Blog › Guide
How to choose a data analyst for your research or NGO project
You have survey returns, interview transcripts or a spreadsheet of records, and you need someone to turn them into findings you can trust. The right data analyst saves you time and protects your credibility with funders, boards and readers. The wrong one can leave you with tables nobody can explain and conclusions the data cannot support.
This guide is for organisations, NGOs, government teams and researchers who are about to hire help with data analysis. It covers what to prepare, what to ask, what you should receive, the warning signs, and what affects the cost.
Start with your question, not the software
Before you contact anyone, write a short brief. One page is enough. It should say:
- the research question or project objectives the analysis must answer;
- the type of data: survey, interviews, focus groups, administrative records or a mix;
- how the data was collected and roughly how many records or transcripts there are;
- who will read the results: managers, a donor, a ministry, a journal;
- the format you need, such as a donor template or report chapter;
- your deadline.
A clear brief gets you clearer quotes. It also shows you quickly which analysts read it carefully.
Check that they understand research design
Good analysis depends on how the data was produced. A capable analyst will ask how respondents were selected, whether the sample was drawn in clusters such as villages or schools, whether some groups were deliberately over-sampled, and what the questionnaire looked like. These details change which methods are valid and how confident you can be in the results.
Be wary of anyone who wants only the dataset and never asks about the study behind it. If your data came from a survey, our guide on designing a baseline survey for an NGO project explains the decisions an analyst will want to know about.
Match the tools to the job: SPSS, Stata, R, NVivo and Excel
Most social research and NGO data can be handled well in one of five common tools. Each has strengths.
- Excel suits data entry, simple cleaning, summary tables and charts. It is a good choice when your own team must update the files later. It is weaker for advanced statistics and for keeping a record of every step.
- SPSS is widely used in the social sciences. It works through menus and through written syntax, and handles survey data, descriptive statistics and common tests well.
- Stata is strong for survey data, regression and data collected over time. Its do-files record every command, so the work can be checked and repeated.
- R is free and open-source software for statistics and graphics. Scripts make the analysis fully reproducible, and it produces high-quality charts.
- NVivo is used for qualitative data such as interview and focus group transcripts. It helps organise coding and themes; the thinking still comes from the analyst.
The software matters less than two things: whether the analyst can explain why a method fits your question, and whether the outputs open on your team's computers. If nobody on your side uses Stata, ask for tables in Word or Excel as well as the do-files.
Questions to ask before you hire
- What do you need from us to start: the questionnaire, the codebook, the sampling notes?
- How will you check and clean the data? What will you do with missing answers and unusual values?
- Which methods would you use for our objectives, and why?
- How will you explain the results to readers who are not statisticians?
- What exactly will you deliver, and in which formats?
- Will we receive the cleaned dataset and the syntax, do-files or scripts?
- How will you keep our data confidential and secure?
- How many rounds of revision are included, and what is the timeline?
- Can you show an anonymised sample of similar work?
Listen for specific answers. "I will run the appropriate tests" says little. "Your sample was drawn by community, so I will account for that clustering when estimating the margins of error" tells you the analyst has thought about your data.
Agree the deliverables in writing
Many disputes start with different ideas of what "the analysis" means. Write down what you will receive. A complete package usually includes:
- a cleaned dataset and a short log of the changes made;
- a codebook explaining every variable and code;
- the syntax, do-files or scripts used;
- tables and charts formatted to your style or template;
- a plain-language interpretation of the main findings;
- a short note on methods and limitations;
- for qualitative work, the coding framework and themes with anonymised illustrative quotes;
- a call or meeting to walk through the results.
Protect your respondents' data
Your respondents trusted you with their information. Share only what the analyst needs. Remove names, phone numbers and addresses before sending files, and use a secure way to transfer them. Agree how the data will be stored and when it will be deleted.
Check the data protection law that applies to you. In Nigeria, for example, the processing of personal data is governed by the Nigeria Data Protection Act 2023. Many funders also set their own data rules, so read your grant agreement.
Red flags to watch for
- Promising "significant" results before seeing the data, or offering to adjust numbers to fit what you hoped to find.
- Giving a fixed quote without looking at the data or asking how it was collected.
- Being unable to explain a method in plain words.
- Refusing to share syntax or scripts, so the work cannot be checked.
- Making cause-and-effect claims from a single survey that can only show associations.
- Being vague about deliverables, revisions or deadlines.
- Being careless with personal data, such as asking for files with names over open channels.
What affects the cost
Prices vary widely, so ask for an itemised quote. These factors drive the cost of data analysis most:
- The state of the data. Cleaning messy data is often the largest hidden task.
- Size and scope. The number of records, variables, objectives and sub-groups to compare.
- Method. Descriptive tables take less time than regression models or weighted survey analysis.
- Qualitative volume. The number and length of transcripts, and whether transcription or translation is needed first.
- Writing. Tables alone cost less than a full written findings chapter or report.
- Time. Short deadlines and extra revision rounds add cost.
A small paid data check before the full job can save money. It shows how much cleaning is needed and lets both sides agree a realistic scope.
Quick checklist
- One-page brief written
- Analyst asked about design and sampling
- Methods explained in plain words
- Software and file formats agreed
- Deliverables, revisions and deadline in writing
- Syntax or scripts and cleaned data included
- Personal data removed and transfer secured
- Itemised quote received
Need a data analyst for your project? We analyse quantitative and qualitative data in SPSS, Stata, R, NVivo and Excel, and explain the results in plain English. See our data analysis and statistical analysis services, then contact us or send a short brief on WhatsApp.