The start of the panel
Like the NIPO in the past, the UvA panel needed the interview program INTERV and the software for the panel communication developed by the SRF. The SRF also had developed many attractive tools for income and expenditure surveys. In order to be able to start its research STP not only had to take over the software but also the people who had worked on these questionnaires. From the SRF they took over the students Robert Voogt , Harm Hartman, Bas van de Putte while I joined them as well.
New coworkers
Besides these people who would be responsible for the questionnaires new staff was needed for other tasks. First of all a secretary was needed and a technician for installing the computers. They were quickly found.
It was more difficult to find a coworker who would check weekly the interviews and a statistician for the sampling issues. For the first position we selected Corry Vis. For the sampling issues we found Adriaan Hogedoorn, a statistician who worked before at the bureau of statistics. With these last appointments the staff was complete and we could start.
Besides these people who would be responsible for the questionnaires new staff was needed for other tasks. First of all a secretary was needed and a technician for installing the computers. They were quickly found.
It was more difficult to find a coworker who would check weekly the interviews and a statistician for the sampling issues. For the first position we selected Corry Vis. For the sampling issues we found Adriaan Hogedoorn, a statistician who worked before at the bureau of statistics. With these last appointments the staff was complete and we could start.
A management program
Building up a panel of 2000 households is much more complicated than a sample of 50-100 households. We needed a complex software program that would store the information of the whole process: the sample drawn, the cooperation or not, the composition of the household to send special information to the different people in the household, the fact if they were willing to join the panel whether they had received a computer, modem and eventually a monitor, whether the computer was installed, whether there were eventual problems, their weekly cooperation and comments, their eventual dropout and then the return of the computer. This was a rather complex program that did not exist.
Harm Hartman had the task to make this program. It will be clear that we could not function without this program so I urged him to work at that program so that it was ready for use when it was necessary. The first part of the program was made so that we could start but the program was not yet complete.
When I thought that it was taking too long because he was also doing other things, which he shouldn´t do, I took him with me on a trip to Crete where we had a meeting about the CASIP project. He had nothing to do with that but I had another plan with him. I hired a car and drove with him to the south coast of the island and was looking for a very small village along the sea where he could not do anything else but work on the program. He needed electricity, something to eat and maybe a possibility to swim occasionally but nothing else. We found such a very small village with a room for him 5 meters from the beach.
I told him that he could only come back when the program was finished. A week later he came back. My approach seemed to have worked. The program looked much more advanced, although we later detected that it was still not completely finished.
Building up a panel of 2000 households is much more complicated than a sample of 50-100 households. We needed a complex software program that would store the information of the whole process: the sample drawn, the cooperation or not, the composition of the household to send special information to the different people in the household, the fact if they were willing to join the panel whether they had received a computer, modem and eventually a monitor, whether the computer was installed, whether there were eventual problems, their weekly cooperation and comments, their eventual dropout and then the return of the computer. This was a rather complex program that did not exist.
Harm Hartman had the task to make this program. It will be clear that we could not function without this program so I urged him to work at that program so that it was ready for use when it was necessary. The first part of the program was made so that we could start but the program was not yet complete.
When I thought that it was taking too long because he was also doing other things, which he shouldn´t do, I took him with me on a trip to Crete where we had a meeting about the CASIP project. He had nothing to do with that but I had another plan with him. I hired a car and drove with him to the south coast of the island and was looking for a very small village along the sea where he could not do anything else but work on the program. He needed electricity, something to eat and maybe a possibility to swim occasionally but nothing else. We found such a very small village with a room for him 5 meters from the beach.
I told him that he could only come back when the program was finished. A week later he came back. My approach seemed to have worked. The program looked much more advanced, although we later detected that it was still not completely finished.
Archangel Michael helped
Marius de Pijper had made in the past for the Telepanel of the SRF a program that had to be installed on the computer of the respondents. That program had to read a message sent from the central computer to the household specifying the tasks the different people in the household were supposed to do or could do. The questionnaires could be different for different people in the household. Besides that computer games were sent to the households for the children to motivate the family to participate in the interviews as well. The message to send to the households could be quite complicated given the tasks they had to do. A problem was that a minor error in such a message would lead to problems in the households and to a lot of telephone calls to the people responsible for this task. We started with simple tasks for all members but later it became more complex. The specification of these tasks was done by the same programmer mentioned above.
Once, we were together at a meeting for CASIP in Barcelona, when there came a phone call from Amsterdam that there was an error in the specification of the tasks. We immediately had to do something. At that time there was no internet so we did not have a print of the text sent to the respondents. Our colleague in Amsterdam spelled the text by the telephone and in the hotel room we looked for the error and the necessary correction. Then we called back to Amsterdam and suggested the correction that we thought was necessary. To be honest we were not 100% sure that it was now correct.
