The value of significance is important n cross-sectional models considering whether you have true or null hypothesis

Saud - A descriptive study can be used among many to: (a) define respondents/study participant characteristics (b) Measure trends in the data collected (c) conduct comparison (d) validate existing conditions etc. To answer your first question, yes- using descriptive research studies we can test a hypotheses/make an inference. The one thing which we can not do with descriptive studies is a conclusion about a CAUSAL relationship. Cross-sectional studies are observational in nature and are often known as descriptive studies. How to calculate the p-value and/or confidence interval depends on the statistical method you use. Each statistical method has its own set of assumptions/procedures that we have to follow so that the p-value or/confidence interval outputted can be interpreted with certain degree of confidence. In case your are looking at particular study and have some specific questions, feel free to let me know.

Thank you very much Cristian and Henok!

I really appreciate your succinct and informative answer Dr. Henok G. Woldu as well as your welcoming response for further clarification!

To express it in a way I understood your reply, for confirmation, you can use significant tests in cross-sectional study design if you have a predefined hypothesis or you need to make an inference. Clear! But on the other hand, if you don't test for significance in such a study design, does it undermine the value of the results or the conclusion?

To bring a specific scenario, the existing condition of a study subjects in relation to a specific disease is studied using certain variables (such as risk factors). Pearson's correlation analysis shows a strong linear correlation between some variables. In such a situation, is it valid to rely on the strength of the association without a test of significance?

Saud - it is a good question!

Think this way- there are many ways you can get a sample of 20 subjects if your entire population is say 100. Lets say Researcher A get a sample of 20 subjects and calculated correlation to get r = 0.90. Researcher B get a sample of 20 subjects and calculated r = 0.60. Two different answers and the conclusions too will be different. Which researcher's result should we believe? Such difference is due to sampling error and it is the main motivation behind the statistical significance/hypothesis testing. Therefore, unless you are getting your results using data collected from the entire population, not conducting significance test will undermine your conclusion. Hypothesis testing can help us to check whether the estimated value we get ( like the sample correlation coefficient is this case ) is purely by chance or its real and not due to chance.

Thank you Henok! Your answers are always to the point and clarifying.

It is better to calculate p-values/ confidence intervals for descriptive studies though these are not meant for drawing inferences. Confidence interval suggests you the degree of confidence of an estimated parameter. Hypothesis test suggest you to check whether the estimated value obtained is by chance or it’s real. Correlation and p-values are to be interpreted carefully. For example, in a study with calculated correlation r = 0.90 and in another r = 0.40, it is possible to get same p value of 0.03. In this case the conclusions will be different.

@kandasamy this holds true & is frequently overlooked, the interpretation of the statistic, for instance in a descriptive study design the interpretation of CI if reported would be very different from an analytical study with a probability sampling approach

Article The use of Bayes factor in clinical cardiology research

Article The Bayes Factor, a Suitable Complement beyond Values of p