Market Research Tutorial: Data Processing

Market research data processing can refer to different aspects of the entire market research analysis process. Most often, data processing and data cleaning are used interchangeably.

The market research tutorial will discuss data cleaning first.

Occasionally a respondent to your survey does not really know the answer to a survey question (and just guesses) or simply makes a mistake in answering, or would just rather not answer the question (make sure you put in a "don't know" category for the respondent to default to).

For example, suppose a survey question asks how many children live in the household. Instead of typing in a "2", the respondent (or the interviewer) slips and types in "22" accidentally. Now, it might be possible for 22 children to be living in a house (not a house we'd want to live in!), but we can probably assume that this was a mistake. Since we don't know the correct response for certain, we would change the response for that respondent from "22" to "missing." That way, this answer is not counted as part of the statistics generated for this question.

If we don't "clean" this market research data, when we calculated the average number of children living in respondents' houses, the number would be inflated. That could easily cause the researcher to make an incorrect conclusion based on the data. By "cleaning" the data, those responses would be "corrected," and the statistical software you are using (or spreadsheet or whatever) would not include that response in the analysis.

Really, it sounds more complicated that it is.

Now you might be thinking, oh #$%@ I have to go through every response from every respondent to clean the data?! No, fortunately you do not.

You generate (or have a professional analyst produce for you) what is called a frequency table or "freqs." (They not at all freaky.)

A freq is simply a count of each response category for each question in the survey. So, in the example above, the freq would list the the response category of 0 and the number of respondents who gave that answer. It would do so for all responses to the question. You can easily go through the freq to see what answers just out as mistakes. You "clean" them in the dataset and presto, you are ready to analyze your market research data.

Now, onward to data analysis, the second part of data processing.

Forward to Market Research Analysis

Back to Data Collection

Top of Market Research Tutorial Data Processing