Online sampling is a rather straightforward process which a few distinctive steps, which will eventually lead to reaching the requested target. The steps are identical for all online quantitative projects with the exception of trackers, IHUT’s and re-contacts. These have proven to be a bit more tricky, but that’s not the subject of interest here. I want to point out common pitfalls in setting up and running and online quantitative sampling project. This first post dives deeper into the set-up process and an upcoming post will focus on the actual sampling process.
We, at Panel Inzicht, have a structured approach on setting up a project. First, we want our panel members to end up at specifically designed pages. For exmaple, when someone completes a survey when want this person to end up at our page stating they indeed have succesfully completed the survey and reward points will be added to their account. The same goes for persons who do not qualify for a particular project and persons who cannot participate (anymore) because the survey is now closed. So we have three links available for all projects and they are project-specific. After panel members have completed the survey (or did not qualify, or could not participate since the survey is already closed), they should be instantly redirected to the corresponding pages (also called endpages or redirects).
- Please don’t mix up the endpages (for obvious reasons)! It’s surprising how often this actually happens. Someone who completes a survey should not end up at a page stating (s)he did not qualify. A rather frustrating event when you’ve just completes a 15 minute survey…
In order to reward our panel members, we need to know who completed which survey. Since our three endpages are project-specific, we know to what project we need to attribute a completed survey. But in order to reward the correct person, we need to know who (s)he is. That’s why all our panel members have a unique ID (sort of a social service number . When someone enters an online survey, this ID is also sent along in the link. When people are sent back to our endpages, we need this ID again to verify who this person is. It also enables us to keep our promise of (instantly) adding reward points when someone completes a survey. Without this ID we cannot know who actually completed the particular survey.
- Kindly send us back the panel member ID so we know who participated (real-time). Someone who ends up at the correct page, only to find out later that (s)he did not get rewarded with points, will be quite frustrated as well…
Once the above two points are covered, we can start testing the survey itself (the questions, translations, routing etc.). Usually, testcases are seperated from the actual completes by using two links: one for testing purposes and one for sending to the panel members. Suffice to say, one can assume that the survey itself is identical, regardless of the link used. Well…, that’s not always the case we found out the hard way. It’s actually possible to succesfully set-up and test a project using the testlink and still run into major problems when using the ‘live link’.
- Make sure the set-up and survey are identical in both the testlink as well as the livelink. Inviting your panel members to an incorrect link/survey is an disturbing experience for panel members.
A project manager also has to check translations, a cumbersome task to say the least. We don’t only face bad translation, but also partial translated surveys or surveys with multiple languages (in random order).
When you are a project manager, don’t assume anything. Test everything all the time. Always. A proper set-up will certainly make your life easier during the fielding of the project (more on that in the next post). And to our clients I would like to say the same: please test everything before sending it over to us. Set the link live, check the quotas, check the redirects, ensure proper translations etc.
Yes, it will take you some extra time.