概率論畢業(yè)論文外文翻譯--統(tǒng)計(jì)假設(shè)檢驗(yàn)

1、<p>　　Statistical hypothesis testingLast updated 44 minutes ago</p><p>　　Adriana Albu，Loredana Ungureanu</p><p>　　Politehnica University Timisoara, adrianaa@aut.utt.ro</p><p>　

2、　Politehnica University Timisoara, loredanau@aut.utt.ro</p><p>　　Abstract In this article, we present a Bayesian statistical hypothesis testing inspection, testing theory and the process Mentioned hypothesis

3、 testing in the real world and the importance of, and successful test of the Notes.</p><p>　　Key words Bayesian hypothesis testing; Bayesian inference; Test of significance</p><p>　　Introduction

4、 </p><p>　　A statistical hypothesis test is a method of making decisions using data, whether from a controlled experiment or an observational study (not controlled). In statistics, a result is called statist

5、ically significant if it is unlikely to have occurred by chance alone, according to a pre-determined threshold probability, the significance level. The phrase "test of significance" was coined by Ronald Fisher:

6、 "Critical tests of this kind may be called tests of significance, and when such tests are ava</p><p>　　Hypothesis testing is sometimes called confirmatory data analysis, in contrast to exploratory data

7、 analysis. In frequency probability, these decisions are almost always made using null-hypothesis tests. These are tests that answer the question Assuming that the null hypothesis is true, what is the probability of obse

8、rving a value for the test statistic that is at least as extreme as the value that was actually observed?) 2[] More formally, they represent answers to the question, posed before und</p><p>　　Statistical hyp

9、othesis testing is a key technique of frequentist statistical inference. The Bayesian approach to hypothesis testing is to base rejection of the hypothesis on the posterior probability.[3][4] Other approaches to reaching

10、 a decision based on data are available via decision theory and optimal decisions.</p><p>　　The critical region of a hypothesis test is the set of all outcomes which cause the null hypothesis to be rejected

11、in favor of the alternative hypothesis. The critical region is usually denoted by the letter C.</p><p>　　One-sample tests are appropriate when a sample is being compared to the population from a hypothesis.

12、The population characteristics are known from theory or are calculated from the population.</p><p>　　Two-sample tests are appropriate for comparing two samples, typically experimental and control samples fro

13、m a scientifically controlled experiment.</p><p>　　Paired tests are appropriate for comparing two samples where it is impossible to control important variables. Rather than comparing two sets, members are pa

14、ired between samples so the difference between the members becomes the sample. Typically the mean of the differences is then compared to zero.</p><p>　　Z-tests are appropriate for comparing means under strin

15、gent conditions regarding normality and a known standard deviation.</p><p>　　T-tests are appropriate for comparing means under relaxed conditions (less is assumed).</p><p>　　Tests of proportions

16、 are analogous to tests of means (the 50% proportion).</p><p>　　Chi-squared tests use the same calculations and the same probability distribution for different applications:</p><p>　　Chi-squared

17、 tests for variance are used to determine whether a normal population has a specified variance. The null hypothesis is that it does. </p><p>　　Chi-squared tests of independence are used for deciding whether

18、two variables are associated or are independent. The variables are categorical rather than numeric. It can be used to decide whether left-handedness is correlated with libertarian politics (or not). The null hypothesis i

19、s that the variables are independent. The numbers used in the calculation are the observed and expected frequencies of occurrence (from contingency tables). </p><p>　　Chi-squared goodness of fit tests are us

20、ed to determine the adequacy of curves fit to data. The null hypothesis is that the curve fit is adequate. It is common to determine curve shapes to minimize the mean square error, so it is appropriate that the goodness-

21、of-fit calculation sums the squared errors. </p><p>　　F-tests (analysis of variance, ANOVA) are commonly used when deciding whether groupings of data by category are meaningful. If the variance of test score

22、s of the left-handed in a class is much smaller than the variance of the whole class, then it may be useful to study lefties as a group. The null hypothesis is that two variances are the same - so the proposed grouping i

23、s not meaningful.</p><p>　　The testing process</p><p>　　In the statistical literature, statistical hypothesis testing plays a fundamental role. The usual line of reasoning is as follows:</p&g

24、t;<p>　　There is an initial research hypothesis of which the truth is unknown. </p><p>　　The first step is to state the relevant null and alternative hypotheses. This is important as mis-stating the h

