For psychological empirical research，Statistical method is an extremely important tool。But behind this kind of tool is behind it，Psychological researchers lack in -depth exploration of the "source and bottom layer" of this tool。This leads to the failure of the statistical method to return to its own value and significance。

　　bet365 best casino games

Psychology as an independent discipline，One of the important research methods support is the statistical method。For more than a hundred years，The development of statistical technology continues to promote the progress of psychological disciplines。In many psychological statistics technology，Statistics based on sample statistical quantity inference overall parameters，Occupy the core position in the application of psychological statistics，provides possible scientific principles for general scientific principles for observation conclusions based on limited samples。

In psychological research，The effectiveness of statistical inference is often measured by P value。for a long time，Statistical inference philosophy preset、The relationship between statistics based on P value and scientific inference，and the possible harm of the abuse of P value，I did not get enough attention from psychologists。The result of this is，"P = 0.05 as a subjective but practical level to judge the significance of the result" advocated by R. Fisher has become the iron law of judging the final value of a study。

Statistics Inference，It mainly includes assumption test、Forecast、Statistics regression and statistical modeling。Among them, the most widely used is Jerzy Bet365 lotto review NEYMAN、Egon S. Pearson et al. Put as a hypothetical test based on objective probability，​​This is usually the basis of other statistical inference theory。Assuming test is in the application of psychological research，There is an extremely important indicator -P value，is statistically significant。Many psychological empirical researchers try to evaluate and test hypothesis through comparative theoretical prediction and observation data，The statistical basis that this work relies on is the significant inspection of zero fakes based on the P value，It mainly includes Z test、T test、Formula analysis、Bayfoli inspection and non -parameter inspection, etc.。

Mirehl (P. Meehl) pointed out，"The research object of psychology is very complicated，It makes the constructive theory very difficult "。He compared the "soft" psychology and other "hard" science how to test their respective theories，Researchers who think that hard science often predict a specific value，They hope to accept zero fake。and psychological researchers usually set up zero fakes，No differences or variables are not related to different conditions，They hope that zero fake can be rejected。When this single point assumes is confirmed，Psychologists think that their theory has been confirmed，or at least to withstand the test。and the "rejection" of zero fake requires a tool，P value is such a tool。

　　P value meets challenges

In recent years，P value is constantly being criticized and questioned，These criticisms and questions are mainly aimed at the limitations of the inherent limit of testing and the misunderstanding of researchers。And these two problems also exist in the application of P value in the application of psychology。

On the one hand，Assuming test as a statistical method，There are problems in itself。J. Mulder and others summarized the three main aspects of the hypothetical test when reflecting on the bet365 best casino games statistical method of psychology，respectively: cannot be used for quantitative support to support zero fake evidence; P value often overestimates evidence that is opposite to zero fake -in view of the current discussion on psychological repetitive crisis; P value is based on extreme results in the sample space in the sample space The distribution calculation above，Its size depends on the sample that is drawn。In addition，Assume that the overall overall is needed according to the sample，And this is likely to make two types of errors，That is, when the zero fake is true, refuse to accept the zero fake，and receive zero fake when the zero vacation is fake。

On the other hand，Assume that the test has misuse problems in psychological research。For example，"P = 0.0002" and "p = 0.02" differences compared with the differences between "p = 0.049" and "p = 0.051"，The former is obviously greater than the latter。But in psychological research，The former is regarded as the level of "statistical significance"，The latter is divided into different prominent types for the boundary of "P = 0.05"，The corresponding research results are also divided into two categories。Next，Affected by "P & LT; 0.05", which is "statistically significant"，Psychological researchers will be more inclined to report the results of "statistical significant"，and worry about the results of the report "not significant"。Another，In psychological research，The statistical significance of the P value and the actual significance of the research are often blurred，People often equate "statistical significance" as "conclusions valuable"。The essence of P value is probability，It can only be used as a reasonable basis for the research that assumes whether it is reliable，instead of the research results itself。The significant statistics cannot measure the meaning and value of the research bet365 best casino games results itself。such as R. Hubbard and others，Scientific inference is a concept that is much more broad than statistics。But in the practical research of psychology，Significant statistics (ie, P & LT; 0.05) is equivalent to the significance of the research results itself。That is to say，In the study of psychology，The research results depend on the statistics itself，and less attention to the nature of the study itself and the compatibility of the statistical tools。

