Universe is the total set of people or objects that are part of a segment to be searched. Sample is a representative sample of the universe, that is, a set of people or objects with the same characteristics as the whole. respondent is the person who fills out a questionnaire, answering the questions formulated in a research.
Polls are opinion collections, usually over the internet, accessed on websites or social networks, whose objective is to obtain answers to single questions, without distinction of public, without collecting profile information such as location, income , gender, social class, schooling, etc. Survey is an opinion study with a small group, whose result is extrapolated to the entire universe, within a degree of confidence. Research may have single questions, but each respondent's social, cultural, economic, and local profile is collected and recorded. The results take into account the margin of error and degree of confidence.
Conducting surveys/researchs dates back centuries, until the 20th century, there was only one way to collect and record statistical information for the purpose of evaluating opinions: the written form by filling out questionnaires, face to face, interviewer and interviewee or via mail, for example. With the advent of the phones and more recently, the cell phone, it was possible to carry out research using these ways of communication called telematics. The internet, a place of few rules, was occupied by polls, which are basically single questions that can be answered indistinctly by anyone, with no profile evaluation for segmentation purposes. data prospect% operates via the internet. We do not approach respondents, they are invited to come spontaneously to the platform to fill in a questionnaire created by a customer.
Probabilistic samples are those in which any respondent in the sample has the same chance as another to respond to a research within the selected universe. Respondents, for example, are drawn among everyone in the universe. Non-probabilistic samples are formed by respondents who are not selected by some non-random criterion. This is the example of the data prospect% platform in which the creator of the research invites people to answer questionnaires formulated by him. Both forms of sampling, probabilistic and non-probabilistic are valid. In the non-probabilistic sample it is not possible to talk about margin of error. However, the researched sample is part of a specific universe, has larger samples and the same validity.
Research institutes and the data prospect% platform use society segmentations defined by some official bodies. They are oriented based on social indices published by the thisw bodies such as race, gender, sexual orientation, marital status, schooling, age group, income, religion, occupation, type of job, special and dietary needs, housing and type of property. And also by the electoral body in the segmentation by municipalities, by gender, age group, and schooling.
Margin of error is the probability of the result obtained in a sample having the same result as the universe researched. Ex.: If there is a result of 80% + or - 2% in a survey, we can have a degree of certainty that if we asked everyone in the universe, we would have the same answer in a range between 78% and 82%.
Level or degree of confidence is the probability of finding the same results by repeating the same questions in several researches. Eg: A 95% confidence level means that if the same research were repeated 100 times, 95 times the results would be the same.
The ideal would be to involve 100% of the universe, but due to the size, the time to obtain a result or the cost involved in collecting the responses, it becomes unfeasible empirically, a formula was developed and from this formula the table below was derived, which shows the sample size and respondents as a function of the expected margin of error.
Important: this table is for probabilistic research only.
Probabilistic sampling | |||||||
---|---|---|---|---|---|---|---|
Universe | Margin of error | ||||||
1% | 2% | 3% | 4% | 5% | 10% | ||
Under de 1.000 | - | - | - | - | 222 | 83 | |
1.000 | - | - | - | 385 | 286 | 91 | |
1.500 | - | - | 638 | 441 | 316 | 94 | |
2.000 | - | - | 714 | 476 | 333 | 95 | |
2.500 | - | 1.250 | 769 | 746 | 333 | 96 | |
3.000 | - | 1.364 | 811 | 517 | 353 | 87 | |
3.500 | - | 1.458 | 843 | 530 | 359 | 97 | |
4.000 | - | 1.538 | 870 | 541 | 364 | 98 | |
4.500 | - | 1.607 | 891 | 549 | 367 | 98 | |
5.000 | - | 1.667 | 909 | 556 | 370 | 98 | |
6.000 | - | 1.755 | 938 | 566 | 375 | 98 | |
7.000 | - | 1.842 | 949 | 574 | 378 | 99 | |
8.000 | - | 1.905 | 976 | 580 | 381 | 99 | |
9.000 | - | 1.957 | 989 | 584 | 383 | 99 | |
10.000 | 5.000 | 2.000 | 1.000 | 588 | 385 | 99 | |
15.000 | 6.000 | 2.143 | 1.034 | 600 | 390 | 99 | |
20.000 | 6.667 | 2.222 | 1.053 | 606 | 392 | 100 | |
25.000 | 7.142 | 2.273 | 1.064 | 610 | 394 | 100 | |
50.000 | 8.333 | 2.381 | 1.087 | 617 | 397 | 100 | |
100.000 | 9.091 | 2.439 | 1.099 | 621 | 398 | 100 | |
Over 100.000 | 10.000 | 2.500 | 1.111 | 625 | 400 | 100 | |
Authors: Arkin, H., & Colton, R. R. (1971). Tables for statisticians. Barnes and Noble |
IMPORTANT: Note that for a universe beyond 100,001 (up to infinity), according to this table, it is enough to get answers from a few people or the selection of a few objects. This table was created in the 50s of the last century and is highly debatable when applied to heterogeneous human universes.
Societies are increasingly complex. Citizens are defining more specific segments based on their personal characteristics. Sexual orientation manifests the citizen's willingness to choose who he wants to have a sexual relationship with, regardless of his genetics. Gender, in turn, is linked to social behavior, function. Because they are different, it is important to collect this information in impartial and independent surveys, when necessary, the data prospect% platform offers this possibility.