Sihai network

Online shopping, reading comments and falling into a hole? One minute to teach you to dump the hole

Double eleven is coming soon. E-commerce is already on the move. It's full of energy. It's estimated that the shopping cart is also full of goods as you love to fight. Then again, we should remind you that shopping, shopping, shopping, shopping, and comparison of goods are three. Don't be in the comment pit. We can teach you how to shop and see the comment pit.

More and more people are proficient in online shopping. They are used to a more 'transparent' shopping environment: whether to buy something or not, first look at the evaluation. How many evaluations are there? How many favorable comments are there? Once you draw your eyes, the quality of a product is in our mind. However, a recent study published in psychological science shows that the way ordinary consumers evaluate commodities may lead to irrational purchase decisions. The first author of the study was Derek Powell, a postdoctoral psychology student at Stanford University.

This study found that people pay too much attention to whether a product has enough evaluation number in online shopping, and how high the score of the product is relatively despised. In other words, under the same conditions, consumers prefer the "hot money" with considerable sales volume (although the number of evaluations is large, but the score may be low), rather than opening the purse for those goods that are "good but not good".

However, there is no direct relationship between the number of commodity evaluation and its quality. On the contrary, a low score commodity with a large number of evaluations is more stable than a low score commodity with only a small number of evaluations. Therefore, too much emphasis on the evaluation number may lead us to buy a bad product.

'popular' or 'good reviews', which one do consumers like?

Before the experiment, the researchers collected 15655439 evaluations of 356619 products from amazon.com, which mainly came from four categories: mobile phone, electrical appliances, kitchen, health and beauty.

By analyzing these real data, the researchers completed two tasks. First of all, they proved that the popularity of a product (i.e. the number of evaluations) has little to do with its own quality and consumer satisfaction (i.e. rating). Secondly, through Bayesian modeling, the researchers calculated what kind of decisions a rational consumer should make on the basis of different evaluation numbers and ratings. In this way, they provide a standard for comparison of different experimental results.

Later, the researchers recruited 138 adult subjects to evaluate which one they would prefer to buy between a series of matching products. One product has only a small number of evaluations (about 25), and its score fluctuates between 2.7 and 4.6 points; the other product has a large number of evaluations (about 150), and its score is higher in half than that of less evaluated products, and lower in the other half.

Subjects need to evaluate which one they prefer to buy among a series of matched products. Each product has two dimensions of 'evaluation number' and 'score'. For example, 145 items are evaluated for commodity h on the left, with a score of 2.7; 20 items are evaluated for commodity f on the right, with a score of 2.4.

What the researchers are concerned about is how the subjects choose between products with different evaluation numbers and scores. The results show that the subjects have a clear preference for the goods with a large number of evaluations.

When the scores of the two matched goods are high (4.6 points), the probability of choosing the goods with more evaluations is close to 95%. This result is more in line with our intuition. We are really more willing to buy those products that have good sales evaluation.

But interestingly, if the scores of both products are very low (2.7 *), the subjects still have a nearly 90% probability of choosing the one with the most evaluations.

*Note: the researchers analyzed the real data on Amazon and found that the average scores of different categories of goods fluctuated between 3.73 and 4.10, including 3.73 for mobile phones, 3.92 for electrical appliances, 4.09 for kitchenware, and 4.10 for health and beauty. Therefore, 2.7 is a relatively low score.

However, 'low scores from 150 evaluations' and' low scores from only 25 evaluations' are essentially different, although the evaluation is not high. According to the principle of statistics, the former can get a very stable 'bad evaluation' due to the large sample size, while the latter can only be caused by errors due to the small sample size. In other words, a bad product based on a large number of consumer evaluations must be very bad; a bad product based on a small number of consumer evaluations may be very bad or OK.

However, the worse situation is that compared with the low score goods with 2.7 points, if the number of evaluation is enough, even if the score is lower (2.4 points), the subjects will still have more than 40% probability to choose it. It can be seen that consumers' preference for hot money in online shopping is more obvious.

Why do people love 'pop money'?

In the whole experiment, the fit degree of Bayesian model to the behavior of the subjects is only 0.17, that is to say, their actual performance in the experiment is far from the answer of 'maximizing the benefits' obtained by the model.

This figure describes the situation that there are more than 150 goods to be evaluated. On the left is the rational decision-making mode based on Bayesian model, and on the right is the actual decision-making mode of the subjects in the task. The abscissa is the product score, and the ordinate is the probability that consumers choose this product. The figure shows the ideal and practical decision-making mode when the difference between this commodity and another commodity with less evaluation is 0.3, 0.1, 0, - 0.1 and - 0.3.

According to Powell, consumers are influenced by social cues that make them follow their peers' choices like sheep, and these cues are how many people evaluate a product.

Social learning theory shows that people will consciously or unconsciously help themselves to solve problems by observing and imitating other people's behaviors. Social learning can not only enable us to quickly obtain more reliable solutions in a short period of time, but also to a certain extent avoid the possible risks in the environment. This is a kind of ability of "not learning but being able", which also has profound evolutionary significance. Therefore, people's first instinct when making decisions is to follow others, which is a relatively safe option. And the anxiety of choosing the path that few people walk will also drive people to turn around and follow others.

The tendency of consumers' preference for evaluation number in online shopping is the result of social learning. They will regard popular goods (with a large number of evaluations) as "good" goods, and ignore the real use of evaluation numbers for consumption decision-making -- to measure the stability of a product's score: the more evaluation numbers the score is based on, the more stable the product's score will be.

Therefore, the next time you cut your hand, you should probably save yourself from the obsession with the number of evaluations, and consider both the number of evaluations and the score. A commodity with a large number of evaluations and a high score is naturally excellent; but a commodity with a large number of evaluations but a low score is certainly really bad.