Evaluations of Alcohol Warning Labels: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Evaluations of Alcohol Warning Labels (NCT06129487), a randomized na study. The primary outcome is Dwell Time.

2. Clinical Question

The study asks whether the narrative content of pictorial alcohol warnings is associated with how long participants attend to the warning areas. Its stated goal is to evaluate how narrative, compared with non-narrative, pictorial warnings communicate cancer risk. The primary endpoint reported in the registry is dwell time, a measure of visual attention, rather than a clinical outcome or a measure of alcohol consumption.

The brief study description also identifies message reactance, risk perceptions, and intentions to reduce or stop drinking as key outcome variables. The posted primary result addressed is specifically dwell time. Results for those other measures are not reported in the registry record, and no conclusion about them follows from the dwell-time comparison.

The randomized comparison concerns the two label formats as assigned in the experiment. It does not, by itself, establish whether either format changes drinking behavior or health outcomes. The registry record reports attention during a one-time online survey experiment; it does not report follow-up evidence about participants' subsequent behavior.

3. Trial Design

The study, Evaluations of Alcohol Warning Labels, is listed as completed. It was a randomized, parallel-group behavioral experiment comparing two types of pictorial warning labels for alcoholic beverages. The record lists 1053 participants enrolled and gives a start date of 2024-01-19 and a primary completion date of 2024-03-06. The phase is listed as not applicable.

Participants were assigned to view either three narrative pictorial warning labels or three non-narrative pictorial warning labels. The registry classifies the study as single-masked, but does not report who was masked or how masking was implemented. Details about the randomization sequence, allocation concealment, or procedures for delivering the online experiment are not reported in the registry record.

The primary endpoint was assessed during a one-time 15-minute online survey experiment. The primary analysis population was restricted to participants whose eye-tracking data were deemed usable by Sticky's internal algorithms. The registry does not give the number initially assigned to each arm, the number excluded from the primary analysis, or the reasons individual records were deemed unusable.

The result entries report 192 usable participants in the narrative arm and 213 in the non-narrative arm. These are the sample sizes attached to the posted primary outcome values, not the overall enrollment. The record lists 10 outcome measures and one analysis for the primary comparison described in the results. The number of secondary results posted is 0.

The sponsor is the University of Connecticut. The registry record does not report the original sample-size justification, detailed analysis plan, or the operational definition of single masking. Those omissions limit what can be said about how the study was planned and implemented beyond the design features explicitly listed.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Sticky's internal algorithms determined cases where the eye tracking data is less than "perfect" and only returns "usable" data. Thus, only usable eye tracking data were analyzed.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Dwell Time, measured over Assessed during the one-time 15-minute online survey experiment..

Dwell time is the total duration of visual attention to an area of interest, measured in seconds. Each warning label contained two areas of interest: an image area and a text area. The registry says dwell time for each area was averaged across the three pictorial warning labels within each arm. The primary result is titled Dwell Time.

Eye tracking provides a behavioral measure of attention to the warning, rather than asking participants to recall or report how much attention they paid. Longer dwell time indicates more time recorded on an area of interest during the measurement task. It does not, by itself, establish whether the participant understood the cancer-risk message, agreed with it, remembered it later, or changed their behavior.

The endpoint is relevant to the study's stated question because visual attention is one aspect of how a warning is encountered. Comparing mean dwell time between the assigned label formats can describe whether the recorded duration of attention differed in this experiment. The record does not provide separate arm-level means for image and text areas, so the posted result should be read as the reported dwell-time measure rather than as evidence about a particular component of the label.

The analysis included only eye-tracking records classified as usable. This restriction means the endpoint describes that analysis population. The registry record does not report how usable records differed from records excluded from analysis, so the result cannot establish whether the same pattern would hold among all enrolled participants.

7. Results Posted in the Registry

• Non-Narrative Pictorial Warning Labels: mean 8.47 versus 8.75 on control; difference -0.28 (95% CI -0.47 to -0.09), standardized effect d = -0.29; z-test p = 0.003.

8. Statistical Interpretation

9. Statistical Methods Explained

What is the mean difference?

A mean is the arithmetic average of a continuous measurement across the observations included in an analysis. Here the measurement is dwell time in seconds. The reported arm means are 8.75 seconds for narrative labels and 8.47 seconds for non-narrative labels. The registry expresses the contrast as non-narrative minus narrative, giving a mean difference of -0.28 seconds. The negative direction means the non-narrative arm's observed mean was lower than the narrative arm's observed mean.

