A survey is a research technique that you will most likely come across if you plan on doing research in the social sciences. A survey is used when a researcher is interested in describing a population that, like almost always, is too large to observe directly. Surveys are also great for figuring what people think about a certain issue or subject. For example, surveys are used in order to find out a political candidate's chances of winning an election by simply asking respondents who they're likely to vote for.
Surveys can be done either verbally or in writing. When it's done writing, it's called a questionnaire. Questionnaires can be done while you are in the respondent's presence, or they can be done remotely (mail or internet). Writing a good questionnaire will determine the validity of your results. Below are a few guidelines for writing a good questionnaire that'll ensure your research will be well received.
Begin your questionnaire with some basic instructions for the respondent (the person taking your survey). This helps make your questionnaire less confusing for the respondent. It also prepares them for whatever comes after. Always provide clear instructions on how to answer questions no matter how obvious you think the answering format is. If you create a section of a survey with different answering format, explain to the respondent how they should answer. For example, if your question asks the respondent to rate from "1-5," make sure to explain what the significance of this scale is and how the respondent should use it when answering.
Make sure you write your survey using a readable, and common, font and font size. You wan to generally make sure that everything about your questionnaire is easy to read and understand. This means paying attention to details and thinking about your questionnaire from the perspective of someone who has never seen it.
Avoid communicating your opinion on whatever subject you want to ask about. The purpose of surveys is to understand how a group of people think about a certain subject. Inserting your opinion might influence them in some way, and therefore, could give you inaccurate results. Remain neutral in every possible way. Avoid words with negative, or positive, connotations.
Make sure to thank your respondent for taking the time to do your questionnaire! It's always good to follow the golden rule with respondents. Be respectful of them, their time, etc, and make sure you don't inconvenience them unnecessarily.
A close ended question is a question in which you give a limited amount of possible answers. For example, asking someone if they had a good day is close ended if the only two possible answers we provide are "yes" or "no." However, if we were to ask the respondent to describe how their day went, that would be an open ended question. It's a question in which the respondent is free to answer in which ever way he/she wants to, and is not limited a range of predetermined answers given by the questionnaire writer, such as "yes" or "no."
Close ended questions are only appropriate when possible answers don't vary much. It's important that you address all, or most, possible responses expected in your close ended questions else you will get results which don't accurately reflect the respondents opinions. For example, if you ask: "what is your opinion on Donald Trump winning the election?" and then you only provide possible answers of "I approve" and "I disapprove," that leaves a lot of possible responses out. It's possible for respondents to have mixed feelings on the matter, or simply be indifferent. For this reason, it would be wise to include a more exhaustive possible range of responses for this particular question in order to get the most accurate results.
Close ended questions are much easier to code than open ended questions due to the limited possible responses. An open ended question requires a lot more work, and interpretation, in order to code. This will be important when you start analyzing your results in order to come to conclusions.
It should never be unclear what the question is asking. All terms and phrases should be clear, and not left up to the interpretation of the respondent. Make sure that your questions are as clear and concise as possible. Avoid necessarily wordy and complex questions.
Specifically, avoid double barreled questions. For example, asking a respondent if he/she agrees with Donald Trump's or Hillary Clinton's policies. It's possible for a respondent to agree with neither. These people would not be able to answer your questionnaire and probably even decide to no longer continue. Also, avoid wording your questions in a negative context. For example, "Do you disagree with Donald Trump's policies?" should be worded as "Do you agree with Donald Trump's policies?" This is a much more direct and less confusing way to answer the question.
Lastly, avoid questions which encourage the respondent to answer a certain way. This may seem obvious, but often we can allow our own biases to slip through without noticing. For example, asking "Do you agree with Donald Trump's destructive international relations policies?" would surely create a bias in which many people might react to the assumption that the policies are destructive. Instead, avoid influencing the respondent by simply removing the word "destructive" and keeping the question as neutral as possible.
