Emotion Regulation Interventions for Alcohol-Related Sexual Aggression: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Emotion Regulation Interventions for Alcohol-Related Sexual Aggression (NCT03159468), a randomized na study. The primary outcome is Sexual Aggression Intentions.

2. Clinical Question

3. Trial Design

The registry record describes a randomized, double-masked, factorial study with enrollment of 209. Its status is completed. The registered phase is NA. The record lists University of Washington as the sponsor, with a start date of 2016-12-31 and primary completion date of 2018-06-29.

The six groups combine three online information or training conditions—cognitive restructuring, mindfulness, and nutrition information—with two laboratory beverage conditions, alcohol and no alcohol. The arm descriptions state that participants received a brief online training or general nutrition information, followed by an alcoholic or non-alcoholic beverage. This arrangement allows comparisons across training conditions and beverage conditions. The registry labels the design factorial, but details such as the randomization procedure, allocation concealment, and the precise way factorial effects were modeled are not reported in the registry record.

The posted outcome counts are 35 in the cognitive-restructuring-and-alcohol group, 35 in the mindfulness-and-alcohol group, 34 in the nutrition-information-and-alcohol group, and 34 in each no-alcohol group. These are the group counts associated with the posted primary outcome. The registry’s safety record uses different at-risk counts for one group, so the outcome counts should not be treated as the safety denominators.

The study’s stated goal was to evaluate two brief online emotion-regulation interventions in heavy episodic drinking young men with a sexual-aggression history. The summary also describes a laboratory alcohol-administration experiment and assessment during both sober and intoxicated states. The registry does not report the original sample-size justification, the planned primary statistical model in enough detail to reproduce it, or whether any interim analyses were conducted.

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.

Before assignment: All enrolled participants were assigned to one of the six arms.

Overall Study

Reasons for not completing this period:

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.

Only the first 6 of 7 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Sexual Aggression Intentions, measured over Within one hour after receiving the intervention.

The primary endpoint is Sexual Aggression Intentions, measured using the Sexual Aggression Intentions Scale. It is a continuous score based on participants’ self-reported likelihood of sexual aggression in a hypothetical scenario. The scale runs from 1, described as “Very unlikely,” to 7, described as “Very likely.” Higher scores indicate a worse outcome. The measurement timeframe is within one hour after receiving the intervention.

This measure was used to capture participants’ stated intentions in a controlled, sexual-aggression-related analogue, rather than a count of subsequent real-world acts. It therefore reflects responses to a hypothetical scenario at the specified assessment time. It does not directly measure actual behavior, the occurrence of sexual assault, or longer-term change. The registry record does not report additional measurement occasions for this primary endpoint, a clinically meaningful score threshold, or a stated minimum important difference.

Posted group results are means and standard deviations. The mean describes the group’s average score; the standard deviation describes the spread of scores around that average. These summaries help compare the reported scores across conditions, but they do not show the individual response distribution or establish why groups may differ. Interpretation also depends on the statistical analysis and on details such as missing outcome data and the analysis population, which are not reported in the registry record.

7. Results Posted in the Registry

• Mindfulness & Alcohol Condition: mean 1.90 versus 1.58 on control; difference 0.32 (95% CI -0.24 to 0.88), standardized effect d = 0.27; z-test p = 0.266.
• Nutrition Information & Alcohol Condition: mean 2.18 versus 1.58 on control; difference 0.60 (95% CI -0.05 to 1.25), standardized effect d = 0.44; z-test p = 0.071.
• Cognitive Restructuring & No Alcohol Condition: mean 1.53 versus 1.58 on control; difference -0.05 (95% CI -0.59 to 0.49), standardized effect d = -0.04; z-test p = 0.856.
• Mindfulness & No Alcohol Condition: mean 1.38 versus 1.58 on control; difference -0.20 (95% CI -0.68 to 0.28), standardized effect d = -0.20; z-test p = 0.413.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary measure is continuous: each participant’s score is on a numeric scale, and the posted summaries use means and standard deviations. A mean difference expresses how far one group’s average score is from the reference group’s average. Because higher values on this scale mean worse scores, the sign matters: a positive difference means a higher average score in the comparison group, while a negative difference means a lower average score.

What does an unadjusted comparison mean?

The recalculations compare each of five groups separately with the cognitive-restructuring-and-alcohol group. They are labeled unadjusted because they are simple pairwise comparisons rather than estimates adjusted for other factors or for a set of simultaneous comparisons. The recalculated estimates use posted group summary statistics. They do not reproduce an unspecified multivariable or factorial model. The registry record does not report whether the sponsor’s analyses used covariate adjustment or how the six groups were represented in those models.

For the five recalculated comparisons, the estimated mean differences are 0.32 for mindfulness with alcohol, 0.6 for nutrition information with alcohol, -0.05 for cognitive restructuring without alcohol, -0.2 for mindfulness without alcohol, and 0.12 for nutrition information without alcohol. Their associated p-values are, respectively, 0.266, 0.071, 0.856, 0.413, and 0.661. These are not the sponsor’s reported ANOVA p-values. Each set of results answers a different statistical question unless the underlying model and contrast are the same, which is not established by the registry entries.

What does a p-value convey?

General statistical explanation: a p-value describes how unusual data at least as incompatible with a specified null hypothesis would be if that hypothesis and the model assumptions were true. It is not the probability that the null hypothesis is true, and it does not measure the size or practical importance of an effect. A small p-value can arise for a small difference when data are sufficiently precise; a larger p-value can occur when estimates are imprecise. Here, the registry does not provide enough detail about the sponsor’s ANOVA hypotheses to assign a specific interpretation to each of its three p-values.

The recalculated p-values are unadjusted. When multiple comparisons are examined, the chance of at least one apparently notable result can increase if no multiplicity procedure is applied. The record lists three sponsor ANOVA analyses and five recalculated contrasts, but it does not report a multiplicity plan, whether the reported tests were prespecified, or how the sponsor handled multiple endpoints or contrasts. The posted primary outcome is singular in the record, and no secondary outcome measures are posted.

Why does the factorial design matter?

General statistical explanation: a factorial design can be used to estimate the separate effects of factors and whether the effect of one factor depends on the level of another. In this trial, the listed conditions combine training or information type with beverage condition. Pairwise comparisons against a single group do not automatically estimate an overall effect of training, an overall effect of alcohol condition, or an interaction between them. Those questions generally require contrasts or a model that explicitly represents the factors and their interaction. The registry labels the trial factorial and reports ANOVA methods, but the specific terms, contrasts, and model assumptions are not reported in the registry record.

ANOVA, or analysis of variance, is a general approach for comparing group means. In a standard formulation it tests whether a set of group means is compatible with a common mean, while accounting for within-group variation. A significant overall test would indicate evidence that not all means are equal, but it would not by itself identify which groups differ or the size of those differences. The registry’s ANOVA p-values do not include enough detail to determine which particular overall or factorial tests they represent.

What assumptions affect these summaries?

General statistical explanation: common mean-comparison methods rely on assumptions about independent observations and the behavior of outcome variation. The scale is bounded from 1 to 7, and the group means and standard deviations are reported, but the registry does not report distributions, residual diagnostics, or a rationale for the chosen analysis. The summaries alone cannot establish whether model assumptions were checked or met. No new distributional conclusion can be drawn from the means and standard deviations alone.

The reported standard deviations range from 0.9 in the mindfulness-and-no-alcohol group to 1.59 in the nutrition-information-and-alcohol group. Standard deviations describe observed variation within the posted groups; they are not standard errors or confidence intervals. A standard error, by contrast, describes the estimated sampling uncertainty of a statistic under a model. The recalculation lists standard errors for the reference and comparison means, but the registry does not describe how the sponsor calculated uncertainty for its ANOVA analyses.

The sponsor’s own analysis takes precedence when it differs from a simple unadjusted recalculation. The registry reports three ANOVA p-values, but no corresponding estimates or intervals. Accordingly, the recalculated mean differences are useful as transparent comparisons of posted summaries, not as a substitute for the sponsor’s full statistical analysis. Information about the original analysis plan, analysis populations, handling of missing observations, and interim analyses is not reported in the registry record.

