Hi,
I want to ask a question about excluding items from a scale. For my research I will use 'Parasocial Interaction Scale. This scale consists of 14 items. Two items of this scale is not relevant for my research purpose. Therefore I want to exclude these two items and use the remaining 12 items.
Is it a problem for the psychometric properties of the scale?
Meryem
Dear Meryem
This is an important question.
It is possible when one borrwo a scale from different culture to find some items dont fit the new setting. Therefore, the scale could be modified. it is prefered to ask permission of the original author on this modifications.
you may before going to delete option, try to modify the 2 items and then discuss and justify this in your paper.
The psychometrics of the scale will be changed. the total score number will not be the same, and in any mean when one use the scale in a new setting, the validity and reliabiltiy will reflect the new sample.
Dear Meryem
This is an important question.
It is possible when one borrwo a scale from different culture to find some items dont fit the new setting. Therefore, the scale could be modified. it is prefered to ask permission of the original author on this modifications.
you may before going to delete option, try to modify the 2 items and then discuss and justify this in your paper.
The psychometrics of the scale will be changed. the total score number will not be the same, and in any mean when one use the scale in a new setting, the validity and reliabiltiy will reflect the new sample.
Dear Muayyad Ahmad,
First of all I want to thank you for your quick response. The scale is already adapted to Turkish language and culture. The problem is that the scale is used for movie stars who are alive and I want to use it for a famous singer who is dead. What should I do?
If I see the 2 items under question, I may help better, but also I may suggest to modify the questions by adding "If the Singer is still alive,...what do you think..etc."
Hello,
another limitations after removing two items is inability to compare your results with past research (mean comparisons, severity thresholds, etc). Maybe ask yourself a question: Is it 'worth' to get rid of two items? Is the 'price' for keeping them high?
Personally, I would rather use them, and PERHAPS exclude them from some analysis later... Other issues have been discussed above by other colleagues.
All the best with your research
Witold
Hello,
I totally agree with Witold and Muayyad, a good solution is probably to keep the items in your assessments and exlclude them afterwards in your analyses.
This way you can study how the 2 items actually influence the score of the scale, can perform a factor analysis to check latent dimensions behind the items, and calculate Cronbachs alphas to check reliability and how much of it you'll loose when excluding the items - and discuss all that in your research.
Regards
Johannes
You could also use a quite different approach.
As mentioned in the former answers, when using the sum score/scale score from a test, you have to think of the tests as a whole. A well-developed test have gone through different phases, resulting in the specific set of questions in the test. If you change the composition of the test by deleting questions, you change the test and its psychometric properties. Naturally, the mean and standard deviations will also be changed.
Depending on the actual questions, maybe you could change the content a bit? In the MMPI-2 there is a question, worded approximately like this “I love my father, or if he is dead, I loved my father”.
Maybe you could just add a clause like this about the dead father?
However, there is another possible way of proceeding. This presupposes that you do not need to compare your results with normative data or other research - in which case you will need a sum score with mean and SD to be comparable. If this is not the case, you could plan to analyze the data using confirmatory factor analysis. In this approach you will get results about how each question performs, you will get results with probably less measurement error, and you may be able to test you research hypotheses without the need for first calculating scale scores and then performing statistical procedures on these. Instead you will skip this step and go directly from the data to the hypothesis testing.
The software program of choice would be Mplus. Learning material for these type of analyses can be found at www.statmodel.com.
Best
Jan Ivanouw
Hello Meryem
1. On a scale of attitudes like any other instrument, the exclusion of an item, usually depends on who (experts or judges) validate the instrument, but also depends on the specific results obtained Reliability Cronbach Alpha with some statistical software such as SPSS (Analyze> Scale> reliability Analysis).
2. should also be considered Table of operationalization of variables, which is conceptually defined variables with their respective dimensions and indicators are operationally defined; each of these indicators the amount (enough) of items are broken down. Therefore, care should be taken not to indiscriminately delete items.
Luck. A hug. George W.
We have just published an article about scale purification, i.e. the process of eliminating items from multi-item scales. We have used the example of SCM, but our framework can be applied to any other discipline. Download: https://doi.org/10.1108/SCM-07-2016-0230 (or request via my ResearchGate page).
Permission from the author regarding modification of the scale is necessary first. after modification according to your research their psychometric properties will also be changed.
- Badges
- Science topic
- Similar topics
- Psychology
- Psychometrics