I find two opposing views that one group of scholars are saying that it is recommended to have reverse coded items to cross-check answer validity; whereas some say it defeats the purpose because it oftentimes confuses the time-strapped respondents. Sometimes, it is also supected that reverse coded items measure entirely different construct! What do you think about reverse coded items in a questionnaire?
There are several arguments against using reversed items, one of which answering even a few items in the wrong direction will reduce your reliability (by interfering with the correlation among items). Another is that it doesn't really solve the problem of acquiescence bias, which is why it is usually recommended (see the attached reference by two well known experts in survey design).
Dear Colleagues,
Good Day,
Attached are a very interesting Youtubes related to the thread, please, see the links (three parts)...
https://www.youtube.com/watch?v=iQRst7_3bhY
https://www.youtube.com/watch?v=IDK_u4l7vJI
https://www.youtube.com/watch?v=IkTvKf5nFsY
There are several arguments against using reversed items, one of which answering even a few items in the wrong direction will reduce your reliability (by interfering with the correlation among items). Another is that it doesn't really solve the problem of acquiescence bias, which is why it is usually recommended (see the attached reference by two well known experts in survey design).
Hello,
Not sure so much about disadvantages, but a positive is you see if your respondent pays attention to the questionnaire content.
Regards, Witold
Thanks for all your contributions. David L. Morgan, thanks for sharing a great reference! Truly appreciate it.
There are advantages when persons answering the questions have a good understanding of the language the text is written. For those that do not have a good understanding of, say English, as it is their second language, the reverse coded question could be confusing and not valid as a check of their understanding. In Africa we often see this..
It is preferable don´t use these kind of reverse code items, because it implies to use usually negative sentences, and people dont feel confortable, and it is common that people are wrong to answer them.
Sometimes researchers need to use negatively worded statement to check on the outlier due to "straight liner" - those who responded in one direction regardless of the item statement.
I have always found the reverse coded items problematic in my research. In most of the cases, I had to exclude these items because the factor loadings of these items are usually poor (in CFA) and Cronbach's alpha increases significantly after removing these items.
As a rule of thumb, we must not confuse our respondents on our worded statement. However, I would like to know what is your objective on using the reverse coded items?
Madonna Sudio, some literature suggests that to prevent "straight line" answering, reverse coding could be used. But I am not sure whether it is worth doing because sometimes it may change the emotional state of respondents to hit a reverse coded question and get confused. Is there any reference favoring "not using" reverse coded questions?
Personally I prefer questions in a rigth worded, because items in negative worded cause confussion to the respondents; and if you dont make the data processing in the statistical software, probably you will have problems, because this kind of questons require special care.
Thanks everyone for this discussion.
I am using a previously validated scale in which some of the items are reversed. Does anyone have any suggestions on what to do in the instance that you suspect one participant has be 'straight line' answering? Would you remove all responses for that participant? Or just the reverse scored responses? Or leave the participant in?
I haven't been able to find anything in the literature on what to do if I suspect acquiescence bias, only how to avoid it.
Many thanks.
Roses,
I think you are accurately describing two options: treat the participant's answers as missing data or "reverse" their scoring. Because the second option requires a judgement on your part, I would prefer to treat it as missing data -- unless you have such a small N that it would affect your sample size.
I dont like reverses codes because they can generate confussion to respondents. It is preferable to use a unique code way to guarantee coherence and increase the level of assurance of right interpretation.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling