Multi-Wave Survey

If you want to survey one and the same respondent over different periods of time, you are usually dealing with a multi-wave survey, a longitudinal study or a diary study which all three face two challenges:

• How to ensure that your respondents complete the questionnaire at the right time?
• How can you combine different datasets of the same respondent at the end of the study?

SoSci Survey is offering a number of functions to ensure and facilitate the implementation of multi-wave surveys.

Technically, there a two possibilities to implement a multi-wave survey with SoSci Survey. Both do have their advantages and disadvantages and for both the following recommendation applies: you should test them carefully prior to your study - for example with shortened field time (a time interval of 5 minutes instead of 7 days) and temporary email addresses.

The first option is to create one questionnaire under Create a Questionnaire. This contains the pages for the first wave, an interruption page and, subsequently, the pages for the second wave. At first, the interruption page shows a Thank-You-note (for the first part of the questionnaire). If the respondent is returning to that page - after a previously defined time interval (for example through mailResume() or via a question type that sends an email to personal contacts) - the interruption page redirects the participant automatically to the next page which starts the second questionnaire. Programming an interruption page is documented and explained under the reference for the command mailResume().

• Advantages
  • All informations, statements and data of participants are stored in one dataset/case/interview.
  • In the second part of the survey, all answers of the first part can easily be accessed through value() or replace() - for example, for filter questions or for adjusting questions to respondents' answers through placeholders.
• Disadvantages
  • In case that questions appear in both parts of the questionnaire they have to be duplicated (otherwise earlier answers will be shown as default and subsequently be overwritten)
  • It is a little more complicated to specifically send reminder-mails to those participants who already received an invitation to the second part of the questionnaire but have not participated yet.
  • If someone leaves the questionnaire during the first part of the questionnaire (without completion), you will have to ensure that this participant is not invited to the second part. Otherwise the participant will continue the questionnaire (without any filters) at the very page where he left off.
  • The progress bar has to be adjusted via option('progress',…) or has to be deactivated completely, because it would count over both parts of the questionnaire otherwise.
  • For surveys with more than two waves it might be complicated if a participant wants to skip a wave or wants to participate later. In this case additional filters are necessary.
  • If a questionnaire is running no more pages can be added or deleted in the first part of the questionnaire (otherwise the participant will return to the wrong page of the second part).

In diary studies usually the same questions are presented in every survey wave - and moreover the data are supposed to be analysed with multilevel analyses. A solution with multiple questionnaires is suggested if…

• the same questions shall be used multiple times
• more than two waves are planned
• data shall be stored per interview (and not per participant) in single datasets

When using multiple questionnaires for every completed questionnaire a new dataset (data row) is stored. For matching reasons every data row needs a distinct participant code. By doing the following one can get such a code:

• If you send mailings, SoSci Survey automatically stores a distinct personal identification under the variable SERIAL (additional variable in the data output). This is possible also without losing any anonymity.
  • Important: The email must contain the placeholder %link% in order to actually send personalized links.
  • Important: All questionnaires must be created within the same survey project. The personal IDs are specific for each project.
• You can send personalized links yourself. The identification of every participant then will be passed by a reference or an access code (see also URL to the Questionnaire). A reference will be stored under the variable REF, an access code will be stored in the variable SERIAL (this one also allows for continuing the questionnaire).
• The participants are asked to create a personal code and enter this to a text gadget (for example with the first letter of their mother's name, the last digit of their birth year and so on). Because of its susceptibility to errors we do not recommend this option. If you want to use such an option it is important to ask the single components of the code in distinct, single text gadgets.

Compared to one questionnaire which is completed step by step (see above), the option of single distinct questionnaires has the following advantages and disadvantages:

• Advantages
  • Technically it is unproblematic if a participant misses or skips a wave.
  • Sending reminders is relatively easy.
  • One question can be used in different questionnaires.
  • One and the same questionnaire can be send to a participant any number of times (for example in diary studies)
• Disadvantages
  • If you want to access earlier answers of a participant you can use the data base of contents (see Using an Answer in Later Questions).
  • You will get multiple datasets (data rows) per participant which have to be matched for data analyses.

Depending on the design of the study, participants might all receive an invitation to the second survey wave at the same point in time or the moment is depending on when they completed to first questionnaire (also, a random assignment of intervals is possible, see MESM below).

Sending an invitation at a previously defined point of time can be obtained directly via the user interface:

• At Invitation Mailings –> Mailings a new mailing is created.
• If the invitation to the first wave was already done via mailings, this one shall be chosen at the tab reminder/follow-up mail (type of follow-up mail: “New wave”) - following this instructions, one can send the new invitation directly to those participants who already participated in the first wave.
• With the button “Prepare Sending” subsequently the sending will be prepared. You can there choose a date and time at which the invitation shall be sent.
• Under “recipients: reminder/follow up mail” you can determine whether or not only certain group of recipients who participated at the first part of the questionnaire shall receive an email (see above).

In case the invitation to the second part is dependend on when recipients participated in the first part, a little PHP-code will be necessary (see also Introduction to PHP). Using the function mailSchedule() it is possible to determine at which point of time an invitation will be sent to certain groups of participants.

In an ideal situation the participant will be invited to the first part of the questionnaire already by mailings. If an email address is is previously unknown the proceedings are described in the chapter Multi-Wave Surveys with Self-Recruited Respondents.

