This type of sample is determined by the researcher's order, ie there is no randomness to choose an element of the population.
Choosing a non-probabilistic method will, as a rule, always encounter disadvantage compared to the probabilistic method. However, in some cases, this option is required. It is observed that when sending questionnaires by mail the method is non-probabilistic (even if the option is by probability sampling). The respondent may not want to answer the questionnaire or even be localized.
Fonseca (1996), warns that there is no way to generalize the results obtained in the sample to the whole population when opting for this sampling method.
Accidental or convenience
Suitable for exploratory studies. Often used in supermarkets to test products.
The interviewer goes to a specific group to get their opinion. For example, when studying a car, the researcher looks for workshops only.
In fact, it is a variation of intentional sampling. Prior knowledge of the population and its proportionality is required. For example, we want to interview only individuals from class A, which represents 12% of the population. This will be the quota for the job. Commonly, a quota is also substratified according to a second proportionality.
Widely used when the choice of sample is disproportionate to the population. We assign weights to the data, and thus obtain representative weighted results for the study. For example, in a mobile phone market, considering a purely illustrative market share, the results were obtained as follows:
|Brands||Market Share||Elements in the sample|
In order to obtain the weights to be assigned to each mobile phone brand, for a joint analysis of all brands in the example above, we obtained the following coefficients:
|Number of elements to be interviewed|
Applied formula: Weight = market share / sample elements (%) Next: Probabilistic sample