25、ypotheses will muddy the rest of the process. Specifically, the null hypothesis allows attaching an attribute: it should be chosen in such a way that it allows us to conclude whether the alternative hypothesis can either

26、 be accepted or stays undecided as it was before the test.[9] </p><p>　　The second step is to consider the statistical assumptions being made about the sample in doing the test; for example, assumptions abou

27、t the statistical independence or about the form of the distributions of the observations. This is equally important as invalid assumptions will mean that the results of the test are invalid. </p><p>　　Decid

28、e which test is appropriate, and state the relevant test statistic T. </p><p>　　Derive the distribution of the test statistic under the null hypothesis from the assumptions. In standard cases this will be a

29、well-known result. For example the test statistic may follow a Student's t distribution or a normal distribution. </p><p>　　Select a significance level (α), a probability threshold below which the null h

30、ypothesis will be rejected. Common values are 5% and 1%. </p><p>　　The distribution of the test statistic under the null hypothesis partitions the possible values of T into those for which the null-hypothesi

31、s is rejected, the so called critical region, and those for which it is not. The probability of the critical region is α. </p><p>　　Compute from the observations the observed value tobs of the test statistic

32、 T. </p><p>　　Decide to either fail to reject the null hypothesis or reject it in favor of the alternative. The decision rule is to reject the null hypothesis H0 if the observed value tobs is in the critical

33、 region, and to accept or "fail to reject" the hypothesis otherwise. </p><p>　　Use and Importance</p><p>　　Statistics are helpful in analyzing most collections of data. This is equally

34、 true of hypothesis testing which can justify conclusions even when no scientific theory exists. Real world applications of hypothesis testing include [7]:</p><p>　　Testing whether more men than women suffer

35、 from nightmares </p><p>　　Establishing authorship of documents </p><p>　　Evaluating the effect of the full moon on behavior </p><p>　　Determining the range at which a bat can detec

36、t an insect by echo </p><p>　　Deciding whether hospital carpeting results in more infections </p><p>　　Selecting the best means to stop smoking </p><p>　　Checking whether bumper sti

37、ckers reflect car owner behavior </p><p>　　Testing the claims of handwriting analysts </p><p>　　Statistical hypothesis testing plays an important role in the whole of statistics and in statistic

38、al inference. For example, Lehmann (1992) in a review of the fundamental paper by Neyman and Pearson (1933) says: "Nevertheless, despite their shortcomings, the new paradigm formulated in the 1933 paper, and the man

39、y developments carried out within its framework continue to play a central role in both the theory and practice of statistics and can be expected to do so in the foreseeable future".</p><p>　　Significan

40、ce testing has been the favored statistical tool in some experimental social sciences (over 90% of articles in the Journal of Applied Psychology during the early 1990s).[8] Other fields have favored the estimation of par

41、ameters. Editors often consider significance as a criterion for the publication of scientific conclusions based on experiments with statistical results.</p><p><b>　　Cautions</b></p><p&

42、gt;　　The successful hypothesis test is associated with a probability and a type-I error rate. The conclusion might be wrong.</p><p>　　The conclusion of the test is only as solid as the sample upon which it i

43、s based. The design of the experiment is critical. A number of unexpected effects have been observed including:</p><p>　　The Clever Hans effect. A horse appeared to be capable of doing simple arithmetic. <

44、;/p><p>　　The Hawthorne effect. Industrial workers were more productive in better illumination, and most productive in worse. </p><p>　　The Placebo effect. Pills with no medically active ingredient

45、s were remarkably effective. </p><p>　　A statistical analysis of misleading data produces misleading conclusions. The issue of data quality can be more subtle. In forecasting for example, there is no agreeme

46、nt on a measure of forecast accuracy. In the absence of a consensus measurement, no decision based on measurements will be without controversy.</p><p>　　The book How to Lie with Statistics is the most popula

47、r book on statistics ever published.[28] It does not much consider hypothesis testing, but its cautions are applicable, including: Many claims are made on the basis of samples too small to convince. If a report does not

48、mention sample size, be doubtful.</p><p>　　Hypothesis testing acts as a filter of statistical conclusions; Only those results meeting a probability threshold are publishable. Economics also acts as a publica

49、tion filter; Only those results favorable to the author and funding source may be submitted for publication. The impact of filtering on publication is termed publication bias. A related problem is that of multiple testin

50、g (sometimes linked to data mining), in which a variety of tests for a variety of possible effects are applied to a </p><p>　　Those making critical decisions based on the results of a hypothesis test are pru

51、dent to look at the details rather than the conclusion alone. In the physical sciences most results are fully accepted only when independently confirmed. The general advice concerning statistics is, "Figures never l

52、ie, but liars figure" (anonymous).</p><p>　　Controversy</p><p>　　Since significance tests were first popularized many objections have been voiced by prominent and respected statisticians. T

53、he volume of criticism and rebuttal has filled books with language seldom used in the scholarly debate of a dry subject. Much of the criticism was published more than 40 years ago. The fires of controversy have burned ho

54、ttest in the field of experimental psychology. Nickerson surveyed the issues in the year 2000. He included 300 references and reported 20 criticisms and alm</p><p>　　Results of the controversy</p><

55、;p>　　The controversy has produced several results. The American Psychological Association has strengthened its statistical reporting requirements after review,[10] medical journal publishers have recognized the obliga

56、tion to publish some results that are not statistically significant to combat publication bias. and a journal (Journal of Articles in Support of the Null Hypothesis) has been created to publish such results exclusively.

57、Textbooks have added some cautions and increased coverage of the too</p><p>　　References</p><p>　　[1] R. A. Fisher (1925). Statistical Methods for Research Workers, Edinburgh: Oliver and Boyd, 1

58、925, p.43. </p><p>　　[2] Cramer, Duncan; Dennis Howitt (2004). The Sage Dictionary of Statistics. p. 76. ISBN 0-7619-4138-X.  </p><p>　　[3] Schervish, M (1996) Theory of Statistic

59、s, p. 218. Springer ISBN 0-387-94546-6</p><p>　　[4] Kaye, David H.; Freedman, David A. (2011). "Reference Guide on Statistics". Reference manual on scientific evidence (3rd ed.). Eagan, MN Washingt

60、on, D.C: West National Academies Press. p. 259. ISBN 978-0-309-21421-6.</p><p>　　[5] C. S. Peirce (August 1878). "Illustrations of the Logic of Science VI: Deduction, Induction, and Hypothesis

61、". Popular Science Monthly 13. </p><p>　　[6] Fisher, Sir Ronald A. (1956) [1935]. "Mathematics of a Lady Tasting Tea". In James Roy Newman. The World of Mathematics, volume 3 [Design of Experi

62、ments]. Courier Dover Publications. ISBN 978-0-486-41151-4.</p><p>　　[7] Box, Joan Fisher (1978). R.A. Fisher, The Life of a Scientist. New York: Wiley. p. 134. ISBN 0-471-09300-9</p>&

63、lt;p>　　[8] Lehmann, E.L.; Romano, Joseph P. (2005). Testing Statistical Hypotheses (3E ed.). New York: Springer. ISBN 0-387-98864-5.</p><p>　　[9] Adèr,J.H. (2008). Chapter 12: Modelling. In H.J.

64、 Adèr & G.J. Mellenbergh (Eds.) (with contributions by D.J. Hand), Advising on Research Methods: A consultant's companion (pp. 183–209). Huizen, The Netherlands: Johannes van Kessel Publishing </p>&l

65、t;p>　　[10] Triola, Mario (2001). Elementary statistics (8 ed.). Boston: Addison-Wesley. p. 388. ISBN 0-201-61477-4. </p><p>　　American Journal of Mathematics, 2007, 126(5): 2387-2425</p&

66、gt;<p><b>　　統(tǒng)計(jì)假設(shè)檢驗(yàn)</b></p><p>　　Adriana Albu，Loredana Ungureanu</p><p>　　Politehnica University Timisoara, adrianaa@aut.utt.ro</p><p>　　Politehnica University Tim

67、isoara, loredanau@aut.utt.ro</p><p>　　摘 要 在這篇文章中，我們給出統(tǒng)計(jì)假設(shè)檢驗(yàn)的貝葉斯檢驗(yàn)，介紹了檢驗(yàn)理論和其過(guò)程。提及了假設(shè)檢驗(yàn)在現(xiàn)實(shí)世界的一些應(yīng)用和重要性，以及成功的檢驗(yàn)的注意事項(xiàng)。</p><p>　　關(guān)鍵詞 貝葉斯假設(shè)檢驗(yàn);貝葉斯推理;顯著性檢驗(yàn)</p><p><b>　　引言</b>&

