Tạp chí đã xuất bản
2004
ISSN
ISSN 2615-9813
ISSN (số cũ) 1859-3682

SỐ 208 | Tháng 7/2023

Giải pháp từng phần cho cuộc khủng hoảng nhân rộng trong kinh tế

William M. Briggs

Tóm tắt:

Mục đích - Nghiên cứu quan trọng từng được cho là không thể công bố đã thất bại trong việc nhân rộng, không chỉ trong kinh tế, mà trong tất cả các ngành khoa học. Do đó, vấn đề này không có gì đáng tranh cãi cũng như là một số nguyên nhân, như thiếu hụt năng lượng, báo cáo có chọn lọc, hiệu ứng ngăn kéo tệp, dữ liệu không được công khai, v.v. Một số giải pháp khá quan trọng đã được đưa ra, chẳng hạn như các giả thuyết đăng ký trước và kế hoạch phân tích dữ liệu.
Thiết kế/phương pháp nghiên cứu /phương pháp tiếp cận - Đây là một bài viết đánh giá về cuộc khủng hoảng nhân rộng, hiện đã được biết đến rất nhiều.
Kết quả - Nghiên cứu này đưa ra một giải pháp từng phần mới, đó là nhắc nhở các nhà nghiên cứu rằng mối tương quan không bao hàm một cách hợp lý mối quan hệ nhân quả. Tác dụng của lời nhắc nhở này là tránh việc kiểm tra "ý nghĩa", dù ở dạng thường xuyên hay dạng Bayes (như hệ số Bayes) và báo cáo các mô hình ở dạng dự đoán, để bất kỳ ai cũng có thể kiểm tra tính xác thực của bất kỳ mô hình nào. Trên thực tế, tất cả các giấy tờ đều có thể trải qua quá trình thử nghiệm sao chép.
Tính mới/giá trị - Tác giả lập luận rằng giải pháp này hoặc bất kỳ giải pháp nào sẽ không bao giờ loại bỏ được mọi lỗi.

 

Tài liệu tham khảo:

