Before I answer the question on how to create a survey question for a cross-sectional study (descriptive study), we need to understand what a cross-sectional study is. A cross-sectional study is one of the observational studies that is also called as a prevalence study, or transverse study. In a cross-sectional study we need to see what is happening at one particular point in time. This is how we actually differentiate a cross-sectional study and a longitudinal study. With later being a study in which we analyze and observe what is happening at multiple points in time. The way we differentiate cross-sectional study from other observational studies is that a cross-sectional study is focused on the entire population, whereas, other observational studies such as case control and cohort are focused on particular people of a population with certain illness or a risk factor.
The most challenging question for a novice researcher is how to conduct a cross-sectional study. Remember, you can create survey or interview questions to conduct a cross-sectional study. However, the validity of the questionnaire is a challenge that you need to surpass. However, the assessment of validity of your survey questions is not a daunting task that only experts can attain. Now lets discuss how to create valid survey questions.
HOW TO DEVELOP VALID SURVEY QUESTIONS:
The most common format of survey questions is the Lickert Scale Questions. You can create around 20-30 or more (there is no limit) questions in a Lickert format and then record the answers. However, this is not the only format of survey questions. You can choose any type of questionnaire. The challenge is not to choose what kind of survey questionnaire you will use. The challenge is
“how can you know if your survey questions are valid?” The validity of questions can be checked by following certain steps described by the expert Dave Collingridge. The steps are as follows:
Dave Collingridge believes, that this is the first step in checking the validity of the research questionnaire for surveys. This is a two step process and involves two individuals or two groups of scientists. The first group of scientists analyzes the scientific aspect of your topic. This group sees if your questions are relevant to your topic. The second group however, looks at structure of the questions and is an expert in question building and looks for possible errors and confusion in your questions.
Conduct a short pilot study with the same questions and some participants who will probably be present in your real pool of people, who will answer the questions. The purpose of this pilot study is to see the results of the study and eliminate irrelevant or useless questions if any. In the pilot study, the scientists want to see if the questions measure what they really wanted to measure. Once your pilot study is a success, it gives you a green signal to start your study. Even a very small sample can give you an idea about the validity of the questions, however, it’s always better to include a bigger sample size.
The next step is data cleaning- this is also achieved by having at least two scientists involved – which is a very important step. One scientist should read the data aloud while the other records the findings on a computer software. This reduces the chances of error.
The next step is to look for possible errors by the survey participants while answering the negatively phrased questions. For example, the questions like “I don’t dislike to eat pizza”. These are the questions that anyone can easily answer incorrectly because of the complexity of understanding the phrase due to the negative words present in the question. The scientists will have to look at these questions and compare them with similar questions with positive phrases to figure out whether these questions really have no error involved.
PRINCIPAL COMPONENT ANALYSIS (PCA)
This helps the scientists to assess if the survey measures what it was supposed to measure. This actually analyzes what is being measured in the survey. This also helps the scientists to decide if certain questions should be removed or added.
You will need an expert of PCA to assist you with this step. You can find many such experts on the websites like freelance or upwork.
The Cronbach’s Alpha (CA) test is usually used for this step and you will need an expert of statistics to help you with this step.
This is the last step in validation of the survey questions. After the PCA and CA you need to revise the survey according to the PCA and CA results. Remove irrelevant questions and keep valid, important, and relevant questions. This is how you maintain the validity of your survey questions. All of these steps will be written and mentioned in the method section of your manuscript to make your article quality high.
Two steps will require a professional experienced statistician to assist you. If you cannot do that on your own, don’t feel bad. That is why professional statisticians are there to help the scientists like us.