Imagine a situation where you have invested your hard earned money on the basis of thorough research. You must be feeling good about yourself, your hard work and your investment. Now you suddenly find out that the data you used for research is inaccurate. What now? You may still hope that your investment strategy works out but you will not be sure. You could have simply let your pet monkey pick stocks for you and ended up in the same situation—with hope and no confidence. Similarly when you embark on a marketing plan or marketing activity without checking your data you may end up with just hope and no confidence. It is necessary to clean your research data to prevent the occurrence of such a situation.
According to SMStudy® Guide-Book 2, Marketing Research, data cleaning is the process of checking the data for omissions, consistency, and legibility. It involves checking for errors and omissions on questionnaires or other data collection forms. When researchers discover a problem or inconsistency, they need to make necessary modifications in order to ensure the data is readable. If a particular data point is beyond comprehension or cannot be corrected (by re-administration), it needs to be excluded from the analysis.
When raw data is collected for the first time, the researcher needs to examine the data points for inconsistencies and determine if an inconsistency is an outlier and should be discarded, or whether the entire questionnaire should be discarded and/or re-administered. The researcher needs to follow established guidelines of data and quality checks to determine which answers are inconsistent and choose an appropriate course of action. Sometimes, the data set or responses may contain only a partial or a vague response. If possible, the researcher should contact the respondent to gain clarity. In some situations, the predefined protocol is to not pursue the missing data and simply leave the question blank. In the case of multiple or duplicate markings on a single question, researchers must carefully examine the answers and keep the most accurate one based on their judgment. If no clear interpretation can be made, the response should be discarded. If a respondent provides a long statement, the researcher may need to decipher the underlying meaning to see if the answer is appropriate for that category. In some instances, researchers may decide to change the category code based on their judgment. This also applies to any out-of-place responses.
Data cleaning is a critical process for any marketing research project. Faulty or suboptimal decisions can be taken if data is not processed and cleaned properly. In such a scenario you may either go ahead with the decision or decide to conduct research again. The only way to avoid this scenario is to become a cleanliness freak….for data.
Please visit www.smstudy.com for more details on Data Cleaning.
