Obviously, if a researcher can do a complete census, this, in quantitative terms, would deliver the most accurate result; however, a single researcher can appropriately take a census of only a very small population---otherwise, for larger populations, samples are the way to go---the most common situation.

I agree with Philip Adams. Refer to this link you can have more clarity. If your target population is 200 or less then go for a Census. Otherwise sample study. To prove that your sample is adequate you have to further calculate statistical power which should be not less than 80 per cent. https://www.tarleton.edu/academicassessment/documents/Samplesize.pdf

Dear Olaolorunpo Olorunfemi,

For computing statistical power please visit this link: https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-und-arbeitspsychologie/gpower.html

First, if you have a census, that means you are not trying to infer to any other cases. So, for example, if you census a certain business expense incurred by all manufacturers of a given type in one geographic region, you are not then going to assume it is the same for another region or for another type of manufacturer. Any such speculation has to be labelled as such, because a census is for the entire population, where there is no inference to any part of that population. Statistics on this are then descriptive statistics. (One might consider this population as one possibility under a superpopulation, but the members are still the same.)

With regard to accuracy, a census will have no sampling error. But it will have nonsampling errors of various types. Frame or coverage errors can mean the population isn't what you think it is. Response or measurement errors can be large, including data reported in the wrong units, which can be a huge problem. These errors are probably more likely to occur if your resources are stretched to produce a census.

A sample will have both sampling and nonsampling error. However, overall accuracy can be better than if a census were done, as control of nonsampling error may require extensive resources.

Census, or enumeration could only be possible if the population of study is manageable, otherwise, it may be difficult

I agree with the duo of Olufemi and Philips Adam on the need to use the Census method for smaller research figures, which will enable you study the population appropriately without mistakes.

Thank you.

The census principle can be use both in survey and experimental studies. As far as the population is within manageable size the census principle can be applied irrespective of the research design.

To my best knowledge as trained as a statistician, we say either 'Census of a study population' or 'A sample survey'. Because "A survey is an investigation about the characteristics of a given population by means of collecting data from a sample of that population and estimating their characteristics through the systematic use of statistical methodology." When people try to talk about "Census survey", it becomes very confused. It is better we all follow the convention in statistics discipline rather than creating some non-standard statistical terms.

I agree with John that it would often be better for communication if we had more standardized language. However, that does not always seem to be the case. As I recall, some time ago at the government statistical agency where I worked, I tried to argue with someone who outranked me, when he insisted that only a census could be a survey. The purpose of the agency was to conduct surveys and produce data, but I wanted him to know that besides the census surveys we did, the sample surveys, which were of primary interest to me, were still surveys. He flatly said "No!"

This happens in other cases. It seems to me that for regression, some refer to epsilon as the "error," and e as the residual, and others, like me, refer to epsilon as the residual, and e as the estimated residual. I know a well-known statistician who insisted on using the term "estimated residual," and since I don't think the word "error" is appropriate for any of this, I go with the latter option.

But there can be misunderstandings when people are not using the same terminology. - Related, I remember two physics books, I think for the same course, where a symbol for one measure in one book was the same symbol for the time derivative of that measure in another book. Confusing.

A better choice might come up for a term, and some things are developed in parallel, so we have to expect a little confusion, but it is often good to minimize this. Still, we have to watch for any miscommunications.

Thank you very much for your responses on census. Kindly is there a specific formula of calculating the number of known respondents when using census or a researcher just uses the specified number that is already known......bearing in mind that there will be no sampling

Ruth,

no specific 'formula' since in a census, we are taking the whole population, or, as Gang (John) says above, a census of a study population.

However, you've actually got the idea of taking a census of a manageable population when you say a researcher just uses the specified number that is already known.

This specified number is---sometimes---a manageable) population that a researcher has chosen to study by census; for instance, all heroin addicts in a particular locality, all twenty-somethings in the village, etc.

In sampling research, this specified number is the sampling frame from which samples are chosen (because the chosen sampling frame, in this case, can be large---hence, usually the practical need for sampling); however, if this specified number is chosen to be manageably small, one can do a census of the study population.

No formula. The nature of the population, population size and research determines the choice of census which can also be referred to as complete enumeration.

I'm conducting research with a population of about 40. All the participants have responded to the questionnaire. Using the census method how do I go about the calculation of the sample size?

How do I I explain this in the methodology

Hello Esther,

There's no calculation required, nor any statistical inference required. Your results represent the full population. You explain that your data collection was a complete enumeration of the population.

It's probably better to post queries that are different from the original post for a thread in a new thread. That way, other RGate users searching for answers related to your new question are more likely to be able to find them.

Good luck with your work.

Hello Esther,

You don't need any statistical procedure to get the sample size because the number is already small. So stick to census principle where all the elements in the population are studied. Thank you. All the best in your research.

a census method is that process of the statistical list where all members of a population are analysed