The University of Bristol study identified these as the three most influential factors when teenagers choose their friends, while income and parents occupational class proved to be insignificant.
Popularity, IQ, bad behaviour such as arriving late and skipping classes and the likelihood of going to university were also found to be characteristics which subconsciously attract friends to each other.
Professor Simon Burgess from the Universitys Centre for Market and Public Organisation and colleagues Eleanor Sanderson and Marcela Umaña-Aponte, looked at an adolescent friendship network of 6,961 links in the West of England.
Their focus was on homophily, which is the tendency to establish relationships among people who share similar characteristics and attributes. This behaviour is important to understand high levels of social segregation, criminal behaviour, the spread of information and the dynamics of the labour market.
Given that high levels of homophily promote consensus in tight-knit but isolated groups, researchers believe the studys findings have implications for the segregation of different groups who will have little insight into each others' views and behaviours.
The study, funded by the Economic and Social Research Council (ESRC), found that personality is a key determinant in the friendship formation process, especially among non-school friends, while physical characteristics such as body fat and weight are not.
Researchers also found that those with longer friendships were more similar, especially when it came to their personality traits such as extraversion, emotional stability, intellect and imagination. Long friendships in adolescents are very solid and those friends are likely to have shaped their personalities together.
The study began in March 2008 when youngsters aged 15 to 17-years-old were asked to nominate a maximum of five best friends. Information about the length of friendship, where they met, how much time they spend together, what they talk about and the activities they do together was collected.
A unique longitudinal dataset, collected as part of the Children of the 90s study, looked at academic achievement, IQ, behavioural problems, health, future aspirations, personality and a familys socio-economic status.
It found that individuals select their friends deliberately, rather than friendships forming as a result of a random social selection process.
Professor Burgess said: Our findings are pertinent for understanding the role of friendships in adolescent society. It is unquestionable that people select and influence each other, which confirms that social networks are powerful in spreading information, beliefs and behaviours.
In our context of a large friendship network of adolescents, the effects of homophily choosing friends similar to ourselves - seem particularly important. These individuals are making a transition between childhood and adulthood and their emerging attitudes and beliefs will be affected by their friendships.
The research, entitled School ties: An analysis of homophily in an adolescent friendship network' by Simon Burgess, Eleanor Sanderson and Marcela Umaña-Aponte from the University of Bristols Centre for Market and Public Organisation, was funded by the Economic and Social Research Council (ESRC).
Explore further: Wives with more education than their husbands no longer at increased risk of divorce