You can remove artifacts like eye and muscle movements with prepossessing, not as well as with electrodes, but still.
I'm more worried about electrode connections and electrode quality in these amateur EEG applications. You also want high quality active electrodes if you take EEG outside laboratory. I'm also skeptical about these dry skin electrodes (no alcohol scrubbing and conductive paste). NeuroSky devices are surprisingly cheap. I would be very interested to know how well they work in practice.
“In October 2006, Potti and Nevins described in Nature Medicine how examining the sensitivity of cell lines to particular drugs could predict patients’ responses to cancer therapies for a range of cancers. [...] We had difficulties pretty early on,” Baggerly says. He and Coombes say they found errors in the 2006 paper, including genes that didn’t seem to belong on the list and tumor samples that were incorrectly labeled. [...] But Potti and Nevins continued to publish papers using the same method. This troubled Baggerly. He became obsessed with determining why the Duke team could make their prediction models work when he and Coombes could not.In subsequent papers by the Duke group, Baggerly says he found new errors and contacted both Lancet Oncology and the Journal of Clinical Oncology (JCO), where they had appeared. Both declined to publish Baggerly’s letters [...] Baggerly then learned that Duke was running three clinical trials using the Potti- Nevins approach to assign patients to treatment. He took a new tack: publishing a paper of his own. He and Coombes shared their critique of several papers published by Potti and Nevins with a “prominent” biological journal, he says, whose editors suggested that the paper was too negative. " http://www.sciencemag.org.edu/content/329/5992/614