Split

Sometimes, on only one individual or one home, we will realize several identic measures. We will ask for example for a consumer to evaluate a set of products on the same criteria. In a home, we will save the identic information (age, sex, class, etc.) for several children. We will have then a questionnaire with first, a set of questions concerning the products for baby, then several parts (three in our example), where we ask the same questions, for other types of products.

If we want to obtain certain results like the average number of children per home, the distribution of different socio-professional categories or even the geographic distribution of the families asked, the statistic unity will be the home.

But if then, we wish to know the most bought brands or the distribution of the types of products per home, we should never reason about the home but about the product. We will have to modify the structure of the base of the responses and operate a « changement of the statistic unit ».

Click on the button Split in the tab Data. You reach the following assistant.

In our example, we think about the product (beverages). We select then all the parts to treat (those ones must be built in the same way). Click OK to validate and quit the assistant, you will be asked then if you wish to save and open the survey created.

Regroup

It is sometimes necessary to regroup several observations concerning the same individual. This notion is frequently used during the analysis  360° in the domain of human resources. The principle is the following : a person is evaluated by a certain number of professional competences by his superior and equally by his colleagues. Otherwise, he could himself be brought to self-assessment. This type of study, once deployed in a company, creates a set of observations concerning a group of collaborators judged several times on the same criteria.

From this data base, and for certain analyses, we wish to regorup the evaluations of the same individual to work for example, only on the average he has obtained on each competence, calcuated on the totality of the questionnaires concerning it (his superior, his colleagues). We will move then from the table of evaluations to the table of the people evaluated. For this, we will make a changement of statistic unity by regrouping the observations.

Here is an example of data collected in the framework of a survey "Human Ressources" :

We can see then that MARTIN Paul for example has been evaluated 4 times : by his superior, by 2 collaborators, and by a (pair). We see then 4 lines in the spreadsheet. But we wish to regroup these 4 evaluations in the same one line : never reason on each evaluation, but on the global evaluation of this person (and this is for all the people evaluated in the data base). For this we must set a changement of the statistic unity by regrouping.

Click on the button Regroup in the tab Data. You reach the following dialog box.

In our example, the key variable corresponds to the people evaluated (we want to regorup all the evaluations in one). We have chosen to conserve all the data by merging them. Otherwise, we calculate the average of all the evaluations of each evaluated individual.

Once this stage is informed, clickr Ok to validate your choices. Another survey will be created : this of regrouping. You must save it and it will be automatically opened.

In the survey ‘regrouping’, the following data are shown in the spreadsheet :

We note that we have henceforth only one evaluation (one line in the spreadsheet) for MARTIN Paul for example. The column ‘Evaluator’ lists all the evaluators of this person. It is therefore possible now to take our analyses on each evaluated person (for example starting from the note average that we have calculated).