Believe it or not, the order in which you ask questions will often significantly alter the results. It could even cause respondents to simply refuse filling out your questionnaire. It's important to place "heavy" questions, which may require the respondent to do a lot of thinking, are not at the beginning of your survey. Instead, place these questions near the middle, or the end, of your survey. Also, try and "pull in" your respondent by putting in some engaging questions near the front. People love to talk, and think, about themselves, so asking some easy personal questions at the beginning might do the job.
If you plan on writing a few questions with similar answering formats, you might also decide to split up some of your questions into a "matrix." This is a good idea if you want to save on some space and at the same time make it easier for the respondent by giving him a quick way to move through similar questions. If doing so, make sure that all questions within the matrix are related to each other. That is, if you're asking the respondent about his thoughts on Donald Trump, don't all of a sudden ask him what his favorite flavor donut is. Keep it consistent. Also, when writing contingency questions, make sure to vary the way you ask the question (positively and negatively). This will avoid a "pattern bias" in which respondents answer a certain way simply because they're being influenced by a repetitious pattern.
Of course, if you're questionnaire has multiple topics you will have to ask questions related to this topics. Instead of just jumping straight into new topic, notify your respondent by writing a transition statement. That is, if you will be switching from Donald Trump related to questions to donut related questions, make sure to write in a transition statement to let the respondent know that the focus of the questionnaire will no longer be about Donald Trump. In other words, section off questions based on topic and remain consistent, and clear, in order to help your respondent answer as easily as possible.
When writing a contingency question, it's important to make the follow up questions clear. A contingency question is any question which has more than two parts. For example, asking "Are you employed" is likely to be followed with, "where are you employed?" This is a contingency because it's possible that the respondent need not answer the latter question if he's not employed. It should be clearly labeled as a contingency so that the respondent knows to skip questions which he/she is not required to answer. This can be done with arrows pointing from the first part of the contingency question to the sub parts which follow. It can also be done with clear wording that lets the respondent know that the following questions will be continent on the answer to the previous one. For example, stating "If yes," or "If you answered no to question X, skip to Question Y."
Surveys can be done either verbally or in writing. When it's done writing, it's called a questionnaire. Questionnaires can be done while you are in the respondent's presence, or they can be done remotely (mail or internet). Writing a good questionnaire will determine the validity of your results. Below are a few guidelines for writing a good questionnaire that'll ensure your research will be well received.
The basics
Begin your questionnaire with some basic instructions for the respondent (the person taking your survey). This helps make your questionnaire less confusing for the respondent. It also prepares them for whatever comes after. Always provide clear instructions on how to answer questions no matter how obvious you think the answering format is. If you create a section of a survey with different answering format, explain to the respondent how they should answer. For example, if your question asks the respondent to rate from "1-5," make sure to explain what the significance of this scale is and how the respondent should use it when answering.
Make sure you write your survey using a readable, and common, font and font size. You wan to generally make sure that everything about your questionnaire is easy to read and understand. This means paying attention to details and thinking about your questionnaire from the perspective of someone who has never seen it.
Avoid communicating your opinion on whatever subject you want to ask about. The purpose of surveys is to understand how a group of people think about a certain subject. Inserting your opinion might influence them in some way, and therefore, could give you inaccurate results. Remain neutral in every possible way. Avoid words with negative, or positive, connotations.
Make sure to thank your respondent for taking the time to do your questionnaire! It's always good to follow the golden rule with respondents. Be respectful of them, their time, etc, and make sure you don't inconvenience them unnecessarily.
Open ended questions versus close ended questions
Close ended questions are only appropriate when possible answers don't vary much. It's important that you address all, or most, possible responses expected in your close ended questions else you will get results which don't accurately reflect the respondents opinions. For example, if you ask: "what is your opinion on Donald Trump winning the election?" and then you only provide possible answers of "I approve" and "I disapprove," that leaves a lot of possible responses out. It's possible for respondents to have mixed feelings on the matter, or simply be indifferent. For this reason, it would be wise to include a more exhaustive possible range of responses for this particular question in order to get the most accurate results.
