There are several issues with this paper.

The first is that neither of the NLP founders state that eye accessing is an accurate measure of lying. I do however accept that with such a low level of quality control around publishing you may be able to find a reference or two elsewhere, but I personally don’t know any NLP practitioners who believe that lying can be reliably measured in this way.

As to the science itself, in NLP the eye movements upwards to the left and right relate to remembered verses constructed images not truth verses a lie. Now in your experiment you told the participants that they were to lie before they were actually interviewed and gave them the opportunity to devise that lie beforehand. Now I’ve never tried this but surely there is a high probability that the person is constructing the ‘lie’ at that point in time and then simply recalling it during the interview? This really doesn’t match how an NLP intervention works at all. You are measuring the second visual access, not the all-important first one. This lying process (for want of a better phrase) in your experiment is a mess.

Another more minor issue I have with the research is the issue of trying to select right-handed people. An NLP practitioner would calibrate the individual. They would not base their expectations on whether someone was right or left handed. So why didn’t you do the same? Simply ask a question that the participant would answer with a remembered visual access such as asking them what colour their front door is. You would then be able to test each participant reliably as their eyes should move to the opposite side when giving a constructed answer.

If the research had involved an NLP practitioner then they could have at least been able to explain that you were not really testing NLP claims at all and in the specific area of interviewing the candidates the experiment could have been better designed without losing its scientific integrity.