The Steve Jobs effect
Around 1980 Steve Jobs brought the Apple II home computer on the market in Europe. At that time we were doing experiments with different response scales on mainframe computers, as I had seen in the USA. When I saw this Apple computer I thought that this was probably the tool that will be used in the future for survey research. Using computers for data collection prevents the printing of questionaires and the coding of the answers before one could do the analysis.
The Apple computer (see photo) could not be used for this purpose, given that one had to carry around the computer , a monitor, a disc drive and the connecting cables. Besides, the programs my colleague Marius de Pijper had made for the mainframe computer had to be rewritten for this Apple II computer.
The Apple computer (see photo) could not be used for this purpose, given that one had to carry around the computer , a monitor, a disc drive and the connecting cables. Besides, the programs my colleague Marius de Pijper had made for the mainframe computer had to be rewritten for this Apple II computer.
A program for computer assisted interviewing
Talking about these future possibilities Marius, Peter Neijens and I decided that it was worthwhile to prepare for this future by rewriting our interview program for this computer and by creating a box in which the different parts mentioned above could be transported. In this way we were able to perform the experiments with continuous response scales using random samples of the population.
Marius rewrote his programs for interviewing on the Apple computer. At that time that had to be done in machine code. He was very curious doing this task and together we developed simple procedures to write questionaires for this computer and to perform interviews. Examples of such question specifications and response forms are presented below. These forms were used for our experiments to measure opinions on continuous scales.
Talking about these future possibilities Marius, Peter Neijens and I decided that it was worthwhile to prepare for this future by rewriting our interview program for this computer and by creating a box in which the different parts mentioned above could be transported. In this way we were able to perform the experiments with continuous response scales using random samples of the population.
Marius rewrote his programs for interviewing on the Apple computer. At that time that had to be done in machine code. He was very curious doing this task and together we developed simple procedures to write questionaires for this computer and to perform interviews. Examples of such question specifications and response forms are presented below. These forms were used for our experiments to measure opinions on continuous scales.
This example shows that we tried to make the writing of questionaires as simple as possible while the author of the question has complete control over the layout of the question. The program controls automatically on numbers outside the bounderies 1 and 1000. In that case a message is given and the question can be answered again.
Question type 2 is used for line production. The lengths of the lines on the screen for the respondent are indicated by L(38) and L(1) in the instruction at the left. The respondent is supposed to answer this question by him/herself. The lines were drawn by pushing the arrow buttons. Correction in both directions was possible. This was done regularly what suggests that people had much more precise information than they could express on the usually used 7 point scale.
The first tests
We asked somebody to make a big box in which component necessary for interviewing outdoors could be transported. Unfortunately at that time our first “portable computer” weighted around 16 kilos and was not easy to transport.
We asked one of my students, Hans Koomen, to go into the town with the interview box, ring at random at a house and ask if the person who opened the door was willing to cooperate in an interview. Hans Koomen was compared with many students at that time a very polite looking student who was not very dangerous (see picture). For unclear reasons he realized only 1 interview in 30 tries. This was a very disappointing result for us, but we were not willing to give up immediately.
I asked a good looking female student whether she was willing to do interviews for us. We suggested the following procedure: By car they should go together to an area for interviewing. She would ask a random person at the door for her/his cooperation and if the person agreed, she would suggest that Hans would do the interview with the computer he had in the box. In this way they got together a response rate of 50% while she also did an interview in the same time at another address. Alone she had an even higher response rate. In this way we could obtain in a more efficient way sufficient data for our experiments.
We asked somebody to make a big box in which component necessary for interviewing outdoors could be transported. Unfortunately at that time our first “portable computer” weighted around 16 kilos and was not easy to transport.
We asked one of my students, Hans Koomen, to go into the town with the interview box, ring at random at a house and ask if the person who opened the door was willing to cooperate in an interview. Hans Koomen was compared with many students at that time a very polite looking student who was not very dangerous (see picture). For unclear reasons he realized only 1 interview in 30 tries. This was a very disappointing result for us, but we were not willing to give up immediately.
I asked a good looking female student whether she was willing to do interviews for us. We suggested the following procedure: By car they should go together to an area for interviewing. She would ask a random person at the door for her/his cooperation and if the person agreed, she would suggest that Hans would do the interview with the computer he had in the box. In this way they got together a response rate of 50% while she also did an interview in the same time at another address. Alone she had an even higher response rate. In this way we could obtain in a more efficient way sufficient data for our experiments.
The results
It became clear that computer assisted interviewing was equally efficient as personal interviewing. At that moment, it was still hard work since interviewers had to carry a box of 16 kilos to the household. However we expected that soon real portable computers would appear on the market which would be much lighter. Except for carrying the box the interviewers had not much to do. Being in the house they had to ask permission to use their electricity to plug in the computer. Then they had to explain the procedure to the respondents and to indicate how they could answer the questions by typing a number, click on a button and draw a line with an arrow key.
This suggested that one source of errors, which is that the interviewers often don´t read the questions as they are formulated by the researchers, could be prevented. This finding helped us later to realize a system where we got rid of the interviewers completely. In 1981 we published our first results about these experiments at a conference in the Netherlands.
It became clear that computer assisted interviewing was equally efficient as personal interviewing. At that moment, it was still hard work since interviewers had to carry a box of 16 kilos to the household. However we expected that soon real portable computers would appear on the market which would be much lighter. Except for carrying the box the interviewers had not much to do. Being in the house they had to ask permission to use their electricity to plug in the computer. Then they had to explain the procedure to the respondents and to indicate how they could answer the questions by typing a number, click on a button and draw a line with an arrow key.
This suggested that one source of errors, which is that the interviewers often don´t read the questions as they are formulated by the researchers, could be prevented. This finding helped us later to realize a system where we got rid of the interviewers completely. In 1981 we published our first results about these experiments at a conference in the Netherlands.