Back to research
At the end of 2019 I got the idea that I had some topics of our research not finished. One of these topics was that in social science research often several questions are asked to measure a concept of interest, while it is not clear if the combination is better than a single question.
For example in 1981 I already complained about the fact that one tried to measure “Job satisfaction” not only with the direct question
- How satisfied are you with your job?
but also with extra questions:
Perhaps you think that these questions are not so bad to get an impression of the job satisfaction of a person, however only the first questions asks directly about job satisfaction while in the other questions also other considerations play a role. For example: how attractive the other jobs are the respondent thinks about or whether the qualifications of his/her friend are good enough or too good for this job. That means that responses to these extra questions not only are determined by job satisfaction but also by other ideas. This implies that the combination of these questions to measure job satisfaction could be even worse that the direct question alone.
For example in 1981 I already complained about the fact that one tried to measure “Job satisfaction” not only with the direct question
- How satisfied are you with your job?
but also with extra questions:
- - Would you like to have another job?
- - Would you recommend your job to your best friend?
- - Would you choose this job again if you had the possibility?
Perhaps you think that these questions are not so bad to get an impression of the job satisfaction of a person, however only the first questions asks directly about job satisfaction while in the other questions also other considerations play a role. For example: how attractive the other jobs are the respondent thinks about or whether the qualifications of his/her friend are good enough or too good for this job. That means that responses to these extra questions not only are determined by job satisfaction but also by other ideas. This implies that the combination of these questions to measure job satisfaction could be even worse that the direct question alone.
Can it be done better?
It can indeed be done better because the linguistic research of Irmtraud had shown that some concepts can be formulated in different grammatical forms without changing the meaning of the questions. For example :
In order to show that this is the case one has to have a procedure to determine the quality of a combination of questions. This was a second project I worked on.
It can indeed be done better because the linguistic research of Irmtraud had shown that some concepts can be formulated in different grammatical forms without changing the meaning of the questions. For example :
- Are you satisfied with your job?
- Is your job satisficing ?
- Does your job give you satisfaction?
In order to show that this is the case one has to have a procedure to determine the quality of a combination of questions. This was a second project I worked on.
A second project
In the past, I had worked with Anna Decastellarnau, a student at RECSM, on the correction for measurement errors in survey research. For that research, one has to know the quality of the questions. We had done this especially for single questions and not for combinations of questions. Given the issue mentioned above, we now wanted to do the same for combinations of questions. Therefore, it was necessary to determine the quality of the combinations of several questions or the composite score as they are called. I asked her to work out this statistical problem under my supervision. She did the statistical derivations very well. We worked on this issue with pleasure together. Later I simplified her statistical derivations and to get the paper published, I had still some work to do in the Corona time. I liked that work.
The paper with Irmtraud was published in 2020 and the second with Anna was published in the first months of 2021.
In the past, I had worked with Anna Decastellarnau, a student at RECSM, on the correction for measurement errors in survey research. For that research, one has to know the quality of the questions. We had done this especially for single questions and not for combinations of questions. Given the issue mentioned above, we now wanted to do the same for combinations of questions. Therefore, it was necessary to determine the quality of the combinations of several questions or the composite score as they are called. I asked her to work out this statistical problem under my supervision. She did the statistical derivations very well. We worked on this issue with pleasure together. Later I simplified her statistical derivations and to get the paper published, I had still some work to do in the Corona time. I liked that work.
The paper with Irmtraud was published in 2020 and the second with Anna was published in the first months of 2021.
This was an inspiration to do some more research work again because in the Corona time not much else could be done but reading or working on scientific problems to keep the mind active.