Oxygen is about 1% of the universe's mass. So that suggests that looking at any random sample of mass throughout the universe you would expect to see 1% oxygen. Now of course you wouldn't be surprised to see higher concentrations and lower concentrations in any given sample, but the larger your sample size the more likely it is that the value will approach 1%.
The atmosphere of a rocky planet like Earth is a tiny fraction of the planets mass (8.6134×10^-5% for Earth). However if we assume that oxygen content in the core and on the surface of a rocky planet without life is similar we can easily assume that any given lifeless planet should have around 1% oxygen in the atmosphere.
Now finding out the variance or standard deviation of oxygen in planetary atmosphere is a problem. I've found this paper that seems to be based on data for a large number of exoplanets, the problem is the data isn't shared, just the results. In Table 1 of that paper data is shown for three different types of exoplanet. We could calculate the variance based on this but variance from 3 data points isn't very accurate.
It comes down to two aspects in my eyes. First can you attribute the oxygen content to another source? Second is the oxygen content high enough to suggest life. It will never be a guarantee from this type of observation, so you really need to know how confident you want to be. Once you know you're confidence level you could calculate your confidence interval based on standard deviation of exoplanets oxygen content. However even this is dependent on how likely you think exoplanets are to support life, if it's very unlikely then you can be more confident if you have just a few similar oxygen content observations; if it's very likely then you would expect the majority of observations to have a high oxygen content.