Close ended questions are much easier to code than open ended questions due to the limited possible responses. An open ended question requires a lot more work, and interpretation, in order to code. This will be important when you start analyzing your results in order to come to conclusions.
What to avoid when writing questions
It should never be unclear what the question is asking. All terms and phrases should be clear, and not left up to the interpretation of the respondent. Make sure that your questions are as clear and concise as possible. Avoid necessarily wordy and complex questions.
Specifically, avoid double barreled questions. For example, asking a respondent if he/she agrees with Donald Trump's or Hillary Clinton's policies. It's possible for a respondent to agree with neither. These people would not be able to answer your questionnaire and probably even decide to no longer continue. Also, avoid wording your questions in a negative context. For example, "Do you disagree with Donald Trump's policies?" should be worded as "Do you agree with Donald Trump's policies?" This is a much more direct and less confusing way to answer the question.
Lastly, avoid questions which encourage the respondent to answer a certain way. This may seem obvious, but often we can allow our own biases to slip through without noticing. For example, asking "Do you agree with Donald Trump's destructive international relations policies?" would surely create a bias in which many people might react to the assumption that the policies are destructive. Instead, avoid influencing the respondent by simply removing the word "destructive" and keeping the question as neutral as possible.
Structure
If you plan on writing a few questions with similar answering formats, you might also decide to split up some of your questions into a "matrix." This is a good idea if you want to save on some space and at the same time make it easier for the respondent by giving him a quick way to move through similar questions. If doing so, make sure that all questions within the matrix are related to each other. That is, if you're asking the respondent about his thoughts on Donald Trump, don't all of a sudden ask him what his favorite flavor donut is. Keep it consistent. Also, when writing contingency questions, make sure to vary the way you ask the question (positively and negatively). This will avoid a "pattern bias" in which respondents answer a certain way simply because they're being influenced by a repetitious pattern.
 |
| Example of a question matrix |
Of course, if you're questionnaire has multiple topics you will have to ask questions related to this topics. Instead of just jumping straight into new topic, notify your respondent by writing a transition statement. That is, if you will be switching from Donald Trump related to questions to donut related questions, make sure to write in a transition statement to let the respondent know that the focus of the questionnaire will no longer be about Donald Trump. In other words, section off questions based on topic and remain consistent, and clear, in order to help your respondent answer as easily as possible.
When writing a contingency question, it's important to make the follow up questions clear. A contingency question is any question which has more than two parts. For example, asking "Are you employed" is likely to be followed with, "where are you employed?" This is a contingency because it's possible that the respondent need not answer the latter question if he's not employed. It should be clearly labeled as a contingency so that the respondent knows to skip questions which he/she is not required to answer. This can be done with arrows pointing from the first part of the contingency question to the sub parts which follow. It can also be done with clear wording that lets the respondent know that the following questions will be continent on the answer to the previous one. For example, stating "If yes," or "If you answered no to question X, skip to Question Y."
 |
| example of contingency question (second part not show) |
Response Rates
A response rate is the number of people who responded to your questionnaire relative to how much you sent out. It's calculated by taking the number of people who took your questionnaire and dividing it by the total amount you sent out. If you're mailing out (or distributing in some other way) your surveys, it's highly unlikely that everyone will respond. In fact, it's highly likely that most people won't. A low response rate could create a bias, so it's important to write a good questionnaire which will have the highest response rate possible. Generally, the higher the response rate, the lower the chance for a non-response bias to occur.
Research has found that compensating individuals for their time is effective for increasing response rates while at the same time creating no bias of its own. Of course, if you don't have the cash this would be impossible. Consider applying for a grant or searching for any source of funding.
Feel free to add any questions, or comments, below.
Comments
Post a Comment