Question 1036654
Typically, we use 95% confidence intervals.  A CI is a per cent, not a probability.  We don't know the parameter of interest and never will, but we can create an interval where we are highly confident (95%) that the parameter will be there.  In reality, the parameter is or isn't, which is a 0/100% probability issue and not useful. If we created 100 such CIs, 95 of them would contain the parameter.  We don't know which 95.  
For a sample, the CI is a +/- t (degrees of freedom=n-1,.975)*s/sqrt(n), where n is the sample size and the t-value may be looked up.  There isn't enough information here but the 95% CI is 69.606 +/- t*(3.01)/sqrt(n).  Anything outside of that would be considered an outlier, although realistically it can happen, just not at the level we consider significant.

Given the definition of an outlier that is >2 SDs from the mean, any data element that is more than 6.02 units (2*3.01) from the mean would be an outlier.  That would be <63.586 or >75.626