Many surveyors probably do not even think that respondents may not be completely honest at times, and may lie intentionally to manipulate the surveyors with the intent to get what they want. Maybe respondents does not understand question or you ask questions socially sensitive questions, even if question is reasonable to ask, respondents may not want to tell you the “painful truth”, these are some of examples where people could give you irrelevant responses about which I wrote in the first part of this article. Read more about: Do respondents honestly answer survey questions? - Part 1.
Here I wanted to give you some advices how to get the best of your survey and how you can prevent getting dishonest or faked answers for more transparent and relevant survey analysis.
How to avoid getting dishonest or fake responses?
Clear language and understandable questions
Clear or plain language is "communication your audience can understand the first time they read or hear it" according to plainlanguage.gov. It is recommended to use as simple words without terms that are specific in some area, for example, technical terms. It is certainly good to avoid jargon. Simple sentences will help respondents in understanding the questions, and instead of using too complicated sentences, use simple.
Avoid ambiguity, do not ask "What do you do?". The respondent will not know the exact answer because he does not know what you mean directly. Were you wondering what job they are doing, what they do in their free time, what are they doing right now? Instead, ask a specific question to get the answer you are looking for.
Also, it would be good to avoid double-barreled questions. Take for example the question "What do you think about our app functionality and design?" You set as a multiple choice question with the scale of "Like it very much" to "Do not like at all". Maybe you have a really good design and everyone likes your app visually and when they open it everyone thinks "Wow, this looks great!" But its functionality is either bad or has a small number of it. In this case, the respondent can not give you a relevant answer because one item is absolutely great and the other is completely different. Instead of two items in one question, ask each question separately.
Furthermore, you can not look for a long-term memory about some subject from respondent. Try to ask questions that are as close to the present as you will get more specific answers. Perhaps the respondent had thought about a subject one year ago somehow, but through time his thinking changed, so now it is completely different. The respondent may not recall what he meant then and give you an irrelevant answer.
Avoid double negative questions! "Is there no way for you to do anything about ...?" What? We have already lost with first negation. Preformulate the question and set it in a positive way. Generally asking questions with negation confuses the respondents.
"How do you like that song?" Who says I like that song? Do not ask leading questions. Instead of leading questions you can ask "What do you think about that song?"
Balanced number of positive and negative answers
There are examples where the likert scale does not have the same number of positive and negative responses. Unbalanced scale soften the truth to surveyors when respondents do not have such a nice opinion about product/service or generally some statement, and according to that surveyors offer less negative responses, while as positive you can choose more different. Surveyor then gets a partially correct answer. Why? If someone chooses "strongly agree", this means that respondent is completely satisfied or agrees with the offered statement. When respondent choose agree, this means there may be some doubts, but generally agrees with the statement. Equally, there are negative opinions. Someone may not be right against the statement, but it is inclined negatively, while there are those who do not agree with statement at all, and there should also be a "strongly disagree" option for them. Balancing the answers helps the respondents to exclaim and decide how much they agree or not. Also, by balancing the answer, a more accurate picture of the respondent's opinion on a certain statement is obtained. If the respondents are not very happy with the asked question, they can not fully determine their dissatisfaction on an unbalanced scale and are forced to choose one and only negative choice. Cover all possible answers.
I have read many articles were authors are discussing whether neutral answers are desirable or not, but I have not found a specific answer. A few advocates the presence of neutral answers, while others are strictly against them. Neutral answer is in use in surveys with odd numbers of options, while surveys with even number of choice do not have neutral option and are usually called as “forced” choice surveys. Neutral responses are there for those respondents who are not heard or are not familiar with item from question or just do not want to answer that question. Instead of ignoring and choosing a random answer, respondents will choose a neutral answer that is more acceptable to your results than to get a bunch of dishonest or fake responses. Some of the neutral answers are: other, none of the above, do not know, no opinion.
Branching helps you segregate your respondents to get more relevant answers. Take for example the question "Are you a dog person?" If the respondent replies that he is not a dog person, it is no need to ask him questions like “Do you own a dog?” because respondent would give a random answer or leave your survey. If respondent is not a dog person it is obviously that he do not own a dog. Using the branching you can divert your respondents to another question that is not related to dogs, or at the end of your survey. If the respondent replied that he is a dog person, it is logical to ask him a few more questions about it. Those who are dog persons will give you the relevant answers that you will be able to use without problems and fear to fulfill the research goal. Branching redirects respondents depending on their fields of interest so it helps you to avoid dishonest or fake responses.
Target audience is one of the most important things when you are planning to do a research. Your targeted audience, for example, someone who is using your product or services, are only people which can give you a specific and valuable feedback. In your survey you should have a few qualifying questions that will help narrow out anyone who doesn’t fit within your target group prior to asking the real questions. Be sure these questions appear completely unbiased and do not sway a person to answer in a specific form.
Why are they going to be motivated to take the survey? Are you offering to respondents who complete the survey a free sample or a coupon? What about a gift card? The more specific the reward is to the target group, the more you’ll be able to decrease the chances that surveyors will try to qualify just for the reward.
The primary factor that affects how accurate the information collected by a given survey is directly relates to the person’s motivation for filling out the survey. For instance, data collection techniques that reward the respondent for completing a survey can sway their answers.This reward may be a gift card to compensate them for their time (which would motivate just about anyone) to a free sample of the product (which helps to only motivate people who actually fit within the survey’s target group).
Unfortunately, it’s impossible to determine which answers are honest or misleading in any one survey you put out. Except motivating the respondents, try to consider other steps to prevent getting dishonest and faked answers. Writing understandable questions is crucial for every survey, such as clean or plain language. You can never completely avoid dishonest or faked answers, but you can definitely reduce the number of such answers.
Allen, E., & Seaman, C. (2007). Likert Scales and Data Analyses. Retrieved February 13, 2018 from the Quality progress website: http://asq.org/quality-progress/2007/07/statistics/likert-scales-and-data-analyses.html
Plainlanguage.gov. (n.d.). What is plain language? Retrieved February 5, 2018 from plainlanguage.gov website: https://www.plainlanguage.gov/about/definitions/
Great and useful articles:
DeMars, D. E., & Erwin, T.D. (2005, August). Neutral or unsure: Is there a difference? Retrieved February 13, 2018 from the Psyc website: http://www.psyc.jmu.edu/assessment/research/pdfs/DemarsAPA05EIS.pdf
Edwards, M. L., & Smith B.C. (n.d.). The Effects of the Neutral Response Option on the Extremeness of Participant Responses. Retrieved February 11, 2018 from the Longwood website: http://blogs.longwood.edu/incite/2014/05/07/the-effects-of-the-neutral-response-option-on-the-extremeness-of-participant-responses/
Vinokurov, J., (n.d.). Jerry Vinokurov’s Guide to Writing Questions. Retrieved February 10, 2018 from the Academic Competition Federation website: https://acf-quizbowl.com/documents/question-writing-guidelines/