Using Counter Attitudinal Advocacy to Change Drinking Behavior: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Using Counter Attitudinal Advocacy to Change Drinking Behavior (NCT04042909), a randomized na study. The primary outcome is Drinks Per Week as Assessed by the Daily Drinking Questionnaire.

2. Clinical Question

3. Trial Design

The registry identifies NCT04042909 as a completed, randomized, double-masked, parallel-group study. The listed phase is NA. Brown University is named as the sponsor. The registry lists enrollment as 591, a start date of 2019-11-12, and primary completion on 2023-01-30. The study title is “Using Counter Attitudinal Advocacy to Change Drinking Behavior.”

Three arms are described: Counter Attitudinal Advocacy, Personalized Normative Feedback, and Assessment-only Control. The advocacy activity involved generating and publicly stating protective strategies. The feedback intervention covered personal drinking, perceived same-sex student drinking norms, and actual same-sex student drinking rates on campus. The assessment-only group received no intervention. The registry does not report further detail about the randomization procedure, masking roles, intervention delivery schedule, or intervention adherence.

The primary outcome was drinks per week at the 6-month follow-up. The registry says the outcome analysis included participants who provided drinks-per-week data at that follow-up. The posted arm summaries list 178.0 participants for Counter Attitudinal Advocacy, 177.0 for Personalized Normative Feedback, and 188.0 for Assessment-only Control. These are outcome-specific arm counts, not the registry’s overall enrollment figure.

The registry record lists 2 outcome measures and 2 analyses. It also includes one sponsor-reported regression analysis, comparing Counter Attitudinal Advocacy with assessment-only control. Further details of the study’s original statistical plan, including its planned analysis population and original sample-size assumptions, are not reported in the registry record.

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: Participants were included in this analysis if they provided drinks per week data at the 6-month follow-up

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 Drinks Per Week as Assessed by the Daily Drinking Questionnaire, measured over 6-month follow up.

The primary endpoint is the mean number of drinks consumed in a typical week over the past 30 days, assessed at the 6-month follow-up with the Daily Drinking Questionnaire. The registry describes the unit as drinks per week and says the score can start at zero but has no upper limit. In the posted results, the endpoint is treated as a continuous measure and summarized by a mean and standard deviation for each arm.

A weekly drinking measure gives a single outcome that can be compared across the randomized groups at follow-up. The mean represents the group’s average reported drinks per week; the standard deviation describes how much individual results vary around that average. Comparing means addresses whether the groups’ average reported drinking differs, while the spread helps convey how varied the responses are within each group.

The endpoint reflects reported drinking behavior over a stated recall period. It is not a direct measure of every alcohol-related consequence, and it does not capture the pattern or timing of each drink. The registry summary identifies negative consequences as a broader research interest, but the primary endpoint is drinks per week. The registry record does not report additional measurement properties or explain why this specific questionnaire measure was selected.

7. Results Posted in the Registry

• Counter Attitudinal Advocacy: mean 6.80 versus 7.10 on control; difference -0.30 (95% CI -1.78 to 1.18), standardized effect d = -0.04; z-test p = 0.691.
• Personalized Normative Feedback: mean 6.00 versus 7.10 on control; difference -1.10 (95% CI -2.34 to 0.14), standardized effect d = -0.18; z-test p = 0.082.

8. Statistical Interpretation

9. Statistical Methods Explained

What does a mean difference describe?

A mean difference compares the average outcome in one arm with the average in a reference arm. A negative value in these posted comparisons means that the intervention arm’s observed mean was lower than the assessment-only control mean. The estimate is expressed in the endpoint’s original unit, drinks per week, which makes it interpretable alongside the arm means. It is a group-level contrast, not a description of the change experienced by each participant.

The re-calculated contrasts are unadjusted: they are based on the posted summary statistics rather than on a model that accounts for additional participant characteristics. The reported values are -0.3 for Counter Attitudinal Advocacy versus control and -1.1 for Personalized Normative Feedback versus control. The registry does not describe the covariates, if any, used in the sponsor’s regression model.

Why are means paired with standard deviations?

The mean gives the arithmetic average of the observed drinks-per-week scores. The standard deviation describes the spread of individual scores around that average. For Counter Attitudinal Advocacy, the mean is 6.8 and the standard deviation is 7.6; for Personalized Normative Feedback, they are 6.0 and 5.2; and for Assessment-only Control, they are 7.1 and 6.8. The spread is important because a mean alone does not show how similar or different participants’ reported outcomes were.

General statistical context: means and standard deviations are most straightforward to interpret when the distribution is reasonably represented by an average and a spread around it. A count-like outcome with no upper bound can have a skewed distribution or unusually high observations. In general, over-dispersion means that observed variability is greater than a model’s assumptions would expect; it can affect standard errors and interval estimates if the model does not accommodate that variability. The registry record does not report distribution checks or an assessment of over-dispersion, so no conclusion about the shape of these data can be drawn from the group summaries alone.

