Response Rate for a Mail Survey as part of Barrier and Benefit Research

Marianne Sawyer Pakenham May 4, 2009 21:46 pm

I am conducting a survey to gain a better understanding as to whether the education campaigns run as part of my project on Community Weed management, and the onground contractor works has been effective. This will guide further strategies to adopt for community engagement tactics, with relevant barriers and benefits infomation.

I am uncertain at this stage how many surveys to mail out as this will depend on what we can expect as a response rate. How many survey responses do I need to have a meaningful outcome to the survey? If I have an estimated % figure, then I can work backwards from there and send out an appropraite amount.
Any comments appreciated !!

Marianne

Marianne Sawyer
Natural Resources Officer
Cardinia Shire Council
Australia

Sign in or Sign up to comment

Oldest First

Marianne Sawyer Pakenham May 6, 2009 22:28 pm

Thanks Chris, your reply is appreciated,we have finalised the survey now and seding out to 2220 residents in our project area. I will let this site know of the outcome. I'm still interested in your surveys so I will check out your site.

regards
Marianne

Marianne Sawyer
Natural Resources Officer
Cardinia Shire Council
Australia

Chris Weston Gordon May 6, 2009 19:16 pm

Hi Marianne;

A good idea is to include an online version of the survey through survey monkey, or similar web based programs. In our community consultation programs, we always include this and they are quite effective, and gets the proportion of the community that don't like paper surveys (like me) or don't have much time to go to the post office. Check out some of the surveys we do at the following link:

http://www.kmc.nsw.gov.au/www/html/2498-have-your-say.asp

Some of these surveys are for all of council services, as it was identified that there was a need for more community engagement. Also consider community members from non-english speaking backgrounds, especially if targeting your survey to a specific location. Many Council's now use id.com.au community profiles and atlas based on cencus information which can be very useful for targeting regional campaigns I have found.

I hope this helps.

Chris Weston
Ku-ring-gai Council
Australia

Marianne Sawyer Pakenham May 5, 2009 23:06 pm

Hi Shannon,
Thanks for yet another helpful response, I will chase up the recomended Dillman and Salant Method ...havnt heard of them!
regards
MARIANNE

Marianne Sawyer
Natural Resources Officer
Cardinia Shire Council
Australia

Marianne Sawyer Pakenham May 5, 2009 19:46 pm

Great information thanks Ingo

Marianne Sawyer
Natural Resources Officer
Cardinia Shire Council
Australia

Shannon Amberg West Lafayette May 5, 2009 11:11 am

Hi Marianne,

I agree with Ingo's comment above and would like to emphasize that you need to be focusing on how to get a representative sample for you survey. Most people, when doing surveys such as this, like to work with a sample of completed surveys that number anywhere between 500-1000 for regional projects. That should provide you with enough statistical power to see any relationships going on. However, the validity of the results really depends on how representative your sample is. Are the people who didn't respond systematically different in some way than those who did respond? Does your sample reflect the census trends for the region you've sent the survey out in? It's hard to estimate these things prior to getting results back, so I can understand your concern about how many surveys to send out.

What I like to do for general surveys is plan on a 20-30% response rate (this is what is typical in the U.S.) and, based on my need for having for example 500 returned surveys, calculate how many initial surveys I would need to send out to achieve this goal considering the anticipated response rate. In terms of the appropriate number of returned surveys for your population, that depends on the size of the population that you're focusing on, but it usually does not require more than 1000 if the sample is truly representative.

Check out the Dillman and Salant method for conducting mail surveys as well. They provide helpful insights into how to conduct the multi-wave procedure for mail surveys, how to calculate the sample size you need, and how to increase your response rate.

Hope this helps!

Shannon

Shannon Amberg
Assistant Professor, Human Dimensions of Natural Resources
Purdue University
United States

Ingo Bensch Madison May 5, 2009 10:32 am

Hello Marianne,

Response rate will depend on numerous factors, including how relevant the topic is for your sampled respondents, respondent recognition of the entity conducting the survey, length and format of the survey, pre-notice of the survey to the sampled population, follow-ups to non-respondents, and the benefit to the respondents for completing the survey (ability to provide useful input for something they care about, an incentive for completing the survey, etc.).

I have gotten response rates of 50+% in similar situations when we have implemented the survey by the book, but its also possible to end up much lower like 10%. When that happens, you have to worry about whether the responses you are getting are truly representative of the population you are surveying. You are best off sampling a reasonable number and putting your efforts into getting a good response rate (with a good survey design and follow ups to non-respondents).

Hope this helps.
Ingo

Ingo Bensch
Energy Center of Wisconsin
United States

...

Sign in or Sign up to comment

Fostering Behavior Change Minute

Advertisement