If the observations are recorded, the aspects of interest can be coded by two or more people. If both (or more) observers give the same scores to the observed material (this can be done by calculating a correlation between the scores) there is high interobserver reliability.
To increase interobserver reliability, the observers must receive clear instructions. They may even need to be trained to correctly assess the material to be observed.
Related topic to interobserver reliability:
- Observation (as a method to gather data)
- Measuring with instruments
- Transferring data
- Generating data