| User | Post |
|
11:19
August 21, 2007
|
suzanne
Guest
| | | |
|
| |
|
|
I need to create a survey for a client, but i’m unsure about two things: how many completed surveys of existing customers do I need for validity, and therefore how many customers do I need to survey for that number.
|
|
|
12:43
August 21, 2007
|
jason
Member
| | | |
|
| posts 159 |
|
|
Suzanne-
Not quite clear on what you’re asking, but would love to help! Are you wondering about SurveyGizmo’s technical features (ie how many responses can the various account levels do) or more generally about what makes for a solid amount of data for a survey?
|
|
|
4:54
September 8, 2007
|
Joe
Guest
| | | |
|
| |
|
|
Suzanne,
there is some debate about the right answer for your questions in the scientific community. The rules of thumb that I go with are
30-50 completed surveys provide a starting point for a valid survey
The response rate of a survey that is send to the general population (i.e. a marketing survey or via a mail shot) is not expected to be much higher than 5%-10%.
Having said that, if you survey a more coherent group (i.e. employees in your department), response rates are normally higher (20%-30%) due to the stronger affiliation with the organisation.
Hope this helps.
|
|
|
|
|
Hi Sussane,
It sounds like you are asking how large your sample size (responses) should be.
To determine that you need to know how many total customers your client has. Then caluclate the sample size based on the condifence interval you want. Generally speaking, if your client has 100 customers, you need at least 80 to respond to get results with a 95% confidence level.
-Christian
|
|
|
6:30
September 10, 2007
|
donna d
Member
| | | |
|
| posts 93 |
|
|
Hi Suzanne -you might also try search the web for a sample size calculator - there are several and it makes the calculation very easy-especially once you know the number of clients you have.
good luck
Donna d
|
|
|
4:16
January 10, 2008
|
Dave
Guest
| | | |
|
| |
|
|
If you have 100 customers and send a survey to them all you are not sampling but taking a census. AS a result you cannot calculate confidence intervals as they only apply to samples. Rather you face the questions of VALIDITY — how well do those why reply compare to those who do not. Simple way to proceed is to compare some respondent characteristics — size, location, how long been a customer, etc. — to your entire customer base.
|
|
|
|
|
Good Post (and Point) Dave.
|
|
Login
Register
Forums
Add a New Topic
Reply to Post
Quote and Reply
Topic RSS 