10. Confidence Intervals and Precision

Confidence intervals describe a range of values for an effect estimate that are compatible with the data and the method used, under the method’s assumptions. They convey uncertainty in a way that a point estimate alone cannot. A narrow interval indicates greater precision than a wide interval, while neither width nor inclusion of zero determines clinical importance. The registry does not state the confidence level for the recalculated intervals, so no level should be inferred from the posted values.

For the five unadjusted recalculations, the interval for mindfulness with alcohol runs from -0.243 to 0.883; for nutrition information with alcohol, from -0.051 to 1.251; for cognitive restructuring without alcohol, from -0.588 to 0.488; for mindfulness without alcohol, from -0.679 to 0.279; and for nutrition information without alcohol, from -0.416 to 0.656. Every interval includes zero. Thus, for each pairwise estimate, the data are compatible with a zero mean difference as well as with values on either side of zero within the interval.

The intervals are not equally wide. For instance, the recalculation for nutrition information with alcohol spans from -0.051 to 1.251, whereas the interval for cognitive restructuring without alcohol spans from -0.588 to 0.488. The estimates should be read together with their uncertainty, not ranked solely by point estimate or p-value. The posted intervals are for unadjusted comparisons and do not account for a reported multiplicity procedure. No sponsor-reported intervals accompany the three ANOVA p-values, and the registry record does not report the precision of those analyses.

Precision is also limited by the group sizes available for these outcome summaries: the comparison groups contain 34 or 35 participants. This description does not establish that the trial was underpowered, because the original design assumptions and planned sample-size justification are not reported in the registry record. Nor does an interval including zero establish equivalence; an equivalence conclusion requires a prespecified margin and an appropriate analysis, neither of which is reported.

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 sample-size illustrations estimate how many participants per arm would be needed under stated planning assumptions to detect a chosen fraction of the observed effect. They are not the trial’s original enrollment target. For these continuous-outcome pairwise comparisons, the figures are participants per arm, not total events. The planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the target effect is the stated fraction of the observed effect. Original trial design assumptions and the original sample-size justification are not reported in the registry record.

The 80% power column represents the planned chance of detecting the specified effect under those assumptions if that effect is truly present. The 90% power column uses a higher target chance of detection, which generally requires more participants for the same effect. For example, for the comparison of mindfulness with alcohol against the reference group, planning for 100% of the observed effect gives 222 participants per arm for 80% power and 297 participants per arm for 90% power. These are planning illustrations based on the observed estimate, not recommended enrollment numbers for a new study.

Each comparison has separate planning values because the observed effect sizes differ. The rows for 75% and 50% of the observed effect ask what would be required if the effect worth detecting were smaller than the observed estimate. Smaller effects need more participants to distinguish them from sampling variation, all else being equal. This is not observed power: it does not describe the probability that the completed trial detected an effect, and it does not change the interpretation of the posted results.

The estimates depend on the stated assumptions and on the observed effect used as a planning reference. They do not incorporate an unreported factorial model, a multiplicity adjustment, missing data, or other design features. In particular, these calculations should not be read as evidence that the trial was designed around the observed results. The registry record does not report the protocol’s original power target, assumed effect, or recruitment calculation.

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 = 0.320; pooled standard deviation σ = 1.203.
2. n = 2 × (1.96 + 0.842)² × 1.203² / 0.320² = 221.7, rounded up to 222 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 <- 0.32                      # observed difference in means
sigma <- 1.20266371026983                     # 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(3159468)
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 = 222 (80% power), 297 (90% power)
 75% of observed effect: n per arm = 395 (80% power), 528 (90% power)
 50% of observed effect: n per arm = 887 (80% power), 1,188 (90% power)
power.t.test: 222.7 
Simulated power at n = 222 per arm: 0.799 (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 registry reports three sponsor analyses using ANOVA, with p-values .041, .160, and .462. It does not provide the associated effect estimates, intervals, tested contrasts, model terms, or enough detail to identify the hypothesis represented by each p-value. Those results therefore cannot be mapped confidently to a particular training effect, alcohol effect, interaction, or pairwise group difference.

The five recalculated comparisons are unadjusted mean differences between the cognitive-restructuring-and-alcohol group and each of the other groups. They are derived from the posted group means and standard deviations, and their p-values are 0.266, 0.071, 0.856, 0.413, and 0.661. They are not a reproduction of the sponsor’s ANOVA unless the sponsor used the same contrasts, data, and assumptions, which the registry record does not establish.

