I wondered if I could slightly change the wording of the question in this questionnaire (Classroom Appraisal of Resources and Demands - CARD; Lambert, McCarthy & Abbott-Shim, 2001) to measure teacher stress. Roeser et al (2013) measured teachers' self-reported job stress from CARD's longer inventory of teacher stress.
From 'how demanding your classroom or teaching responsibilities are in these areas?' to 'Do you find the following classroom or teaching responsibilties stressful/demanding?'
Sample statements are paperwork completion, children with problem behaviours (1= not demanding, 5 = extremely demanding).
In Roeser's paper, they wrote 'items tapped teachers level of agreement that particular aspects of their work "puts a lot of stress on me".
SCORING
In CARD, they have two scales resources and demands, and they work out the stress score by determining the difference between the two. As I am only using one item from the resource scale, I was thinking of calculating the stress score by '(sum of items from demand scale) - (item from resource scale)/mean of all items i.e. from both resource and demands scale.
I hope all is clear - thanks so much!
Lydia - yes - you are at liberty to 'adapt' existing questionnaires. Small adjustments should not make much difference - but there is always the risk that any modification can potentially alter the instrument's accuracy and validity - especially if the instrument is well 'tried and tested'. Altering 'wording' would pose a lesser risk than changing the scoring system. It would need your (or colleagues) expert insight as to whether that might constitute piloting and/or re-testing of the instrument - or not.
I agree with Dean - one main issue is *framing* when you change the wording of the question you also change the frame of the answer. In this case the difference is potentially very big and might have a serious impact on the answers.The original question is inviting the respondent to think about demand (where stress is expressed voluntarely) - however if you *mention* stress in the question - you lead the respondent to think about *stress* explicitly - this then is not any more the same question. It is not simply a re-wording of the same question - because it is in-fact a completely different question due to change of framing. As such the results may not be comparable without a significant exploration of what the change of framing means in the particular context - so it would *not* be straightforward to compare with previous investigations.
So the question is *why* do you want to change the question? Because you can (obviously) change the question *on purpose* where the change of framing is intentional (e.g. - you want to lead the participants especially to think about stress and *not* on other potential experiences of demand. Then it would be OK - but you would need to justify this change of agenda. IF you wish to discuss *stress* in relation to other demands it may not be a good idea to frame the question to highlight stress to the participants -= as it is a leading question and you might be limiting their responses unwittingly. It is dependent on what your agenda is with your research. If you want to discover stress as experience among other - because your framing is exclusive and not promoting other - the responses could become irrelevant for your investigation.
forgot to mention - lack of demand is also stressful for many people (e.g.boredom). SO there is not always a direct correlation - it depends on the investigation and context.
For the portion of your questions related to calculating the difference between two aspects of your scale, this is actually a fairly primitive approach that limits the results you can get. The preferred approach would be to enter both scales into the same regression equation, and determine what the effect of each is, after controlling for the other.
In your case, you could simply control for the single item that you using.
Rewording should not result in leading to an answer bias, as it is predicted here by other members. Questionnaire pretesting may be done before administering among respondents.
*Rewording should not result in leading to an answer bias* >> I do not know what *answer bias* could possibly mean. However *every* question is *always* framed and thus biased. By definition there is a purpose with the question. So the wording is influenced by the purpose and intention that the researcher has with the question, what they want to achieve with it. This also means that if the researcher wants to change the question (rewording) there is a purpose and reason for this change. In other words *the bias of the question is intentionally changed*. So bias is always involved - it is also dependent on setting, tone, style, environment - in other words both physical, emotional and socio-cultural context this includes conscious and subconscious bias (and framing). Whether or not any framing is introducing *unintentional* bias which is not taken into consideration by the researcher is the real issue that matters here.
Thanks guys! I decided not to reword it.
I also wanted to ask if you could randomise the questions of the survey though to make it less obvious.
For example, in the Maslach Burnout Inventory, they have different sections and I think it's very clear what it is asking - it goes from different scales 1. Emotional exhaustion, 2. Depersonalisation, 3. Personal accomplishment. Is it possible to randomise the statements so they are in different orders?
I personally favor what is known as "transparent ordering" where you do keep the questions grouped according to their sub-scales. The practical effect is that people understand what the clusters of questions are about and thus answer them more consistently.
The opposite approach is to worry about "response bias" which conceptualizes answering questions according to a consistent perception of the topic as a problem rather than a benefit. One thing is certain, however, which is that the more inconsistent the responses are, the lower the reliability will be.
I think you have already decided, but one more reason not to change the wording is that answering your research question would depend on your subjects all understanding the word "stress" to mean the same thing as you do. This is unlikely, so better not to introduce the problem. The more measure items are grounded in terms that are clear and understood to mean the same thing across people, the better. The original item has the problem that people may understand "demanding" in different ways. For some who like challenges, demanding may have a positive connotation, while others think of "demanding" as something that is "too much" for them. Variations in interpretation of key terms add error into responses.
Hello Lydia,
Rewording of the well established questionnaire is always a "risky business". If this is exercise that is ok, but in terms of publishing that would be a limitation in your research.
Good luck with your work.
Dear Lydia,
You can, because what you want measure is different from what the original question investigates. However you need to test the validity and reliability of the new scale/questionnaire.
Let us know!
Anna Maria
Lydia,
I believe that you can, but at a risk of altering the instrument. I strongly encourage you to consult with the test developer(s) first.
It will be interesting to know what the developer(s) say.
Good luck
Enobong Inyang
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning