Temel İstatistik Hataları

Özet

İstatistiksel analizlerin doğru uygulanması kadar, sonuçların doğru yorumlanması ve şeffaf biçimde raporlanması da bilimsel çalışmaların güvenilirliği açısından kritik öneme sahiptir. Buna karşın yetersiz örneklem büyüklüğü, uygun olmayan araştırma tasarımı, yanlış test seçimi, çoklu karşılaştırmaların ihmal edilmesi, p-değerine aşırı odaklanma ve korelasyon–nedensellik yanılgısı gibi istatistiksel hatalar literatürde sıkça görülmektedir. Bu bölümde bilimsel makalelerde en yaygın karşılaşılan temel istatistiksel hatalar ele alınmış ve bu hataların sonuçların yorumlanması ile editöryal değerlendirme sürecine etkileri tartışılmıştır. Ayrıca etki büyüklükleri ve güven aralıklarının önemi vurgulanmış, istatistiksel olarak anlamlı olmayan bulguların nasıl raporlanması gerektiği açıklanmıştır. Son olarak CONSORT, STROBE ve PRISMA gibi uluslararası raporlama kılavuzları doğrultusunda istatistiksel hataların önlenmesine yönelik pratik öneriler sunulmuştur.

Correct application of statistical analyses is as critical to the reliability of scientific studies as the accurate interpretation and transparent reporting of results. Nevertheless, statistical errors such as inadequate sample size, inappropriate study design, incorrect test selection, neglect of multiple comparisons, excessive focus on p-values, and confusion between correlation and causality are frequently encountered in the literature. This section addresses the most common fundamental statistical errors observed in scientific manuscripts and discusses their impact on the interpretation of results and the editorial evaluation process. In addition, the importance of effect sizes and confidence intervals is emphasized, and appropriate reporting of statistically non-significant findings is explained. Finally, practical recommendations for preventing statistical errors are presented in line with international reporting guidelines such as CONSORT, STROBE, and PRISMA.

Referanslar

Slutsky DJ. Statistical errors in clinical studies. J Wrist Surg. 2013;2(4):285–287.

Fernandes-Taylor S, Hyun JK, Reeder RN, Harris AH. Common statistical and research design problems in manuscripts submitted to high-impact medical journals. BMC Res Notes. 2011;4:304. doi:10.1186/1756-0500-4-304

Makin TR, Orban de Xivry JJ. Ten common statistical mistakes to watch out for when writing or reviewing a manuscript. Elife. 2019;8:e48175. doi:10.7554/eLife.48175

Simmons JP, Nelson LD, Simonsohn U. False-positive psychology: undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychol Sci. 2011;22(11):1359–1366. doi:10.1177/0956797611417632

Messerli FH. Chocolate consumption, cognitive function, and Nobel laureates. N Engl J Med. 2012;367(16):1562–1564. doi:10.1056/NEJMon1211064

Pearl J. Causal diagrams for empirical research. Biometrika. 1995;82(4):669–688.

Hill AB. The environment and disease: Association or causation? Proc R Soc Med. 1965;58(5):295–300.

Rosnow RL, Rosenthal R. Statistical procedures and the justification of knowledge in psychological science. Am Psychol. 1989;44(10):1276–1284. doi:10.1037/0003-066X.44.10.1276

Wasserstein RL, Lazar NA. The ASA’s statement on p-values: Context, process, and purpose. Am Stat. 2016;70(2):129–133. doi:10.1080/00031305.2016.1154108

Hopewell S, Chan AW, Collins GS, Hróbjartsson A, Moher D, Schulz KF, et al. CONSORT 2025 statement: Updated guideline for reporting randomised trials. BMJ. 2025;389:e080987. doi: 10.1136/bmj-2024-081123

von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. Ann Intern Med. 2007;147(8):573–577. doi:10.7326/0003-4819-147-8-200710160-00010

Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. PLoS Med. 2021;18(3):e1003583. doi:10.1371/journal.pmed.1003583

Gelecek

17 Şubat 2026

Lisans

Lisans