Longitudinal Survey Tracking Template
Two documents and three sheets that lock every anchor item's wording, then trace it to every trend series it feeds before a change is approved
Free download · No account needed
Comparable Series Register
Corrigan Outdoor Brand Tracker, Wave 6
| Item | Series it feeds | Status |
|---|---|---|
| Q1 recommend-intent | Brand NPS | Bridged (adjusted) |
| Q1 recommend-intent | Category-Relative NPS Gap | Bridged (adjusted) |
| Q2 purchase intent | Purchase Intent Trend | Comparable |
One proposed wording edit to Q1 reached two named series before it ever fielded. A register kept one row per item, not per series, would have caught only the first.
Lock every core question's wording before Wave 1, and if a change is truly unavoidable, run the old and new wording in parallel for one wave to build a bridge. That is the advice behind most brand-tracking guidance and behind the General Social Survey's own methodology, whose stated ideal is to phase in wording changes by conducting split-sample experiments administered to random subsets of respondents in the same year.
None of that guidance says what a same-wave split-sample test should do when it does not clear significance at ordinary sample sizes, which is common. Nor does it name which of a tracker's other published series that same item quietly feeds beyond its own. This space fixes both gaps. Comparable Series Register traces every item to every named series it feeds, directly or through a derived calculation, before a change is approved. The bridge test itself is then measured against that series' own historical wave-over-wave movement, not against a p-value alone.
In the worked example, Thornbury Research's Corrigan Outdoor Brand Tracker proposes widening one recommend-intent item's wording at Wave 6. The register finds two series it feeds; a split-sample test estimates a 4.3-point effect against a 1.3-point historical baseline, a ratio large enough to bridge both series despite p=.170. Wave 9's Attrition Analysis separately shows the matched original cohort's satisfaction rising 0.73 points against 0.44 for the full replenished panel, the gap panel studies lose to attrition alone. Locking and validating each item, before it reaches this pack, is a separate, earlier step.
What's in the pack
Wave Protocol
Every anchor item's exact wording, scale and mode locked before a second wave fields, the panel's retention floor and replenishment rule, and the only path a locked item can take to change.
Item Consistency Across Waves sheet
Every anchor item checked wave by wave against its locked wording, with any break logged as a bridge test rather than described as if the new wording had always been there.
Comparable Series Register sheet
Every named trend series and every item feeding it, direct or derived through a composite or a benchmark gap, so a proposed change's full reach is visible before it is approved.
Attrition Analysis sheet
The tracked metric reported two ways every wave, across the original cohort and across the full replenished panel, so the gap between them is stated rather than absorbed into one number.
Change Control Note
The closed record of a wording change actually proposed: the wording on both sides, which series matched, the bridge test's result, and the adjustment applied with its own stated uncertainty.
A wording change is approved against the series it feeds, not against how small it looks
The standing space rule every prompt reads first, including the 2x-baseline test for a bridge result that does not clear significance.
How to use it
- 1
Open in River, or download it
Open the pack in River and the agent locks your anchor items and runs the bridge-test procedure against your own wave data, or download the two documents and three CSV sheets instantly, filled in with the worked example.
- 2
Send your instrument and prior wave's data
The questionnaire you are already fielding or about to field, and whatever prior-wave data exists, so the anchor items get locked against something real rather than a blank protocol.
- 3
Every anchor item gets locked before a second wave fields
Wave Protocol states each item's exact wording, scale and mode, the panel's retention floor, and the only procedure a locked item can take to change.
- 4
A proposed change is checked against the register, not against how it reads
Comparable Series Register gets checked first; a match triggers a same-wave bridge test measured against the series' own historical movement, and the result closes out in Change Control Note.
Frequently asked questions
Is this template free?
Yes. Download the whole pack as Word documents and CSV sheets with no signup and no credit card, filled in for the Corrigan Outdoor Brand Tracker worked example. Edit with AI is a separate, optional path that locks your own anchor items and runs the procedure against your real wave data.
What format are the downloaded files?
Word documents for Wave Protocol and Change Control Note, and CSV for the three sheets, all in one zip. They open in Word, Pages, Google Docs, Excel, Numbers and Sheets with nothing to convert. Inside River the same content opens as live Docs and Sheets.
Does this draft or validate the survey items themselves?
No. It assumes the instrument already exists, drafted and piloted, and locks it for repeated fielding. Whether an item's psychometrics hold up is a separate, earlier validation pass run before the item is locked as an anchor here.
What happens when a bridge test doesn't clear statistical significance?
It gets compared against the series' own historical wave-over-wave movement rather than treated as clearance to swap the wording silently. An estimate at or above twice that historical baseline is practically meaningful and triggers a bridge adjustment even without a low p-value, because a same-wave split sample rarely has the power a dedicated study would.
How is attrition different from an item's wording changing?
They are two separate risks to the same trend line. A wording change is a break you introduce; attrition is a bias that accumulates even when every item's wording has stayed identical, because who stays in a panel is rarely a random subset of who started. Attrition Analysis tracks it independently every wave.
Does it also weight each wave's sample?
No, that is a separate document. Once a wave's data is in, naming the benchmark behind each weighting variable is its own note, run alongside this pack rather than folded into it, since the two documents answer different reviewer questions.
Find out what a wording change would actually break
Send your instrument and any prior-wave data. Every anchor item gets locked, and any proposed change gets checked against the series it feeds before it fields.
Edit with AI