RESEARCH DISCIPLINE
General Social Survey Analysis
Public-Use Files, Analyzed on Their Terms
Public-use survey files handled as their designers documented them: weights and design variables applied together, items checked for comparability across rounds, special missing codes handled by a stated rule, and estimates reported with intervals.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
What Does General Social Survey Analysis Involve?
General Social Survey analysis is a secondary analysis: it answers a new research question with data that another team collected, designed and documented, so the analyst inherits its sampling plan, its question wording and its missing-value codes. The same holds for European Social Survey analysis, and both open with the codebook, not the data editor: which weights and design variables apply, whether an item kept its meaning from one round to the next, and who was never asked. At OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, an in-house PhD-level statistician reads that documentation before any model is chosen. Which hypothesis is tested, and what the estimates say about social change, are the researcher’s to decide.
WHEN TO ENGAGE
Choosing, Stacking and Defending a Public File
A reanalysis has no fieldwork to plan, so the decisions that matter most are which file, which years and which respondents, and they fall early.
Survey and Rounds to Choose A research question is set and a survey must be chosen, but candidate files differ in which items were asked, in which years and of whom, so the choice is made against the codebook.
Rounds to Be Stacked Several waves or rounds are to be combined into one file, and changes of wording, response options or interview mode have to be checked before any trend is read from them.
Outside Data to Attach Contextual measures such as regional unemployment are to be linked to respondents, and the geographic detail released, any restricted-use conditions and the number of areas decide what the merged file can support.
Estimates Under Review A reviewer asks whether a trend is real change or a change in the question, or why estimates differ from published tables, and the answer must come from the file and its documentation.
DESIGN-BASED REANALYSIS
Working Through a Public-Use Survey File
The work follows the order in which a public file can be trusted: what the survey measured, whom it reached, how it must be weighted and then what the model can claim.
Codebook and Universe
Wording, routing and universe are read for every variable used, and codes for don't know, refused and not asked are kept apart, so absent respondents are never counted as answering no.
Weights and Design
The weight and the stratum and cluster variables the documentation names are applied together, since a weight alone adjusts the point estimates for unequal selection but leaves the standard errors unreliable.
Comparing Rounds
Items are traced across rounds to see whether wording, response options or interview mode changed, and only items that kept their meaning enter a trend.
Yes-or-No Outcomes
Holding an attitude or belonging to a group is modeled with Logistic Regression under the declared survey design and shown as marginal predictions with intervals.
Pooled Rounds
When many rounds are stacked, Linear Mixed Models (Multilevel Modeling) separate variation among respondents from variation among rounds, which a handful of rounds cannot estimate reliably.
Reporting With Provenance
Estimates are reported with the survey, release, years and weight named, in the conventions of Sociology research within Education & Social Sciences.
FILE-LEVEL PITFALLS
Faults That a Survey Codebook Would Have Caught
Each fault below arises from treating a public file as a plain spreadsheet, and each can be seen in the documentation before it reaches a table.
Weight Used, Design Ignored Point estimates use the survey weight while standard errors are computed as if the sample were simple random, so intervals are narrower or wider than the design supports.
Not Asked Read as No Respondents routed past a question, or given it on only one form of the questionnaire, are coded as answering no, which changes both the denominator and the result.
Trend Without Item Check A rising or falling line is read as social change, though the wording, the response options or the mode of interview changed somewhere along the series.
Subgroup Cut Before the Design Respondents outside the group of interest are deleted before the design is declared, so strata and clusters lose members and the standard errors of the subgroup estimate are misstated.
Special Codes Averaged Values such as 98 or 99 for don't know or refused stay in the variable and enter the mean as if they were real answers, shifting it without any warning in the output.
Two Surveys, One Variable Items from two surveys that ask about one concept are relabeled onto a shared scale, though their wording and response options differ, so a gap may reflect the instrument.
DELIVERABLES
Provenance Files for Reanalysis Results
Each file records a choice that the survey documentation left open, so a reviewer can retrace how a published figure came from the downloaded file.
Source and Version Note The survey, release, years and countries used, with the download date and file version, so that the analysis extract can be rebuilt from the same source.
Variable Crosswalk For each concept, the variable used in every round, its wording and scale, and the recode that places rounds on one footing where that is defensible.
Weight and Design Specification The weight, stratum and cluster variables applied to every estimate, with the reason each was chosen for the question and the sample.
Estimate Tables With Intervals Weighted estimates and differences with 95% confidence intervals and the unweighted number of respondents, by year or country as the question requires.
Comparability Memo A short memo on items whose wording or mode changed, which rounds were kept or set aside, and how far the conclusions depend on that choice.
Rebuild Script for New Releases One script that reads the downloaded files, builds the analysis sample, applies the recodes and fits each model, so it can be rerun when a new release appears.
