been published in the Journal of Sex Research.