By: Caroline Boone
This week I am in the business of interpretation. I am working with a set of survey responses from a professional development training for which the Voinovich School is doing an evaluation. The end goal is to summarize the responses and analyze them for trends and themes so that this information can be included in our final report.
But first, someone had to go through the hard copies of the surveys and record the responses in an Excel spreadsheet. This means that someone read the handwriting of about 25 doctors, nurses, and healthcare providers (to call it “handwriting” is generous). That “someone” being me, I worked through the stack until only the most hieroglyphic markings remained — these I took to my supervisor who is fluent in Chicken Scratch.
With the responses now neatly entered and organized, I can begin to look for themes in the data.
Going through each question at a time, I note responses which appear more than once. These are not necessarily the same word-for-word, but key phrases and responses with the same meaning emerge. Typically, there are two or three responses that will stand out as having been mentioned with greater frequency (like hashtags trending on Twitter). This sort of data interpretation is useful for getting an idea about the group overall. But it is also a method of interpretation that is easily influenced by the researcher because, to a certain extent, the research decides what the respondent means in their answer. “Are those two responses saying the same thing?” For this reason, it is important to have multiple people involved in interpreting the data. Thankfully, my supervisor is a pro at this sort of interpretation too and has provided some helpful guidelines for assessing themes in the data.
Once I have gone through each part of the survey, I can start the write-up. Since this is a summary of the results of the survey, it is important to discuss each part, and to note the most prominent responses, without excluding other responses. As with any kind of interpretation, be it data, language, or chicken scratches, interpreter discretion should err on the side of preserving the original