The graph illustrates information about the results of a poll of theatregoers regarding disturbances during theatre performances. In general the 4 most disturbing problems are coughing, rustling sweet papers, whispering and arriving late. Their percentages stand at above 50%.
Thanks to the context setting at the start of the essay, I can see that 50% means 50% of the theatregoers who took part in the poll. But it’s not immediately obvious and I had to read the opening a second time to make sure I understood what you meant. If a text is difficult to understand then it will receive a low score in IELTS for coherence and cohesion (CC). In the example above it can also affect your score for task achievement (TA), because you don’t really say anything meaningful about ‘50%‘. (See IELTS Task 1 Writing public band descriptors)
A good strategy to introduce and develop percentages in IELTS Task 1 can be seen in this extract from an article in The Economist (analysis below):
Data collected by USC Annenberg (University of Southern California) demonstrate that the “hyper-sexualisation” of men in films has increased substantially in less than a decade. Of the 100 top-grossing films at the US box-office in 2007, 4.6% of male characters  were seen dressed in “sexualised attire” and 6.6%  were shown “with some nudity”. In 2014 those figures stood at 8.0% and 9.1% . 2013 marked the highest point of this trend (the year that “Man of Steel”, featuring Mr Cavill, was released), with 9.7% of male characters  shot in sexually alluring clothing, and 11.7%  taking some—or all—of their kit off on film.
That said, Mr Cavill and Mr Harington would do well to remember that these figures are paltry when compared to those of actresses. In 2014, 27.9% of female characters  wore ‘sexy’ clothing and 26.4%  exposed their chests, legs, or other body parts on camera: they are roughly three times more likely to be objectified on screen than men.
What is ‘good’ about this reporting of percentages?
 The first percentage is expressed using the following pattern:
x% + of + noun (‘male characters’) + verb (‘were seen dressed..’)
Notice that the reader knows exactly what is meant by ‘male characters‘ thanks to the clear context setting of the opening sentence. Setting a context like this makes your writing coherent. When you use this structure you explicitly state the ‘whole’ – in this case ‘male characters’.
 The second percentage obviously also applies to ‘male characters’, and so there is no need to repeat ‘..of male characters’.
 ‘Those figures’ signals back to the previous two percentages, which we understand refer to ‘male characters’. Notice the structure:
past time expression (‘In 2014’) + subject (‘those figures’) + ‘stood at‘ + x%
 ..follows the structure:
x% + of + noun (‘male characters’)
 is in the same sentence as , and so we can assume the figure also refers to ‘male characters’.
 ..follows the structure:
x% + of + noun (‘female characters’).
We’re not surprised to read ‘female characters’ because this new context was set in the opening sentence of the paragraph. Again, this context setting makes your writing both cohesive and coherent since you explicitly state the whole (now ‘female characters’).
 is in the same sentence as , and so we can assume the figure also refers to ‘female characters’, which is the new context of this second paragraph.
Whatever you do..
Make sure your first mention of a percentage includes an explicit reference to the whole:
x% + of + noun (the whole)
Would anybody like to try and re-write the text about theatregoers to make the ‘50%’ figure mean what it’s supposed to mean? Answers in comments below! 🙂
PS. Another example of what I’m talking about just came to my attention:
Researchers identified 990 fatal shootings in 2015 – more than twice as many as had ever been recorded in a single year by the federal government – and Washington Post data journalists and graphic designers built an interactive, searchable database detailing those incidents.
A team of Washington Post reporters dug into the data and revealed that most of those who died were white men armed with guns who were killed by police in threatening circumstances. But The Post also uncovered some troubling patterns: A quarter of those killed were suicidal or had a history of mental illness. More than 50 of the officers involved had killed before. And while only 9 percent of people killed by police were not armed, unarmed black men were seven times more likely than whites to die by police gunfire.
source: Washington Post