Preparing for survey analysis

Survey analysis, as the name suggests is to analyze the survey data to draw meaningful conclusions. There are few issues to be managed before survey analysis is started. Let us have a look at the process that precedes survey analysis. This process is the actual survey process.

Survey is all about collecting data and information from the respondents to accomplish research objectives. One of the popular instruments to collect data is deploying a questionnaire and having it filled from the target population.. The questions in the questionnaire are such that these facilitate collecting data / information that will lead to draw conclusions, in turn help in accomplishment of research objective.

Questions in the questionnaire are of two types - open and close ended. Open-ended questions do not offer any response categories to the respondents. Close ended questions, on the other hand use pre-defined response categories.

Once the survey process is through (i.e., all the questionnaires are filled in and submitted in the office), the questionnaires are checked for validity and legibility. A process called 'editing' is deployed to complete responses (in case of semi-filled questionnaires) or writing is made legible (in case of illegible writing of the respondent). Once this is done, coding process starts.

Coding process is used for open as well as close-ended questions. Coding is nothing but using numerals to replace text. For example, brand A, brand B and brand C will be treated as 1, 2 and 3 respectively for data entry (and later for survey analysis). Therefore, it is apparent that close-ended question coding is easy. The complication starts with open-ended questions. Because the respondents are free to respond to a question in their own words, almost each open-ended question in all the questionnaires will have different responses. This is called ‘verbatim’. The coder converts all these open-ended question responses into manageable number of categories. For example, if the sample size of a survey is 1,000, one could expect 1,000 almost different responses to that question. However, the coder’s job is to ‘classify’ these 1,000 responses to may be 10 to 15 response categories. Questions getting long answers from the respondents need lot of time for coding.

Once the coding has been done, the data is entered into a software application. The data is mostly entered in spreadsheet format (row-column) where one row represents a record (one respondent) while the column represents one question. Thus a look at the spreadsheet data will show all the questions in columns while all the respondents in rows. Some of the popular software tools used are SPSS, Statistica, Sphinx Survey, etc.

Now, you are ready for the data (survey) analysis. Check out my other article on basics of survey analysis.