Data errors can be broadly categorized as — sample and non-sample errors. Sample errors are the inaccuracies arising from taking only a sample of the target population. These errors, as we have seen, may be reduced by increasing the sample size or improving the sample design, and accuracy standards can be set in terms of acceptable error levels at some confidence interval.
Apart from sample errors, research is subject to a variety of other errors commonly referred to as non-sampling errors. They arise in all forms of studies (including census) both from systematic and random causes. These errors tend to be difficult, if not impossible, to measure.
Non-sampling random errors (Exhibit 33.8) are the unpredictable errors occurring during data collection and data processing. For instance, incorrect recording of data by an interviewer, or incorrect response by a respondent, or incorrect computation of some metric during processing. They lead to an increased variability in the results, though they tend to cancel out if the sample is large.
Non-sampling systematic errors (Exhibit 33.6) tend to accumulate over the entire sample, making the results unrepresentative of the population. For example, poor wording of a question may inadvertently influence responses, or for example, coverage issues where some targets with different properties cannot be reached. These types of errors are of greater concern since unlike random errors, they often lead to a bias in the final results. Moreover unlike sampling inaccuracies, bias resulting from systematic errors cannot be reduced by increasing the sample size.
The nature of a study needs to be carefully considered, in order to minimize bias. For instance, for shopper trend studies, respondents need to be selected in a manner that neutralizes the bias resulting from their proximity to shopping outlets; this because shopping habits are so heavily dependent on store location. It will usually result in significant increase in the total sample size.
Non-sampling error can be reduced by maintaining reliable survey frames, well designed and properly tested questionnaires, good training of the data collection team (interviewers, auditors), monitoring and call-backs, and high standards in measurement and processing systems.
Note: To find content on MarketingMind type the acronym ‘MM’ followed by your query into the search bar. For example, if you enter ‘mm consumer analytics’ into Chrome’s search bar, relevant pages from MarketingMind will appear in Google’s result pages.
In an analytics-driven business environment, this analytics-centred consumer marketing workshop is tailored to the needs of consumer analysts, marketing researchers, brand managers, category managers and seasoned marketing and retailing professionals.
Is marketing education fluffy too?
Marketing simulators impart much needed combat experiences, equipping practitioners with the skills to succeed in the consumer market battleground. They combine theory with practice, linking the classroom with the consumer marketplace.