Elklit argues in his “rebuttal” that purposive sampling was the only feasible option for IREC, given its  mandate. KPTJ disagrees with that statement, and will provide a better way forward below. However, let’s begin by assuming that Elklit is correct, and that purposive sampling was the only way to go.

Download Full Response Here

 What  implications does his assertion have on the credibility of the IREC report? First of all, section 6.5.1 (p. 129), entitled “The random nature of the errors affecting the presidential election in the eighteen constituencies analysed,” should be struck from the report. In his response, Elklit states “KPTJ claims that IREC should not infer population parameters from results based on analysis of a non-random sample. Theoretically, KPTJ has a point, but only theoretically.” Our “theoretical” point has a clear implication: we cannot reliably estimate  population parameters from a non-random sample. Section 6.5.1 makes implicit claims about a specific  population parameter: the average difference (for each candidate) between the ECK tallies and IREC’s  re-tallies. IREC appears to claim that, because the differences between ECK numbers and IREC’s numbers do not systematically favor any one candidate, that we can conclude that discrepancies did not favor any one candidate.

Download Full Response Here

Tweet