Different types of bias 1

Different biases in research might relate to the data collection, analyses of data, dissemination of results, and assessment of findings or arguments. 

During data collection, sampling bias of measured (often sampled) units might provide a skewed picture of an entire targeted population, for instance when a pollster samples only urban voters in an effort to predict an upcoming election while urban voters are known to prefer other candidates than rural voters or voters from sub-urban areas. 

Allocation bias can emerge in experimental studies (such as clinical trials) when participants differ systematically in relevant regards (e.g., in severity of disease) between the different groups or conditions (treatment vs. control). 

Attrition bias might emerge when a selective group of participants drop out during the study (e.g., when the most strongly afflicted patients drop out disproportionally over the course of a study of a certain medical treatment). 

Measurement bias might emerge when measurement errors are not random but systematic, which might be problematic for instance in educational testing when certain groups (e.g., minority students) systematically score lower than is expected based on their genuine level of academic proficiency (e.g., because of language problems).