>>12684837>This may sound like a really stupid question, but would a distribution be considered skewed right if there's only like one outlier. Like hypothetically if one of the values on the x axis of pic related was like 400 or something.No, not if the rest of the data keeps a roughly normal shape.
>>12685008>You can test for it trough the Shapiro-Wilks test, which assumes normality under the null.
In this case, it might be a good bet. In most scenarios, Shapiro-Wilks is a trap. Most people learning stats don't understand how or why it works.
Example: If you generate 500,000 data points from a normal distribution (say using rnorm in R), and you use Shapiro-Wilks, it will reject your assumption of normality. This is by design. If you don't understand how or why, I'd suggest not using it.
When the sample size is small, large departures from normality will not be detected (i.e., the test will say "its normal!" even if its clearly not), while when your sample size is large, the absolute smallest deviation from perfect randomness leads to rejection of the null (normality).
It's a terrible test, imo, because so many poor bastards starting to learn stats learn about assumption of normality, and say "oh! I know! I'll test my data for normality before using ANOVA/t-test/whatever", without realizing that 1) you are inflating your error with 2x p-value based tests, and 2) the assumption of normality can be broken quite severely, and most generally used tests will still be robust
