3 ms·
In summary: Make sure your sample size is large enough so that it is possible to get a statistically significant result.
by jedberg 14y ago
In summary: Make sure your sample size is large enough so that it is possible to get a statistically significant result.
- keesschippers 14y agoIndeed, or, if that is not possible, consider lowering reliability. Business situations are different from academic settings where an alpha of 5% is more or less the rule. We do not look for publications in scientific journals, nor do we engage for instance in medical research – and even there -, we are in a business situation trying to optimize decision making. Up to us to decide for an alpha of 10% or 20%, as long as we are fully aware of the consequences of such choices. I am not advocating ‘going easy’ on statistical significance, I am in favour of weighing consequences and such can only be properly done if one masters the dynamics of reliabilty and power of a test. That could also mean you wanted a reliability of 99% and a power of 99%, it all depends on the risks you want take with your specific research questions.