Our hotel was opposite to the Cathedral of Barcelona. As a “good catholic” I suggested to the programmer that we should go to the Cathedral and light a candle at an altar of a saint in the hope that this will help that our suggestion would be good. The programmer, who was grown up in a strict protestant family, thought that I meant what I said. I also raised the question which saint we should choose because it should be a saint with influence, but not too popular because then he/she had too much requests and possibly would ignore our request. So we went to the cathedral, looking for an influential saint. The programmer followed me completely confused. In the end we selected the Archangel Michael for support because “he” had a lot of influence and very few people had paid for candles in that chapel, much less than in other chapels. It turned out to be the right choice because the correction was indeed the right one and the problem was solved.
Marius de Pijper had made in the past for the Telepanel of the SRF a program that had to be installed on the computer of the respondents. That program had to read a message sent from the central computer to the household specifying the tasks the different people in the household were supposed to do or could do. The questionnaires could be different for different people in the household. Besides that computer games were sent to the households for the children to motivate the family to participate in the interviews as well. The message to send to the households could be quite complicated given the tasks they had to do. A problem was that a minor error in such a message would lead to problems in the households and to a lot of telephone calls to the people responsible for this task. We started with simple tasks for all members but later it became more complex. The specification of these tasks was done by the same programmer mentioned above.
Once, we were together at a meeting for CASIP in Barcelona, when there came a phone call from Amsterdam that there was an error in the specification of the tasks. We immediately had to do something. At that time there was no internet so we did not have a print of the text sent to the respondents. Our colleague in Amsterdam spelled the text by the telephone and in the hotel room we looked for the error and the necessary correction. Then we called back to Amsterdam and suggested the correction that we thought was necessary. To be honest we were not 100% sure that it was now correct.
Our hotel was opposite to the Cathedral of Barcelona. As a “good catholic” I suggested to the programmer that we should go to the Cathedral and light a candle at an altar of a saint in the hope that this will help that our suggestion would be good. The programmer, who was grown up in a strict protestant family, thought that I meant what I said. I also raised the question which saint we should choose because it should be a saint with influence, but not too popular because then he/she had too much requests and possibly would ignore our request. So we went to the cathedral, looking for an influential saint. The programmer followed me completely confused. In the end we selected the Archangel Michael for support because “he” had a lot of influence and very few people had paid for candles in that chapel, much less than in other chapels. It turned out to be the right choice because the correction was indeed the right one and the problem was solved.
Another risk
A similar problem could occur due to errors in the questionnaires. If there are no complex routings in the questionnaires not much can go wrong. However, if a set of questions depends on previous answers serious problems can arise. If the branching was not specified correctly the respondents get questions which are not for them and sometimes they can even not answer these questions. If this error occurs for many respondents there are at least a number who start to call the help desk to complain that something is wrong. Of course as research organization one has to avoid such problems. The control of all possible response patterns is sometimes impossible because there are too many possibilities. Although Corry Vis tried to check as much as possible all questionnaires before they went into the field there occurred still occasionally errors.
Therefore we had every weekend a help desk which could be called if there was a problem or a question. The tasks were sent to the households on friday afternoon and there were always some people who immediately started to fill in the questions and started to complain if there was something wrong. In that case we could repair the problem before the large crowd started to fill in the questionnaires in the weekend.
After some months the work became more like a routine business although there were occasionally situations that we got far too many phone calls. The problem was that some parts of the process we could not check using automatic procedures.
A similar problem could occur due to errors in the questionnaires. If there are no complex routings in the questionnaires not much can go wrong. However, if a set of questions depends on previous answers serious problems can arise. If the branching was not specified correctly the respondents get questions which are not for them and sometimes they can even not answer these questions. If this error occurs for many respondents there are at least a number who start to call the help desk to complain that something is wrong. Of course as research organization one has to avoid such problems. The control of all possible response patterns is sometimes impossible because there are too many possibilities. Although Corry Vis tried to check as much as possible all questionnaires before they went into the field there occurred still occasionally errors.
Therefore we had every weekend a help desk which could be called if there was a problem or a question. The tasks were sent to the households on friday afternoon and there were always some people who immediately started to fill in the questions and started to complain if there was something wrong. In that case we could repair the problem before the large crowd started to fill in the questionnaires in the weekend.
After some months the work became more like a routine business although there were occasionally situations that we got far too many phone calls. The problem was that some parts of the process we could not check using automatic procedures.