research bias

research bias

1. aim for high participation rates

   • people who respond will have dif. characteristics than those who don't
   • samples are as representative as possible- applicable to generalize to society
   • ensure strict, explicit criteria
     • another party to collect data so the researcher doesn;t conciously or unconciously treat participants differently
   • enuring standardised data collection forms and procedures
   • enuring these are used uniformly accross all participants

2. confounding

   • too confuse
   • a variable or factor that confuses the relationship being examined
   • controlling confounding
     • done through the design of study methods
       • happens before the data is collected or while the data is analysed
     • study methods used
       • randomisation
       • restriction
       • matching
       • analyses stage of research

3. if there is bias the finding may not be reliable

4. selection bias

   • bias in the procedures used to select particpant
   • bias in the factors influencing participation
   • problem- differences in characeristics between study participants and non study participants
   • problem- people selected for one group in a study differe from people in another comparison group
   • self selection bias
     • participants in a study select themselves
       • may have many characteristics which differ between people who participate and people who don't
         • outcome does not reflect the whole population
   • healthy worker effect
     • population contains people who cannot work due to ill health
     • similarly the population has many people who are not well enough to particpate in a research study
     • particpant are likely to be healthier than the general population

5. measurement bias

   • occurs when individual measurements of disease or exposure are;
     • inaccurate
     • do not measure what they are supposed to measure
   • misclassification bias
     • occurs when people are miclassified according to an exposure of outcome
     • participant may not have been truthful
     • error by researchers
     • disease outcome misclassification
       • a person may have several symptoms of a disease but not formally diagnosed due to not meeting all of the diagnosis criteria
   • recall bias
     • occurs when information obtained from participants is not accurate because they have difficulty recalling events in their past
     • people with the disease may recall differently to people without the disease