General statistical explanation: a difference in means puts the two group averages on the original measurement scale. That makes it possible to describe the size and direction of the contrast in seconds. A mean difference does not show the full distribution of individual observations. Participants can have different dwell times within either arm, and group averages do not describe every participant's attention pattern.

What do the standard errors describe?

The registry reports a standard error of 0.069 for the narrative-arm mean and 0.066 for the non-narrative-arm mean. A standard error describes the statistical uncertainty in an estimated mean, not the spread of individual dwell-time observations. The endpoint entry identifies the reported dispersion as standard error. The registry does not give standard deviations for the two arm distributions in the primary result.

General statistical explanation: standard errors are used to express how much an estimate might vary across repeated samples under an analysis framework. They tend to be smaller when estimates are more precise, but they are not a direct description of how varied participants' measurements are. In particular, a standard error should not be interpreted as saying that most individual dwell times fall close to the arm mean.

How should the confidence interval be understood?

The registry reports a confidence interval for the mean difference. A confidence interval communicates uncertainty about an estimated contrast by giving a range of values compatible with the analysis. The posted interval is entirely below zero, consistent with the negative point estimate. Its exact confidence level is not reported in the registry record, so it should not be labeled with a specific confidence percentage based only on this entry.

General statistical explanation: the interval is not a range containing a known proportion of individual participant values, and it is not a probability statement about a fixed effect after the data have been observed. It is a way to show the uncertainty associated with the estimate under a statistical procedure. The interval's position relative to zero helps describe whether values representing no difference are included by that procedure.

What does the p-value add?

The registry reports an unadjusted p-value of 0.003 for the mean comparison. The word unadjusted matters: it indicates that the posted value is not presented as corrected for a set of multiple comparisons. The record lists 10 outcome measures, but does not report a multiplicity adjustment or clarify whether the full outcome set was incorporated into a single testing strategy.

General statistical explanation: a p-value describes how surprising data at least as incompatible with a specified null model would be, if that null model and the test assumptions held. It is not the probability that the null hypothesis is true, nor does it measure the importance of the observed difference. The exact test and model assumptions are not reported in the registry record. Accordingly, the reported p-value can be quoted as posted, but the precise testing procedure cannot be reconstructed from the record alone.

Why does over-dispersion matter in general?

General statistical explanation: over-dispersion occurs when the observed variation is greater than a statistical model expects. When a model understates variation, its standard errors can be too small, which can make confidence intervals too narrow and p-values too small. A model that allows for appropriate variation can yield more reliable uncertainty estimates. This is a general issue for statistical analysis, not a reported feature of this trial. The registry record does not describe a distributional model or report an over-dispersion assessment, and the primary outcome is presented as a continuous measure with means and standard errors.

What does randomization contribute?

General statistical explanation: random assignment is intended to make the comparison between assigned groups less vulnerable to systematic differences at baseline. It supports interpreting an outcome contrast as an effect of assignment when the study is implemented and analyzed appropriately. It does not ensure that every randomized participant contributes usable data, nor does it resolve bias that may arise from exclusions after assignment. In this study, only usable eye-tracking data were analyzed, and the registry does not report enough participant-flow detail to evaluate the consequences of that restriction.

The record does not report the detailed randomization procedure, model specification, covariate adjustments, handling of missing observations, or a prespecified statistical analysis plan. Those features should not be inferred from the fact that the study is listed as randomized or from the numerical results. The posted statistics support a description of the analyzed-group difference, while the omitted methodological details limit a fuller reconstruction of the analysis.

10. Confidence Intervals and Precision

The primary mean difference is -0.28 seconds, with a confidence interval whose reported limits are both below zero. This interval conveys uncertainty around the contrast and is consistent in direction with the lower observed mean in the non-narrative-label arm. The registry's unadjusted p-value is 0.003. Taken together, these statistics indicate that the estimated direction is not resting solely on a point estimate that lies close to zero under the reported analysis.

Precision is not the same as importance. A confidence interval can help show how uncertain an estimated mean difference is, but it does not identify a threshold at which a dwell-time difference would matter for communication, understanding, or behavior. No such threshold is reported in the registry record. Nor does the interval describe the range of individual participants' dwell times; it concerns the estimated difference between arm means.

The record does not state the confidence level associated with the interval. It also does not provide a separate interval for image-area and text-area dwell time, despite defining both as areas of interest. Therefore, the interval supports interpretation of the posted overall dwell-time contrast, not a claim about which warning component attracted more attention.

