Different kinds of survey questions yield data in different forms. These different forms of survey data can impact the type of analysis that you can do and the richness of the personas that you can create. Survey2Persona prefers Likert scale survey question.
Here is a quick guide to a few of them.
Categorical (nominal) data: exists in categories that have no hierarchical or scale relationship to each other. No item is more or less, better or worse than the others. Examples would be primary biological sex (e.g., male, female, other), age groupings (e.g., 18-24, 55-64), or countries (e.g., China, India, USA). Multiple choice questions and drop-down list questions are often used to produce this kind of data. This data is usually analyzed via counts – e.g., number of males, number of people 18-24, or number of people from the USA.
Natural language data: are words that a respondent writes. This type of textual response is usually collected in an open field (text box) question format. It is always good practice to end a survey with an open-ended question, allowing respondents to present relevant information that the survey may have missed. This data is usually analyzed using natural language processing (NLP), such as topics, sentiment, or stance.
Ordinal data: has an intrinsic rank, such as quantity, quality, degrees, or preference – but there is no external comparison scale. Levels of agreement questions are ordinal data questions (i.e., how strongly a person agrees or disagrees with a statement). Likert scales or ranking questions are often used to collect this type of data. This data is valuable as one can use statistical methods for analysis.
Scalar data: deals with relative quantity and quality, including ranking, what comes with established scale, such as age (expressed as a number), temperature (in Celsius), percentages (out of 100), or time (such as in minutes). Drop-down, sliding scale questions, and even open-ended questions can collect scalar data. Again, this data is valuable as one can use statistical methods for analysis.
