3 Tips to make sure the data you collect are usable
It might seem obvious, but it isn't always clear whether the data you plan to collect will answer your research questions in the way you expect. Some strategies to help you get good data are below.
- Document the following and discuss with your advisor:
- The research questions
- The analytical process for answering the questions above. (i.e., Which statistical tests will you use? What method will you use to code the data?)
- Describe the characteristics of the data required to conduct those analyses (i.e., Do you need ordinal or ratio data elements? Will you code phrases, sentences, or paragraphs? What coding schema will you use?)
- Make sure the tool, instrument, or protocol you use to collect data produces it in a format that you can easily process and analyze.
- Before you start collecting data for real, test (pilot) the entire process for data collection, processing, analysis, and visualization to make sure your plan will actually work. A small sample of data usually works fine. Your classmates can be great test subjects.
Take snapshots of your data at these key points, at least. A snapshot is just a (locked) copy of your data files, saved to a backup storage location. These files should not be changed; they are for reference only.
- raw data, as collected (before it is cleaned and processed)
- your cleaned and processed data (before it is analyzed)
- the data used for your analysis, any analytical scripts or procedures used, and detailed notes about why data were selected for analysis or excluded
