Duplicate response IDs
6Compared timestamps and source fields; retained one authorised record per confirmed duplicate pair.
DEMONSTRATION CASE USING SYNTHETIC DATA · SYNTHETIC CASE STUDY · SPSS
The important work happened
before the first p-value.
A transparent example of turning a messy satisfaction survey into a labelled, auditable SPSS-ready dataset without hiding exclusions or manufacturing complete records.
Which service experiences are associated with overall satisfaction, and is the dataset trustworthy enough to support that comparison?
ISSUE REGISTER
Every change needs a reason.
The working file was derived from an untouched raw copy. Original values, rule, decision and reviewer status were retained in the cleaning log.
Invalid age codes
9Checked the codebook; 999 represented missing and was recoded to system-missing—not treated as age 999.
Broken skip logic
14Set downstream responses to not applicable only where the parent answer proved the question was not asked.
Inconsistent scale direction
2 variablesReverse-scored the specified items before calculating the scale; preserved original fields.
SPSS WORKFLOW
A file another analyst can reopen.
Cleaning was expressed as reviewable transformations, not a trail of invisible spreadsheet edits.
Dictionary
Variable names, labels, types, measurement levels, value labels and user-missing codes were checked against the instrument.
Syntax
RECODE and COMPUTE steps preserved original fields, created derived variables and documented reverse-scored scale items.
Quality
FREQUENCIES, DESCRIPTIVES and CROSSTABS exposed invalid codes, sparse cells, unusual ranges and missingness patterns.
Gate
Reliability, group sizes, outcome distribution and test assumptions were checked before comparative analysis was released.
INSPECTABLE SPSS SYNTAX
Cleaning decisions can be rerun.
The downloadable syntax preserves the imported fields, recodes the documented age-missing value, reverse-scores the specified item and derives a scale only when enough items are observed.
Download the complete .sps syntaxCOMPUTE age_years = age_code.
IF age_code = 999 age_years = $SYSMIS.
COMPUTE experience_q2_scored = 6 - experience_q2_reverse.
COMPUTE experience_scale = MEAN.3(
experience_q1, experience_q2_scored,
experience_q3, experience_q4
).
EXECUTE.ILLUSTRATIVE OUTPUT
Descriptive context stays beside the test.
Synthetic results demonstrate reporting structure; they are not claims about a real service or population.
Six confirmed duplicate rows excluded with a documented rule.
Interquartile range: 3.2–4.4; distribution shown before modelling.
Internal consistency for the specified four-item experience scale.
Kept distinct from refused, unknown and genuinely missing responses.
ANALYSIS GATE
The initial chi-square table contained sparse expected counts, so categories were not merged merely to obtain significance. The report retained descriptive counts and used an exact or design-appropriate alternative only where the research question and category meaning allowed it.
HANDOVER PACKAGE
The deliverable is more than an .sav file.
Untouched source with import date and scope.
Labels, types and derived fields ready for analysis.
Record, field, original value, rule and outcome.
Re-runnable cleaning and analysis commands.
HAVE A MESSY SURVEY EXPORT?
Start with an audit—not an assumption.
Share a de-identified codebook or structural sample first. We will identify the cleaning risk and scope the smallest responsible next step.