68、lt;/p><p>　　統(tǒng)計(jì)假設(shè)檢驗(yàn)是一種利用數(shù)據(jù)做決策的方法，無(wú)論是在有控制的實(shí)驗(yàn)還是在沒有控制的觀察性研究中都有實(shí)用。在統(tǒng)計(jì)學(xué)中，如果一個(gè)結(jié)果不可能根據(jù)預(yù)先確定的閾值的概率，顯著性水平，單獨(dú)的發(fā)生，那么就說(shuō)這個(gè)結(jié)果有統(tǒng)計(jì)學(xué)意義。那句“有意義的測(cè)試”是由羅納德·費(fèi)希爾所說(shuō)的：“這種關(guān)鍵測(cè)試可能被稱為有意義的測(cè)試，當(dāng)這種測(cè)試是可接受的，并且我們可以發(fā)現(xiàn)另一個(gè)例子和第一個(gè)有顯著性的不同。</p>

69、<p>　　假設(shè)檢驗(yàn)有時(shí)也被稱為驗(yàn)證性數(shù)據(jù)分析 ，它與探索性數(shù)據(jù)分析相對(duì)而言。在頻率的概率中，這些決定幾乎總是用零假設(shè)檢驗(yàn)。</p><p>　　有些測(cè)試回答了這個(gè)問(wèn)題 ，聲稱零假設(shè)是正確的，它是一個(gè)觀測(cè)一個(gè)測(cè)試統(tǒng)計(jì)價(jià)值至少是一個(gè)是否確實(shí)被觀測(cè)到的價(jià)值的概率。更普遍的，他們?cè)谶M(jìn)行實(shí)驗(yàn)之前對(duì)問(wèn)題提出一個(gè)結(jié)論再根據(jù)實(shí)驗(yàn)的結(jié)果和一定的概率判斷所推測(cè)的結(jié)論是否正確。假設(shè)檢驗(yàn)的用途之一就是去決定實(shí)驗(yàn)的結(jié)果是否有

70、足夠得信息去懷疑傳統(tǒng)的智慧。</p><p>　　統(tǒng)計(jì)假設(shè)檢驗(yàn)時(shí)概率統(tǒng)計(jì)涉及的關(guān)鍵技術(shù)，假設(shè)檢驗(yàn)的貝葉斯方法是立足于拒絕后驗(yàn)概率的假設(shè)。其他的方法，通過(guò)決策理論和最優(yōu)決策達(dá)到通過(guò)數(shù)據(jù)分析得出結(jié)論的目的。其他地區(qū)的假設(shè)檢驗(yàn)的關(guān)鍵是贊成替代假說(shuō)拒絕零假設(shè)的所有結(jié)果形成集合，通常有字母表示C臨界域。</p><p><b>　　介紹</b></p><p

71、>　　當(dāng)一個(gè)樣本正在同來(lái)自假設(shè)的人口對(duì)比時(shí)，單個(gè)樣本測(cè)試是可取的，人口的特征通過(guò)理論可知或通過(guò)人口能夠被計(jì)算。</p><p>　　兩個(gè)樣本測(cè)試用于比較兩個(gè)樣本，通?？茖W(xué)的控制實(shí)驗(yàn)實(shí)驗(yàn)組和對(duì)照組樣品。當(dāng)不可能控制重要變量時(shí)，配對(duì)測(cè)試適用于比較兩個(gè)樣本。而不是比較兩套，樣本成員進(jìn)行配對(duì)以至于成員之間的不同變成樣本。通常情況下成員之間的差異相比為零。</p><p>　　常態(tài)和已知標(biāo)準(zhǔn)差

72、的條件下比較適合應(yīng)用Z-測(cè)試。</p><p>　　T-檢驗(yàn)是適用于比較寬松的條件 下（較少假定）的手段。</p><p>　　類似的測(cè)試手段（50%的比例）的測(cè)試。卡方檢驗(yàn)，適用相同的計(jì)算和不同的應(yīng)用程序相同的概率分布：</p><p>　　卡方檢驗(yàn)用于檢驗(yàn)正常人群中是否有一個(gè)指定的方差，零假設(shè)就是這個(gè)方差。</p><p>　　卡方獨(dú)立性

