Data Types

Data type. The raw data the method produces is analyzed in the results section. The analysis done depends on the data type, which can be qualitative or quantitative. Qualitative data requires human interpretation, e.g. an interview record. Qualitative analyses include Content analysis, Narrative analysis, Discourse and Framework analysis, and Hermeneutics. Quantitative data in contrast is numeric, so statistics can be used. The statistics available depend on whether the data is:

  • Categorical. A set of values with no particular order, e.g. PC, Laptop, Tablet, Handheld.
  • Ordinal. A set of values with a rank order, e.g. income levels of Low, Medium and High.
  • Interval. A set of numbers where the interval between each consecutive point of measurement is equal to every other. The values can be discrete, e.g. number of children, or continuous, e.g. time of day, 1pm, 2pm etc.
  • Ratio. A set of interval numbers that have a zero point, e.g. height.

The data type affects the amount of information, e.g. measuring job experience in years gives more information than self-ratings of Expert, Competent or Novice but the latter may have more validity as years doesnt always make an expert. Note one can convert interval data to ordinal, e.g. call 1-3 years Novice, 3+ to 6 years Competent and 6+ years Expert, but not vice-versa. In general, one can always convert more information to less but not vice-versa. Analyze data according to data type.

Data conversion. Data conversion turns raw data into the base data set that gives the findings. It can take some time, e.g. transcribing an interview audio into text. Only collect the raw data you need, e.g. don’t record an interview on video unless you are doing a video analysis. Data conversion raises two issues:

  • Missing data. If parts of an audio are unclear or a questionnaire wasnt answered that is missing data. Research doesn’t throw information away so missing data is recorded. Record different types of missing data differently, e.g. an unclear audio section is coded differently from a recording failure, and not answering a question at all is coded differently from ticking a Don’t Know box. If missing values are coded as 0 and 9, these values are excluded from any calculations.
  • Outliers. An outlier is a data value that is significantly different from all the others. If attributed to an equipment failure it may be removed but in large data sets outliers are expected, e.g. some people are in fact eight feet tall.

Briefly state how raw data was converted and how missing values were handled.

Science  Writing  Review  Glossary Checklist  Next