Research data is prone to integrity risk

If you ever commissioned a research survey and you never questioned the integrity of the data you received, unknown to you, you might have been a victim of data fraud.

A UK-based research agency says data fraud in survey research has in recent years gone up to as high as 30 per cent!

This statistic is alarming, considering that research findings are at the heart of innovation, evidence and decision-making.

Picture this: Company X operating in Kenya is in the business of fast-moving consumer goods.

The business development manager is considering packaging, labelling and selling goods under the company’s name.

Own-label has been tried and tested in overseas markets with resounding success.

Given that it offers the same or even better quality than regular brands but at a lower price, it’s expected to fly off the shelves. The boss commissions a research study.

NEW PRODUCT

The research methodology entails 100 intercept interviews with customers at each of its 30 branches countrywide using a quota sampling strategy so that different demographics are adequately represented.

Enumerators complete the surveys. The data is analysed and reports churned out.

Overall, 50 per cent like the idea and say they would definitely purchase the own-label and make it their regular brand.

Another 20 per cent are somewhat convinced, saying that they would try it first, while a similar proportion adopts a wait-and-see stand. Only 10 per cent reject the idea outright.

Company X invests heavily in the label. But six months later, the product is stuck on the shelves. The BDM is at pains to explain what went wrong. He commissions another study.

FORGED DATA

The new research consultant requests access to the dataset from the previous study to interrogate it.

Data diagnostics combined with telephone back-checks reveal a cocktail of fraudulent elements related to sampling, interview duration, and interview location.

It is established that age and gender was forged, with more men interviewed than required but recorded as women.

The age demographic was also much lower than required — younger people interviewed but recorded as older with the more mature consumer left out.

The time stamps show that some respondents were only interviewed for shorter than was required.

Some enumerators asked respondents only a couple of introductory questions then filled in the rest of the form.

RESPONSIBILITY

Finally, there are ‘ghost’ respondents. GPS coordinates reveal that some interviews were conducted away from the outlets with fictitious respondents.

The fraudulent 30 per cent are a sub-set of the enthusiastic 50 per cent who said they would definitely buy and switch to label X! Only 20 per cent are truly positive about the idea.

The damage is far-reaching in terms of millions of shillings invested in the research and packaging, branding and distribution of the product.

Simple control measures like supervision of enumerators, telephone back checks and scrutiny of surveys are missing.

The onus is on you who commissions research to hold the consultants you hire accountable so as to secure the veracity of your data.

Mr Masita is the CEO of Research Plus. [email protected].