How do you know what types of statistical methods to use when you have a pile of survey data? You have set out with worthy goals of comparing, contrasting, analyzing, correlating, summarizing, predicting, examining, and describing? What next?
To determine the appropriate statistical methods, you must know what type of measurement scales your data derives from: nominal, ordinal, or numerical. You must also know how many independent and dependent variables you have. It is also important to determine if the data fit the survey method.
A review of terms to determine where your survey fits to apply the appropriate methods:
Nominal data has no numerical value. Try remembering this by the definition of the word “nominal”, which has a root word of “nomin” or “name”, as in not number. If you were going to “nominate” someone for an award, you would select a name. One example is data such as male or female.
Ordinal data has order to it, which again is easy to remember by the name of the scale. If you have a Likert scale survey, this is your area. The order may go along the lines of strongly agree, agree, neither agree nor disagree, disagree, strongly disagree.
Numerical data is associated with numbers, but these numbers do not have to have a given order. If you are looking at temperature, age, BMI, or other exact measurements, then numerical data is your focus.
Independent and dependent variables are easy to confuse. The dependent variable is the result. I remember it this way: When I measure the Dependent variable, I will have my Result and be DR. Craddock. D for Dependent, R for Result, DR for Doctor. Any other variable is independent. However, officially the independent variables are those that predict what the response of an intervention will be.
Now that your terms are straight, we will take the next post (or more….) to look at what kinds of statistical methods are appropriate for examining the survey data based on the kind of data you have and the number of variables.