STATISTICAL METHOD
Missing Data Analysis
Statistical Consulting for Incomplete Data
We carry out missing data analysis within your study design, assessing the amount and pattern of missingness, stating what each assumption allows, and choosing a handling method that protects your estimates, reported in publication-ready form.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
METHOD OVERVIEW
What Is Missing Data Analysis?
Missing data analysis examines how much of a dataset is missing, why it is missing and what that implies for valid inference, before a handling method is chosen. Data are missing completely at random (MCAR) when missingness is unrelated to any variable, missing at random (MAR) when it depends only on observed data, and missing not at random (MNAR) when it depends on the unobserved values themselves. Complete-case analysis is unbiased only under MCAR or special conditions, likelihood-based methods and multiple imputation are valid under MAR, and MNAR calls for sensitivity analysis. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides missing data analysis within your study design, with the relevant checks, effect estimates, uncertainty measures, interpretation and publication-ready reporting.
WHEN TO ENGAGE
When Does Incomplete Data Need Attention?
Incomplete data analysis is needed whenever missing values could change the answer, and these are the research situations in which researchers most often ask us to assess and handle them.
Substantial Item Nonresponse A sizeable share of records lack values on key variables, and the amount, pattern and likely reasons need assessing before any method is chosen.
Dropout From Follow-Up Participants leave a trial or cohort before the final visit, and whether leaving relates to the outcome decides which analysis stays valid.
Codes Treated as Real Values Sentinel codes such as 99 or 999 entered the dataset as measurements, and they must be recoded as missing before any modeling begins.
Case Deletion Challenged A reviewer questioned dropping incomplete records. Independent Statistical Review checks the missingness assessment and the handling.
ANALYTICAL APPROACH
How We Assess and Handle Missing Data
Within the family of Bayesian & Computational Methods, the missingness assessment comes first and the handling method follows from it.
Amount and Pattern We tabulate missingness by variable, case and occasion and relate it to observed data, to inform which assumption is plausible, not to settle it.
Reasoning About the Mechanism A test of MCAR cannot prove MAR or MNAR. We reason from the design, subject-matter knowledge and observed patterns, and state what each assumption allows for the analysis.
Choosing the Handling Method Complete-case analysis can change the target population and lose precision when missingness is systematic, so we weigh it against multiple imputation or likelihood-based models.
Sensitivity to MNAR Where missingness may depend on the unobserved values, we vary the assumptions, for example by delta-adjusted or pattern-mixture scenarios, and show how far the conclusions move.
DELIVERABLES
What You Receive
Within our Quantitative Data Analysis service, the work is delivered with the assessment, checks and files needed to report and defend it.
Data & Variable Review Sentinel codes and blanks are identified and recoded as missing, and the unit of analysis, variable types and the variables that may explain missingness are documented.
Method-Specific Checks Missingness patterns, its relation to observed variables and the plausibility of MAR are examined, and results are compared across the chosen method and complete-case analysis.
Statistical Results A missingness summary by variable and occasion, the final estimates with their 95% confidence intervals, and the sensitivity results under the alternative missingness assumptions.
Publication-Ready Tables & Figures A missingness table and pattern figure, with results tables formatted to APA 7, STROBE or CONSORT, or the journal's requirements, ready to insert.
Interpretation The findings are written with the missingness assumptions stated openly, with the limits of what the observed data can say about the missing values.
Reproducible Analysis Files SPSS syntax, R scripts, Stata do-files or SAS code are provided where applicable, so every recode, model and estimate can be traced and rerun.
ALTERNATIVE METHODS
When Another Method May Be More Appropriate
The assessment decides how incomplete data are handled. These methods fit when the model itself can absorb the incompleteness, or the missing part has a particular structure.
01
Linear Mixed Models (Multilevel Modeling)
For repeated outcomes with missing follow-up, a likelihood-based mixed model uses every observed visit and is valid under MAR when the model is well specified.
02
Generalized Estimating Equations (GEE)
Standard GEE assumes data are MCAR, so with dropout related to observed variables, a weighted GEE or another approach is needed.
03
Survival Analysis
When the unobserved part is the time of an event beyond follow-up, censoring methods are built for it, assuming censoring is not informative.
04
Structural Equation Modeling (SEM)
Full information maximum likelihood estimation in a structural model handles incomplete indicators under MAR, without a separate imputation step.
CASE EXAMPLE
How We Work in Practice
An illustration of how work on incomplete data moves from an assessment of missingness to a reportable, defensible result.