What does an interval add to the estimate?

An interval around a mean difference communicates the range of values reasonably compatible with the estimate under the statistical method used. It indicates precision: a narrower interval conveys more precise estimation than a wider interval, all else equal. If an interval includes zero, the data do not exclude a zero mean difference at the interval’s stated confidence level. The recalculated intervals are -1.78 to 1.18 for advocacy versus control and -2.338 to 0.138 for feedback versus control. The registry does not state the confidence level for these re-calculated intervals. The sponsor’s interval is explicitly reported as 95%.

General statistical context: an interval is not a range containing most individual participants’ outcomes. It concerns uncertainty about a population-level contrast under the analysis assumptions. It also does not account for every possible source of uncertainty, such as bias from missing outcomes or departures from a model’s assumptions.

How do the p-values relate to the estimates?

The unadjusted p-values are 0.691 for advocacy versus control and 0.082 for feedback versus control. A p-value describes how unusual data at least as incompatible with a null comparison would be if that null model and the analysis assumptions held. It is not the probability that the null hypothesis is true, and it does not indicate the probability that a finding will replicate. The p-values should be considered with the estimated mean differences and intervals rather than used as a stand-alone measure of evidence.

The registry’s sponsor-reported analysis is a linear regression for Counter Attitudinal Advocacy versus control. Its coefficient is -0.309, its 95% confidence interval is -1.632 to 1.013, and its p-value is .646. A regression coefficient is the model’s estimate of the contrast on the outcome scale, under the model as fitted. The registry reports the method and result but does not give the regression specification, included variables, or diagnostic checks. It is therefore not possible to characterize the model as adjusted for particular factors or to assess its assumptions from the record.

What is the role of randomization and masking?

Random allocation is intended to make comparison groups more alike, on average, before intervention, reducing systematic differences as an explanation for later outcome contrasts. It does not guarantee balance in every measured or unmeasured feature, and incomplete follow-up can affect how well the randomized comparison is preserved. The registry identifies the study as randomized but does not report the allocation method or concealment details.

Double masking indicates that the study was designated as masked, but the registry does not say which people were masked or how masking was maintained. The assessment-only comparator provides a reference for the intervention arms, but the registry does not report whether participants or staff could distinguish assignments. These missing details limit assessment of possible expectancy, implementation, or measurement effects.

10. Confidence Intervals and Precision

The re-calculated mean difference for Counter Attitudinal Advocacy versus assessment-only control is -0.3 drinks per week, with an interval from -1.78 to 1.18. The interval spans zero and includes values on either side of it. The result therefore does not precisely determine whether the average outcome was lower or higher in the advocacy arm than in control. The unadjusted p-value is 0.691.

For Personalized Normative Feedback versus control, the re-calculated mean difference is -1.1 drinks per week. Its interval extends from -2.338 to a value above zero, so it also includes zero. The unadjusted p-value is 0.082. The estimate points toward a lower average in the feedback arm, but the uncertainty interval does not rule out no mean difference.

These intervals are about uncertainty in the estimated group contrast, not the range of individual drinking outcomes. The arm standard deviations—7.6 for advocacy, 5.2 for feedback, and 6.8 for control—show that participant outcomes vary around the means. The registry does not report the distribution of individual scores or the confidence level used for the re-calculated intervals. The sponsor’s separate regression result for advocacy versus control is reported with a 95% confidence interval, -1.632 to 1.013, which also spans zero.

In general, precision depends on the amount of information in the data and the variability of the outcome, among other factors. A confidence interval is not proof that every value inside it is equally plausible, nor does it account automatically for missing-data bias, multiple comparisons, or model misspecification. Those issues require details beyond the interval itself. Here, the registry does not report the missing-data methods or a multiplicity strategy, so the intervals should be interpreted as estimates for the posted analyses rather than as a complete accounting of all uncertainty.

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 for 80% power and participants per arm for 90% power. They are not total events: the registry’s planning assumptions specify participant counts per arm for these continuous-outcome comparisons. The calculations use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They use the observed effect as a starting point and consider detecting 100%, 75%, or 50% of that effect. The trial’s own original design assumptions are not reported in the registry record.

For the advocacy comparison, planning for 100% of the observed effect gives 9,043 participants per arm for 80% power and 12,105 participants per arm for 90% power. Planning for 75% gives 16,076 and 21,520 participants per arm, respectively; planning for 50% gives 36,169 and 48,420. For the feedback comparison, planning for 100% gives 480 participants per arm for 80% power and 642 for 90% power. Planning for 75% gives 852 and 1,141; planning for 50% gives 1,917 and 2,566.

