Probability and Evidence

 How can we collect evidence to form a conclusion? Evidence guide our thinking or our thinking guides collection of evidence. How much is sufficient evidence and what are relevant evidence both in the context of legal inquiry and auditing. As I am more in practice of auditing and also auditing exercises not exactly like advocating in favor of client, it is more unbiased perspective to look for evidence. 

Being skeptical about client professionally can be said to be little negatively biased yet we are not actively seeking for frauds, we are more of neutral evidence collector. Then it is perfectly suitable place to be to seek neutral evidence. Auditors are both fact finders and judge for collected evidence given the benchmark of financial reporting framework (with high degree of subjectivity). 

Going back to our first quest, whether evidence leads to thinking or thinking process leads to evidence collection. Let us take an example of bank branch with 23000 deposit holders, should we proceed it with a framework of this this and this area is high risk areas and we should focus only on them or we should go through whole data set with fresh insight and then proceed accordingly. 

For first approach we take previous years results and adjust it with recent developments or measures taken against past errors in terms of control development. But problem with this approach is, it narrow down sample size and all previously ignored issues will be ignored by recent evidence collector too. Sample size will be narrowed down as say previously 3000 accounts were opened and faulty account codes were used in 3 accounts out of 500 accounts sampled and hence concluded that 3/500 times only accounts will be faulty. Low risk but somehow erroneous finding of previous period will compel evidence collector to sample in similar way and due to low risk and consequence of previous period, for this period sample size will also be around 500 or lower than 500 or promotion of 500/3000 and previous periods 20,000 and unaudited 2500 accounts will go unaudited again.

let us take another approach, more inductive and from the whole dataset instead of using prior period knowledge. it is more costly approach and somehow ignoring efforts of previous periods. But the truth must be the first priority for evidence collector. which evidence to seek after under this approach will be determined by studying the whole 23000 accounts their changes and any risk identified should be limited to this period only as different economic environment for different period translate same accounts with different attributes of risk for the information use.