From raw records to tables you can put your name on
Bharat Survey offers survey data analysis services for market research firms, CSR and NGO teams, academics and departments who have survey data, or are about to collect it, and need it cleaned, weighted, tabulated and interpreted properly. We work on data from our own fieldwork and on datasets you bring, and every transformation from raw file to final table is logged so a reviewer can retrace it.
- Sampling design before fieldwork
- Data entry for paper schedules
- Cleaning with a written log
- Weighting, tables and significance tests
Why analysis starts before fieldwork
The quickest way to improve a survey's analysis is to involve the analyst while the sample is still on paper. We work out what the study needs to report, and for which groups, and build the sample from there: strata, clusters, sample size per reporting cell, selection method and the weights the design will need later.
What a sampling note covers
A short document your team and reviewers can check, written in plain language.
- Target population and sampling frame, with its known gaps
- Strata and allocation, including any oversampling
- Clusters per stratum and interviews per cluster
- Expected design effect and resulting margin of error
- Replacement and non-response rules
- How weights will be calculated
When the sample is already fixed
If fieldwork is done, we assess the achieved sample against the plan, explain what it can and cannot represent, and recommend weights or reporting limits rather than pretending the gaps are not there.
Common design questions
- How many interviews per district?
- Should we oversample a small group?
- How many villages vs households?
- What margin of error is realistic?
Five stages, each producing a file you keep.
- 01
Intake and codebook check
We compare the data against the questionnaire: variable names, value labels, skip patterns. Mismatches go into an intake note before anything is changed.Stage 13,000Target outlets2,184Visited1,962Audited1,871AcceptedDaily accepted auditsLast 12 days - 02
Cleaning with a log
Out-of-range values, broken skips, duplicates and inconsistent answers are resolved by written rules. Every change is logged with the rule, original value and new value.Stage 2Outside area · 14.2 km away · in review - 03
Coding open-ended answers
Verbatim responses are coded into a frame agreed with you, with a second coder checking a sample for consistency.Stage 3Survey 3/6 · AmenitiesOfflineHow is drinking water in your area?- Comes regularly
- Comes sometimes
- Rarely / never
- Don't know
✓ Section saved on phone · autosaveBackNext - 04
Weighting
Design weights, non-response adjustment and, where justified, calibration to known totals such as Census population, with the effect on precision reported.Stage 4OfflineNo network6Safe on phone8Synced to serverSyncs when network returns - 05
Tables and analysis
Banner tables with bases, significance testing, and deeper analysis where the question needs it: regression, segmentation or index construction.Stage 53,000Target outlets2,184Visited1,962Audited1,871AcceptedDaily accepted auditsLast 12 daysQualityAccepted1871In review64Outside area19Weak signal8Top gaps · by territoryOut of stock62%Planogram48%Price label31%Visibility24%
- Comes regularly
- Comes sometimes
- Rarely / never
- Don't know
Paper is still common in some government and academic studies. Data entry done carelessly adds errors that no analysis can detect.
Entry screens that enforce the questionnaire
We build entry forms that mirror the schedule, with the same skips and valid ranges, so an operator cannot type an impossible value without a warning.
Schedules are logged in and out by ID, so none is entered twice or lost.
- Range and skip checks
- ID-based batch tracking
- Operator-level error rates
- Comes regularly
- Comes sometimes
- Rarely / never
- Don't know
Double entry and reconciliation
For studies that need it, a share of schedules or all schedules are entered twice by different operators and differences are resolved against the paper.
- Independent second entry
- Mismatch report
- Resolved against source
- In areaInside approx. boundary
- Accuracy±12 m
- Time09:42 · fresh fix
- Mock GPSNot detected
- Interview length11 min
Data files
- Raw data as received
- Clean dataset with labels
- Weights with documentation
- Coded open-end file
Documentation
- Codebook
- Cleaning log
- Sampling and weighting note
- Syntax or scripts used
Outputs
- Banner tables with bases
- Charts for key indicators
- Analysis note
- Limitations stated plainly
Not the right fit if
- Reweight a sample to force a preferred result
- Report sub-groups with bases too small to trust
- Hide cleaning decisions from reviewers
- Analyse personal data without a lawful basis
A good fit if
- Research teams short of analyst capacity
- Studies needing a documented audit trail
- Paper surveys needing careful digitisation
- Sampling design reviews before tender or fieldwork
Corporate research solutions
All market and customer research services.
Learn moreConsumer research
Usage, attitude and segmentation studies.
Learn moreRural market research
Village and small-town fieldwork.
Learn moreBrand tracking
Waves designed for consistent comparison.
Learn moreIVR survey
Short automated phone surveys.
Learn moreStraight answers to the questions people ask most about this topic.
Can you analyse data collected by another agency?
Yes. Share the dataset, questionnaire and any sampling documentation. We start with an intake check and tell you what the data can support before any analysis, including gaps in the sample or skips that were not followed.
Which formats and tools do you work with?
Common formats such as CSV, Excel, SPSS and Stata files, and exports from ODK-based tools. Deliverables come in the format your team uses, along with the scripts so the work can be rerun.
What does survey data analysis cost?
It is quoted per project. Cost depends on dataset size, cleaning effort, number of tables and banners, open-ended coding volume and whether advanced analysis is needed. Send the questionnaire and a sample of the data for a quotation.
How is data kept confidential?
Access is limited to the named project team, personal identifiers are separated from analysis files where possible, and the data hosting location and retention period are stated in writing in the proposal.
Send the questionnaire, a data sample and what you need to report. We will reply with a scope and quotation.