The simple explanation here is that Google’s search robots crawl at differing frequencies to different web locations for different users — This is done to make web searches faster, but they also build citation counts this way (see http://bit.ly/1cMBuuW). Some journal’s servers are more amenable to crawling than others — I recently had a paper from 2012 through Wiley which finally appeared on my Google Scholar page. For your own record, I would use the highest citation count (the citation numbers and titles from Keith’s scholar page) assuming that index was the most recently crawled.