In scoring polygraph charts, polygraphers typically assign numeric scores to reactions on each channel: "1" for a small reaction, "2" for a mid-size reaction, and "3" for a large reaction. These numbers are then added up. For relevant questions a negative (-) value is assigned. For "control" questions, a positive (+) value is assigned. Scores for both relevant and control questions are combined, and if the final tally is a positive number, it indicates that overall, reactions to the control questions were greater (and the examinee "passed"). If the final value has a negative value, then the examinee "failed." (However, if the final tally is close to zero, say between -3 and +3, the result may be deemed "inconclusive.")

For a more detailed explanation of polygraph chart scoring, see Department of Defense Polygraph Institute Numerical Evaluation Scoring System (376 kb PDF).

While AntiPolygraph.org has received many complaints of boorish behavior by polygraphers, it is certainly not the case that they are all unpersonable. On the contrary, the ability to gain the trust and confidence of others is an important trait in an interrogator (and polygraphers are interrogators).

Yes, a polygrapher's first impressions, as well any bias (for or against a person), can influence outcomes.

With regard to polygraph instruments, there's no evidence that one type is better than the other for detecting lies. Polygraphy is junk science. Computerization adds no more to polygraphy than it does to astrology.