  1. Baker, M. (2016), “1,500 scientists lift the lid on reproducibility”, Nature, Vol. 533, pp. 452-454.
  2. Banerjee, A., Duflo, E., Finkelstein, A., Katz, L.F., Olken, B.A. and Sautmann, A. (2020), In Praise of Moderation: Suggestions for the Scope and Use of Pre-analysis Plans for Rcts in Economics, Working Paper 26993, National Bureau of Economic Research.
  3. Bernardo, J.M. and Smith, A.F.M. (2000), Bayesian Theory, Wiley, New York.
  4. Bose, S. (2004), “On the robustness of the predictive distribution for sampling from finite populations”, Statistics and Probability Letters, Vol. 69 No. 1, pp. 21-27.
  5. Briggs, W.M. (2016), Uncertainty: The Soul of Probability, Modeling and Statistics, Springer, New York.
  6. Briggs, W.M. (2019), “Reality-based probability and statistics: solving the evidential crisis”, Asian Journal of Business and Economics, Vol. 3 No. 1, pp. 37-80.
  7. Briggs, W.M. and Nguyen, H.T. (2019), “Clarifying asa’s views on p values in hypothesis testing”, Asian Journal of Business and Economics, Vol. 3 No. 2, pp. 1-16.
  8. Briggs, W.M. (2019), “Everything wrong with p values under one roof”, in Kreinovich, V., Thach, N.N., Trung, N.D. and Thanh, D.V. (Eds), Beyond Traditional Probabilistic Methods in Economics, Springer, New York, pp. 22-44.
  9. Bruns, S.B. and Ioannidis, J.P.A. (2016), “p-curve and p-hacking in observational research”, PLoS ONE, Vol. 11 No. 2, e0149144.
  10. Camerer, C.F., Dreber, A., Forsell, E., Ho, T.H., Huberand, J., Johannesson, M., Kirchler, M., Almenberg, J. and Altmejd, A. (2016), “Evaluating replicability of laboratory experiments in economics”, Science, Vol. 351 No. 6280, pp. 1433-1436.
  11. Camerer, C.F., Dreber, A., Holzmeister, F., Ho, T.H., Huber, J., Johannesson, M., Kirchler, M., Nave, G., Nosek, B.A., Pfeiffer, T., Altmejd, A., Buttrick, N., Chan, T., Chen, Y., Forsell, E., Gampa, A., Heikensten, E., Hummer, L., Imai, T., Isaksson, S., Manfredi, D., Rose, J., Wagenmakers, E.J. and Wu, H. (2018), “Evaluating the replicability of social science experiments in nature and science between 2010 and 2015”, Nature Human Behaviour, Vol. 2 No. 9, pp. 637-644.
  12. Charlton, A. (2023), “Replications of marketing studiesass”, available at: https://openmkt.org/research/replications-of-marketing-studies/
  13. Clarke, B.S. and Clarke, J.L. (2018), Predictive Statistics, Cambridge University Press, Cambridge.
  14. Fanelli, D. (2017), “Is science really facing a reproducibility crisis, and do we need it to?”, PNAS, Vol. 115 No. 11, pp. 2628-2631.
  15. Hájek, A. (1996), “Mises redux—redux: fifteen arguments against finite frequentism”, Erkenntnis, Vol. 45 Nos 2-3, pp. 209-227.
  16. Harvey, C.R. (2019), “Replication in financial economics”, SSRN, available at SSRN: https://ssrn.com/abstract=3409466
  17. Horton, R. (2015), “Offline: what is medicine's 5 sigma?”, The Lancet, Vol. 385 No. 9976, p. 1380.
  18. Hossenfelder, S. (2017), “Science needs reason to be trusted”, Nature Physics, Vol. 13 No. 4, pp. 316-317.
  19. Ioannidis, J.P.A. (2005), “Contradicted and initially stronger effects in highly cited clinical research”, JAMA, Vol. 294 No. 2, pp. 218-228.
  20. Klein, R.A., Vianello, M., Hasselman, F., Adams, B.G., Adams, R.B. Jr, Alper, S.,Aveyard, M., Axt, J.R., Babalola, M.T., Bahník, Š., Batra, R., Berkics, M., Bernstein, M.J., Berry, D.R., Bialobrzeska, O., Binan, E.D., Bocian, K., Brant,M.J., Busching, R., Rédei, A.C., Cai, H., Cambier, F., Cantarero, K., Carmihael, C.L., Ceric, F., Chandler, J., Chang, J.H., Chatard, A., Chen, E.E., Cheong, W., Cicero, D.C., Coen, S., Coleman, J.A., Collisson, B., Conway, M.A., Corker, K.S., Curran, P.G., Cushman, F., Dagona, Z.K., Dalgar, I., Rosa, A.D., Davis, W.E., de Bruijn, M., De Schutter, L., Devos, T., deVries, M., Doğulu, C., Dozo, N., Dukes, K.N., Dunham, Y., Durrheim, K., Ebersole, C.R., Edlund, J.E., Eller, A., English, A.S., Finck, C., Frankowska, N., Freyre, M.Á., Friedman, M., Galliani, E.M., Gandi, J.C., Ghoshal, T., Giessner, S.R., Gill, T., Gnambs, T., Gómez, Á., González, R., Graham, J., Grahe, J.E., Grahek, I., Green, E.G.T., Hai, K., Haigh, M., Haines, E.L., Hall, M.P., Heffernan, M.E., Hicks, J.A., Houdek, P., Huntsinger, J.R., Huynh, H.P., Ijzerman, H., Inbar, Y., Innes-Ker, Å.H., Jiménez-Leal, W., John, M.-S., JoyGaba, J.A., Kamiloğlu, R.G., Kappes, H.B., Karabati, S., Karick, H., Keller, V.N., Kende, A., Kervyn, N., Knežević, G., Kovacs, C., Krueger, L.E., Kurapov, G., Kurtz, J., Lakens, D., Lazarević, L.B., Levitan, C.A., Lewis, N.A., Jr, Lins, S., Lipsey, N.P., Losee, J.E., Maassen, E., Maitner, A.T., Malingumu, W., Mallett, R.K., Marotta, S.A., Mededovic, J., MenaPacheco, F., Milfont, T.L., Morris, W.L., Murphy, S.C., Myachykov, A., Neave, N., Neijenhuijs, K., Nelson, A.J., Neto, F., Nichols, A.L., Ocampo, A., O'Donnell, S.L., Oikawa, H., Oikawa, M., Ong, E., Orosz, G., Osowiecka, M., Packard, G., PérezSánchez, R., Petrović, B., Pilati, R., Pinter, B., Podesta, L., Pogge, G., Pollmann, M.M.H., Rutchick, A.M., Saavedra, P., Saeri, A.K., Salomon, E., Schmidt, K., Schönbrodt, F.D., Sekerdej, M.B., Sirlopú, D., Skorinko, J.L.M., Smith, M.A., Smith-Castro, V., Smolders, K.C.H.J., Sobkow, A., Sowden, W., Spachtholz, P., Srivastava, M., Steiner, T.G., Stouten, J., Street, C.N.H., Sundfelt, O.K., Szeto, S., Szumowska, E., Tang, A.C.W., Tanzer, N., Tear, M.J., Theriault, J., Thomae, M., Torres, D., Traczyk, J., Tybur, J.M., Ujhelyi, A., van Aert, R.C.M., van Assen, M.A.L.M., van der Hulst, M., van Lange, P.A.M., van ’t Veer, A.E., VásquezEcheverría, A., Vaughn, L.A., Vázquez, A., Vega, L.D., Verniers, C., Verschoor, M., Voermans, I.P.J., Vranka, M.A., Welch, C., Wichman, A.L., Williams, L.A., Wood, M., Woodzicka, J.A., Wronska, M.K., Young, L., Zelenski, J.M., Zhijia, Z. and Nosek, B.A. (2018), “Many labs 2: investigating variation in replicability across samples and settings”, Advances in Methods and Practices in Psychological Science, Vol. 1 No. 4, pp. 443-490.
  21. Knuteson, B. (2016), “The solution to science's replication crisis”.
  22. Lancaster, T. (2004), An Introduction to ModernBayesianEconometrics, Blackwell Publishing, New York.
  23. Macleod, M. and the University of Edinburgh Research Strategy Group (2022), “Improving the reproducibility and integrity of research: what can different stakeholders contribute?”, BMC Research Notes, Vol. 15 No. 1, p. 146.
  24. Mede, N.G., Schäfer, M.S., Ziegler, R. and Weißkopf, M. (2021), “The “replication crisis” in the public eye: Germans' awareness and perceptions of the (ir)reproducibility of scientific research”, Public Understanding of Science, Vol. 30 No. 1, pp. 91-102, PMID: 32924865.
  25. Miyakawa, T. (2020), “No raw data, no science: another possible source of the reproducibility crisis”, Molecular Brain, Vol. 13, pp. 1-6.
  26. Page, L., Noussair, C.N. and Slonim, R. (2021), “The replication crisis, the rise of new research practices and what it means for experimental economics”, Journal of the Economic Science Association, Vol. 7 No. 2, pp. 210-225.
  27. Reed, W. (2017), “Replication in labor economics”, IZA World of Labor, Vol. 413.
  28. Sharpe, D. and Poets, S. (2020), “Meta-analysis as a response to the replication crisis”, Psychologie canadienne, Vol. 61 No. 4, pp. 377-387.
  29. Smaldino, P.E. and McElreath, R. (2016), “The natural selection of bad science”, Royal Society Open Science, Vol. 3 No. 9, doi: 10.1098/rsos.160384.
  30. Smith, R.L. (1998), “Bayesian and frequentist approaches to parametric predictive inference (with discussion)”, Bayesian Statistics, Oxford University Press, Vol. 6, pp. 589-612.
  31. Spanos, A. (2022), “Frequentist model-based statistical induction and the replication crisis”, Journal of Quantitative Economics, Vol. 20 No. 1, pp. 133-159.
  32. Trafimow, D. (2018), “An a priori solution to the replication crisis”, Philosophical Psychology, Vol. 31 No. 8, pp. 1188-1214.
  33. Williams, C.R. (2019), “How redefining statistical significance can worsen the replication crisis”, Economics Letters, Vol. 181, pp. 65-69.


A Partial Solution for the Replication Crisis in Economics

Abstract:

Purpose - Important research once thought unassailable has failed to replicate. Not just in economics, but in all science. The problem is therefore not in dispute nor are some of the causes, like low power, selective reporting, the file drawer effect, publicly unavailable data and so forth. Some partially worthy solutions have already been offered, like pre-registering hypotheses and data analysis plans.

Design/methodology/approach - This is a review paper on the replication crisis, which is by now very well known.

Findings - This study offers another partial solution, which is to remind researchers that correlation does not logically imply causation. The effect of this reminder is to eschew “significance” testing, whether in frequentist or Bayesian form (like Bayes factors) and to report models in predictive form, so that anybody can check the veracity of any model. In effect, all papers could undergo replication testing.

Originality/value - The author argues that this, or any solution, will never eliminate all errors.