73、測(cè)試用于決定是否兩個(gè)變量關(guān)聯(lián)或者是獨(dú)立的。零假設(shè)變量是獨(dú)立的。在計(jì)算中使用觀察的數(shù)據(jù)預(yù)計(jì)事件的發(fā)生頻率。</p><p>　　卡方檢驗(yàn)用來(lái)確定適合數(shù)據(jù)的曲線充足。零假設(shè)是，曲線擬合是足夠的。確定曲線形狀，以盡量減少均方誤差這是常見的。所以它是適當(dāng)?shù)暮玫臄M合計(jì)算方差的方法。</p><p>　　F檢驗(yàn)（方差分析）是常用的，在決定是否按類別的數(shù)據(jù)分組是有意義的。如果左手中的一類考試成績(jī)的差異是

74、比全班方差小，那么它可能是有用的，零假設(shè)兩個(gè)差異是相同的，因此，擬合的分組是沒有意義的。</p><p><b>　　測(cè)試的過(guò)程</b></p><p>　　在統(tǒng)計(jì)學(xué)中統(tǒng)計(jì)假設(shè)檢驗(yàn)做了一個(gè)基礎(chǔ)的角色。通常的推理思路是下面這樣的：</p><p>　　有一個(gè)初步的研究假說(shuō)，總體情況是未知的。</p><p>　　第一步是去

75、聲明相關(guān)的零假設(shè)和被擇假設(shè)。具體來(lái)說(shuō)，零假設(shè)允許附加屬性：應(yīng)該選擇這樣一種方式，它可以讓我們得出結(jié)論，是否可以被接受的替代假說(shuō)或保持未定，因?yàn)樗窃跍y(cè)試之前定下的。</p><p>　　第二步是去考慮統(tǒng)計(jì)假說(shuō)關(guān)于正在做的測(cè)試的統(tǒng)計(jì)假設(shè)的制定，舉個(gè)例子，關(guān)于統(tǒng)計(jì)獨(dú)立性的假設(shè)或關(guān)于觀測(cè)值的分配形式的假設(shè)。具體的說(shuō)，這是同樣重要的因?yàn)闊o(wú)效的假設(shè)將意味著測(cè)試結(jié)果是無(wú)效的。</p><p>　　決定

76、哪個(gè)測(cè)試是適當(dāng)?shù)?，說(shuō)明有關(guān)的檢驗(yàn)統(tǒng)計(jì)量T。</p><p>　　從零假設(shè)的假設(shè)下得出的檢驗(yàn)統(tǒng)計(jì)量的分布，在標(biāo)準(zhǔn)情況下，這將是一個(gè)眾所周知的結(jié)果。檢驗(yàn)統(tǒng)計(jì)量可以按照學(xué)生的t分布或正態(tài)分布。</p><p>　　選擇一個(gè)顯著性水平（a），將拒絕零假設(shè)的概率置于他之下，一般選擇5%和1%。</p><p>　　零假設(shè)下統(tǒng)計(jì)檢驗(yàn)的分布把T的可能值分布到零假設(shè)被拒絕的區(qū)域，這

77、就是關(guān)鍵域，他不是T的可能值，臨界域的概率是a.</p><p>　　觀測(cè)計(jì)算檢驗(yàn)統(tǒng)計(jì)量T的觀測(cè)值t。</p><p>　　決定是否拒絕零假設(shè)接受被擇假設(shè)。如果觀測(cè)時(shí)值落在了臨界域則拒絕零假設(shè)HO，接受或拒絕其他的假設(shè)。</p><p><b>　　應(yīng)用和重要性</b></p><p>　　假設(shè)檢驗(yàn)對(duì)于分析大部分的收集的

78、數(shù)據(jù)是有幫助的。這同樣是真正可以證明的結(jié)論，即使沒有科學(xué)理論存在的假設(shè)檢驗(yàn)。 </p><p>　　假設(shè)檢驗(yàn)現(xiàn)實(shí)世界的應(yīng)用包括：</p><p>　　測(cè)試是否男性比女性更容易做惡夢(mèng)。</p><p><b>　　建立文件的著作權(quán)。</b></p><p>　　評(píng)估滿月對(duì)行為的影響。</p><p>

