{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T15:58:29Z","timestamp":1791475109587,"version":"4.3.4"},"reference-count":29,"publisher":"Institute of Mathematical Statistics","issue":"5","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Ann. Statist."],"published-print":{"date-parts":[[2020,10,1]]},"abstract":"<jats:p>Modern data-driven approaches to modeling make extensive use of covariate\/model selection. Such selection incurs a cost: it invalidates classical statistical inference. A conservative remedy to the problem was proposed by Berk et al. (Ann. Statist. 41 (2013) 802\u2013837) and further extended by Bachoc, Preinerstorfer and Steinberger (2016). These proposals, labeled \u201cPoSI methods,\u201d provide valid inference after arbitrary model selection. They are computationally NP-hard and have limitations in their theoretical justifications. We therefore propose computationally efficient confidence regions, named \u201cUPoSI\u2019 (\u201cU\u201d is for \u201cuniform\u201d or \u201cuniversal.\u201d) and prove large-$p$ asymptotics for them. We do this for linear OLS regression allowing misspecification of the normal linear model, for both fixed and random covariates, and for independent as well as some types of dependent data. We start by proving a general equivalence result for the post-selection inference problem and a simultaneous inference problem in a setting that strips inessential features still present in a related result of Berk et al. (Ann. Statist. 41 (2013) 802\u2013837). We then construct valid PoSI confidence regions that are the first to have vastly improved computational efficiency in that the required computation times grow only quadratically rather than exponentially with the total number $p$ of covariates. These are also the first PoSI confidence regions with guaranteed asymptotic validity when the total number of covariates $p$ diverges (almost exponentially) with the sample size $n$. Under standard tail assumptions, we only require $(\\log p)^{7}=o(n)$ and $k=o(\\sqrt{n\/\\log p})$ where $k$ ($\\le p$) is the largest number of covariates (model size) considered for selection. We study various properties of these confidence regions, including their Lebesgue measures, and compare them theoretically with those proposed previously.<\/jats:p>","DOI":"10.1214\/19-aos1917","type":"journal-article","created":{"date-parts":[[2020,9,18]],"date-time":"2020-09-18T22:02:27Z","timestamp":1600466547000},"page":"2953-2981","source":"Crossref","is-referenced-by-count":27,"title":["Valid post-selection inference in model-free linear regression"],"prefix":"10.1214","volume":"48","author":[{"given":"Arun K.","family":"Kuchibhotla","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lawrence D.","family":"Brown","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Andreas","family":"Buja","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Junhui","family":"Cai","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Edward I.","family":"George","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Linda H.","family":"Zhao","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"108","reference":[{"key":"20","doi-asserted-by":"publisher","unstructured":"Lee, J. D., Sun, D. L., Sun, Y. and Taylor, J. E. (2016). Exact post-selection inference, with application to the lasso. <i>Ann. Statist.<\/i> <b>44<\/b> 907\u2013927.","DOI":"10.1214\/15-AOS1371"},{"key":"10","doi-asserted-by":"publisher","unstructured":"Cui, Y., Leng, C. and Sun, D. (2016). Sparse estimation of high-dimensional correlation matrices. <i>Comput. Statist. Data Anal.<\/i> <b>93<\/b> 390\u2013403.","DOI":"10.1016\/j.csda.2014.10.001"},{"key":"14","doi-asserted-by":"publisher","unstructured":"Hjort, N. L. and Claeskens, G. (2003). Frequentist model average estimators. <i>J. Amer. Statist. Assoc.