Page 314 - CyberSecurityEssentialsEnglish

TABLE 7-11: FIRST QUESTIONS ABOUT RISK OF BIAS BY TYPE OF SOURCE

Type of source Appropriate questions about bias

A trial or intervention Was allocation randomized? Was attrition large? Was the
study outcome measured the same way in both arms?

Was confounding addressed? Could the selection of the sample
An observational study
or the measurement change the result?

Is the sample justified? Is the coding deep? Is there reflexivity
A qualitative study
and support from quotations?

Which body produced it? What was the purpose? How was the
An institutional report
indicator defined? Is there an interest in a particular result?

TABLE 7-12: FACTORS THAT LOWER CERTAINTY IN A BODY OF EVIDENCE
Factor Its practical meaning

Could the design or the execution push the result in a particular
Risk of bias
direction?

Do the findings agree across studies, or diverge without
Inconsistency
explanation?

Is the evidence about a population, intervention, or outcome
Indirectness
different from our question?

Do the sample sizes or the intervals make the conclusion
Imprecision
unstable?

Publication bias Might negative or unpublished studies be missing?

Say how certain the evidence is in modest language: “the evidence indicates”, “the evidence
is limited”, “current evidence does not support wide generalization”, rather than in

categorical language the sources do not carry.

313