Because the record flags a discrepancy, the sponsor’s own analysis takes precedence over the simple recalculations. The recalculated results can still help readers understand the direction and approximate size of the posted pairwise mean differences, but they should not be used to override the sponsor’s analysis. A fuller reconciliation is not possible from the registry record: the details needed to identify the discrepancy and reproduce the sponsor’s model are not reported.

14. Multiplicity, Missing Data and Analysis Populations

The registry posts one primary outcome and three sponsor-reported ANOVA analyses. The analyses list the same six groups, but the registry does not specify the exact hypotheses tested, whether the tests concern main effects or interactions, or whether they were part of a prespecified testing hierarchy. No secondary outcome measures are posted. Five additional unadjusted pairwise comparisons are available as recalculations against the cognitive-restructuring-and-alcohol group; these should be distinguished from the sponsor’s three ANOVA analyses.

Multiplicity matters when several statistical hypotheses are tested. General statistical explanation: if many tests are conducted, the chance of at least one result appearing notable by chance can increase. Procedures such as family-wise error control or false-discovery control can address different multiplicity goals, but their appropriateness depends on which comparisons are considered one family and on the analysis plan. The registry record does not report whether any adjustment was used, how the testing family was defined, or whether the three sponsor p-values were adjusted. The unadjusted recalculated p-values should therefore be read as individual comparisons rather than as a multiplicity-controlled set.

The registry record does not report the number of participants included in each sponsor analysis, the analysis population definition, or whether all randomized participants were included. It also does not report how missing primary outcome scores were handled, whether any observations were excluded, or whether sensitivity analyses were conducted. The posted outcome group counts give the numbers associated with the outcome summaries, but they do not answer those analysis-population questions.

Other details that affect interpretation are also absent. The statistical analysis plan, any interim analyses, and the protocol’s handling of deviations from planned treatment or assessment are not reported in the registry record. Their absence from the posted record does not demonstrate that no such procedures existed; it means their use cannot be assessed from these results. Consequently, readers cannot determine from the registry alone how fully the reported analyses follow the original analysis plan.

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 registry’s adverse-event record lists zero serious adverse events and zero deaths in each of the six groups. It also lists zero other adverse events in each group. The at-risk counts are 35 for each alcohol intervention group, 37 for the nutrition-information-and-alcohol group, and 34 for each no-alcohol group. These are the safety denominators recorded for adverse-event reporting, not necessarily the counts used for the primary outcome analyses.

The safety counts are descriptive reports in the registry. Zero recorded events means no events are listed in these categories for the stated at-risk groups; it does not establish that there was no risk, nor does it rule out unrecorded events. The registry does not list serious adverse event terms, and the top serious-event terms list is empty. The record does not report additional details about how adverse events were solicited, monitored, classified, or followed over time.

One safety denominator differs from the corresponding primary-outcome group count: the nutrition-information-and-alcohol safety group lists 37 at risk, while its posted outcome group count is 34. The registry record does not explain the difference. Safety counts should therefore be read using their own at-risk figures rather than inferred from the outcome summaries.

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

The results describe short-term average scores on a self-report measure of intentions in a hypothetical scenario, assessed within one hour after the intervention. Across the six groups, the posted means range from 1.38 to 2.18 on a scale where higher scores indicate a worse outcome. The recalculated mean differences against the cognitive-restructuring-and-alcohol group are modest in direction and uncertain, with every recalculated interval including zero. The sponsor’s registry-listed ANOVA p-values are .041, .160, and .462, but the tested effects are not identified in the record.

These findings do not establish that any intervention changes actual behavior, prevents sexual assault, or produces lasting effects. The endpoint is a reported intention score, not an observed behavioral outcome. Nor do the available results establish that an intervention works or does not work. The recalculated comparisons are unadjusted and do not replace the sponsor’s own analysis, which takes precedence when there is a discrepancy.

The groups are combinations of training or general information and beverage condition. A comparison of one group with the reference group cannot be interpreted automatically as the isolated effect of mindfulness, cognitive restructuring, nutrition information, or alcohol. Separating those effects and evaluating whether one factor modifies another requires clearly specified factorial contrasts or a model; the registry’s detailed analysis specification is not reported.

The study population and setting also delimit interpretation. The registry summary describes heavy episodic drinking young men with a sexual-aggression history, and the outcome was assessed in a laboratory-related analogue. The record does not establish how results would generalize to other populations, settings, or outcomes. The findings should be understood as reported group summaries and analyses for this study, not as a general estimate of future behavior or a clinical recommendation.

17. Why This Trial Matters Statistically

Statistically, the study brings together two factors—emotion-regulation training or general information, and alcohol or no alcohol—in a factorial arrangement. That makes it important to distinguish the question “Do the group means differ?” from more specific questions about the contribution of each factor or their interaction. The sponsor’s analyses are labeled ANOVA, but the registry does not identify the tested effects. The five recalculated pairwise comparisons answer narrower questions against one reference group and cannot stand in for a fully described factorial analysis.

The record also illustrates why effect estimates and uncertainty matter alongside p-values. The recalculated mean differences range in direction, and their intervals all include zero. The sponsor-reported ANOVA p-values do not include effect estimates or intervals that would show the size and precision of the tested effects. Without those details, readers cannot connect the sponsor’s significance tests to a clear magnitude of difference.

Finally, the endpoint is a short-term self-reported intention measure rather than a measure of subsequent conduct. That choice is relevant to the study’s experimental question, but it sets boundaries on what statistical results can support. Understanding the outcome, design, comparison, and analysis together prevents a group-mean contrast from being mistaken for evidence about broader behavioral outcomes. Missing information about the original analysis plan, multiplicity, and missing data further limits how completely the reported inference can be 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.

Several trial-specific details needed to assess the results are not reported in the registry record. These include the full eligibility criteria, recruitment and setting details beyond the laboratory-based description, the randomization procedure, allocation concealment, and the precise masking procedures. Although the trial is labeled randomized and double-masked, the record does not explain which participants or study personnel were masked, or how masking was maintained across the intervention and beverage conditions.

The registered outcome is a self-reported score on a hypothetical scenario measure, collected within one hour after the intervention. The registry does not report a longer-term assessment of this primary endpoint, a threshold for a clinically meaningful change, or a measure of actual subsequent behavior. The scale’s score range and direction are described, but the record does not provide its validation details or evidence about how sensitive it is to change in this study population.

The available statistical reporting is incomplete for interpreting the sponsor’s analysis. Three ANOVA p-values are posted without the corresponding estimates, intervals, model terms, degrees of freedom, or identified hypotheses. The original analysis plan, multiplicity handling, sample-size justification, interim analyses, missing-data procedures, and analysis populations are not reported. The discrepancy flagged in the record cannot be resolved from the posted information, so simple unadjusted recalculations should be treated as supplementary comparisons rather than a substitute for the sponsor’s analysis.

The outcome group counts and safety at-risk counts are not identical for the nutrition-information-and-alcohol group, and the registry does not explain why. The safety record lists no events, but does not provide further event-monitoring details. Finally, the population described is specific to young men with heavy episodic drinking and a sexual-aggression history; the registry does not establish how well the findings generalize outside that study context.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Emotion Regulation Interventions for Alcohol-Related Sexual Aggression — NCT03159468.
• Registry record last updated 2019-07-26; status last verified 2019-07.
• Results first posted 2019-07-26.
• 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

NCT03159468 is a completed, randomized, double-masked factorial study sponsored by University of Washington, with enrollment of 209. It evaluates brief online cognitive-restructuring or mindfulness training, alongside a nutrition-information condition, across alcohol and no-alcohol laboratory conditions. The primary outcome is a self-reported Sexual Aggression Intentions Scale score measured within one hour after the intervention; higher scores indicate a worse outcome.

Posted group means range from 1.38 to 2.18. Five unadjusted recalculations compare other groups with the cognitive-restructuring-and-alcohol group, and all their intervals include zero. The registry also reports three ANOVA p-values, but does not identify the tested effects or provide accompanying effect estimates. The sponsor’s analysis takes precedence over the recalculations. Details about the original statistical plan, multiplicity, missing data, and analysis populations are not reported in the registry record.