Finally, the interval and p-value pertain to the usable eye-tracking records included in the comparison. The registry record does not report the number or characteristics of excluded records. As a result, the displayed statistical precision does not account for possible selection effects arising from usability classification, and it should not be taken as precision for every enrolled participant.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are participants per arm, not total events. They are illustrations based on the observed effect, rather than the trial's own original design assumptions; those original assumptions are not reported in the registry record. The calculation uses a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each scenario targets the stated fraction of the observed effect.

The 80% power column describes the number of participants per arm that the illustration estimates would be needed to have 80% power under its assumptions. The 90% power column gives the corresponding per-arm planning count for 90% power. Power is a property of a planned design under specified assumptions; it is not a measure of whether the completed study's result is valid or important.

For planning at 100% of the observed effect, the illustration gives 185 participants per arm for 80% power and 247 participants per arm for 90% power. Planning for 75% of the observed effect gives 328 participants per arm for 80% power and 440 participants per arm for 90% power. Planning for 50% of the observed effect gives 738 participants per arm for 80% power and 988 participants per arm for 90% power.

These scenarios show the general relationship between effect size and required sample size: smaller effects need more participants to detect under otherwise matching assumptions. The calculations use the observed effect as their starting point, so they are planning illustrations, not evidence that the completed experiment had a particular probability of producing its result. This is not observed power. The registry record does not report whether the original study planning used these assumptions or figures.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -0.280; pooled standard deviation σ = 0.960.
2. n = 2 × (1.96 + 0.842)² × 0.960² / -0.280² = 184.5, rounded up to 185 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -0.279999999999999                      # observed difference in means
sigma <- 0.95985799197812                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(6129487)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 185 (80% power), 247 (90% power)
 75% of observed effect: n per arm = 328 (80% power), 440 (90% power)
 50% of observed effect: n per arm = 738 (80% power), 988 (90% power)
power.t.test: 185.4 
Simulated power at n = 185 per arm: 0.800 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists 10 outcome measures and one analysis for the primary result. It reports no secondary results. The study description identifies visual attention, message reactance, risk perceptions, and intentions to reduce or stop drinking as key outcomes, but the record does not report results for the latter measures. The relationship between the listed outcome measures and the single posted analysis is not explained in enough detail to reconstruct a complete analysis plan.

The p-value for the dwell-time comparison is explicitly unadjusted. The registry record does not report a method for controlling the chance of false-positive findings across multiple outcomes or comparisons, nor does it say whether any outcomes were designated as secondary or exploratory in a prespecified plan. Therefore, the p-value should be understood as the posted unadjusted value for this comparison, not as a multiplicity-adjusted result.

Missing-data handling is also not described in detail. The primary analysis included only eye-tracking observations classified as usable by Sticky's internal algorithms. The record does not report the number of unusable records, why records failed usability criteria, whether the criteria were set before analysis, or how many participants were assigned to each arm before exclusions. It also does not describe imputation or another method for handling unavailable measurements.

The analysis population is consequently narrower than the enrolled population, but the magnitude and implications of that restriction cannot be assessed from the registry record. The results pertain to the usable eye-tracking data in each arm. Whether this subset differs systematically from excluded records is not reported. Details of covariate adjustment, the statistical model, and the original analysis plan are likewise not reported in the registry record.

14. What the Result Does, and Does Not, Mean

The posted result describes a difference in visual attention duration between participants assigned to narrative and non-narrative pictorial alcohol warning labels. Among usable eye-tracking records, mean dwell time was 8.75 seconds for narrative labels and 8.47 seconds for non-narrative labels. The mean difference was -0.28 seconds when calculated as non-narrative minus narrative. The confidence interval lies below zero, and the unadjusted p-value is 0.003.

This is evidence about the measured duration of attention during a one-time online survey experiment. It does not show that participants understood the warning, retained the cancer-risk message, or changed their intentions or behavior. Although the study description names reactance, risk perceptions, and intentions to reduce or stop drinking as outcomes of interest, results for those measures are not reported in the registry record. No conclusion about those outcomes should be drawn from the dwell-time comparison.

The result does not establish that narrative labels are better or worse as a public-health intervention. The registry reports an attention measure, not a validated threshold for message effectiveness or a clinical endpoint. It also does not report later alcohol consumption, cancer incidence, or other health outcomes. The data therefore address one intermediate aspect of exposure to a warning, rather than the full chain from attention to comprehension, behavior, and health.

