CATI surveys (fieldwork CATI)

We carry out CATI surveys since the introduction of this methodology in Italy, in the second half of the 80 years, when IT innovations allowed us to abandon the PATI method (Paper And Telephone Interview) to move to a video management of the questionnaires.

Until the early 2000s, the CATI methodology reigned supreme in the panorama of statistical surveys in Italy, but for ten years the spread of the web and the advent of mobile telephony have introduced decisive innovations. In 2003, when the available technology allowed it, we started to carry out CAWI surveys and from the 2009 - CAMI research.

Today the CATI surveys represent one of the methods of conducting a survey in the composite framework of mixed-mode that employs all the available tools: landline phone, mobile phone (smartphone included), online panels and campaigns on social networks.

Our structure

For our work we use:

•  28 CATI stations with Linux Debian 7.0 operating system, 160 GB hard disk, with WiFi connection, equipped with Plantronics Audio 355 headphones
•  70 stations at home, remotely monitored, of interviewers with at least 5 years of shared experience
•  3 VoIP provider able to guarantee up to 10.000 simultaneous calls
•  3 physical servers  for the management of mails and for the management of private instances of the CloudResearch software
•  27 cloud server intended for the management of blocks of functions and instances of different customers
•  3 workstations dedicated to supervision activities that use CloudResearch instances for monitoring
•  6 workstations with programs for statistical processing and software development

The phases of a CATI research

Sampling

To start off, the universe to be contacted is defined - population or companies in possession of the information that must be collected through a questionnaire. Once the universe has been defined, the sampling method is defined in order to extract a sample that is representative of the population: capable of providing us with an estimate of behavior and opinions.

Sampling can be done in different ways:

• List: A random sample is extracted from a list of telephone numbers and inserted into the CATI software. The software extracts a random sample proportional (for example) to the territorial area.
• Parameters: We aim to interview a certain number of units with a certain characteristic (for example n males and m females). The contact phase continues until the number of predetermined units with the defined characteristics have been interviewed. Typically, you want the sample dimensions to mirror the universe.
• Random Digit Dialing: a random procedure that generates the telephone number.
  With this method, the sample unit is the family that corresponds to that telephone number.
• From the family to the individual: If the sample unit is an individual, a method is needed to ensure the randomness of the extraction of the individual within the family:
  1. if there is only one component, interview that one;
  2. otherwise, the number of males and females in the family is asked and a family element is chosen through a predefined algorithm;
  3. last birthday method: interview the person whose date of birth is closest to the day of the interview.

Scripting of the questionnaire / Pilot Survey

Programming of the questionnaire

Once the questionnaire has been defined, it is programmed via the CATI software. Remember that the software passes the call to the interviewer when someone answers. The interviewer's task is to read the questions that appear on the screen and then depending on the answers of the Respondent, input them into the computer.

Pilot survey

Before carrying out the actual survey, a pilot survey is carried out in order to minimize wording errors typically caused by: duration, understanding of the questions, their sequencing. If the pilot test gives satisfactory results, proceed with the survey, otherwise we intervene with the appropriate changes.

Procure cooperation

From the experience of the scientific community, as from common sense, it is clear that sending information before the interview increases the response rate. A letter, an email or a text message, all this can be useful!

Presentation e-mail: assuming that the interviewees have an e-mail, the contact procedure consists of contacting the sample unit to which the e-mail address is requested and to which the presentation is therefore always sent.

Presentation SMS: incredibly efficient if we have mobile numbers.

Interviewer training

The training of the interviewers is a phase of decisive importance for the correct collection of information.

Here it is a question of providing detailed explanations of the purposes of the research and of the implicit and explicit meaning of the individual questions in order to provide the interviewers with the cognitive horizon of the survey. This is necessary to minimize the so-called interviewer error.

As a data collection company for market research, we pay particular attention to the training of our interviewers. In fact, our interviewers have to go through several stages of training and testing before they can start taking surveys. In addition, they are continuously supervised, supported and corrected to continuously improve the data collection process and reduce distortions.

Survey, supervision and data files

Survey and supervision / monitoring

The actual survey takes place under the constant supervision / monitoring of a researcher who is able to detect the need to provide further explanations to the interviewers. The CATI software allows you to verify the veracity of the interview. During the survey, the quotas (typically gender and age) are checked (in the surveys where it is necessary) in order to guarantee the representativeness of the sample.

Data file

We deliver the data file to the client, in the requested format with the appropriate labels specifying the variables and the response methods.

Need a hand with a CATI survey, or may be you prefer that we do it for you and provide you with the collected, weighed and processed questionnaires?