When I looked at the Minor Allele Frequences of some SNPs and I compared them to the Population Diversity Overview below, I noticed that some MAF were not correct.
But when we look at its population diversity, you can see that C is totally not the minor allele, but G. According that population diversity, C has an allele frequency of 0.9.
The confusion here is because dbSNP are displaying the reverse stranded alleles (see table of genomic positions on their page), whereas 1000 genomes have reported the MAF for the forward strand allele. Here's the same variant in Ensembl, where all variant alleles are reported as the forward strand. As you can see, the alleles are listed as G/C with the MAF reported as C 0.04. If you click on population genetics, you'll see the frequency of the two alleles across the populations reflects this MAF.
Both Ensembl and 1000 genomes only report alleles on the forward strand, as this makes everything more consistent and clear. Unfortunately, dbSNP have many old records in their database which are reported on the reverse strand, which can be very confusing. When these are imported into Ensembl and 1000 genomes, they are moved onto the forward strand.