In the first questionnaire, preferably at the end, on a page that dropouts usually do not reach, the command mailSchedule() is positioned, for example:

mailSchedule(false, 2, 7 * 24 * 3600);

The first parameter “false” says that an invitation shall be sent to a participant who just completed the questionnaire. The second paramter “2” determines that the invitation with the number 2 shall be sent. The third parameter schedules the invitation to be sent in exactly 7 days, indicated in seconds (an hour has 3600 seconds, following this “7*24*3600” gives you the number of seconds for 7 days). Important: Every invitation via mailings can only be sent once to every participant. If you want to send one and the same email several times, you have to copy this mail. The corresponding PHP-code might look as following (here for 5 days every day an invitation will be sent):

mailSchedule(false, 2, strtotime('+1 day')); mailSchedule(false, 3, strtotime('+2 day')); mailSchedule(false, 4, strtotime('+3 day')); mailSchedule(false, 5, strtotime('+4 day')); mailSchedule(false, 6, strtotime('+5 day'));

In the example above the function strtotime() was applied. It creates an Unix-timestamp (number of seconds since 1.1.1970) out of a string-input. “+1 day” will be become “tomorrow at the same time”. “+1 day 8:00” would become “tomorrow at 8 o'clock in the morning”.

When working with previously determined intervals (for sending an invitation), it is easy to create a reminder mail as a new mailing and send it.

When working with relative intervals, you will have to prepare the invitation to the second part of the questionnaire as well as the reminder mail during the first questionnaire using the command mailSchedule(). Consequently, without any further ado, both mailings will be sent:

mailSchedule(false, 2, strtotime('+7 day')); Invitation the the second questionnaire mailSchedule(false, 3, strtotime('+8 day')); Reminder mail one day later

In a second step, you will have to ensure that the reminder will not be sent if the participant completes the second questionnaire. In order to do so, you will have to position the command mailRevoke(). This function will delete the planned sending from the schedule.

mailRevoke(false, 3);

Multi-wave studies allow for multi-variant designs. The following paragraph is dedicated to specific cases of application:

For multi-wave studies it is of advantage that email addresses of participants are known. With these you can send them personalized invitations to the second, third,… part of the questionnaire. In case the email addresses are previously unknown, the following chapter describes some solutions: Multi-Wave Surveys with Self-Recruited Respondents.

In a diary study one usually creates three questionnaires:

• Previous Questionnaire
• A Questionnaire that shall be completed every day
• Follow up Questionnaire

If email addresses are unknown you might need a fourth questionnaire for registration of the participants (Multi-Wave Surveys with Self-Recruited Respondents).

A mailing is created for every email or reminder that a participant shall receive. For example, two mailings for the previous questionnaire (invitation and reminder) and one per day. In the list of mailings is a button which easily and fast creates copies of mailings (button: duplicate).

Subsequently, for every email with “Preparing Sending” it will be determined at which point of time it will be send.

Tip: To which addresses an email will be send, will identified right at time of sending - in case it is a standardized interval for all addresses. Consequently, addresses can be added even after determining the time and date of sending.

Important: Please choose the questionnaire to which a mail shall refer to only in the tab “link to questionnaire”, if you are using placeholders like %link% and %link.html% in the invitation. The placeholder ensures that data rows will contain personal identifications in the variable SERIAL.

Mobile Experience Sampling is a special type of diary studies. Thereby participants receive at random intervals (within a certain timeframe) SMS messages on their smartphone. Those messages ask the participants to complete a preferably short questionnaire.

• SoSci Survey supports the sending of SMS via the Mailings-function. In case the SMS-function is unlocked, you can leave the text- or html-version of an invitation empty and instead enter some text for an SMS.
• In case you enter email-text as well as SMS-text, SoSci Survey sends messages via both channels (if both have a valid address or phone-number).
• The List of Contacts takes email-addresses as well as phone-numbers (instead or additionally).
• The Opt-In-question takes email addresses as well as mobile-phone-numbers. For a validation of a number an additional email address is necessary.
• Important: providers of mobile phone network charge a fee for sending SMS (per SMS costs incur by 0.10 Euros (plus sales taxes).
• SoSci Survey supports the sending of mailings (or SMS) at random intervals. At “Preparing Sending” you can choose the option “time-randomized sending” and subsequently determine the interval.
• Important: For every invitation a participant shall receive, a new distinct mailing must be created. For example, if during the period of 7 days 3 SMS per day shall be sent, you have to create 21 mailings (you can easily duplicate an email). For the first email you could then for example choose and random point of time on monday between 8 and 11 o'clock, for the second a random point of time between 11 and 2 o'clock and so on.
• Important: Time-randomized sending assigns a specific point of time to every participant. When subsequently adding addresses those participants will not receive an invitation, because for them no point of time was determined.
• Questionnaires that were created using SoSci Survey can usually be completed via smartphone, but for a MESM-study it is worth the effort to optimize the questionnaire for smartphones, among others with an optimized layout.

The counterpart to experience sampling is that respondents call up a questionnaire independently when a specific event occurs. Ideally, they do not have to enter a personal code each time, but can use a personalized link directly. There are several ways to do this:

• You can use mailings (Send Mailings) to send personalized links that allow multiple completion of a questionnaire. To do this, set the option “Multiple completion with the personalized link allowed” in the questionnaire link tab for the multiple participation setting in the mail merge.
• You can also create links yourself that carry a marker and send them to respondents: Mark a link. Note that in that case the pseudonym will appear in both the invitation and the dataset.

Search terms: panel, panel studies, experience sampling, ambulatory assessment.