Research Methods: web survey (CAWI)
A Web survey (CAWI – Computer Assisted Web Interview) is part of a methodology based on a questionnaire provided to the respondent with a link, in a panel, or a website. It’s considered the most economical way to collect survey data, because you don’t need interviewers, devices or extra tools.
For all these reasons web surveys, also called online questionnaires, are one of the most common methods to collect information. When you’re opting for CAWI, all the attention is on the questionnaire design because the response rate is directly linked to the quality of the questionnaire itself.
Main advantages of a web survey
CAWI can be convenient for several reasons. Here’s a list of them:
tailored questionnaire: if the questionnaire is well designed, it will automatically manage questions path using logic conditions like display or skip logic;
more clarity: the questionnaire can be structured to facilitate comprehension and increase the response rate;
guides: you can insert instructions to help the respondent to understand and complete the survey. This should obviate for the lack of an interviewer presence;
reduced costs: there’s no costs involved to buy electronic devices;
reduced time: the time needed to analyze the data is reduced because those data are available in real time to the admin database. Some software, like IdSurvey, allow you to have a real time survey reports and statistics of collected data and email campaign.
Main disadvantages of web surveys
When you reduce survey costs you have to reckon some inevitable disadvantages that can somehow jeopardize your survey research quality. One of those could be the lack of an interviewer as a help and guide for the survey respondents. Other disadvantages to keep in mind:
not everybody has an internet connection, so the target will be wide but not universal;
with the online survey you may lose the attention of the respondent increasing the possibilities to get low quality or partial surveys.
The perfect online survey questionnaire
The most important step of a web survey is the questionnaire design. It has to be carefully studied and tested before sending it out to respondents. In this paragraph we’ll give you some practical tips on designing the perfect questionnaire.
Data collection and priority: you have to define your research goal.
Questionnaire format: the questionnaire should start with a short introduction to inform the respondent on survey purposes. A clean and engaging layout and the use of multimedia may increase the number of completed interviews. IdSurvey makes you choose and customize optimized ready to use templates.
Questionnaire length: as a general rule, the questionnaire shouldn’t exceed 5 minutes.
Test: it’s important to get feedback on the questionnaire to make it better, easier, or more elaborated.
Question types: it’s crucial to pick the best type of answer. There are numerous options you can choose from:
- single select: the respondent has to click on the button correspondent to the desired answer. All other options will be excluded;
- multiple choice: the question allows multiple answers;
- dropdown menu: clicking on the answer area you’ll be able to display a dropdown menu containing all possible answers. Just one answer is possible;
- open answer: they allow the interviewees to write their answers.
- rating scale: a really simple scale is used (for example: Poor, Fair, Good, Very good, Excellent).
- “smart” answers: a lot of software, like IdSurvey, allow you to use sliders, comparing sliders, rating, and drag&drop.
Univocal link distribution
Univocal distribution requires a different link for each respondent. The online survey software is then able to link respondents profiles in the database with interviews completed. This procedure is useful to get information as who completed the survey and who didn’t complete it. Tracking survey responses is important to perform mass actions on the database, for example to send a reminder to all those who didn’t answer the questionnaire.
Univocal link and quotas
Univocal link distribution is also useful to create surveys with quotas based on data already known. Let’s see an example: we want to complete 1000 interviews, half for females and half for males. When a quota is full, the online survey tool will stop unnecessary interviews (called over quota). If we know the gender in advance, the software will automatically stop respondents of a completed quota with an alert, avoiding useless interviews.
When quotas are based on existing data, the univocal distribution is essential for the system to discard irrelevant contacts in real time. Sending email to irrelevant respondents can be also avoided using a unique feature in IdSurvey. This characteristic optimizes survey distribution and saves you a lot of time and money.
In anonymous distribution there’s no connection between the respondent and the interview. This method is useful when you have to guarantee anonymity to your respondents. In this case all the contacts will use the same link to log in. It can be published in a website, in social media or sent via email. It’s good to remember that anonymous distribution doesn’t prevent the same respondent to answer the questionnaire multiple times corrupting final data.
How is the distribution made with online survey software?
Email and SMS
You can send invites to a list of respondents with email or SMS. A software with an integrated sender also gives you information on opened emails or, for example, on people who didn’t receive the email because of a full mailbox or a wrong email, etc…
Carefully selecting the online survey software is really important as just a few of them have the integrated email feature. This integration is crucial to easily manage the entire survey. In fact, in some of these software you can also schedule reminders to all respondents who didn’t answer the survey. Discover CAWI software.
You can publish the link in any website or social media. This method is frequently used along with anonymous distribution.
A panel is a container of outlined respondents that can be periodically interviewed for market research. We usually consider a representative sample of a specific category to be a panel and panels can really useful to make frequent surveys. The term panel is used today to define any list of respondents available to join and answer surveys in exchange for prizes or money. There are several services of this kind as Toluna and Research Now that offer their respondents for surveys of any nature.
There are also specific software to manage your own panel. For example, a company can create a panel with their own customers to design marketing strategies, check customer feedback or to test a new product or brand. Surveys based on panels perfectly integrate with CAWI methodology because they can considerably increase response rate. Almost any CAWI system is now able to interact with panel services as that’s more a procedure than a proper feature. But you have to keep in mind that just some of those can really make the procedure simple allowing you to set everything up in a few clicks.
CAWI web surveys on mobile
Another important issue is the compatibility with most common devices: more than 35-40% of respondents open the questionnaire directly from their smartphone or tablet. To improve their experience we recommend designing matrix questions with just a few columns to avoid unpractical (not to say impossible) questionnaire compilation. The responsive feature is not provided by all CAWI surveys system. In this case IdSurvey guarantees the best performance: the entire system is natively responsive.
How to manage the contacts list
The quality of the contacts is important when you’re doing CAWI online surveys. Just think about the fact that you may require to send more than 100,000 invitations to complete 1,000 interviews. To overcome this problem there are a lot of strategies.
For example, if the contacts are part of a customer list, they’ll probably be more likely to complete the interview because they’re aware of the importance of their answers. Panels are another valid option: more and more companies are using them. They will help you to get a list of profiled contacts (e.g. tv service subscribers, supermarket chain customers) and to manage prizes to assign to people who complete the questionnaire. Using these two strategies you may be able to cut down the invitations from 100,000 to 5,000 to get your 1,000 completed interviews.
It’s also undeniable that administering questionnaires to loyal customers will improve the quality of collected data. The increasing value assigned to contact lists quality generated a real profiled contacts trade. This business gave life to services like Toluna or Swagbucks. But you have to be careful: using these services doesn’t always guarantee a good quality of data collected. We’ll see this topic more in detail in one of our following posts.
Data quality of a web survey
In CAWI surveys it’s crucial to monitor data quality and reliability. While the survey is ongoing, it is advisable to check complete and partial interviews. So you can verify if complex (or long) questions may cause higher levels of drop out or may lead the respondent to random and rushed answers.
Questionnaire quality can be evaluated also during the data analysis process: examining compilation time you will be able to exclude data that will compromise the survey result. Another important information is the level of satisfaction of respondents. You can get an idea of that inserting a rating question at the end of the interview or comparing the number of completed questionnaires to the number of drop out.
Monitoring and fixing questionnaire deficiencies is decisive to get reliable data for high-quality analysis.
To sum up, to create an effective web survey you have to be highly meticulous and to follow specific rules that will help you obtain great results at a very low cost.