# Survey Response Problem

#### Blogma

##### New Member
I hope someone will help me with this. The problem I have to work out is a scenario involving a survey.

Let's say a survey is administered to 985 people and 38 respond to a question which states "how many people found out about your party". THe answers range between 0 and 200.

1. Is it possible to estimate what the average response to this question would be if all 985 people answered? 2. After that, is it possible to estimate what the sum of all the responses would be?

:wave:

#### divinityofnumber

##### New Member
I would hesitate to make judgments about the entire 985 people based on the 38 who responded.

Whenever a survey is voluntary it suffers greatly from the voluntary response bias.

http://en.wikipedia.org/wiki/Biased_sample

In short....

1. Is it possible to estimate what the average response to this question would be if all 985 people answered?

No. The problem is that all 985 did not answer. And the 38 who did are a biased portion. They are people who are likely to volunteer.

The randomness condition is not met. The 38 are not an unbiased and representative sample of the 985.

2. After that, is it possible to estimate what the sum of all the responses would be?

See Above.

I can make no inferences about those whom you sampled or attempted to sample. However, as an example of the type of biases that I mention above; suppose that only those who had a high turn out for their party were willing to submit the survey. Maybe those with low turn out were embarrassed or angry and may have not responded for those reasons. It could be that those who responded all have certain things in common, and in that case they ARE NOT representative of the whole 985. So not only can you not draw any scientific conclusions about the 985, but you cannot even use the data from the 38 respondents for anything other than complete entertainment, as it is not valid.

Last edited: