MHB Statistics and customer ratings

  • Thread starter Thread starter find_the_fun
  • Start date Start date
  • Tags Tags
    Statistics
AI Thread Summary
Surveys that only target willing participants often suffer from selection bias, as those who choose to respond may not represent the overall population. This bias can lead to skewed customer ratings on platforms like Amazon, where only individuals with strong opinions are likely to leave feedback. To mitigate this issue, it's suggested that companies report the number of participants versus non-participants in their surveys. Offering incentives for survey completion could also help achieve a more representative sample. Awareness of selection bias is crucial for accurate data interpretation and customer insights.
find_the_fun
Messages
147
Reaction score
0
When conducting a survey it is poor choice to only solicit the survey to people who feel like taking it. This is a bad idea because a person who goes out of their way to do a survey may be different than the rest of the population in some respect thus their answer is biased. I know this is elementary, but what is the name of this concept?

Anyways my point: online retailers such as Amazon.com must have non-representative customer ratings since only people who feel like it (probably because they had a really good or really bad experience) choose x amount of stars. Is there anyway around this aside from giving people incentives like money if they fill out a survey?
 
Physics news on Phys.org
Re: Statistics and customer raitings

find_the_fun said:
When conducting a survey it is poor choice to only solicit the survey to people who feel like taking it. This is a bad idea because a person who goes out of their way to do a survey may be different than the rest of the population in some respect thus their answer is biased. I know this is elementary, but what is the name of this concept?

It's called bias, or more specifically selection or selection bias.

Anyways my point: online retailers such as Amazon.com must have non-representative customer ratings since only people who feel like it (probably because they had a really good or really bad experience) choose x amount of stars. Is there anyway around this aside from giving people incentives like money if they fill out a survey?

I believe the important thing is to be aware of it, and preferably to include the number of participants that took the survey versus the ones that didn't in a report.
 

Similar threads

Back
Top