Tips for Conducting a Community Survey

As geographers, we are often required to conduct surveys in communities to gauge their views on a particular issue such as poverty, climate change or other topics. Surveys are a great way to gain insight into the views of a population and designing them was already covered in a previous post. In this post I'd like to focus on tips for conducting community surveys.

Know your Research Area:

Conduct background research on the community before going in. Use GIS software to get the layout of the community. Learn as much history and local customs that you can. Know the demographic aspects of your community. Ethnic composition, religion even the political views of the majority population can be important to your survey. So be thorough in your background research. It can go a long way in influencing your success.

Develop a Clear, Systematic Plan for Sampling: 

While in theory it may be great to use stratified sampling where you select community members to do surveys, more often than not you will find a large amount of people who will simply refuse to answer your survey, which can destroy your pre-defined sampling strategy. Additionally, it is often difficult to get information on the population of a community so it may not be possible to use a stratified sampling approach in the field as determining a representative sample will be impossible. Working from the national averages which may be available is not a good idea on the community scale as age, gender and other demographic distributions may not be homogenous throughout the country.
Quite often, we are forced to resort to simply, "taking what we can get," when it comes to sampling due to the above factors. Still there are methods to be systematic about it. Below are two of the methods I use for sampling.

Method 1:

Split the community into several, roughly equal parts and select an equal amount of participants from each of them. This is a variation on the cluster sampling technique ("Cluster Sampling" 2016) where instead of only selecting a subset of the clusters for use in sampling, one uses all.

For Example:

During a recent project I did on communities surrounding the Camden Airfield in Couva, Trinidad, I split the communities into four parts (Balmain, Central Park, Exchange and Mcbean) based on their sizes and bearing from the airfield (Fig. 1).

Fig. 1:- Map of the Survey Area Around Camden.

Now bear in mind that a lot of the times your population is unknown so therefore in order to get an amount for each cluster one should use a, "take what you get approach," in the first cluster and then use that a benchmark for the others.

Method 2:

Method two is almost identical to method one, except instead of aiming for a minimum number of participants in each cluster, you use a time limit for sampling instead. For example, you decide that you spend 6 hours maximum in each cluster. While less systematic than the previous method, it may be better to use this method if you are pressed for time (or if the sun happens to be blazing hot that day).

Try to Avoid a Skewed Sample:

Sample skewing is likely to occur if you simply take a random approach. You run the risk of skewing the sample in favour of some parameter eg. age if you collect too much names from a specific age category or gender if you collect too much from one particular gender group. You should try to keep your sample representative.
To avoid this, keep a running check on your surveys. Try to see if you collected too many or too little in a category. Compare your dataset to a previous, recent dataset for that community like a census or previous study to see how it stacks up. It should reveal whether your sample is skewed or not and allow you to take corrective action by being more discerning in your participant selection.
There is a caveat to this though in that it assumes you know the distribution of demographic characteristics. As mentioned before, this may not be the case for that community and you may be, "working blind." Attempting to equalize the sample by taking equal numbers in each category may not always be the best approach as the community may be naturally skewed and doing this may actually make your sample less representative of the community. Again, using a national census for comparison may not be the best idea either as the distribution may not be homogeneous throughout the country may not hold at the community level.
The best approach in this case if you can't source comparison data is to simply accept the possibility that your sample may be skewed. You have no way of determining how accurate it is or not. Its one of the truisms of doing research.

Chart a Designated Route:

Use maps or GIS software to plot a route for you to follow throughout the community. This will help you make sure you cover it in its entirety and prevent you from getting lost. It might be a good idea to mark out points where you have passed as you go along to avoid back-tracking. You might want to keep a sketch map of your route on you, or your Google Maps open on your phone if you have access to data or wi-fi.

Safety First:

BE SAFE! Before venturing out into the field make sure you locate the nearest police station, hospital and fire station. Have a fully-charged cellphone with sufficient credit on you at all times should you need it. Know emergency contact numbers by heart (should you lose your phone). Make sure that at least one other person knows that you are out in the field and is able to maintain contact with you at all times. If possible, get a friend to accompany you, two heads are always better than one (plus you have someone to help carry equipment). Maintain situational awareness at all times and be wary of distractions and anything out of the ordinary. Don't like something about the area or feel threatened by someone? LEAVE!!! Research is not worth compromising your personal safety!

Clearly Identify Yourself:

Make sure to carry proper identification on you at all times. This may seem obvious but is often overlooked. Proper identification would give any respondent you meet some sense of security and will make them distrust you less (not trust you more as people are typically distrustful of strangers randomly approaching them with clipboards). If you are conducting research as part of a university course, then carry your student ID. National ID such as driver's permit may also be appropriate in lieu of this.
Identification will also help you out should you be stopped by law enforcement and questioned for skulking about and area in a suspicious manner. You will be able to almost instantly legitimize your presence there.

Be Polite:

One would be surprised as to how far being nice to people will get you in life. Be polite to your participants. Greet them first, do not launch into the purpose of your research immediately. This will make you seem like less of an impersonal robot. Remember these are people you are dealing with, not lab mice. Aside from being the nice thing to do, being polite to people can pay off as it may actually sway them into responding to your survey.

Respect Local Cultural Norms and Customs:

This tip is more applicable if operating in a foreign country. Respecting cultural norms is not only good manners, but it stands to reason that if you offend your participants in any way you may not get the co-operation you need. Diplomacy is an asset to any researcher. That being said, there are limits to what you do to accommodate local customs especially if they contradict your ethical mandates as a researcher . For example, some tribes in South America require you to drink alcohol as part of any greeting. Ethically, you should not be drinking while conducting research so you may not want to respect that particular local custom.

Conclusion:

To summarize, community surveys can be complicated affairs. There is no one standard formula for conducting a community survey and every situation is unique and presents a different set of challenges. You cannot prepare for every scenario and you may have to rely on your intuition in the field to make split-second decisions demanded by a situation you did not anticipate.

Thanks for Reading,

Barindra

References

"Cluster Sampling". 2016. Wikipedia. https://en.wikipedia.org/wiki/Cluster_sampling.