In general, the 80% power column represents a planning target designed to detect the stated effect under the assumptions, and the 90% column represents a higher target for detecting that effect. Smaller effects require more participants because they are harder to distinguish from random variation. These are planning illustrations based on the observed estimates, not observed power and not evidence that the trial had a particular probability of detecting its true effect. In particular, an observed effect close to zero can make planning figures based on that estimate unstable or unusually large.

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 = -0.300; pooled standard deviation σ = 7.200.
2. n = 2 × (1.96 + 0.842)² × 7.200² / -0.300² = 9,042.2, rounded up to 9,043 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 <- -0.3                      # observed difference in means
sigma <- 7.20012209908681                     # 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(4042909)
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 = 9,043 (80% power), 12,105 (90% power)
 75% of observed effect: n per arm = 16,076 (80% power), 21,520 (90% power)
 50% of observed effect: n per arm = 36,169 (80% power), 48,420 (90% power)
power.t.test: 9043.2 
Simulated power at n = 9,043 per arm: 0.801 (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. Analyses Reported by the Sponsor

The sponsor-reported analysis compares Counter Attitudinal Advocacy with Assessment-only Control using linear regression. It reports a beta coefficient of -0.309, a 95% confidence interval from -1.632 to 1.013, and p = .646. The interval includes zero, and the p-value is consistent with substantial uncertainty about the contrast.

The unadjusted re-calculation from the posted arm means and standard deviations gives a mean difference of -0.3 drinks per week, an interval from -1.78 to 1.18, and p = 0.691. Both approaches estimate a contrast close to zero relative to their uncertainty, and both intervals include zero. The estimates and p-values are not identical, as can occur when an analysis method differs from a simple comparison of group summaries.

The registry does not report enough information about the regression specification to identify the source of the difference, including whether covariates were included or how standard errors were calculated. The discrepancy indicator is false, so the record does not flag a discrepancy between the sponsor analysis and the re-calculation. The sponsor’s regression result remains the trial’s reported model analysis; the re-calculation is a separate, unadjusted analysis of the posted summary statistics.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists 2 outcome measures and 2 analyses, but only 1 sponsor-reported analysis is described in the available record. The record lists no secondary posted results. It does not report a multiplicity strategy, such as whether outcomes or comparisons were prioritized in advance, whether p-values were adjusted, or whether a single primary hypothesis governed interpretation. The re-calculated p-values are explicitly unadjusted. Without a reported plan, they should not be treated as adjusted for the number of outcomes or comparisons.

For the primary outcome, the registry says that participants were included in the analysis if they provided drinks-per-week data at the 6-month follow-up. The posted outcome counts are 178.0 for Counter Attitudinal Advocacy, 177.0 for Personalized Normative Feedback, and 188.0 for Assessment-only Control. The record does not explain how many randomized participants lacked outcome data, why data might be missing, or whether missingness differed by arm.

Methods for handling missing data are not reported in the registry record. It is not clear whether the analysis used complete cases only, imputation, weighting, or another approach. The record also does not state whether the primary analysis followed an intention-to-treat principle or define another analysis population. Thus, the posted group summaries describe participants with follow-up outcome data, but the registry does not provide enough information to judge how missing outcomes may have affected the randomized comparisons.

For the sponsor’s regression, covariates, model diagnostics, and handling of missing data are not reported. The registry also does not report interim analyses, stopping rules, or a detailed statistical analysis plan. These omissions do not change the numerical results, but they limit evaluation of how the analyses were selected and whether assumptions were appropriate.

15. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The adverse-event record lists zero serious adverse events in each arm. The serious-event at-risk and affected figures are 191 and 0 for Counter Attitudinal Advocacy, 196 and 0 for Personalized Normative Feedback, and 200 and 0 for Assessment-only Control. The record also lists zero other adverse events in each arm, with at-risk and affected figures of 191 and 0 for advocacy, 196 and 0 for feedback, and 200 and 0 for control. Deaths are listed as 0 in all three arms.

These counts describe what was recorded in the registry’s adverse-event fields. They do not establish that no adverse experiences occurred outside the definitions, reporting procedures, or observation period used by the study. The timing and duration of safety monitoring, the definitions applied to serious and other events, and the completeness of event collection are not reported in the registry record. No serious adverse-event terms are listed.

The safety figures are separate from the primary drinking-outcome analysis. They should be read as event reporting counts, not as evidence about the comparative change in drinks per week. The at-risk counts are not the same as the primary outcome analysis counts, and the registry does not explain the difference between those analysis populations.