79、;　　確定蝙蝠可以用回聲捕捉昆蟲的范圍。</p><p>　　確定是否醫(yī)院的地毯導(dǎo)致了更多的感染。</p><p>　　選擇戒煙的最佳手段。</p><p>　　檢查是否保險(xiǎn)杠貼紙反應(yīng)車主的行為。</p><p>　　測(cè)試筆跡分析師的索賠。</p><p>　　統(tǒng)計(jì)假設(shè)檢驗(yàn)在整個(gè)統(tǒng)計(jì)和統(tǒng)計(jì)推斷中起著重要的作用。舉個(gè)例子

80、，萊曼（1992）在關(guān)于奈曼和Pearson（1933）的一篇基礎(chǔ)文件的審查中說(shuō)：“不過(guò)，盡管他們的缺點(diǎn)，一個(gè)新的典范在1993年的文件中形成，許多新的發(fā)展著利用它的框架繼續(xù)在統(tǒng)計(jì)的理論和實(shí)踐中發(fā)揮著中心作用，并可以期望在可預(yù)見的將來(lái)也會(huì)這樣做。</p><p>　　顯著性檢驗(yàn)時(shí)一直青睞的統(tǒng)計(jì)工具，在一些實(shí)驗(yàn)性的社會(huì)學(xué)（超過(guò)90%，在20世紀(jì)90年代初，在應(yīng)用心里學(xué)雜志上的文章）等領(lǐng)域有利于參數(shù)的統(tǒng)計(jì)，編輯經(jīng)常考

81、慮出版基于實(shí)驗(yàn)的統(tǒng)計(jì)結(jié)果的科學(xué)結(jié)論出版的意義。</p><p><b>　　注意事項(xiàng)</b></p><p>　　成功的假設(shè)檢驗(yàn)是與概率和第一類錯(cuò)誤率相聯(lián)系的，結(jié)論可能是錯(cuò)誤的。檢驗(yàn)的結(jié)論是基于它所使用的樣本的，樣本不同結(jié)果可能不同，這個(gè)設(shè)計(jì)是實(shí)驗(yàn)的核心，已觀測(cè)到的一些意想不到的效果包括：</p><p>　　聰明的漢斯效果。一匹馬似乎是能夠做

82、簡(jiǎn)單的算術(shù)題。</p><p>　　霍索恩效果。產(chǎn)業(yè)工人更多更好的照明生產(chǎn)，最糟糕的生產(chǎn)。</p><p>　　安慰劑效應(yīng)。沒有醫(yī)療活性成分的藥片是非常有效的。</p><p>　　一個(gè)誤導(dǎo)性的數(shù)據(jù)統(tǒng)計(jì)分析產(chǎn)生誤導(dǎo)性的結(jié)論。數(shù)據(jù)質(zhì)量問(wèn)題，可以更加微妙。例如，在預(yù)測(cè)中，有沒有協(xié)議的預(yù)報(bào)準(zhǔn)確率的措施。在一個(gè)共識(shí)測(cè)量情況下，沒有基于測(cè)量的決定是毫無(wú)爭(zhēng)議的。</p&g

83、t;<p>　　這本書如何用統(tǒng)計(jì)說(shuō)謊是曾經(jīng)出版最流行的一本關(guān)于統(tǒng)計(jì)的書，它沒有過(guò)多的考慮假設(shè)檢驗(yàn)，但它的注意事項(xiàng)是適用的，包括：一些論斷是在樣本太小不能說(shuō)服問(wèn)題的情況下做出的，如果報(bào)告沒提到樣本大小是值得懷疑的。</p><p>　　假設(shè)檢驗(yàn)充當(dāng)統(tǒng)計(jì)結(jié)論的過(guò)濾器，只有符合概率閥值的結(jié)果是能發(fā)布的。經(jīng)濟(jì)還充當(dāng)出版物的過(guò)濾器，只有那些有利于作者和資金來(lái)源的結(jié)果可能會(huì)被提交出版。出版物過(guò)濾器的影響被稱為出

