Garbage In, Garbage Out: A Guide to Good Data Collection

Before any powerful analysis can be done, you need good data. For many NGOs and research organizations, surveys are the primary tool for data collection. However, poorly designed surveys can lead to biased, inaccurate, and ultimately useless results. This is the principle of "garbage in, garbage out." Here’s how to ensure you're collecting quality data.

1. Define Your Research Question First

Don't start by writing questions. Start by clearly defining what you want to learn. Every single question in your survey must directly contribute to answering your core research question. If it doesn't, remove it. A shorter, more focused survey will yield higher quality responses.

2. Avoid Leading and Loaded Questions

The way you phrase a question can dramatically influence the answer.

Leading Question: "Don't you agree that our new health program has been a huge success?" This question pressures the respondent to agree. A better question is: "How would you rate the success of our new health program on a scale of 1 to 5?"
Loaded Question: "Where do you enjoy drinking your evening beer?" This question assumes the respondent drinks beer. A better approach is to first ask a filter question, like "Do you drink alcoholic beverages?"

A person carefully crafting questions on a clipboard.

3. Understand Sampling

You can't survey everyone. Sampling is the process of selecting a representative group from your target population.

Random Sampling: This is the ideal. Every person in the population has an equal chance of being selected. This minimizes bias and allows you to generalize your findings to the entire population.
Convenience Sampling: This involves surveying people who are easy to reach (e.g., standing outside a supermarket). While easier, it is highly prone to bias and you cannot confidently generalize the results.

The method you use to sample is often more important than the number of people you survey. A small, well-chosen random sample is far more valuable than a large convenience sample.

4. Pilot Test Your Survey

Before deploying your survey to hundreds of people, test it on a small group first. Are the questions clear? Is the language easily understood? Is it too long? This pilot test will help you identify and fix problems before they compromise your entire dataset.

A diagram showing how a small, random sample can accurately represent a large population.

Collecting high-quality data is a discipline. By following these principles, you can ensure that the data you gather is a solid foundation for generating credible and actionable insights.