Political Science

Developed in the 1970s by the Central Intelligence Agency, the ACH or Analysis of Competing Hypotheses is a reliable unbiased methodology for assessing multiple competing (or contradictory) for observed data. In general, this method helps the analyst overcome or at least minimize some of the cognitive constraints inherent in any intelligence analysis. According to Heuer and Richards (1999), there are five processes involved in ACH. Here are as follows

Hypothesis. Hypothesis is defined as constructive guess based on preliminary observed data.

Evidence. This step involves listing or classifying evidences and arguments for and against each of the considered hypothesis

Diagnostics. This step is also called elimination. The analyst, with the help of a complex matrix, applies evidence against each hypothesis in an attempt to refute as many considerations as possible.

Refinement. The analyst evaluates the findings. A thorough identification of gaps and overestimate is done to eliminate as many of the remaining hypotheses as possible.

Inconsistency. Inconsistency is defined as the over-duration of expected value against actual value. The analyst draws tentative conclusions about the likelihood of each hypothesis.

Sensitivity. This step involves general testing of conclusion. With sensitivity analysis, each hypothesis is weighted  key evidences are introduced to test the stability of the hypotheses.

Evaluation. The last step involves a general assessment of alternatives. A list of the rejected hypotheses is prepared to provide a detailed description of the process.

ACH reduces biases in the research process by 1) removing hypotheses which do not meet assumptions, 2) utilizing non-variable matrices, and 3) providing venue for group analysis. By removing faulty hypotheses, the analyst will likely to choose reliable and valid hypotheses (which follow from preliminary findings). Utilizing non-variable matrices allows the analyst to view each hypothesis in different contexts (this prevents the analyst from making faulty assumptions). Individual analysis is often vulnerable to measurement error. Group analysis allows the research process to be both valid and suited to preexisting conditions.

Bias reduction is not sufficient though to meet the requirements of a good intelligence analysis (Jones, 1998). Validity and reliability are the other requisites. Reducing bias directly translates to increased validity (as conclusion follows from factsassumptions). The question is are the factsdetails reliable In the diagnostic process (elimination), data are screened based on their authentic applicability. Evidence is weighed against contradicting evidences to measure the likelihood. If the evidence is likely to be true then it is reliable.

Thus, in the whole process, bias is reduced while validity and reliable are optimized.