The comparison is based on participants with usable eye-tracking data. The registry does not report how many observations were unusable or whether usability was related to arm assignment or participant characteristics. This limits the extent to which the result can be generalized to every enrolled participant, and still more to populations or settings not represented by the online experiment.

In short, the posted statistics show a lower mean dwell time in the non-narrative arm in the analyzed records, with an estimated contrast that is below zero under the reported comparison. They do not resolve the practical significance of that difference, the results for other stated outcomes, or the effect of either label format on drinking behavior.

15. Why This Trial Matters Statistically

This experiment is statistically informative because its primary endpoint is a directly measured behavior during exposure to the warning: time spent attending to areas of interest. That differs from relying only on a participant's later recollection of whether a label was noticed. The comparison also places two message formats within the same randomized, parallel-group design, making the contrast specifically about assigned narrative versus non-narrative pictorial labels in this experimental setting.

The result illustrates the importance of distinguishing a precisely described statistical contrast from a broader claim about communication effectiveness. A difference in dwell time is interpretable in seconds, but attention duration is only one part of how a health message may be processed. Statistical evidence about attention does not automatically establish comprehension, persuasion, intention, or behavior. Keeping the outcome tied to its measurement definition helps prevent a result on one construct from being treated as proof about another.

The usable-data restriction is another important statistical feature. Eye-tracking outcomes depend on data quality, and the registry states that internal algorithms classified records as usable or not. Because the record does not give the number or characteristics of unusable observations, readers cannot fully assess the consequences of analyzing only the usable subset. This highlights why participant flow and transparent reporting of exclusions matter, particularly for outcomes requiring specialized measurement.

Finally, the record's limited reporting of multiplicity and analysis details cautions against treating one unadjusted p-value as a complete account of all study evidence. The posted estimate and interval can be interpreted for the reported comparison, while the other outcomes and full analysis strategy remain unclear. Statistically, the trial is most useful as a focused example of how assignment, measurement quality, analysis population, and endpoint choice shape interpretation.

16. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The main limitation specific to this result is that eye-tracking data were analyzed only when classified as usable by Sticky's internal algorithms. The registry does not report the criteria in operational detail, how many records were classified as unusable, or whether usability differed between the randomized arms. Without that information, the possibility that the analyzed subset differs from all enrolled participants cannot be evaluated.

The outcome description includes both image and text areas of interest and says dwell time for each area was averaged across three warning labels within each arm. The posted primary result does not provide separate arm estimates for image and text areas. Consequently, the comparison does not reveal whether the observed pattern was similar for each component or driven more by one area.

The experiment was a one-time 15-minute online survey experiment. The registry record does not report later follow-up for the dwell-time endpoint, so the result does not establish persistence of attention or subsequent behavior. Details of the online setting, participant characteristics, recruitment process, and representativeness are not reported in the registry record, limiting assessment of how broadly the result applies.

Further methodological details are also absent: the exact randomization procedure, who was masked, the statistical model, any covariate adjustment, and the original sample-size justification are not reported. The record does not explain whether or how the listed outcomes were prioritized or whether multiplicity was addressed. These gaps do not invalidate the posted numbers, but they constrain evaluation of the full design and analysis context.

The registry lists zero in the reported adverse-event counts for both arms. Those entries do not provide detailed information about how adverse events were solicited or assessed, and should not be treated as a comprehensive account of participant experience. The posted endpoint remains a measure of visual attention, not a direct assessment of the acceptability or consequences of the warning messages.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Evaluations of Alcohol Warning Labels — NCT06129487.
• Registry record last updated 2025-11-14; status last verified 2025-10.
• Results first posted 2025-11-14.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

20. Record Summary

Evaluations of Alcohol Warning Labels is a completed, randomized, parallel-group behavioral study sponsored by the University of Connecticut. Participants were assigned to view three narrative or three non-narrative pictorial warning labels for alcoholic beverages. The primary endpoint was dwell time in seconds during a one-time online survey experiment.

The posted primary result included 192 usable eye-tracking records in the narrative arm and 213 in the non-narrative arm. Mean dwell time was 8.75 seconds and 8.47 seconds, respectively. The reported non-narrative-minus-narrative mean difference was -0.28 seconds, with an unadjusted p-value of 0.003 and a confidence interval entirely below zero.

The result pertains to usable eye-tracking data and does not establish effects on the other stated outcomes or on drinking behavior. The registry record does not report the original sample-size justification, detailed analysis plan, or results for secondary outcomes.