A key feature of any data collection exercise is to create a medium for quality data to be collected and processed for insightful and impactful decision-making. There are some steps that researchers can undertake to ensure that the data received is of the highest quality possible:
The process of selecting the sample is crucial to the resulting data since it directly impacts its credibility and applicability. Sampling strategy is guided by research objectives, key research questions, and corresponding research design. To understand the prevalence of a certain indicator, the participants need to be selected on a more random level. This ensures generalizability and also eliminates selection bias.
But how does a researcher create an optimal sampling strategy?
In more focused studies, such as one on pregnant women in a catchment area, screening questions can be administered to identify the households where such a respondent may be found. For example, the question in this particular case would be, “Is there any member of your house who is pregnant?”. If the answer is yes, that household becomes part of the sample.
In order to get the most accurate data, it is crucial that the enumerator is well-versed with the language of the respondent; they should be able to speak the same language and develop a rapport with the participant to ensure their comfort. Proper pauses and breaks, if administered throughout the survey, avoid fatigued data.
During a survey, human error is expected and must be dealt with immediately. Technological resources such as computer-aided programming intervention (CAPI) softwares help minimise errors while punching data. They carry out range, consistency, and validity checks, producing error-free data points, resulting in more reliable results.
During a survey, collected data tends to follow a certain pattern. A feedback or reporting system can be developed in order to assess the quality of this data. This can help the researcher understand possible reasons for any data deviations and make necessary adjustments to the survey parameters to ensure consistency. Hence, concurrent feedback is a key feature of ensuring data quality.
At Sambodhi, the researchers check the data for the first 7 days on a regular basis, after which the field researchers communicate any adjustments that need to be made to the module to enumerators at the time of data collection.
Any researcher must benchmark their questionnaire and compare it with questionnaires from similar studies. Pretesting their tools also helps gain feedback and improves the quality of the material. Standard modules can also be referenced while developing the tool kit for the research study to be efficient and valid.
The collected data can be assessed for quality control on a weekly, fortnightly, or monthly reporting format. It must be followed by feedback; if the error rate exceeds 5-10%, remedial measures should be used in order to produce quality data.
Remedial measures must be put in place before the study begins; the researcher can decide whether they drop the numbers that produce an error rate surpassing 5-10% or they can choose to analyse this data separately.
Many of these protocols are set before they survey, but because the data collection exercise is subjective, there can be changes in the benchmarking standards depending upon the learnings of the field.
Aishwarya Bhatia, Sambodhi