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

The posted results show that the mean drinks-per-week values at the 6-month follow-up were 6.8 for Counter Attitudinal Advocacy, 6.0 for Personalized Normative Feedback, and 7.1 for Assessment-only Control. The unadjusted mean-difference estimates were -0.3 for advocacy versus control and -1.1 for feedback versus control. Both corresponding intervals include zero, so the results leave uncertainty about the direction and size of the average group differences.

These findings do not prove that either intervention changed drinking behavior, and they do not prove that the interventions had no effect. The observed estimates are compatible with a range of possible contrasts. The group means also do not describe every participant’s experience: the standard deviations indicate variation in reported drinks per week within each arm.

The primary measure is self-reported drinks in a typical week over the prior 30 days, not a direct count of every drinking occasion or an assessment of all alcohol-related harms. Although the study summary describes negative consequences as an area of interest, the primary result is drinks per week. The record does not provide detailed results for other outcomes.

Interpretation is also limited by missing information about follow-up completeness, the statistical analysis plan, multiplicity handling, and the sponsor regression specification. The outcome analysis included participants with follow-up data, but the registry does not report missing-data methods. These uncertainties matter when considering how well the posted estimates represent the randomized groups as originally assigned.

The results therefore support a careful description of the reported means and their uncertainty, not a clinical recommendation. They should be understood as registry-reported outcomes from this trial, with the scope and limitations of the recorded analyses.

17. Why This Trial Matters Statistically

This trial illustrates a common statistical challenge in behavioral research: an intervention may be designed to shift behavior through brief activities, while the outcome is variable and reported over a recall period. In this study, the endpoint is drinks per week, and the reported standard deviations are substantial relative to the means. That variability makes it important to interpret estimates alongside their uncertainty rather than judging a comparison by its direction alone.

The two intervention strategies also differ in content. One asks participants to develop and state protective strategies; the other presents personalized drinking and normative information. A randomized comparison with an assessment-only control can place each intervention’s observed outcome in context, but the results still depend on follow-up availability, measurement, analysis choices, and implementation. The registry record does not provide enough detail to evaluate all of those features.

Statistically, the results emphasize the distinction between an estimated contrast and a precise conclusion. The point estimates are negative, but both recalculated intervals include zero. The sponsor’s regression comparison for advocacy also has a 95% interval spanning zero. This is a useful reminder that a directionally lower mean is not itself a definitive estimate of a real group difference.

The trial is also a case for transparent reporting. Details on missing data, the analysis population, multiplicity, model specification, and original planning assumptions would help readers assess the strength and scope of the results. Their absence does not invalidate the posted numbers, but it constrains what can be inferred from them and how fully they can be independently evaluated.

18. 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 primary outcome analysis is restricted to participants with drinks-per-week data at the 6-month follow-up. The registry does not report the amount or pattern of missing follow-up data, the reasons for missingness, or the methods used to address it. As a result, the posted summaries cannot by themselves show whether missing outcomes affected the comparisons.

The registry does not provide detailed eligibility criteria, participant characteristics, or an explanation of how representative the analyzed participants were of the young adults discussed in the study summary. It also does not report intervention adherence, fidelity, or the extent to which participants completed the advocacy or feedback activities. These details would help assess the population and intervention exposure to which the results apply.

Analysis reporting is limited. The sponsor’s regression specification and diagnostics are not described, and the registry does not report a multiplicity strategy or the original analysis plan. The confidence level of the re-calculated intervals is not stated in the comparison details. The available group summaries also do not reveal the shape of individual outcome distributions, which is relevant for interpreting a mean-based analysis of an unbounded count-like measure.

The record reports zero adverse events in its listed fields but does not provide safety-monitoring timing, event definitions, or ascertainment procedures. Finally, the planning illustrations are based on observed effects and stated assumptions, not on the trial’s original design rationale. The original sample-size justification and other design assumptions are not reported in the registry record.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Using Counter Attitudinal Advocacy to Change Drinking Behavior — NCT04042909.
• Registry record last updated 2025-05-16; status last verified 2025-05.
• Results first posted 2025-05-16.
• 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.

22. Record Summary

NCT04042909 is a completed, randomized, double-masked, parallel-group study sponsored by Brown University. Its listed arms are Counter Attitudinal Advocacy, Personalized Normative Feedback, and Assessment-only Control. The registry lists enrollment as 591 and the primary completion date as 2023-01-30.

The primary outcome is mean drinks per week at the 6-month follow-up, measured with the Daily Drinking Questionnaire. The posted means were 6.8 for advocacy, 6.0 for feedback, and 7.1 for control. Re-calculated unadjusted comparisons have intervals that include zero. The sponsor reported a linear regression comparison for advocacy versus control, also with a 95% confidence interval spanning zero.

The registry does not report the original sample-size assumptions, multiplicity approach, missing-data methods, or detailed regression specification. The safety fields list zero serious adverse events, zero other adverse events, and zero deaths in each arm.