84、版偏見。一個(gè)相關(guān)的問(wèn)題是多次測(cè)試（有時(shí)與數(shù)據(jù)挖掘相聯(lián)系），各種測(cè)試各種產(chǎn)生的影響被應(yīng)用到一個(gè)單獨(dú)的數(shù)據(jù)集，僅僅那些有意義的結(jié)果能夠被報(bào)道。</p><p>　　基于假設(shè)檢驗(yàn)結(jié)果的關(guān)鍵決策更著重細(xì)節(jié)的觀察而不僅僅是結(jié)論本身。在物理科</p><p>　　學(xué)中，大部分結(jié)果只有被獨(dú)立證實(shí)時(shí)才能被完全接受。關(guān)于統(tǒng)計(jì)通常的建議是，數(shù)字不會(huì)說(shuō)謊，但騙子會(huì)數(shù)字。</p><p>

85、<b>　　爭(zhēng)議</b></p><p>　　由于顯著性檢驗(yàn)，首先由著名的受人尊敬的統(tǒng)計(jì)人員的反對(duì)而流行起來(lái)，批評(píng)和反駁量填補(bǔ)了少量的用于學(xué)術(shù)辯論的語(yǔ)言書籍。許多評(píng)論出版了超過(guò)40年。爭(zhēng)論的火焰已經(jīng)在實(shí)驗(yàn)心理學(xué)領(lǐng)域燃燒到最旺。尼克森在2000調(diào)查了這個(gè)問(wèn)題。他包括300篇引用，報(bào)告了20篇評(píng)論，和差不多一樣多得建議，替代和補(bǔ)充。以下的部分極大的凝聚了尼克森的討論，省略了許多問(wèn)題。</p

86、><p><b>　　爭(zhēng)論的結(jié)果</b></p><p>　　爭(zhēng)議已經(jīng)產(chǎn)生了一些成果。通過(guò)審查，美國(guó)心理協(xié)會(huì)已經(jīng)加強(qiáng)了對(duì)統(tǒng)計(jì)報(bào)表的要求，醫(yī)學(xué)雜志出版商已經(jīng)意識(shí)到有義務(wù)出版一些沒有統(tǒng)計(jì)學(xué)意義的結(jié)果以打擊發(fā)表偏倚。期刊（雜志的文章支持零假設(shè)）已經(jīng)建立專門公布這樣的結(jié)果。教材增加了一些注意事項(xiàng)和增加覆蓋必要的工具來(lái)估算產(chǎn)生所需樣品的大小。主要組織沒有放棄使用顯著性檢驗(yàn)，雖然 他

87、們已經(jīng)討論了這樣做。</p><p><b>　　參考文獻(xiàn)</b></p><p>　　[1] R. A. Fisher (1925). Statistical Methods for Research Workers, Edinburgh: Oliver and Boyd, 1925, p.43. </p><p>　　[2] Cramer,

88、 Duncan; Dennis Howitt (2004). The Sage Dictionary of Statistics. p. 76. ISBN 0-7619-4138-X.  </p><p>　　[3] Schervish, M (1996) Theory of Statistics, p. 218. Springer ISBN 0-387-94546-6</p&

89、gt;<p>　　[4] Kaye, David H.; Freedman, David A. (2011). "Reference Guide on Statistics". Reference manual on scientific evidence (3rd ed.). Eagan, MN Washington, D.C: West National Academies Press. p.

90、60;259. ISBN 978-0-309-21421-6.</p><p>　　[5] C. S. Peirce (August 1878). "Illustrations of the Logic of Science VI: Deduction, Induction, and Hypothesis". Popular Science Monthly 13. </p>

91、;<p>　　[6] Fisher, Sir Ronald A. (1956) [1935]. "Mathematics of a Lady Tasting Tea". In James Roy Newman. The World of Mathematics, volume 3 [Design of Experiments]. Courier Dover Publications. ISBN 

92、;978-0-486-41151-4.</p><p>　　[7] Box, Joan Fisher (1978). R.A. Fisher, The Life of a Scientist. New York: Wiley. p. 134. ISBN 0-471-09300-9</p><p>　　[8] Lehmann, E.L.; Romano, Joseph P

93、. (2005). Testing Statistical Hypotheses (3E ed.). New York: Springer. ISBN 0-387-98864-5.</p><p>　　[9] Adèr,J.H. (2008). Chapter 12: Modelling. In H.J. Adèr & G.J. Mellenbergh (Eds.) (wit

94、h contributions by D.J. Hand), Advising on Research Methods: A consultant's companion (pp. 183–209). Huizen, The Netherlands: Johannes van Kessel Publishing </p><p>　　[10] Triola, Mario (2001). Elementar