　　Reflection P value use

In recent years，American Statistical Society (ASA) reflects on P value and related issues。In the article "Age to Begame" P & LT; 0.05 "，The American Statistics Institute puts forward four suggestions on the use of P values: accepting uncertainty、Think about it、Open and humble。These four suggestions also have a reference meaning for the use and development of statistical methods in psychology。

　　First, accept uncertainty.We need to recognize it again，Statistical inference is never the same as scientific inference。Psychological research report should refuse to use "P & LT; 0.05" as an absolute criterion for judging the results of judgment research，"P & LT; 0.05", "Statistical is significant", "*" or other similar unsatisfactory expression。"P & LT; 0.05" can easily confuse statistical significance and actual language significance，The illusion of "statistical significant statistics" that is, "experimental assumption is true"。When the researchers are no longer superstitious "P & LT; 0.05"，Statistics can play its due role in psychology。The significance of psychological empirical research is not judged by whether a statistical measurement is less than a specific threshold，Instead, it is judged by the entire design of the researcher and research。

　　Second, think deeply.Before doing research，Psychological researchers should clearly grasp the status quo of the field of research，Determine on the purpose of research，Strict、Carefully design research plan，and on the basis of existing theory，Through literature review、Expert Opinions、Alonge review and other methods to analyze the feasibility and results of the study，To enhance the objectivity and effectiveness of research。B. Mcshanne and others believe，Thinking about research should consider "related prior evidence、Reasonable mechanism、Research design and data quality、cost and income of the real world、The novelty discovered，and other factors that are different from the field of research ... instead of prioritizing the P value or other statistical quantities "。

　　Third, open.Psychological research due to the particularity of its research objects，Even if the same research method is used and the various variables are controlled as much as possible to copy the same study，Its results are also difficult to the same。So，In psychological research, "one phenomenon，Multiple research methods "，This is more conducive to us to examine the current research from different angles。When the report is reported,，Researchers not only need to report p value，Also report statistical inspection、Effective amount and confidence interval, etc.;，It is also necessary to report unfavorable but important data; not only need to report all data analysis，Also requires all the results obtained。For the integrity of scientific publishing and research and communication，When deciding to present or highlight which results，You should not consider whether the P value exceeds a specific threshold。The comprehensive report of the research results will help enhance the study of the subsequent reference to the research results、Reports and News Objectiveness。

　　Fourth, modesty.Psychology as a subject that studies human nature，Its research results have received widespread attention from the media and the public。Researchers must clearly realize that different readers will understand the results of research results from different interest positions。So，When writing a paper, it should be used as a neutron person to carefully report the limitations of research and research results，to avoid misleading the public。Another，Statistics tool itself has limitations，p value、The confidence interval and other statistical measurements are uncertain。Psychological researchers cannot get a certain conclusion through a single study，should be modest，Encourage others to repeat their own research in different ways。

Overall，What we advocate is，Researchers do not need to stop using the P value index or any statistical method in psychological research，Instead, the limitations of the statistical method itself used before use and the place where you need special attention in psychological applications，Therefore, the "source and bottom layer" of psychological research is clearer。Only this，Psychological research can be more able to withstand the test of time and practice。

Of course，The development of everything is a spiral rising，instead of a straight line。Make change not only takes time，Also requires psychological society、Psychological semester journals and publishers edit、Psychological researcher、Collaborative cooperation with teachers' groups。Society of Psychological Society and the Branch Society should do a good job of "uploading and getting"; the editor of the psychological semester and publishers should strictly control the quality of the paper; psychological researchers should accept "accepting uncertainty when writing papers、Think about it、Open、Humor "; Teachers' groups should adjust their teaching bet365 Play online games content appropriately，Therefore, "We use‘ P & LT; 0.05 ’because the teacher teaches it，Teachers like this because we do this "cycle。Only the institutions are working together，Only can the statistical method return to this position in the field of psychology。

　(This article is the theoretical root of psychological discipline division crisis and the philosophical basic research of the subject split crisis of the National Social Science Foundation "(18bzx048) phase results)

(Author Unit: School of Psychology, Northwest Normal University)