Due to shortcomings in the method of selecting research subjects, there are differences in certain characteristics between those who are selected and those who are not selected, resulting in the selected research subjects not being representative of the population, thus leading to bias in the connection between exposure and disease.

Admission rate bias (Berkson bias)

Current case-new case bias (Neyman bias)

Detection of symptoms bias (exposure bias)

attrition bias

non-response bias

During the research implementation phase, due to flaws in the methods of measuring exposure and outcomes, research subjects were misclassified, causing the research results to deviate from the true situation.

indifferent misclassification

differential misclassification

from research subjects

from researchers

from measuring instruments

Use blind methods to collect data whenever possible

Use objective indicators whenever possible

Improve inquiry skills or investigative techniques

Standardization of measuring instruments

is a risk factor for disease

Statistically related to the exposure studied

Not an intermediate cause in the causal chain between exposure factors and disease

Strictly follow the principle of random sampling to select research subjects

Strict inclusion criteria

Use a matching approach to select controls

Stratified analysis, standardized or multi-factor models are used for data analysis.

step

Experimental research > Observational research