Neurocognitive Factors in EdTech Intervention Response: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Neurocognitive Factors in EdTech Intervention Response (NCT04568824), a randomized na study. The primary outcome is Test of Word Reading Efficiency (TOWRE II), Total Word Reading Efficiency (TWRE) Index.

2. Clinical Question

3. Trial Design

The registry describes a randomized, parallel-group trial with no masking. It lists the study as completed, with a recorded enrollment of 315. The listed start date is 2020-11-15 and the primary completion date is 2025-08-29. The trial compared a GraphoLearn reading intervention with an active math control. Both were technology-based activities intended for use at home.

The brief summary describes recruitment of children with reading difficulty from the Healthy Brain Network and gives ages 6.0-10.0 for the target group. It also describes the Healthy Brain Network as an ongoing study with broad mental health, learning, cognitive, EEG and MRI characterization. The summary's plan to recruit 450 children differs from the recorded enrollment of 315; the registry record does not explain the difference or give a separate account of enrollment over time.

The intervention period was 12 weeks. Reading and related language skills were to be assessed before and after intervention, with the assessment after intervention designated as the primary outcome time point. The summary also describes a 12 week retention period, with assessment before and after retention, to examine the stability of gains. The posted primary outcome population notes that participant numbers are smaller for that measure because of study attrition.

The listed primary endpoint is the Total Word Reading Efficiency index from the Test of Word Reading Efficiency, Second Edition. Secondary measures include additional achievement and phonological-processing assessments and EEG measures. The registry record does not report allocation concealment, the randomization procedure, the masking of outcome assessors, or the detailed schedule for all assessments. It also does not report the trial's original sample-size justification or a complete protocol-level analysis plan.

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: There were no significant events between enrollment and assignment. Participants who consented but did not initiate the study generally cited one of three reasons: (1) scheduling conflicts, (2) loss of interest, or (3) assignment to the math rather than reading game arm. Although all participants were ultimately offered the reading game, delayed access may have reduced participation among those assigned to the math game.

Overall Study

Analysis population for the primary outcome, as defined in the registry: T2 Total Word Reading Efficiency (TWRE) is the outcome measure, which is measured after the 12-week intervention. Participant numbers are smaller for this measure relative to the baseline measures because of study attrition.

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 Test of Word Reading Efficiency (TOWRE II), Total Word Reading Efficiency (TWRE) Index, measured over Assessment is given before and after the intervention, ~12 weeks apart for the primary outcome; the assessment will also be re-administered after a 12 week retention period. T2 (after the intervention) is the primary outcome measure..

The primary endpoint is the Total Word Reading Efficiency index, abbreviated TWRE, from the Test of Word Reading Efficiency, Second Edition. It combines Phonemic Decoding Efficiency and Sight Word Efficiency, timed measures of word and nonword reading. The registry describes higher scores as better performance and greater change as a better response. Scores are expressed in units on a scale.

The primary assessment is T2, after the 12-week intervention. The registry also describes an assessment after a 12 week retention period, but that later assessment is not identified as the primary time point. Focusing on the post-intervention score makes the endpoint a direct measure of reading efficiency at the end of the intervention period. The registry lists the scale range as 53-147.

Word-reading efficiency is relevant to the study question because the intervention targets reading, including matching auditory language elements with written forms. The endpoint captures timed reading of words and nonwords rather than every possible aspect of reading or language. The registry lists secondary measures that extend assessment to achievement, phonological awareness, phonological memory, rapid naming and EEG-based processing. Those measures provide a broader set of outcomes, but they do not change the definition of the primary endpoint.

The registry notes that the primary-outcome participant numbers are smaller than the baseline-measure numbers because of attrition. It does not report the reasons for missing primary assessments or the method used to handle missing endpoint data.

7. Results Posted in the Registry

• GraphoLearn Reading Intervention: mean 83.12 versus 80.28 on control; difference 2.84 (95% CI -0.94 to 6.62), standardized effect d = 0.21; z-test p = 0.141.

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports two kinds of statistical results for the primary outcome: group-level summary statistics and a Linear Mixed-Effects Regression analysis. They should be read as separate analyses because they do not estimate exactly the same quantity.

What does the unadjusted mean comparison describe?

A mean is the arithmetic average of a continuous measurement. The TWRE group means are 83.12 for GraphoLearn and 80.28 for math control. The simple difference between these means is reported as 2.84. This is an unadjusted comparison of the posted endpoint summaries. It describes the observed separation between the group means at T2 without accounting for other variables or the pattern of measurements over time.

The standard deviations are 13.82 for GraphoLearn and 13.18 for math control. A standard deviation describes how dispersed individual scores are around a group mean; it is not the uncertainty in the mean itself. The standard errors reported for the two means are 1.362 and 1.367, respectively. Standard errors describe the sampling uncertainty of estimated means under the assumptions of the calculation. They are not measures of how much individual scores vary.

What does a confidence interval add?

A confidence interval gives a range of values reasonably compatible with an estimate under the statistical method used. For the unadjusted mean difference, the reported interval runs from -0.941 to 6.621. Because the lower limit is negative and the upper limit is positive, the interval includes zero, the value representing no mean difference. The interval also shows that the estimate is not precise enough to isolate a narrow range of possible differences.

A p-value summarizes how unusual the data would be under a specified null model and its assumptions. The unadjusted comparison has p = 0.141. The p-value is not the probability that the null hypothesis is true, and it does not measure the size or practical importance of the difference. A p-value should be interpreted together with the estimate, interval, design and analysis choices.

Why use a mixed-effects model?

General statistical explanation: a mixed-effects model can represent repeated measurements within participants and account for the fact that observations from the same person are related. It can also estimate group, time and group-by-time terms, depending on the model specification. The registry reports a Linear Mixed-Effects Regression group-by-time estimate of -0.018, with a 95% confidence interval from -0.119 to 0 and p = 0.728. It does not report the model's full formula, covariance structure, included covariates, coding conventions, or handling of missing observations. Those details are needed to interpret precisely what the interaction coefficient represents.

The interaction term is not interchangeable with the difference between T2 means. A group-by-time term usually addresses whether change over time differs by group, whereas a single-time-point mean contrast compares group averages at one assessment. The registry record does not provide enough detail to reconstruct a particular change contrast from the reported coefficient. The model result should therefore be stated as reported rather than converted into a different effect measure.

Why can dispersion matter?

General statistical explanation: larger variability among individual measurements makes a group mean less precisely estimated when other design features are unchanged. The standard deviations show the observed spread in each group, while the confidence interval for a contrast reflects uncertainty in that contrast under its calculation. If observations are more variable than a model assumes, or if repeated observations are handled incorrectly, standard errors and confidence intervals can be misleading. Over-dispersion is especially relevant for count data when variability exceeds what a chosen count model expects; the TWRE endpoint is continuous, and the registry does not report a count model.

The registry characterizes the endpoint as a mean of a continuous measure and reports its spread as standard deviation. The calculations for the unadjusted comparison use the posted group means and standard deviations. No adjustment is described for baseline score, demographic characteristics or other prognostic variables. Whether an adjusted model was prespecified or used elsewhere is not reported in the registry record.

How should the two analyses be put together?

The unadjusted summaries provide a transparent description of endpoint scores and the simple mean contrast. The mixed-effects result is the analysis identified in the registry as the reported model analysis. They are not conflicting numerical versions of a single statistic: one is a T2 contrast, and the other is a group-by-time coefficient. The registry record does not report a discrepancy flag, but it also does not provide enough details to fully reconcile the estimates. The model-based result and the simple recalculation should consequently retain their distinct labels and interpretations.

10. Confidence Intervals and Precision

The unadjusted mean difference in T2 TWRE scores is 2.84 points, with a 95% confidence interval from -0.941 to 6.621. The interval crosses zero, so it is compatible with no mean difference as well as with differences in either direction. Its breadth indicates uncertainty about the magnitude of the contrast. The reported p-value for this simple comparison is 0.141, which should not be read as a probability that the groups are equal.

General statistical explanation: a confidence interval describes the range of effect values compatible with the observed data and the assumptions of the method. It is not a guarantee that the true effect lies within the particular interval, and it does not describe the range of individual participants' scores. The group standard deviations, 13.82 and 13.18, describe individual-score dispersion; they answer a different question from the interval around the mean difference.

The registry's mixed-effects estimate is -0.018, with a 95% confidence interval from -0.119 to 0 and p = 0.728. This interval belongs to the reported group-by-time model term, not the unadjusted T2 mean difference. Because the registry does not report the full model specification or parameter coding, the interval's clinical or scale-based interpretation cannot be made more specific than the reported coefficient.

Precision is also affected by the number of participants with endpoint measurements. The registry notes that fewer participants contributed to the primary outcome than to baseline measures because of attrition. It does not report the missingness pattern or the reasons for missing assessments. The confidence intervals quantify statistical uncertainty under their respective analyses, but they do not remove concerns about missing data or unreported modeling choices.

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 are participants per arm, not total event counts. They use the observed unadjusted effect as a planning reference and examine planning for 100%, 75% and 50% of that observed effect. The assumptions are a two-sided significance level of 0.05, equal allocation between the two arms and a normal approximation. The original trial design assumptions and its own sample-size justification are not reported in the registry record.

For planning for 100% of the observed effect, the illustration gives 356 participants per arm for 80% power and 477 participants per arm for 90% power. For planning for 75% of the observed effect, it gives 633 participants per arm for 80% power and 847 participants per arm for 90% power. For planning for 50% of the observed effect, it gives 1,424 participants per arm for 80% power and 1,906 participants per arm for 90% power.

In plain terms, the 80% power column is the modeled number of participants per arm needed to have an 80% chance of detecting the specified effect if the planning assumptions hold. The 90% power column applies the same idea at a higher target chance of detection. Smaller effects need more participants to distinguish them from random variation, which is why the illustrations increase as the assumed effect becomes a smaller fraction of the observed effect.

These numbers are planning illustrations, not observed power. They do not establish what the trial's power was, and they are not a record of the original design calculation. The accompanying curve should be understood within the same planning assumptions; the registry record does not report additional assumptions for it.

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 = 2.840; pooled standard deviation σ = 13.520.
2. n = 2 × (1.96 + 0.842)² × 13.520² / 2.840² = 355.8, rounded up to 356 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 <- 2.84                      # observed difference in means
sigma <- 13.5202722171095                     # 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(4568824)
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 = 356 (80% power), 477 (90% power)
 75% of observed effect: n per arm = 633 (80% power), 847 (90% power)
 50% of observed effect: n per arm = 1,424 (80% power), 1,906 (90% power)
power.t.test: 356.7 
Simulated power at n = 356 per arm: 0.802 (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 the sponsor's Linear Mixed-Effects Regression analysis as a group-by-time estimate of -0.018, with a 95% confidence interval from -0.119 to 0 and p = 0.728. This is the reported model-based result. It differs in form from the simple unadjusted recalculation using the posted T2 means, which gives a mean difference of 2.84 and a 95% confidence interval from -0.941 to 6.621, with p = 0.141.

The two estimates address different contrasts. The recalculation compares the observed endpoint means at T2. The sponsor's group-by-time term concerns the relationship between group and time in a mixed-effects model. The registry does not report enough information about the model, including its coding and full specification, to reconstruct the sponsor's estimate from the summary statistics or to express it as the same T2 contrast.

The record does not flag a discrepancy between the analyses. Their numerical differences should not be treated as evidence that one analysis is an erroneous version of the other. The sponsor's model result should be reported as the registry's stated analysis, and the simple calculation should be identified as an unadjusted comparison of posted summary statistics. The registry record does not report a further explanation of the relationship between them.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists four outcome measures and three posted secondary outcomes. The secondary measures include Woodcock Johnson achievement subtests, phonological-awareness, phonological-memory and rapid-naming composites, and EEG measures of phonological discrimination and letter-sound processing or integration. Multiple outcomes and assessment times create opportunities for many statistical comparisons. When many hypotheses are tested, the chance of at least one apparently unusual result can increase unless the analysis plan addresses multiplicity.

The registry record does not report whether the trial used a family-wise error adjustment, a false-discovery procedure, a hierarchy of outcomes, or another multiplicity strategy. It does not report which secondary comparisons were designated confirmatory or exploratory. The reported unadjusted p-value should therefore be read as an unadjusted result, not as evidence that multiplicity was handled.

The primary endpoint population notes that participant numbers are smaller for this measure than for baseline measures because of study attrition. The registry does not report how missing primary assessments were handled, whether all randomized participants were included in the analysis, or whether the analysis followed an intention-to-treat principle. It also does not give missing-data counts by reason or group for the primary endpoint.

The mixed-effects model is named, but the registry record does not state the assumptions used for missing observations, the estimation method, or sensitivity analyses. General statistical explanation: mixed-effects models can use observed repeated measurements under particular assumptions about missingness, but that does not by itself establish that missing data are ignorable. Different missing-data mechanisms can affect the interpretation of an estimated group difference.

The record does not report analysis-population definitions, protocol deviations, per-protocol results, or a detailed plan for handling participants who contributed to more than one intervention period. These omissions limit assessment of how closely the posted analyses follow the original trial plan. The absence of these details in the registry record is not evidence that no such procedures were used; it means their use cannot be established from the record.

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 reports serious adverse event figures separately for GraphoLearn Reading Intervention, Math Control, and GraphoLearn Reading Intervention Post Math Control. The respective numbers at risk for serious adverse events are 166, 149 and 17. Serious adverse events affected 0 participants in each listed group. The reported death count is 0 for each group.

These figures describe the safety information posted in the registry record. They do not establish that no events occurred outside the record's reporting definitions or observation periods. The record lists no serious adverse event terms. It also reports 0 other adverse events in each listed group, with the corresponding at-risk figures being 166, 149 and 17.

The registry record does not explain why a post-math GraphoLearn group is listed separately, nor does it provide further detail about the timing or duration of safety follow-up. The at-risk numbers are reported by group and are not interchangeable with the primary-outcome group counts. Safety figures should be read as reported event counts, not as a comparison of risk adjusted for follow-up or participant characteristics.

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

The posted primary endpoint summaries show a higher mean TWRE score in the GraphoLearn group than in the math control group at the post-intervention assessment: 83.12 compared with 80.28. The simple unadjusted mean difference is 2.84. Its 95% confidence interval extends from -0.941 to 6.621, so the data are compatible with a range that includes no difference. The unadjusted p-value is 0.141.

The registry also reports a group-by-time coefficient from a Linear Mixed-Effects Regression analysis, with an estimate of -0.018, a 95% confidence interval from -0.119 to 0 and p = 0.728. This coefficient is not the same as the T2 mean difference. The registry does not give sufficient model details to translate it into a direct comparison of post-intervention scores.

These results do not establish that the interventions are identical, and they do not show that the observed difference is clinically important. The registry record does not report a threshold for meaningful change. The mean contrast is a description of the posted scores, while its confidence interval expresses uncertainty about the contrast under the calculation used. Neither alone determines practical importance.

The study was designed to assess reading and related language outcomes, with a retention assessment also described in the brief summary. The posted primary results concern the T2 score after intervention. The registry record does not provide results for the later retention assessment in the registry record reported for this analysis, so the stability of gains cannot be assessed from these primary endpoint summaries.

Finally, the primary-outcome description notes attrition and smaller participant numbers than at baseline. The available summaries do not describe missing-data patterns or the analysis population in sufficient detail to determine how attrition may have influenced the estimates. The conclusions should remain tied to the recorded outcome summaries and their uncertainty.

17. Why This Trial Matters Statistically

This trial illustrates why an endpoint, an estimand and a model coefficient need to be distinguished. The posted TWRE means support a straightforward description of group scores after the intervention, and the unadjusted contrast makes that comparison explicit. The separately reported group-by-time estimate comes from a mixed-effects model and addresses a different statistical quantity. Treating these as interchangeable would make the result harder to interpret rather than clearer.

The confidence interval is important because the point estimate alone does not communicate the range of uncertainty. Here the unadjusted interval includes zero and extends to positive values, while the reported model interval belongs to a different term. Readers therefore need both the effect estimate and the analysis definition to understand what the results say.

The registry also describes repeated assessments and a retention period. That structure raises practical analytic questions: how repeated observations were modeled, how attrition was handled, and whether outcomes across time and domains were treated as a family of tests. The registry names a mixed-effects method but does not give the full specification, missing-data approach or multiplicity strategy. These are important reporting details for interpreting a longitudinal intervention study.

Finally, the record distinguishes a stated recruitment target of 450 children from recorded enrollment of 315 and notes smaller numbers for the primary endpoint because of attrition. These details matter for understanding the study's observed information and the precision of its estimates, without implying a particular explanation for the difference. The sample-size illustrations are useful for future planning but are not a retrospective measure of trial power.

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 registry record contains limited information for evaluating how closely the posted analyses follow the original protocol. It does not report the original sample-size justification, detailed randomization procedure, allocation concealment, masking of outcome assessors, or a complete statistical analysis plan. No covariate list or full mixed-effects model specification is given, making it difficult to assess the sponsor's group-by-time coefficient in detail.

The brief summary describes a recruitment target of 450 children, while recorded enrollment is 315. The record does not explain this difference. It also notes that fewer participants contributed to the primary outcome than to baseline measures because of attrition, but does not give reasons, timing, or group-specific missingness for the primary assessment. These limitations constrain assessment of representativeness and possible bias from missing outcomes.

The primary endpoint summary is for the post-intervention T2 assessment. Although the brief summary describes a retention period and later assessment, the registry entries for this analysis do not include those retention results. The record also lists several secondary outcome domains without reporting a multiplicity strategy, so the statistical interpretation of a set of outcome comparisons cannot be fully assessed from the record.

The trial is described as randomized and parallel, with no masking. The record does not provide enough detail to assess implementation fidelity, adherence, contamination between activities, or the extent of outcome-assessor awareness. The descriptions of both intervention arms are truncated, limiting assessment of their precise content and delivery. These are limits on what can be determined from the registry record, not proof that the underlying study lacked such procedures or information.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Neurocognitive Factors in EdTech Intervention Response — NCT04568824.
• Registry record last updated 2026-10-20; status last verified 2026-09.
• Results first posted 2026-10-20.
• 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

The registry describes a completed randomized, parallel-group trial of a home-based GraphoLearn reading intervention compared with an active math control. It records enrollment of 315 and identifies children with Reading Disability as the study population; the brief summary describes a target age range of 6.0-10.0 and a recruitment plan of 450 children.

The primary endpoint is T2 Total Word Reading Efficiency after the 12-week intervention. Posted means are 83.12 for GraphoLearn and 80.28 for math control. The unadjusted mean difference is 2.84, with a 95% confidence interval from -0.941 to 6.621 and p = 0.141. The sponsor's reported mixed-effects group-by-time estimate is -0.018, with a 95% confidence interval from -0.119 to 0 and p = 0.728.

The primary measure has fewer participants than baseline measures because of attrition. The registry record does not report the missing-data method, full model specification, original sample-size justification, or multiplicity strategy. Serious adverse events and deaths are reported as 0 in each listed group.