The SNP genotypes you are seeing in Browse Raw Data might not match what you learn about the SNP from other sources such as dbSNP. This is because every SNP can be represented using either of the two DNA strands and this representation will often differ from database to database or publication to publication.
For example, 23andMe might report that a SNP has two versions, G and A. But other sources may report that the versions are C and T. Because of the double-stranded nature of DNA, both ways of reporting the SNP are correct: G pairs with C on the opposite DNA strand, while A pairs with T.
All of the genotypes displayed in Browse Raw Data are oriented with respect to the positive strand on the reference assembly of the human genome (build 37). Note that this could be different from how the SNP is oriented in dbSNP or how it might be presented in a publication.