Adjustment
Despite all the efforts deployed to select the individuals, during the phases of e-mailing for example, a sample may be "biased", which means that its composition is not satisfactory, as it doesn't correspond to the criteria of representativeness we have previously defined. For example, we think that we have reached 60% of women and 40% of men but it's not real. We must then « adjust » the sample.
This operating mode is accessible in two ways :
- In the tab Data of the Spreadsheet, click on the button Adjust.
- From the home panel, click on the button Adjust of the stage Data management.
You reach the dialog box below :
1 Click Add to select the variable on which we wish to adjust the sample.
2 Indicate the percentage we wish to obtain for each modality.
In our example, we have now 17,26% of the respondents French as our objective is to have 20% of the probed people French. The same for the other nationalities. As the goal is to make the sample representative of the population surveyed on this criterion.
Two methods of calculation are possible :
- If you select Weight the observations, this allows to assign a weight to each observation (for example, if your sample doesn't contain enough French respondents, it will be assigned to each observation which contains the value France for the variable (relative to the country of origin a weight superior to 1). On the other side, as we have 33% of Spanish respondents, a weight inferior to 1 will be assigned to observations which contain the value Spain for the (variable relative to the country of origin).
- If you select Extract a sapmle, this consists in extracting a sample which allows to assign the modalities 1 or 0 to the new variable Adjustment, (1 corresponding to observations retained to represent the sample).
By clicking Finish, a new variable « Adjustment » will be created (this variable may be used as a sub-sample (profile) in the board : you can therefore visualize the boards on the new sample created representative, by adjustment, of the population studied.