RESEARCH SUPPORT
Data Analysis for Dissertation Research
A Screened Dataset Before Any Model
Dissertation data screening and cleaning comes before the models of the proposal: variables coded and labeled, ranges and duplicates checked, missing values profiled and every cleaning decision recorded, so that the analysis starts from a file you can defend.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
What Comes Before the Main Analysis of a Dissertation?
The analysis of a dissertation begins before the first model is fitted: the dataset is examined for coding errors, impossible values, duplicate records and patterns of missing data, and the variables the proposal described are built from the raw items. This stage decides what the later estimates can mean, since a model fitted to an unexamined file inherits its errors silently. At OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, an in-house PhD-level statistician carries out the agreed dissertation data screening for doctoral and master’s researchers, documents what was found and explains each cleaning decision. What the cleaned data say about the program is for you to argue.
WHEN TO ENGAGE
A Dataset That Is Not Yet Ready to Analyze
A file straight from a survey platform, a clinic export or a fieldwork spreadsheet seldom matches the variables the proposal describes.
Raw Export in Hand The data have just come out of a survey platform or a laboratory file with default labels and unnamed codes, and the analysis variables do not exist yet.
Several Sources to Combine Waves, sites or instruments sit in separate files, and the merge has to be built and checked so that no case is lost or duplicated unnoticed.
Results That Look Wrong A first run gave a mean outside the possible range or counts that differ from the number recruited, and the cause has to be found in the file first.
Gaps Left Unexamined Many cases have blanks and the analysis dropped them without a look, so the pattern of missingness and the handling chosen both need to be examined.
WHAT THE PREPARATION COVERS
Stages of Preparing a Dissertation Dataset
Coding decisions change what screening can find, and screening changes what a model may assume, so the sequence is part of the method.
Variable Construction
Raw items become the variables of the proposal, with scales scored by a stated rule and categories labeled, as part of dissertation dataset preparation under Quantitative Data Analysis.
Range and Duplicate Checks
Values outside the possible range, repeated participant records and impossible combinations are listed and traced to the source file instead of being corrected by guesswork.
Missingness Profile
The amount and pattern of missing values are tabulated by variable and by group, and the handling is justified, with the methods described under Missing Data Analysis.
Distribution Review
Each analysis variable is reviewed for its shape and extremes, completing the dissertation outlier and missing data checks begun with the missingness profile, so that unusual cases are understood before any is kept or removed.
Alignment With the Plan
The finished variables are compared with the proposal, and where a variable cannot be built as described, the plan itself is reconsidered through Research Planning & Study Design.
Documentation
Each step is recorded in a cleaning log and a codebook, so that the results chapter can be reported through Results Reporting & Interpretation from a file whose history is clear.
COMMON DATA PROBLEMS
What an Unscreened Dataset Does to a Dissertation
Problems in a file rarely announce themselves, and they surface later as an implausible estimate or a committee question that the analysis cannot answer.
Scale Totals on Partial Items A composite is computed from whichever items were answered, so scores built from two items sit beside scores built from twelve under one label.
Duplicated Participants A respondent who submitted a survey twice, or a record repeated by a merge, counts as two people and narrows the interval around every estimate.
Value Labels Mismatched Category codes are attached to the wrong labels after an export, so one group is reported under the name of another.
Outliers Deleted on Sight Extreme values are removed because a plot looks untidy, though some are valid observations, and the exclusion changes the result without a recorded reason.
Original File Overwritten The cleaned file is saved over the raw one, and no one can say which cases were removed or which variables were recoded.
Scoring Differs From the Source A scale is scored differently from the instrument's published rule, and the chapter reports it as if the standard scoring had been used.
DELIVERABLES
The Files That Document a Prepared Dataset
Taken together, the files are the history of the dataset, from the raw export to the variables the models use.
Cleaning Log A dated record of every correction, exclusion and recode, with the rows or variables affected and the reason for each.
Annotated Codebook Variable names, labels, value codes, missing-value codes and scoring rules for every item and scale in the analysis file, so that any variable can be traced to its source.
Screening Summary A short report on ranges, duplicates, missing-value patterns and distributions, with the decision each finding led to.
Analysis-Ready File The dataset in the format of your software, ready for the planned models, with the raw data kept unchanged beside it for comparison.
Preparation Script The syntax or code that turns the raw file into the analysis file, so that the preparation can be rerun from the start.
Sample Description Tables Tables describing the prepared sample, formatted to your style guide, with the number analyzed in each row and the handling of missing values stated.
ILLUSTRATIVE EXAMPLE
A Merge That Duplicated Participants
An illustration of how screening finds a problem that would otherwise have entered every model of a dissertation.
ILLUSTRATIVE EXAMPLE · LIBRARY LITERACY PROGRAM RESEARCH · UNITED STATES
A doctoral researcher in public administration was evaluating a library literacy program using attendance records from two program years, exported from different systems and due to be combined before any model was fitted.
The screening began with the identifiers, because their format changed between the exports. A first merge produced more participants than the program had enrolled, and the excess was traced to records that appeared twice under different formats of the same identifier. Match rules were stated and applied, the unmatched records were listed for the researcher to resolve with the program office, and the analysis file was rebuilt from the raw exports with every step logged. The distribution of attendance was then examined, and a few extreme values were checked against the source before the model planned in the proposal was fitted.
Outcome: The researcher received the cleaning log, the annotated codebook, the screening summary, the preparation script and the analysis-ready file, with the sample description tables for the methods chapter.
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
Preparing Data in the Program You Will Use
Preparation is carried out in the program the analysis will run in, so that the cleaned file and the later models stay together.
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
Preparing the Data, Not Deciding What They Mean
Dataset preparation is statistical work carried out under the agreed scope, while the meaning of the data and the study around them remain the researcher's.
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 dissertation. Authorship remains the researcher's, including the decisions on how your data are described.
Your Research Decisions What the variables represent, how participants were recruited, what the findings mean and the conclusions you draw are yours, with overall responsibility for the research.
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
Questions About Preparing a Dissertation Dataset
Three decisions recur in most datasets: when a cleaning rule may change, what happens to the raw file and how partial responses are counted.
What does preparing a dissertation dataset involve before any model is fitted?
Preparation starts from the raw data, and OnlineSPSS handles that stage by coding and labeling the variables, screening ranges, duplicates and missing values, and recording each decision in a cleaning log and a codebook. The planned models are fitted afterward, from a file in which every change can be traced to the original export.
Is it a problem to change a cleaning decision after seeing results?
It can be, if the change is made because it moves the estimate. Cleaning rules are best set from the structure of the data before any model is run, and a rule changed afterward is reported with its reason, with the analysis shown both ways. A cleaning log makes the sequence visible, so that a committee can see what was decided before the results and what after.
Should I keep the raw data file after cleaning?
Yes. The raw file stays unchanged, and every change is made in a copy by a script or a logged step, so that the preparation can be reproduced and a committee member can see exactly what differs from the original. Saving the cleaned version over the original is the one step that cannot be undone.
How should a respondent who left half the questionnaire blank be treated?
First against a rule set in advance for what counts as a usable response, for example a minimum share of items answered. Beyond that, the question is whether those who stopped differ from those who finished, which the screening examines by comparing the two groups on what was measured. The rule and the number of cases it affects are reported.
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 and the data as they stand, such as a raw export, a variable list or a codebook, and a statistically trained Research Coordinator reviews the request and responds the same day. The preparation steps, 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
Open the Dataset
Before You Model It
A duplicate or a mislabeled code found before the first model is one line in a cleaning log; found after the results chapter is drafted, it can change every table that follows.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014