Section 8: This section will teach novice researchers:

1. How to produce documents that will improve the reliability of their study.
2. How to ensure the internal and external validity of the study through the involvement of key stakeholders, community members (as in community-based participatory research (CBPR)), and external auditors.
3. Know what information to add to any report on findings to increase study validity.

Example of what you will find in this section:

Reliability

If the implementation team has followed all the steps in this training curriculum and filled out all the worksheets under the categories of “quality control” and performing tasks “systematically,” documentation needed for study replication is already in place. The team has worksheets on the research questions, pre-research, the target communities, the sampling plan[s], the research design, all forms of data collection, and all forms of data analysis. The team has documented all ways that the research was done systematically and each process was re-visited during specific intervals. The team has created a paper trail for replication of the study.
The team should now create a comprehensive file of these worksheets. The team may design any type of file that seems appropriate. A simple strategy would be to organize the sheets chronologically and keep them in a ring binder with dividers that separate the worksheets by relevant subject matter (e.g., pre-research, sampling, ethics, data collection, design, data analysis). An introduction to the file could then be added at the beginning with a table of contents. (Always keep photocopies of originals.)
In quantitative research measurement, other strategies can be added to increase reliability. These include:

• Test-retest comparisons. Here the implementation team can use the same measure (such as survey questions) on the same people several weeks apart, to see if the same results emerge.
• Interviewer comparisons. Here the team can compare results from separate raters who are observing or examining the same data independently.
• Internal consistency. The implementation team can compare multiple items that are expected to measure the same construct (or general category of information) in a scale.