There is no doubt that merchants spend money to buy reviews on the Internet. This is beyond doubt. In order to minimize this situation, many websites have made corresponding regulations and revised the settings. In fact, for some more formal and popular websites, the power of the business is limited, and most of the comments are uploaded by real users. The question is, can real netizens be reliable? Are they likely to be influenced by others?
HerdEffect
There is a term in the field of psychology called "HerdEffect", which is specifically used to describe this herd mentality of human beings. Anyone who has let go of the sheep knows that the flock has a habit, and that is to follow the head sheep. No matter how wide the front pit is, as long as a sheep jumps first, the sheep in the back will follow it. The herding effect has always been one of the hotspots of psychology research. The emergence of social networking sites provides an excellent experimental field, and researchers can use big data research methods to eliminate the impact of cases.
As we all know, it is unscientific to study a case alone in the field of psychology. In terms of the herd effect, it is impossible to study the evaluation of a certain product of netizens alone, because this product may indeed be of good quality. To avoid this problem, Dr. Sinan Aral of the Sloan Business School affiliated with the Massachusetts Institute of Technology and his colleagues decided to conduct a large-scale randomized controlled trial using natural science research methods. They worked with a social networking site and spent more than five months experimenting with all the netizens on this site and published the results of the experiment this year.
This is a comprehensive social networking site where users first upload articles and then other users make comments. The content of the comment can be scored, either like (thumbs up) or bad reviews (thumbs down). The number of praises minus the number of bad reviews is the final score for each comment. The researchers and the station reached an agreement to score 101,281 comments in the past five months. That is to say, every time a comment appears, a netizen comment is automatically generated by the machine. Of course, whether it is a praise or a bad review is randomly determined, and has nothing to do with the content of the post. There are also a number of comments that are not scored in advance as a control group.
It is worth mentioning that the total number of random and bad reviews is not the same, because, under normal circumstances, the website likes more than bad reviews. The researchers then set the proportion of random and bad reviews based on past conditions. As can be seen from this detail, the researchers are very careful in the selection of experimental materials, try not to affect the intuitive feeling of netizens.
The results show that the herding effect on social networking sites is quite obvious. The researchers collected a total of 308,515 netizens and found that the comments that were previously praised were more likely to be positively scored 32% higher than the control group, and the final score was 25% higher than the average. In contrast, comments that were previously rated as bad were not affected and were no different from the control.
“After a negative evaluation, there are often many people trying to correct this result, which ultimately leads to negative evaluations that have no effect on the total score of a comment. The positive evaluation does not have this situation, indicating that people are not The attitudes of positive and negative evaluations that are consistent with their own opinions are different, and they are more tolerant of the former."
Professor Bernie Hogan of the Oxford Internet Institute agrees with this judgment. He believes that positive evaluation is often an advertisement. People who make positive comments want more netizens to like something. Negative evaluation is a kind of personal emotional venting, which is why the average social networking site likes more than bad reviews.
In addition, the content of the comments also has an impact on the experimental results. News, social, political, and commercial news reviews are susceptible to the herd effect, while general news and economic news have less impact, presumably because the former is more subjective, influenced by personal opinions, and tends to go to extremes, while the latter Factual news is much more objective than the former and is not affected by personal factors.
The reason why Alar is doing this research is not to smash social networking sites or to sing the wisdom of the community. Instead, he has always believed that the Internet is a great invention and is very helpful in breaking the monopoly of information. But the results of this study show that group intelligence is flawed and decisions based on public opinion are likely to need to be revised. For example, he believes that social networking sites should be redesigned so that netizens can't see the opinions of other netizens before making comments. Only in this way can we prevent the impact of the herd effect on the evaluation results and avoid making wrong decisions.