ILLUSTRATIVE EXAMPLE
Tracing Interpersonal Trust Over Decades
An illustration of how a long survey series is analyzed when an item changed partway through and age, period and cohort overlap.
ILLUSTRATIVE EXAMPLE · GENERATIONAL TRUST TREND RESEARCH · UNITED STATES
A doctoral researcher in sociology wanted to know whether interpersonal trust had declined among younger adults over several decades, using a long-running national cross-sectional survey series.
The plan recorded the item, the years and the weight and design variables before any trend was drawn. The first complication was in the codebook: the trust question had been put to only part of the sample in some years, and a related item changed its wording midway, so the analysis kept the unchanged item and showed the years with the smaller base separately. The second was conceptual, since age, survey year and birth cohort cannot all be separated without an added assumption. The researcher therefore described trust by age group over time and by cohort at stated ages, and the findings were worded as patterns in repeated cross-sections, not as changes within individuals.
Outcome: The researcher received the variable crosswalk, the weighted trust estimates by year and age group with their 95% confidence intervals, the comparability memo, the weight and design specification and the rebuild script.
This example does not describe a real client or project. We never share details of actual client work, in line with the confidentiality we promise every researcher. This illustrative example is included to show the type, scope, and standard of work involved in this kind of analysis.
SOFTWARE
Survey-Aware Software for Public Files
Public survey files arrive in several formats, and the program is chosen for how it carries weights and design variables into every estimate and for what your department reads.
HOW WE ASSESS YOUR PROJECT
We Review Your Actual Research
Your assessment is based on a review of your actual research materials and analytical requirements, not a generic project category.
You Submit Your Project Details
Share whatever is available at the current stage, such as the research aim, protocol or proposal, dataset, variable list, manuscript or statistical requirements.
We Review Your Research and Define the Scope
Your statistically trained Research Coordinator reviews the materials to understand the study design, research questions, data structure, analytical requirements and expected deliverables. If clarification is needed, additional documents or information can be requested before the assessment is finalized.
Your Analysis Is Completed and Delivered
Once the scope and fixed project fee are agreed, the in-house statistician completes the work and the Research Coordinator remains the communication point through delivery, with the analysis, deliverables and timeline as agreed.
RESEARCH RESPONSIBILITY
The Research Question Behind the Reanalysis
The survey's designers collected the data and you pose the question, so the statistical work joins the two while the sociological reading of the result is yours.
Agreed Statistical Support OnlineSPSS provides the agreed methodological, analytical and statistical reporting support, defined in a scope that is confirmed before any work begins.
No Intellectual Contribution OnlineSPSS does not claim any intellectual contribution to your research. Authorship of the dissertation, article or report remains yours.
Your Research Decisions The research question, the choice of survey, the theoretical framing, the interpretation of the estimates and the conclusions remain yours, with overall responsibility.
Confidential Handling Research data and project details are handled confidentially, project information is not published, and an NDA is available where required.
FREQUENTLY ASKED QUESTIONS
Public Survey Files, Weights and Comparable Items
Which weight to use, whether years can be pooled and why estimates differ from published tables all shape how a reanalysis should be read.
What does General Social Survey analysis cover?
General Social Survey analysis covers the reanalysis of public-use survey files: reading the documentation, applying the weights and design variables, checking that items are comparable across rounds, and estimating the quantities your question needs with confidence intervals. The survey’s sampling and fieldwork were fixed by its designers, so the analysis is built around them.
Which weight applies to a European Social Survey or General Social Survey file?
The documentation for the release decides, and it can change between rounds. The European Social Survey supplies a design weight, a post-stratification weight and a population-size weight: the post-stratification weight serves a single country or a comparison across countries, and pooling countries to describe a wider region combines it with the population-size weight. The General Social Survey documents different weights for different years and supplies stratum and cluster variables for the standard errors.
Can survey years be pooled to obtain more respondents?
Sometimes. Pooling adds respondents for a small group, but it assumes that the item meant the same thing in every year and that the association of interest did not change, and neither is automatic. Year can be entered as a predictor or the association allowed to vary by year, and a pooled estimate is described as an average over the years included, not as the position in any single year.
Why do the estimates differ from the survey's published tables?
The usual reasons are a different weight or none, a different universe, special codes treated as values, or a revised release. Reproducing a published figure first confirms the pipeline before new models are built. Any difference that remains is reported with its documented reason, so that a reviewer sees it was understood and not hidden.
How do you protect my research data and confidentiality?
Your data and project details are handled confidentially, and project information is not published. Researcher names and institutions are not used publicly without permission. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, such as the survey and years used, a variable list and any reviewer comments, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Questions a Survey Codebook
Can Already Answer
A reanalysis stands or falls on weights, universes and changes of wording, and your Research Coordinator reads the survey documentation against your question when you request an assessment.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014