It probably has to do with differrent measurements. There is journal impact (reflecting citations), eigenvalue (weighed factors for incoming citations), H-index (productivity of scientists and journals) etc. etc. So depending on what instrument they use, it might differ.

Another reason can be the difference in frequency of measurement (if researchgate updates once a week but other sites once per quarter, you can expect discrepencies). Yet another reason can be that number of citations is kind off vague. If they measure citations of other journals, it means that it depends on which journals are measured. Especially with temporary journals, single-edition journals and internet-journals, the amount of citations per article may vary between the extensiveness of the measurement.

The scientific impact of a journal can be measured in many different ways.

The Scimago Journal Ranking is based on an iteration method that takes into account the prestige of journals in a given field. Details on this method can be found in details on the SCImago web page (see link bellow).

An Impact Factor is a way of measuring a journal's impact as measured by Thomson Reuters using data from ISI Web of Knowledge. It calculates the mean sites per article published during the two previous years. Apparently, Researchgate reports this value. More details on how impact factors are calculated can be found on the second attached link. 

http://www.scimagojr.com/SCImagoJournalRank.pdf

http://admin-apps.webofknowledge.com/JCR/help/h_glossary.htm#JCR_year_g

Paul is right. Further, Impact Factor is a simple ratio used by Journal Citation Reports a Thomson Reuters trademark.

SJR is a more complex indicator developed by SCImago Labs to Elsevier's Scopus, and is based on an Page Rank style algorithm.

http://wokinfo.com/essays/impact-factor/

http://www.scimagolab.com/

http://www.scimagojr.com/

thank you very much for the links.... , so now can i use research gate impact factor to show impact of publications in my C.V. ?