BehaviorLens Workbook: Implicit Test with Products

BehaviorLens Workbook: Implicit Test with Products

Overview

The BehaviorLens® Workbook  when using products captures instinctive versus deliberate reactions after panelists have tried the products. The analysis is organized two ways in parallel: by sample and by attribute. This article documents what each sheet does and how the Implicit/Explicit Score is built.


Why would I use it?

Use the Implicit test with products when panelists have actually tried the products and you want to compare how they react across several samples. Because every stimuli is shown for every sample and each answer is timed, you see which stimuli associations come from a fast, instinctive response or a slower, considered one.

Every answer is timed, so the report separates a fast, instinctive response from a slow, considered one and places each response on the 0 to 100 Implicit/Explicit Score. That lets you:

  • compare samples on each attribute, and read one attribute across the whole range;
  • see which reactions are instinctive (system 1) and which come only on reflection;
  • test whether differences between samples are real, and compare audiences using the descriptive splits.

How is the Implicit / Explicit Score calculated?

BehaviorLens separates an instinctive (implicit / fast) reaction from a deliberate (explicit / slow) one by timing every Positive/Negative answer. The response time is compared against a per-panelist threshold, and the Positive/Negative answer combined with implicit/explicit maps onto the 0 to 100 Implicit /Explicit Score:

Response typeScoreMeaning
Fast Positive100Positive selected in the implicit (fast) time: at or below the threshold
Slow Positive75Positive selected in the explicit (slow) time: above the threshold
Slow Negative25Negative selected in the explicit (slow) time: above the threshold
Fast Negative0Negative selected in the implicit (fast) time: at or below the threshold


The calibration threshold

Each panelist completes 18 calibration trials (10 numbers + 8 colours). To set their threshold, the second slowest trial becomes the implicit/explicit (IE) threshold between an implicit (fast) and explicit (slow) response. The threshold is capped at the max time (default 6,000 ms). A response at or below the threshold is Fast; above it, Slow.


Character-length adjustment

From the Raw Data sheet onward, the threshold is adjusted for stimulus length using the Character-length adjustment set on the Calibration block (default 0.015 s/char, 0 disables): the per-response threshold = calibration threshold + (0.015 s × character count). Longer statements get proportionally more time before a response counts as explicit/Slow. This adjustment applies across all analysis sheets and to every test type.


What’s included in the workbook?

The workbook flows from calibration and raw data into per-panelist scores, then into mean, composition, statistical and descriptive views (each organised by sample and by attribute) plus a PCA view.

SheetWhat it does
Calibration18 calibration trials per panelist (10 numbers + 8 colours). The threshold is found by taking the second slowest trials, capped at the max time (default 6,000 ms). Includes a Status Summary (OK / Max time reached) and a Distribution of Calibration Thresholds plotted in 500 ms intervals.
Raw DataOne row per panelist × sample × attribute: Sample Set Number, Sample, panelist UI fields, Attribute, Response, Response time (ms), Threshold (ms), Fast/Slow, Implicit score, and demographics. The single source of truth for every other sheet.
Implicit ScoresOne column per attribute with samples grouped together; panelists in rows.
Mean Scores - SampleHeat-map of mean implicit scores by sample, colour-coded by zone, showing all data combined then one table per sample. Rows: Category, n, Implicit Score, Min, Max, %Fast Positive, %Slow Positive, %Slow Negative, %Fast Negative.
Mean Scores - AttributeThe same means re-pivoted so attributes lead: all data combined then one table per attribute.
Fast/Slow - SampleThe n and four-type composition (% Fast Positive / Slow Positive / Slow Negative / Fast Negative) grouped by sample, with stacked bars; all data then one block per sample.
Fast/Slow - AttributeThe same composition grouped by attribute.
Stats - Within SamplePer sample: Table 1 (attributes in columns) n, Mean, SD, SE, CI lower, CI upper and post-hoc letters; Table 2 the pairwise comparison matrix (* marks significance); Table 3 the fast/slow Positive/Negative responses with post-hoc lettering. Repeats for each sample.
Stats - Within AttributeThe mirror analysis, repeating the three tables for each attribute (samples in columns).
Nudge vs DisruptClassifies each stimulus (or sample × attribute cell) as Nudge / Nudge & Disrupt / Disrupt / Weak from its Fast and Slow shares. Thresholds are adjustable; this sheet is deselected by default in the report options.
Descriptive Splits - By SampleTable 1 shows the base n; a further table gives the fast/slow breakdown, mean and post-hoc letters per sample across the demographic subgroups, with a score-means graph.
Descriptive Splits - By AttributeThe same split organised by attribute.
PCAA principal components analysis of the implicit scores for stimuli and samples: attribute loadings on Factor 1 and Factor 2 with correlations. A minimum of 3 samples and 6 stimuli are required for the PCA to run.


How do I read the results?

Calibration summary

The Calibration sheet reports a Status Summary counting panelists as OK (below the max time) or Max time reached (≥ 6,000 ms by default), and a Distribution of Calibration Threshold plotted in 500 ms intervals so you can see the spread of thresholds across the panel.

Zone colours on the heat-maps

The mean of all the implicit scores are taken and assigned a zone.

Mean-score views are colour-banded so the whole matrix can be read at a glance:
ZoneMean implicit score
Strong Positivemean ≥ 75
Positive50 to 74
Negative25 to 49
Strong Negativemean < 25


Response composition (Fast / Slow breakdown)

The breakdown sheets split each stimulus into its four response types. Reading the composition, not just the mean, tells you whether a middling score comes from broad indifference or from a genuine split between fast acceptance and slow rejection.


Nudge vs. Disrupt

Each stimulus (or sample × attribute cell) is classified by how its Positive or Negative (depending on selection made) responses split between the fast and slow routes. The thresholds are adjustable and the sheet is off by default in the report options:

MechanismRuleRead
NudgeFast ≥ 45% and Slow < 25%Instinctive appeal: reinforce it.
Nudge & DisruptFast ≥ 30% and Slow ≥ 25%Both instinctive and reflective pull.
DisruptFast < 30% and Slow ≥ 25%Works on reflection: needs a reason to believe.
Weakotherwise (low fast + low slow)Low pull on either route.


Statistical analysis

The statistical sheets carry a sample breakdown and an attribute breakdown. Within each, Table 1 gives n, Mean, SD, SE and the 95% CI lower/upper with post-hoc letters; Table 2 is the pairwise comparison matrix where * marks significance; and Table 3 covers the fast/slow Positive/Negative responses with post-hoc lettering. The three tables repeat for every sample (or attribute). Overlapping confidence intervals, or a shared post-hoc letter, are the first check before calling any pair different.


PCA

The PCA sheet reduces the attribute scores to two factors and reports each attribute’s loading on Factor 1 and Factor 2. Attributes that move together sit close on the biplot, and samples can be projected into the same space to see which attributes each one is associated with. A minimum of 3 samples and 6 stimuli are required for the PCA to run.


Descriptive Splits

The Descriptive Splits sheets (by sample and by attribute) show the panelist demographics: one table gives the base n, another the fast/slow breakdown and mean implicit score by subgroup with post-hoc letters, plus a score-means graph.