3.3 Collecting
3.3.1 Data collection methods

The main modes of data collection for surveys are in-person interviews, telephone interviews and self-completed questionnaires. Interviews can be computer-assisted or not, and self-completed questionnaires can be on paper or digital support. To choose the best method, factors such as the characteristics of the sample frame, the target population, the survey budget, the desired level of accuracy, the sensitivity of the information to be collected or the complexity of the survey concepts must be taken into account. Here is an overview of the main collection modes.

In-person interview

Interviewers visit people in their environment to survey them. It’s a good way to get higher response rates to a sample survey or census and, because of the interviewers’ work, the quality of collected data is higher. However, travel costs can be high. When the interview takes place at the place of residence and nobody in the household is present or available to reply to the survey, the interviewer might have to come back multiple times. 

When the interview is computer-assisted, the interviewer brings a laptop or electronic tablet and capture the data directly in a database or with the help of an electronic questionnaire designed for this purpose. This method reduces the time needed for data processing and the interviewer avoid carrying a large amount of paper questionnaires which have to be stored safely to protect confidentiality of respondents. However, it is more expensive and longer to implement because computer hardware must be provided to interviewers and the computer systems need to be developed and tested before the start of data collection. When the interview is not computer-assisted, the interviewer writes down the answers on a paper questionnaire. This method requires fewer resources and preparation before collection, but increases data processing time since the answers must be captured in a digital support to be processed and analyzed with software.

Telephone interview

Interviewers call people to survey them. This collection mode is faster and less expensive than an in-person interview. But you need a sample frame with the phone numbers of the population units and the observed population exclude households without a telephone. It is also very easy for the person called to hang up the phone … or simply, not to answer the phone in the first place if they don’t know the incoming telephone number. For this reason, response rates are lower than in-person interviews. It is also important to note that fewer and fewer households have a landline nowadays which is likely to make it more difficult to reach them in the future.

Like for an in-person interview, the telephone interview can be computer-assisted or not. Computer-assisted means more preparation before collection, but less processing data after data collection.

Self-completed questionnaire

A questionnaire is provided to the respondent who has to fill and return it. It can be printed or it can be a hyperlink and access code to fill online. In both cases, it can be handed directly, sent by mail or delivered at the dwelling. To send the questionnaire by mail you need a list of addresses in the sampling frame. Home delivery can be useful when dwellings are not listed in the sampling frame.

It’s the least expensive mode, and it can be used to reach a very large number of people. The respondent can fill the survey at the moment of his choice. Self-completed questionnaires may be easier when the survey questions are sensitive. The drawbacks of this mode are that response rates are lower than for other collection mode and the quality of collected data is not as good. It is also the slowest method because there is no control over the moment at which the respondent will fill and return the questionnaire.

The questionnaire must be simple with very precise instruction because the respondent cannot ask for clarification if they don’t understand the question. Respondent with limited capacities in reading or writing in French and English can have difficulties in filling the questionnaire.

When a digital support is used, often called Electronic Questionnaire (EQ), the web interface can be more convivial for the respondent and allow him to answer the survey faster. It is possible to notify the respondent that a question has not been answered and to indicate to him his progression in the questionnaire. Return of the questionnaire is almost instantaneous, and the respondent doesn’t have to remind himself to return it by mail. Like for computer-assisted interviews, processing time is reduced because the answers are captured directly in a machine-readable format. However, people with no access to the internet or with less computer knowledge will be less likely to participate unless they are given the option to answer on a printed form.

Nowadays, EQ is used as the first option before telephone interviews by many surveys, including business and household surveys.

Other methods

Other methods are direct observation or direct measurement. Here are two examples. In the context of data collection for the Consumer Price Index, surveyors visit points of sale to note the prices of a predetermined list of items. This data allows the estimation of the inflation rate in Canada. In the Canadian Health Measures Survey, some physical measures are taken from respondents like weight, blood pressure and blood measures. These data are gathered using mobile examination centres or mobile clinics.

Combining methods

The most efficient strategy will often use a combination of methods that were just described. The Canadian Census of the population is a good example of a multimode collection strategy. During Census 2016, a letter was sent by mail to a majority of households, with a hyperlink and a secured access code to fill the census form online, along with a free phone number to request a printed form. In areas where not all dwellings are listed on the census frame, printed forms were delivered by census enumerators responsible for the listing operations. Along with the print form, a hyperlink and a secured access code to fill online were also provided. In Indigenous reserves and hard-to-reach regions, like Nunavut for example, data collection was carried with in-person interviews. Finally, in-person interviews and telephone interviews were used for nonresponse follow-up. The similar strategy has been applied to Census 2021.

A communication strategy used along with the data collection strategy can be useful to make the survey known to the public and increase response rates. It can be a letter sent by mail to initiate contact with the household, a press release in local media or, more rarely, a national communication strategy like in the case of the census of the population.

What is paradata?

Paradata is data collected during the collection operations, but which is about the collection itself and not about the survey subject. It can be the time at which the interview was done, the length of the interview, the number of times it took to reach the respondents, etc. It can be useful in the conception of adaptive survey designs or as auxiliary variables to adjust results for nonresponse.