ILLUSTRATIVE EXAMPLE · SURVEY RESEARCH · UNITED STATES
A survey research organization found codes such as 99 and 999 among the reported values in a household panel dataset, with missing income and attitude items.
We recoded the codes as missing and tabulated missingness by item, wave and respondent group, which showed that nonresponse on income was related to age and region. Whether it depended on income itself could not be tested. Before any estimates were examined, we specified MAR given the observed covariates as the working assumption, with multiple imputation using the analysis variables and auxiliary variables as the primary approach and complete-case analysis as a comparison. Because high earners may be less willing to report income, we added a prespecified delta-adjusted sensitivity analysis that shifted the imputed incomes downward and upward.
Outcome: The organization received the missingness assessment and pattern tables, the recoded dataset documentation, imputation-based estimates with 95% confidence intervals beside the complete-case comparison, the sensitivity results and the code.
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
Statistical Software for Incomplete Data
Missingness is assessed and handled in IBM SPSS, R, Stata or SAS, each with established imputation and likelihood-based routines, with code provided where applicable.
RESEARCH DISCIPLINES
Incomplete Data Across Research Fields
Incomplete data arise in every research field, each with its own reasons for missingness, study designs and reporting conventions.
A Defined Process for Your Analysis
We review your dataset, research question and design before any work begins, so the scope, deliverables, timeline and fixed project fee are clear from the outset.
01
Share Your Research Materials
Send your dataset, research question, study design and any existing analysis plan or output. We will let you know if anything else is needed to assess the project.
02
Receive Your Assessment
Your statistically trained Research Coordinator reviews the project and confirms the required analysis, deliverables, timeline and fixed project price before work begins.
03
Analysis & Delivery
Once you confirm the scope, an in-house statistician completes the agreed analysis and delivers the outputs, interpretation and reporting files.
RESEARCHER VOICE
What Researchers Say
Feedback from a researcher whose missing-value codes had been treated as real observations, and who received an analytical summary instead of software output.
“We came to the team with a fairly technical Missing Data Analysis problem: missing-value codes that had been treated as real observations. They worked through the assumptions and reporting choices with us, then delivered a clean analytical summary rather than pages of software output. The final output was concise, clearly interpreted, and ready to use in our research reporting.”
RESEARCH RESPONSIBILITY
Statistical Support Without Taking Over Your Research
We provide the statistical work for your incomplete data, and the study, its meaning and its authorship stay entirely with you.
Agreed Statistical Support We provide the agreed methodological, analytical and statistical reporting support for your incomplete data, within the scope confirmed before work begins.
No Intellectual Contribution OnlineSPSS claims no intellectual contribution to your research. Our work is statistical support, and authorship remains yours.
Your Research Decisions The research question, your knowledge of why values are missing, interpretation and conclusions remain with you, and our statistical interpretation stays within what the data and the design support.
Confidential Handling Research data and project details are handled confidentially and are not published. An NDA is available where required, and can be arranged before sensitive materials are shared.
FREQUENTLY ASKED QUESTIONS
Questions About Missing Data
Short answers to what researchers ask before starting on incomplete data, from MAR and deletion to confidentiality and the assessment process.
Can a test show whether my data are MAR or MNAR?
No. A test such as Little’s can reject MCAR, but whether missingness depends on the unobserved values cannot be established from the observed data, which is what separates MAR from MNAR. We reason from the study design, subject-matter knowledge and the observed patterns, and use sensitivity analysis to show how much the conclusions depend on the assumption.
Is it acceptable to delete cases with missing values?
Only in limited circumstances. Complete-case analysis is unbiased under MCAR, or in special conditions such as missingness depending only on covariates in a regression, and otherwise it can shift the estimates toward the people who happened to respond and lose precision. We compare it with a method that suits the assumed mechanism and report both.
Is multiple imputation or a likelihood-based model better for my data?
Both are valid under MAR when the model is correctly specified. A likelihood-based mixed model suits repeated outcomes with missing follow-up and uses every observed visit, while multiple imputation suits missing covariates or analyses that lack a likelihood route. Neither removes MNAR bias, so we match the approach to your design and the sensitivity analysis it needs.
Can you review an analysis that has already been completed?
Yes, we can review a completed analysis of incomplete data before submission or a response to reviewers. We check how missingness was assessed and coded, the assumption stated, the handling method and its model, the sensitivity analysis and the reporting, and confirm what stands or needs to change.
How do you protect my dataset and research confidentiality?
Your dataset, codebooks and any protocol or analysis plan are handled confidentially, and project information is not published. 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, 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
Need Support With
Incomplete Data?
Share your research question, study design and data. We assess how your data are missing and confirm the approach before any work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014