Perfect vote
Often on various sites, users are asked to rate (vote for) one or another object - anecdote, quote, post, author, person, etc. At the same time, unfortunately, one or more errors are almost always made.
I will only talk about when users are asked to rate the object positively or negatively. Votes with a choice of one of several answer options are not considered here.
')
Often it is proposed to rate the object on a scale, for example, from 1 to 5.
This is very inconvenient, because it is simply not clear what grade to put - 3, 4 or 5? A person is very difficult to choose. Teachers, for example, have been specially trained for a long time on how to give marks to students, there are strict formal criteria on this score, etc.
The user does not have all this. He can only say whether he likes it or not. In practice, in most cases, the user chooses either the maximum (if you like) or the minimum (if you do not like) assessment.
There should be only two possible answers - âplusâ and âminusâ.
It often happens that voting is done by radio switches, which work like ordinary buttons, that is, they send a voice immediately after clicking on them. This is very bad (as in all cases when interface elements are used for other purposes). Buttons should be only buttons, not switches or drop-down lists.
Radio switches should always be accompanied by a button that confirms the selection. But, since we found out that there should be only two options, such an interface would be redundantly complex. The correct approach: two buttons for voting for and against.
There is another problem: buttons for selecting options are often too close together. Some users (their number can sometimes be significant) just miss.
For voting, there should be two buttons (âgoodâ and âbadâ) that are not next to each other.
If the user mistakenly pressed the wrong button; hurried with the answer; changed his mind - he should be able to re-vote (unless, of course, the voting results have not yet been summed up). Many sites do not provide such an opportunity.
The effect of re-voting should not differ from the effect of voting - this is also a common error: for example, on âHabrahabrâ when voting for karma, the effect of voting âforâ is higher than from voting âagainstâ and then again âforâ.
Sometimes it is possible to vote for an object only for a limited time, after which it is summed up. This is never justified for permanent objects, such as articles on the site.
Voting for permanent facilities should be open as long as the facility exists.
Not so rarely in polls there is no protection against cheating. It is really not easy to take any measures: for example, the restriction âone voice from one IP addressâ does not take into account users of networks with one external address; permission to vote only to registered users is not always acceptable and not insured against creating virtuals. At the same time, absolutely no defense is possible.
Very often, the site is allowed to vote one and all. The results of such errors are tragic. Here you can remember and "schoolchildren" and "trolls." If we are interested in their opinion - the qualification is not needed. If we are interested in the opinion of only adequate visitors - a few simple actions will allow us to allocate them quite effectively. The Turing test, after all, separates people from computers - something similar is appropriate in this case.
This is perhaps the most important mistake. Knowing the number of people who voted âforâ and the number of people who voted âagainstâ, it is necessary to somehow calculate the final grade (rating) of the object based on these data.
Most often there is a simple subtraction of the minuses from the pros. This is a very, very, very bad decision. Suppose we have two articles; the first was rated at 1000 pros and 500 minuses; the second is 200 pluses and only 2 minuses. Obviously, the second article is much better; however, the rating of the first will be 500, and the second - 198. Obviously, this approach to processing the results is unacceptable.
To obtain an adequate result, it is enough to divide the difference between the pros and cons of their sum (that is, the total number of votes). As a result, we obtain the estimate from â1 to +1. To avoid division by zero, in the absence of votes, the estimate can be taken as zero.
For new objects (for example, only two people voted for the article just published, and both are âforâ - the rating goes out to the maximum), a qualification should be provided: for example, until the total number of votes reaches a hundred, the rating is divided by 20.
An ideal vote is one in which the errors listed above are missing. There are exactly two voting buttons, they donât stand close, you can re-vote, voting is permanent, protected from markups, inadequate visitors are eliminated, the rating is calculated taking into account the total number of voters.
I will only talk about when users are asked to rate the object positively or negatively. Votes with a choice of one of several answer options are not considered here.
')
Typical mistakes
Rating Scale
Often it is proposed to rate the object on a scale, for example, from 1 to 5.
This is very inconvenient, because it is simply not clear what grade to put - 3, 4 or 5? A person is very difficult to choose. Teachers, for example, have been specially trained for a long time on how to give marks to students, there are strict formal criteria on this score, etc.
The user does not have all this. He can only say whether he likes it or not. In practice, in most cases, the user chooses either the maximum (if you like) or the minimum (if you do not like) assessment.
There should be only two possible answers - âplusâ and âminusâ.
Interface inconvenience
It often happens that voting is done by radio switches, which work like ordinary buttons, that is, they send a voice immediately after clicking on them. This is very bad (as in all cases when interface elements are used for other purposes). Buttons should be only buttons, not switches or drop-down lists.
Radio switches should always be accompanied by a button that confirms the selection. But, since we found out that there should be only two options, such an interface would be redundantly complex. The correct approach: two buttons for voting for and against.
There is another problem: buttons for selecting options are often too close together. Some users (their number can sometimes be significant) just miss.
For voting, there should be two buttons (âgoodâ and âbadâ) that are not next to each other.
Inability to re-vote
If the user mistakenly pressed the wrong button; hurried with the answer; changed his mind - he should be able to re-vote (unless, of course, the voting results have not yet been summed up). Many sites do not provide such an opportunity.
The effect of re-voting should not differ from the effect of voting - this is also a common error: for example, on âHabrahabrâ when voting for karma, the effect of voting âforâ is higher than from voting âagainstâ and then again âforâ.
Closing the vote
Sometimes it is possible to vote for an object only for a limited time, after which it is summed up. This is never justified for permanent objects, such as articles on the site.
Voting for permanent facilities should be open as long as the facility exists.
Cheat
Not so rarely in polls there is no protection against cheating. It is really not easy to take any measures: for example, the restriction âone voice from one IP addressâ does not take into account users of networks with one external address; permission to vote only to registered users is not always acceptable and not insured against creating virtuals. At the same time, absolutely no defense is possible.
Lack of electoral qualifications
Very often, the site is allowed to vote one and all. The results of such errors are tragic. Here you can remember and "schoolchildren" and "trolls." If we are interested in their opinion - the qualification is not needed. If we are interested in the opinion of only adequate visitors - a few simple actions will allow us to allocate them quite effectively. The Turing test, after all, separates people from computers - something similar is appropriate in this case.
Incorrect processing of results
This is perhaps the most important mistake. Knowing the number of people who voted âforâ and the number of people who voted âagainstâ, it is necessary to somehow calculate the final grade (rating) of the object based on these data.
Most often there is a simple subtraction of the minuses from the pros. This is a very, very, very bad decision. Suppose we have two articles; the first was rated at 1000 pros and 500 minuses; the second is 200 pluses and only 2 minuses. Obviously, the second article is much better; however, the rating of the first will be 500, and the second - 198. Obviously, this approach to processing the results is unacceptable.
To obtain an adequate result, it is enough to divide the difference between the pros and cons of their sum (that is, the total number of votes). As a result, we obtain the estimate from â1 to +1. To avoid division by zero, in the absence of votes, the estimate can be taken as zero.
For new objects (for example, only two people voted for the article just published, and both are âforâ - the rating goes out to the maximum), a qualification should be provided: for example, until the total number of votes reaches a hundred, the rating is divided by 20.
Perfect vote
An ideal vote is one in which the errors listed above are missing. There are exactly two voting buttons, they donât stand close, you can re-vote, voting is permanent, protected from markups, inadequate visitors are eliminated, the rating is calculated taking into account the total number of voters.
Source: https://habr.com/ru/post/30324/
All Articles