For honest scientific work to be done, all gathered data should be considered. You cannot just pick and choose data and manipulate the data in order to get the desired results.
Since 2017, when scientists were able to get access to the raw C-14 data and analyze it, it has been discovered the labs had manipulated their data.
According to their own published reports, they [Dr. Tite’s C-14 researchers] discarded readings that didn’t fit what they wanted. From their own figures, they were as much as 400 years off on the low end, and on the high end 1500 years off, which is pretty significant.
To be clear, the manipulation of the data does not explain why a 1st century cloth would be dated to the 13th or 14th century. Even if the data was not manipulated, it would’ve still dated the samples to be medieval by C-14 dating.
The results show that the different assessments produced by the same laboratory (raw vs. Nature) are not statistically signficant, whereas the analysis of the raw radiocarbon dates confirmed that the different laboratories produced different assessments and that these differences are, in most cases, statistically signficant.
So, why would they want to manipulate the data?
The manipulation of the data seemed more for the purpose of not invalidating the test results from the labs.
Harry E. Gove, the inventor of the AMS method, affirmed that ‘[if] one of the three laboratories obtained an outlier result … it would be impossible statistically to identify it and the three measurements would all have to be included in the average thereby producing an incorrect result’ (Gove 1989, 237). Our statistical analysis confirms that this criticism was warranted.
By discarding the low end of the data, it made the distribution curve tighter and allow it to pass the chi-square test.
After the change in two uncertainties from Arizona Raw 1 to Arizona Raw 2, the chi-square becomes almost acceptable (Table 3). This adjustment is unusual, since none of the radiocarbon dates of the control samples were modiﬁed by Arizona.
With the result of the entire raw data, it would produce less than a 1.36% probability the composite dating is reliable. However, even with outlier data being discarded, it would only produce a 4.176% significance level. This was rounded to 5% to meet the minimum 5% threshold level in order to be considered a valid test.
More on the statistical analysis:
“It’s a well-known fact that scientists can produce whatever result they want. If you believe that passionately in something, you can steer the results. My God, we’ve all been guilty of that.”
Harry Gove, May 1988