Surveys and other kinds of quantitative data are often already in forms suitable for algorithmic data analysis. That is, as the surveys are presented so that each variable (like a survey question item) is clearly identified and has documented values, the data are ready. To build intuition, any data presentation that could be used to run statistical hypothesis testing or regression analysis is also suitable for algorithmic data analysis.
Often survey researchers focus on the differences between the levels of measurementâif the data are in a nominal, ordinal, interval or ratio scale. This is essential for these approaches as different measurement levels allow for different mathematical operations and, thus, different statistical operations. For example, a nominal scale only allows examination if two objects are the same or different but does not allow sorting them. An apple cannot be compared with an orange. Other kinds of scales measure more extensive analysis. For example, the commonly used Likert five-step scale (e.g. strongly disagree, disagree, neither disagree nor agree, agree and strongly agree) allows clear comparisons in surveys and often is used to measure means between the groups. This means that it is seen as an interval-level measurement. However, there is an endless debate about if this is true or if a Likert scale should be seen as an ordinal or even nominal variable. add some references The importance of these to algorithmic data analysis depends on the algorithmic analysis approach, but most often these are not as essential as in traditional statistical tests.