<\/i> <b>98<\/b> 879\u2013899.","DOI":"10.1198\/016214503000000828"},{"key":"5","doi-asserted-by":"publisher","unstructured":"Berk, R., Brown, L., Buja, A., Zhang, K. and Zhao, L. (2013). Valid post-selection inference. <i>Ann. Statist.<\/i> <b>41<\/b> 802\u2013837.","DOI":"10.1214\/12-AOS1077"},{"key":"27","doi-asserted-by":"crossref","unstructured":"Tibshirani, R. J., Taylor, J., Lockhart, R. and Tibshirani, R. (2016). Exact post-selection inference for sequential regression procedures. <i>J. Amer. Statist. Assoc.<\/i> <b>111<\/b> 600\u2013620.","DOI":"10.1080\/01621459.2015.1108848"},{"key":"9","doi-asserted-by":"publisher","unstructured":"Claeskens, G. and Carroll, R. J. (2007). An asymptotic theory for model selection inference in general semiparametric problems. <i>Biometrika<\/i> <b>94<\/b> 249\u2013265.","DOI":"10.1093\/biomet\/asm034"},{"key":"22","doi-asserted-by":"crossref","unstructured":"Olshen, R. A. (1973). The conditional level of the $F$-test. <i>J. Amer. Statist. Assoc.<\/i> <b>68<\/b> 692\u2013698.","DOI":"10.1080\/01621459.1973.10481406"},{"key":"26","unstructured":"Tian, X., Bi, N. and Taylor, J. (2016). Magic: A general, powerful and tractable method for selective inference. Preprint. Available at <a href=\"arXiv:1607.02630\">arXiv:1607.02630<\/a>."},{"key":"2","doi-asserted-by":"publisher","unstructured":"Bachoc, F., Leeb, H. and P\u00f6tscher, B. M. (2019). Valid confidence intervals for post-model-selection predictors. <i>Ann. Statist.<\/i> <b>47<\/b> 1475\u20131504.","DOI":"10.1214\/18-AOS1721"},{"key":"15","unstructured":"Kivaranovic, D. and Leeb, H. (2018). Expected length of post-model-selection confidence intervals conditional on polyhedral constraints. Preprint. Available at <a href=\"arXiv:1803.01665\">arXiv:1803.01665<\/a>."},{"key":"28","doi-asserted-by":"publisher","unstructured":"Tibshirani, R. J., Rinaldo, A., Tibshirani, R. and Wasserman, L. (2018). Uniform asymptotic inference and the bootstrap after model selection. <i>Ann. Statist.<\/i> <b>46<\/b> 1255\u20131287.","DOI":"10.1214\/17-AOS1584"},{"key":"8","doi-asserted-by":"publisher","unstructured":"Chernozhukov, V., Chetverikov, D. and Kato, K. (2017). Central limit theorems and bootstrap in high dimensions. <i>Ann. Probab.<\/i> <b>45<\/b> 2309\u20132352.","DOI":"10.1214\/16-AOP1113"},{"key":"4","doi-asserted-by":"publisher","unstructured":"Belloni, A., Rosenbaum, M. and Tsybakov, A. B. (2017). Linear and conic programming estimators in high dimensional errors-in-variables models. <i>J. R. Stat. Soc. Ser. B. Stat. Methodol.<\/i> <b>79<\/b> 939\u2013956.","DOI":"10.1111\/rssb.12196"},{"key":"1","doi-asserted-by":"publisher","unstructured":"Bachoc, F., Blanchard, G. and Neuvial, P. (2018). On the post selection inference constant under restricted isometry properties. <i>Electron. J. Stat.<\/i> <b>12<\/b> 3736\u20133757.","DOI":"10.1214\/18-EJS1490"},{"key":"3","doi-asserted-by":"publisher","unstructured":"Bachoc, F., Preinerstorfer, D. and Steinberger, L. (2016). Uniformly valid confidence intervals post-model-selection. Preprint. Available at <a href=\"arXiv:1611.01043\">arXiv:1611.01043<\/a>.","DOI":"10.1214\/19-AOS1815"},{"key":"6","doi-asserted-by":"publisher","unstructured":"Buehler, R. J. and Feddersen, A. P. (1963). Note on a conditional property of Student\u2019s $t$. <i>Ann. Math. Stat.<\/i> <b>34<\/b> 1098\u20131100.","DOI":"10.1214\/aoms\/1177704034"},{"key":"7","doi-asserted-by":"publisher","unstructured":"Buja, A., Brown, L. D., Berk, R. A., George, E. I., Traskin, M., Pitkin, E., Zhao, L. H. and Zhang, K. (2019). Models as approximations, part I: Consequences illustrated with linear regression. <i>Statist. Sci.<\/i> To appear. Available at <a href=\"arXiv:1404.1578\">arXiv:1404.1578<\/a>.","DOI":"10.1214\/18-STS693"},{"key":"11","unstructured":"Deng, H. and Zhang, C.