Common source bias

She had seen the phone reviewed positively three times on different platforms. Each video featured the same influencer, holding the same device, making the same points about battery life against the same background. She took the repetition as confirmation: multiple independent voices agreeing on the same product.

She had not noticed that all three videos used the same script. The repetition was not three opinions arriving at the same conclusion. It was one promotional contract distributed across three channels.

The evidence that felt like convergence was a single source, multiplied.

What It Is

Common source bias is the tendency to combine or compare pieces of evidence as though they were independent when they in fact share a common origin - the same dataset, the same measuring instrument, the same methodological pipeline, or the same source organisation.

The result is an inflation of the apparent evidential weight. Common source bias refers to the inflation of observed relationships between variables due to shared measurement method or source. When two pieces of evidence share a source, their agreement tells you something about the consistency of that source - not about the truth of the claim they both make.

Why It Feels Like Evidence

The mind favours familiarity and coherence. Evidence that arrives from multiple directions feels more credible than evidence from a single point, because genuine replication - independent experiments or observations arriving at the same result - is one of the most reliable signals in empirical reasoning.

But the appearance of replication is not the same as replication. When multiple pieces of evidence share an origin, they inherit each other's limitations. Errors introduced by a shared methodology, shared dataset, or shared measurement approach are repeated rather than cancelled out. The agreement is real; what it means is not what it appears to mean.

Common source bias can lead to inflated correlation coefficients, potentially altering substantive conclusions. In research synthesis, a meta-analysis drawing on studies that use the same dataset may produce results that look more robust than they are.

Where It Shows Up

In research evaluation. A systematic review that draws heavily on studies from a single laboratory, using a single instrument, or drawing on a single large dataset may show strong consistency - not because the finding is reliable, but because the source is consistent. The consistency is methodological, not evidential.

In business and policy decisions. When a proposal is supported by multiple reports that all draw from the same survey, the same vendor's data, or the same model, the apparent weight of evidence is misleading. The agreement among reports reflects the single input, not independent validation.

In consumer decisions. Product reviews, endorsements, or testimonials that appear across multiple platforms can stem from a single promotional arrangement, editorial decision, or coordinated campaign. The volume is real; the independence is not.

What It Is Not

It is not the same as publication bias. Publication bias is the tendency for statistically significant results to be published and non-significant results not to be - which skews the visible evidence base. Common source bias is a different error: it concerns the independence of evidence, not its statistical threshold.

It is not limited to the same authors. Studies can share a common source through shared datasets, shared instruments, or shared methodological conventions even when conducted by entirely different research teams. Authorship overlap is one signal, not the only one.

It does not automatically invalidate findings. Recognising common source bias does not mean the evidence is wrong - it means its independence is limited. The appropriate response is to adjust the weight given to the combined evidence and seek genuinely independent sources before drawing strong conclusions.