Create a sample source
1
Open the Sample source tab
Go to the Sample source tab in data collection.
2
One sample source per origin
Create one sample source per respondent origin. The simplest approach: one sample source per type of panel or per data source.
3
Set the provider
Set the provider: Consumer panel, Own contact file or Social networks. For a contact file, invitations are sent from the E-mailing page.
4
Set the sample size
Enter the sample size you are targeting for this sample source (e.g. 200 interviews).
5
Panelist ID is required
State whether a panelist ID is needed. This ID links each respondent to their panel or, for a survey run on a contact file, to their entry in that file. On a panel survey, it can be passed back through the redirection URL, as a parameter of the link (see Redirections). It also governs how duplicate responses are blocked (see below).
Avoid duplicate responses
The panelist ID is what allows poll&roll to recognize the same person from one visit to the next. The Panelist ID is required setting therefore directly changes the level of protection against multiple responses:A panelist ID must not contain
? or /. A link that appends a path or an extra parameter after its value produces an invalid ID, and the interview is not created.Rename a sample source
You can rename each sample source to tell them apart, for example an internal panel and an external panel.Combine several sample sources
On multi-source fieldwork, each sample source generates a separate URL (retrieved in the Launch tab) and separate progress monitoring: completes, screen outs and incidence per sample source. You can therefore steer each sample source independently and fine-tune fieldwork for each of them (chase a lagging panel, slow down a source that is running ahead, and so on). From the moment you create them, anticipate the distinctions you will need at the analysis stage: this is the right time to plan how you will want to separate your targets and your different launches.The sample source does not appear in the results, but it is present in the interview list. You can therefore recreate a “sample source” variable to crosstab it at the analysis stage (e.g. to tell a panel send apart from a boost on social networks).