-H. (2017). Beyond Gaussian approximation: Bootstrap for maxima of sums of independent random vectors. Preprint. Available at <a href=\"arXiv:1705.09528\">arXiv:1705.09528<\/a>."},{"key":"12","unstructured":"Fithian, W., Sun, D. L. and Taylor, J. (2014). Optimal inference after model selection. Preprint. Available at <a href=\"arXiv:1410.2597\">arXiv:1410.2597<\/a>."},{"key":"13","doi-asserted-by":"publisher","unstructured":"Freedman, D. A. (1983). A note on screening regression equations. <i>Amer. Statist.<\/i> <b>37<\/b> 152\u2013155.","DOI":"10.2307\/2685877"},{"key":"16","unstructured":"Kuchibhotla, A. K., Brown, L. D. and Buja, A. (2018). Model-free study of ordinary least squares linear regression. Preprint. Available at <a href=\"arXiv:1809.10538\">arXiv:1809.10538<\/a>."},{"key":"17","unstructured":"Kuchibhotla, A. K. and Chakrabortty, A. (2018). Moving beyond sub-Gaussianity in high-dimensional statistics: Applications in covariance estimation and linear regression. Preprint. Available at <a href=\"arXiv:1804.02605\">arXiv:1804.02605<\/a>."},{"key":"18","unstructured":"Kuchibhotla, A. K., Brown, L. D., Buja, A., George, E. I. and Zhao, L. (2018). A model free perspective for linear regression: Uniform-in-model bounds for post selection inference. Preprint. Available at <a href=\"arXiv:1802.05801\">arXiv:1802.05801<\/a>."},{"key":"19","doi-asserted-by":"crossref","unstructured":"Kuchibhotla, A. K., Brown, L. D., Buja, A., Cai, J., George, E. I. and Zhao, L. H. (2020). Supplement to \u201cValid post-selection inference in model-free linear regression.\u201d <a href=\"https:\/\/doi.org\/10.1214\/19-AOS1917SUPP\">https:\/\/doi.org\/10.1214\/19-AOS1917SUPP<\/a>.","DOI":"10.1214\/19-AOS1917SUPP"},{"key":"21","doi-asserted-by":"crossref","unstructured":"Liu, R. Y. and Singh, K. (1995). Using i.i.d. bootstrap inference for general non-i.i.d. models. <i>J. Statist. Plann. Inference<\/i> <b>43<\/b> 67\u201375.","DOI":"10.1016\/0378-3758(94)00008-J"},{"key":"23","doi-asserted-by":"publisher","unstructured":"Rencher, A. C. and Pun, F. C. (1980). Inflation of $R^{2}$ in best subset regression. <i>Technometrics<\/i> <b>22<\/b> 49\u201353.","DOI":"10.2307\/1268382"},{"key":"24","doi-asserted-by":"crossref","unstructured":"Rinaldo, A., Wasserman, L., G\u2019Sell, M.,and Lei, J. (2019). Bootstrapping and sample splitting for high-dimensional, assumption-free inference. <i>Ann. Statist.<\/i> To appear. Available at <a href=\"arXiv:1611.05401\">arXiv:1611.05401<\/a>.","DOI":"10.1214\/18-AOS1784"},{"key":"25","doi-asserted-by":"crossref","unstructured":"Simmons, J. P., Nelson, L. D. and Simonsohn, U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. <i>Psychol. Sci.<\/i> <b>22<\/b> 1359\u20131366.","DOI":"10.1177\/0956797611417632"},{"key":"29","unstructured":"Zhang, X. and Cheng, G. (2014). Bootstrapping high dimensional time series. Preprint. Available at <a href=\"arXiv:1406.1037\">arXiv:1406.1037<\/a>."}],"container-title":["The Annals of Statistics"],"original-title":[],"link":[{"URL":"https:\/\/projecteuclid.org\/download\/pdfview_1\/euclid.aos\/1600480938","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,8]],"date-time":"2026-10-08T15:31:40Z","timestamp":1791473500000},"score":1,"resource":{"primary":{"URL":"https:\/\/projecteuclid.org\/journals\/annals-of-statistics\/volume-48\/issue-5\/Valid-post-selection-inference-in-model-free-linear-regression\/10.1214\/19-AOS1917.full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2020,10,1]]},"references-count":29,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2020,10,1]]}},"URL":"https:\/\/doi.org\/10.1214\/19-aos1917","relation":{},"ISSN":["0090-5364","2168-8966"],"issn-type":[{"value":"0090-5364","type":"print"},{"value":"2168-8966","type":"electronic"}],"subject":[],"published":{"date-parts":[[2020,10,1]]}}}