Eliminate "Interference" in random experiments
January 12, 2022 09:20 Source: "Chinese Social Sciences" January 12, 2022 Total 2328 Author: Wang Youran/Compilation

  bet365 Play online games According to the official website of the University of Texas University Austin, January 5th,The school's McCums Business School Financial Clinical Assistant Professor David Puelz and others developed a new method,To solve the problem of "interference" in business experiments。

Current,Many companies collect data through random control experiments,To make a strategy。For example,The social media platform can understand whether the changes in news push can be strengthened by random experiments to strengthen user participation,Bet365 app download Banks can understand whether the new preferential policies can increase the savings rate through random experiments。but,"Interference" phenomenon will cause experiments that are difficult to display clear causality。For random control experiments,The ideal state is: While changing the behavior of the experimental group,Keep the behavior of groups outside the experimental group as much as possible。The so -called "interference" refers to,During the experiment,,Experimental group (subject to the object of the experimental variable processing) has also changed。Especially when there is a certain connection between the experimental group and the control group,"Interference"。For example,Members of the experimental group and control group members bet365 Play online games are friends on social networks or reality、In the same trading network、From the same family or live in the same community。"If you don't consider the effects of 'interference",The causal conclusion obtained by the experiment will be serious errors。But if you incorporate ‘interference’ into statistical analysis in a proper way,It can bring help。"Polz said。

Researchers have designed a new method,Used to directly test "interference",For example, the overflow effect and the same group effect in social networks。Social media platform can use this method to test the news push change not only affects users,It also affects users of users; policy makers can use this method to test whether wearing masks can help control epidemic dissemination;。This new study focuses on "Space bet365 best casino games Interference",Data comes from another large -scale police experiment conducted by another research team in 2015。The research team at the time inspected the criminal activities of all blocks in Madalin,Identify the "Hot blocks" with frequent crimes,and tested whether the police patrol caused an impact on the neighboring neighborhood in the "hot block","Increasing police patrol" is to reduce criminal activities as a whole,Still transfer the criminal activity from the "hot block" to other blocks。

Projects for 2015 research,Polz and others made a chart,Among them, the interference structure information of various blocks in Madelelin and all the combinations corresponding to the police patrol are。They used this chart to create an algorithm,Divide data into two subsets: bet365 best casino games non -experimental group but may have an overflow group associated with the experimental group;,It is also a pure control group that has nothing to do with the experimental group。On the chart,Researchers performed a random test of Fisher,Where to determine where the overflowing effect appears。

This research verified the conclusion of 2015 research: strengthening police patrol in a block can reduce criminal activities in neighboring neighborhoods。Polz said,This new method is applicable to any research on the phenomenon of "interference" in the mixed network,Internal experiments including enterprises to improve operations。

 (Wang Youran/Compilation)

Editor in charge: Chen Jing
QR code icon 2.jpg
Key recommendation
The latest article
Graphics
bet365 live casino games
Video

Friendship link: Official website of the Chinese Academy of Social Sciences |

bet365 live casino games Website filing number: Jinggong.com Anmi 11010502030146 Ministry of Industry and Information Technology:

All rights reserved by China Social Sciences Magazine shall not be reprinted and used without permission

General Editor Email: zzszbj@126.com This website contact information: 010-85886809 Address: 11-12, Building 1, Building 1, No. 15, Guanghua Road, Chaoyang District, Beijing: 100026