Data quality is one of the most important, if not essential, aspects of our business. Therefore, we carry out market research always aiming for the highest level of quality, using innovative methodologies and best practices.
Our method for data quality
What matters are the premises. Having a clear idea of where we start and how we intend to act, allows us to define credible and achievable goals. For us, doing research means to be aware of the pitfalls of the path and to guard the most vulnerable joints of the system. We have chosen to adopt the approach of the scientific community of statistical survey methodologists as our stantard, which is the result of decades of experience and reflection. It is called Total Survey Error and it's a map that reproduces the territory of research, our field, the one where the meta data of a statistical research can be collected.
Total Survey Error
Total survey error (TSE) is a term used to refer to all sources of bias (systematic error) and variance (random error) that can affect the validity (accuracy) of the survey data.
The first such approach was developed by Robert M. Groves in the late 80s in an attempt to pinpoint all sources of possible errors in statistical surveys. Lavrakas has extended this cognitive map to qualitative research by enriching it with contributions from the last 25 years of research. The result is a detailed and effective recognition system that divides the possible errors mainly into two types: Measurement & Representation. By analyzing the various phases of the research in more detail, the two types can be further divided into sub-categories. Regarding the Errors of Measurement we have: specification, measuremet, processing and inferential errors. While for what the Errors of Representation are concerned we have: coverage, sampling, non-reponse and adjustmenterrors.
Our approach to Total Survey Error
What we do is develop (manual or automated) control / monitoring procedures on every single node of the system.
For example, regarding the Measurement Error of the column Errors of Measurement we introduced the standardized survey interviewing. This allows you to promptly verify the learning of the detectors during the training phase. The randomization of the answers in a questionnaire implemented on Limesurvey, also has the purpose of reducing this type of errors. The same applies to the responsiveness of the questionnaires, that is to say their adaptability to the various devices which allows a precise reading of the questionnaire by the respondent and a consequent reduction of the Measurement Error.
All automated survey monitoring procedures implemented on Cloudresearch have as their purpose the control of Processing Error. The adoption of the mixed-mode instead allows you to reduce both the Coverage Error and the Non response Error, as far as the Errors of Representation are concerned. Finally, the sampling management of our software keeps the Sampling Error under control.
The quality of our management system for social research, opinion and market activities has been certified in compliance with the regulations UNI EN ISO 9001: 2015 by the following organizations: