Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

On the choice of computational unit in statistical analysis.

N Blomqvist

    Journal of Clinical Periodontology
    |November 1, 1985
    PubMed
    Summary

    When analyzing data from multi-level trials, always use the highest-level unit for statistical calculations. Using lower levels underestimates standard errors and significance, leading to incorrect conclusions in research.

    Related Concept Videos

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    Psychosocial stress causes endothelial injury in cynomolgus monkeys via beta1-adrenoceptor activation.

    Atherosclerosis·1998
    Same author

    Estimating confidence intervals for cost-effectiveness ratios: an example from a randomized trial.

    Statistics in medicine·1998
    Same author

    Characterization, standardization and experiences with KONE ISE for Mg2+.

    Scandinavian journal of clinical and laboratory investigation. Supplementum·1994
    Same author

    On the bias caused by regression toward the mean in studying the relation between change and initial value.

    Journal of clinical periodontology·1987
    Same author

    Analysis of change--are base-line measurements needed? Some statistical comments on a common experimental design.

    Journal of clinical periodontology·1985
    Same author

    Sampling of subgingival microbiota for dark-field microscopy.

    Journal of clinical periodontology·1985

    Area of Science:

    • Biostatistics
    • Clinical Trials Methodology
    • Statistical Inference

    Background:

    • Multi-level experimental designs are common in scientific research, involving hierarchical data structures.
    • Accurate statistical inference relies on appropriate selection of the computational unit.
    • Previous analyses may have incorrectly utilized lower-level units, potentially biasing results.

    Purpose of the Study:

    • To emphasize the critical importance of selecting the correct computational unit in multi-level trials.
    • To demonstrate how using the highest-level unit impacts standard error and P-value calculations.
    • To prevent underestimation of statistical significance in hierarchical data analyses.

    Main Methods:

    • The study highlights the principle of using the highest-level experimental unit for statistical inference.
    • It explains the impact of unit selection on standard error estimation.
    • A numerical illustration is provided to elucidate the concept.

    Main Results:

    • Utilizing the highest-level unit as the computational unit is essential for valid statistical inference.
    • Employing a lower-level unit leads to an underestimation of the standard error.
    • This underestimation results in an inflated level of significance (P-value), potentially leading to false positive findings.

    Conclusions:

    • Correct identification and use of the highest-level unit in multi-level trials are crucial for accurate statistical analysis.
    • Failure to do so can lead to erroneous conclusions regarding treatment effects or observed phenomena.
    • Researchers must adhere to this principle for robust and reliable scientific reporting.

    Related Experiment Videos