No, no chance. Not without substantial knowledge. Typically you will also need to know details about the data generation that are often not part of the data table that is provided to the analyst.

Just to expand Jochen Wilhelm 's answer, there are many parts of the analyses that require knowledge of how the data arose. When I am working on a project where the other authors are the substantive experts, I still need to know about the area and am emailing them daily with questions to help guide the analyses. There are things you can learn from the codebook (if you don't have that, ask for it), but a lot of the analytic decisions require discussion with those experts. And just to point out the trivial, if the codebook says the variable is from a 1-5 scale, 33 is an error, or ages that you know are wrong (in free response I see people giving their year of birth when asked for age and vice versa ... but this requires [limited] knowledge of the data). If you provide more information about your situation you might get more specific comments.

The analysts should always know the background knowledge of the data. You need to know how data are collected and which mechanisms might cause missing or invalid data.

As others have pointed out, there may be situations in which you are 100% certain that this data point is invalid. For example, history of menstruation in a man, symptoms collected on a day after a patient has died, evaluation of hand function in a patient who has lost their hand, a value of 33 in a scale of 1-5. Those can be safely considered "error".

But those situations are rare. In most situations, you need to directly observe how people collect data, what the problems may result from, and you may even need to do a validation study to quantify the bias. Also, you need to verify your collected data with the source document - for example, compare case report forms vs. medical charts, and compared data entered into an electronic database vs. the paper forms.

Outliers should never be treated as invalid data. The term "outlier" mostly refer to Tukey's definition (< Q1 - 1.5IQR or > Q3 + 1.5IQR), and the number that he proposed is not mandatory for every analysis (I feel we now use it for convenience). Therefore, it is simply a mathematical way to define which values look "okay" and which may warrant more checking.

Thank you so much for sharing your insights.. looking forward to learn more from you all. Heartfelt gratitude..