==== Observational Bias ====

**Observational bias** (also called **observation bias** or **ascertainment bias**) refers to systematic errors in the measurement, recording, or interpretation of data that occur due to the **observer's expectations**, **knowledge**, or **study design**.

It can affect the validity of results in both **clinical** and **epidemiological** research, especially in **non-randomized** or **open-label** studies.

=== Types of Observational Bias ===
  * **Detection bias**: Outcomes are more likely to be observed in one group due to **increased monitoring or surveillance**.
  * **Observer bias**: The person collecting data **intentionally or unintentionally distorts** measurements due to prior beliefs or expectations.
  * **Reporting bias**: Selective recording or emphasis of certain outcomes over others.
  * **Recall bias** (in self-reported data): Patients may remember or report information differently depending on exposure or outcome status.

=== Example ===
In an unblinded clinical trial, a physician who knows which patients are receiving the active drug may **more closely monitor** them and detect side effects that go unnoticed in the control group — artificially inflating adverse event rates.

=== Prevention Strategies ===
  * **Blinding** of participants and investigators
  * **Standardized protocols** for data collection
  * **Objective outcome measures**
  * Use of **independent adjudicators**

=== Related Concepts ===
  * [[selection_bias|Selection Bias]]
  * [[information_bias|Information Bias]]
  * [[confirmation_bias|Confirmation Bias]]