|In the past 24 hours 2716 questions|
have been answered.
Take part in online surveys on current topics.
Survey tip 9: Make sure that the context of a question does not have an (uncontrolled) effect on the answer!We now come to the rule which in practice causes the most difficulties and is least controllable. The fact that questions and their corresponding answers can have an effect on follow-up questions is indisputable and fully proven. But which questions have an effect on follow-up questions? The answer is that we can often only speculate about this when formulating questionnaires. Only a pretest or in the worst case scenario only the data of the questionnaire provide information, provided that mechanisms (e.g. different versions of the questionnaire with different preliminary questions inserted before the question you are interested in)have been built into the questionnaire, which can control context effects.
We can demonstrate this with an example. In order to do this we must make an exception and give a reference, the following example can be found in Schwarz & Bless (1992)1:
"What is your general opinion of the CDU (Christian Democratic Union)?"
with an answer scale from 1 = "very negative" to 11 = "very positive"
we obtain quite different average values depending on the preliminary question:
A) "Do you know which position Richard von Weizsäcker holds outside party politics?" - average value 3.4
B) Preliminary question without political content – average value 5.2
C) "Do you know which party Richard von Weizsäcker has been a member of for over 20 years?" - average value 6.5
Depending on whether the well-known and popular former President of the Federal Republic of Germany Richard von Weizsäcker is excluded from the CDU(variation A) or included in the CDU (variation C) the CDU is rated with or without Weizsäcker – with a remarkably different result. The example is persuasive, but it is elaborately constructed and cannot be generalised. Context effects in questionnaires can be anticipated by reflection, but can only be proven by systematic tests.
Make sure when designing your questionnaire that some questions do not influence other questions; in case of doubt a systematic test is advisable.
1: Schwarz, Norbert & Herbert Bless (1992): Assimilation and Contrast Effects in Attitude Measurement: An Inclusion/Exclusion Model. P. 72 - 77 in: Advances of Consumer Research 19
|10 survey tips|
Formulation of questions
• Survey tip 1
• Survey tip 2
• Survey tip 3
• Survey tip 4
• Survey tip 5
• Survey tip 6
• Survey tip 7
• Survey tip 8
• Survey tip 9
• Survey tip 10
• Book tip
Q-Set.de Germany | Q-Set.ch Switzerland | Q-Set.co.uk United Kingdom | Q-Set.it Italy | Q-Set.fr France
Q-Set.es Spain | Q-Set.nl Netherlands | Q-Set.be Belgium | Q-Set.li Liechtenstein | Q-Set.cz Czech Republic
Q-Set.pl Poland | Q-Set.eu Europe | Q-Set.tk Tokelau | Q-Set.ru Russia | Q-Set.in India | Q-Set.us USA
© 2008-2019 Q-Set.at
Last updated: 07.12.2019