Question
Asked 1st Jan, 2016

How to create a quantitative research questionnaire correctly?

I just creating a questionnaire about relationship between two variables for quantitative research. Is the questionnaire parted by each variable or every single it's question have to representation of the relationship between two variables?
For example, the title is: relationship between knowledge management (KM) with librarian competencies.
The question from librarian competencies indicators: I could operate the computer well.
Should I change into: I could operate the computer well because I did KM.
Just like it or not?

Popular Answers (1)

1st Jan, 2016
Han Ping Fung
HP Inc.
How to create a quantitative research questionnaire correctly?
Before creating your own quantitative questionnaire representing a variable or construct e.g. KM or Librarian Competencies etc. - try to do rigorous literature review from academic articles / theses etc. on what are the popular / established questionnaire already available because you don't want to "re-invent the wheel".  Reason being established questionnaire had been used in many research & it has certain validity & reliability.  What you need to do is to get permission from the original questionnaire owner to adopt / adapt the questionnaire for your research's usage.  Some renown questionnaire you might need to purchase.
Only after rigorous literature review & there is no quantitative questionnaire fits your needs, then you can develop your own questionnaire.  Some considerations include:
  1. Ensure you develop your operational definition first for each variable & construct.
  2. Use simple language & words for each questions & when all the questions group together should referring to one variable / construct.
  3. Ensure there is no double / multi-barrels question i.e. a question ask more than 1 thing that respondents are confused not sure which thing the researcher is asking & when they responded, the researcher not sure which thing the respondents are answering (because too many things are asked in 1 question).
  4. Ensure you use formative or reflective questions as appropriate to represent a variable or construct - formative questions are several questions in which each has its own unique attribute / characteristic & when all questions group together to form / represent the variable.  Reflective questions refer to several questions whereby each question is reflecting a variable from different angle for several times.  Reason being formative / reflective questions can affect what data analysis modeling you need to use e.g. Partial Least Squares-Structural Equation Modeling (PLS-SEM) vs Covariance-based SEM etc.
  5. Since it is a new questionnaire developed, you need to do a pilot test to evaluate its Cronbach Alpha / Composite Reliability etc.
  6. Perform Exploratory Factor Analysis (EFA) on the variable / construct so that all factors generated are mapping to your operational definition e.g. if your operational definition for a construct consists of 3 attributes, there should be 3 factors surfaced after the EFA.
Is the questionnaire parted by each variable or every single it's question have to representation of the relationship between two variables?
Each question within a group of questions should focus on each variable.  Each question shouldn't link up 2 variables together i.e. questions should be "decoupled" / grouped easily & only represent a variable - that's the purpose we do EFA.  Also when questions are not linked up the 2 variables, the questions within a variable will be good to test on the relationship between the same variable with other variable which might outside your current research scope.
Sorry for the long answer.  Wishing you all the best.
11 Recommendations

All Answers (10)

1st Jan, 2016
Allah Nawaz
Gomal University
Use the existing research for extracting the variables and attributes of each variable relating to your topic. Then ask a question about each attribute using one or another level of measurement.
1 Recommendation
1st Jan, 2016
Han Ping Fung
HP Inc.
How to create a quantitative research questionnaire correctly?
Before creating your own quantitative questionnaire representing a variable or construct e.g. KM or Librarian Competencies etc. - try to do rigorous literature review from academic articles / theses etc. on what are the popular / established questionnaire already available because you don't want to "re-invent the wheel".  Reason being established questionnaire had been used in many research & it has certain validity & reliability.  What you need to do is to get permission from the original questionnaire owner to adopt / adapt the questionnaire for your research's usage.  Some renown questionnaire you might need to purchase.
Only after rigorous literature review & there is no quantitative questionnaire fits your needs, then you can develop your own questionnaire.  Some considerations include:
  1. Ensure you develop your operational definition first for each variable & construct.
  2. Use simple language & words for each questions & when all the questions group together should referring to one variable / construct.
  3. Ensure there is no double / multi-barrels question i.e. a question ask more than 1 thing that respondents are confused not sure which thing the researcher is asking & when they responded, the researcher not sure which thing the respondents are answering (because too many things are asked in 1 question).
  4. Ensure you use formative or reflective questions as appropriate to represent a variable or construct - formative questions are several questions in which each has its own unique attribute / characteristic & when all questions group together to form / represent the variable.  Reflective questions refer to several questions whereby each question is reflecting a variable from different angle for several times.  Reason being formative / reflective questions can affect what data analysis modeling you need to use e.g. Partial Least Squares-Structural Equation Modeling (PLS-SEM) vs Covariance-based SEM etc.
  5. Since it is a new questionnaire developed, you need to do a pilot test to evaluate its Cronbach Alpha / Composite Reliability etc.
  6. Perform Exploratory Factor Analysis (EFA) on the variable / construct so that all factors generated are mapping to your operational definition e.g. if your operational definition for a construct consists of 3 attributes, there should be 3 factors surfaced after the EFA.
Is the questionnaire parted by each variable or every single it's question have to representation of the relationship between two variables?
Each question within a group of questions should focus on each variable.  Each question shouldn't link up 2 variables together i.e. questions should be "decoupled" / grouped easily & only represent a variable - that's the purpose we do EFA.  Also when questions are not linked up the 2 variables, the questions within a variable will be good to test on the relationship between the same variable with other variable which might outside your current research scope.
Sorry for the long answer.  Wishing you all the best.
11 Recommendations
1st Jan, 2016
A.H. Sequeira
National Institute of Technology Karnataka
  • Conduct an in-depth related  literature review after posing important relevant research questions.
  • Draw a literature tree if possible ; Identify the research gaps to focus on  the research area/ title .
  • Write the research objectives to fill the gaps .Write the operational definitions for the key terms .
  • Identify the variables and the levels of measurement . Identify the scales if opinions or attitudes are involved.
  • Draw the constructs to get clarity wherever important intangible factors /concepts  are involved. Include a  mix of scales and also different levels of measures for capturing the  variables .
  • Frame a few hypotheses based on self experience or related literature review or pilot study , for testing .
  • Draw a conceptual frame work mapping the research questions/objectives and hypotheses on the framework if possible.
  • Check for validity and reliability of the research tool and for internal consistency of the items through a pilot test .
  • Now compile  the  questions which capture the variables  in a logical  way to arrive at the  final research questionnaire for quantitative research.
4 Recommendations
1st Jan, 2016
Thoriq Tri Prabowo
Universitas Islam Negeri Sunan Kalijaga Yogyakarta
Thank You Mr. Allah Nawaz, Mr. Han Ping Fung, and Mr. Aloysius Sequeira.
Mr. Han Ping Fung you answered what I need to know, it was very helpful. If you have some paper that explain your last answer about the 'focus of question' I need it. My Thesis Supervisor ask it to show on my research method thesis, thank You.
1st Jan, 2016
Han Ping Fung
HP Inc.
Hi Thoriq,
Group of questions should only focus on a variable / construct is the logic behind Construct Validity.  Construct Validity consists of Convergent Validity (related data of questions within a construct should converge / focus / high correlated) & Discriminant Validity (related data of questions from one variable should be different / uncorrelated with questions' data from another variable).  You can find out more about Construct, Convergent & Discriminant Validity from typical books on research method or quantitative data analysis.
2 Recommendations
1st Jan, 2016
Thoriq Tri Prabowo
Universitas Islam Negeri Sunan Kalijaga Yogyakarta
Thank You Mr. Han, then I looking for the paper/book that relate with it.
1st Jan, 2016
Avishag Gordon
Technion - Israel Institute of Technology
Hi,
There are many tools on the net that can advise you on how to build your questionnaire. It is also advisable to approach the online software "SurveyMonkey" and if you subscribe to this tool, they could help you  build the right questionnaire for your research.
1 Recommendation
1st Jan, 2016
David L Morgan
Portland State University
If you have two separate concepts (KM and competency), then you need to have separate measures for each of them. That is necessary to test whether they are related to each other or not.
As others have suggested, you want to have more than one item to measure each concept. A common approach here is to create a scale out of multiple, related items. For competencies, they could be Yes/No questions about their ability to do things, or Likert-scored items about how comfortable they feel in doing those activities (for example, 5 scores that range from "Not all" to "Very Much").
2nd Feb, 2016
Allah Nawaz
Gomal University
The questionnaire must be GROUNDED in the literature. It means that in every research the researcher begins with the existing knowledge on the topic in the form of Books, Research Papers and other secondary sources. There are multiple purposes of this practice because literature review is what creates the SPADEWORK or Foundation of all the forthcoming research activities however following are main purposes of this PRELIMINARY STUDY:
1. Extraction of WORKING CONCEPTS [the VARIABLES and ATTRIBUTES] of each variable.
2, The RELATIONSHIP between the variables.
These Variables, Attributes and Relationships are then used to develop a THEORETICAL FRAMEWORK, Which can be presented textually as well as in the form of a SCHEMATIC DIAGRAM.
This then works as the guideline for further data collection as well as analysis.
It should however be clear that its time consuming and intellectually very demanding activity to locate and pick the most relevant working concepts (both variables and their respective attributes). So .get ready for a rigorous activity.
Hope it Helps....
2nd Feb, 2016
Khandoker Mahmudur Rahman
United International University
Dr. Han Ping Fung,
Good summary to save for future reference for any researcher!
In addition, I would like to refer to scale and questionnaire adaptation under international context. Literature oftentimes reports a level of internal consistency for a particular scale, which, when converted to another language (even with the help of experts), returns a different level of internal consistency, sometimes substantially lower in particular context (traditional cronbach alpha measure). What I found that besides translation by language experts, researchers must pre-test the same scale multiple times with a sample of actual respondents and adapt language for clarity and face validity (this is a standard procedure and nothing new about it). This helps to refine scale item language in a such a way that it measures what it purports to measure (validity). Oftentimes, I found that certain changes in linguistic expression is needed to accurately measure certain constructs, and such linguistic adaptation might well be  opposed by language experts, saying that the expression might be grammatically incorrect from that particular language context. What  I am trying to say is this: while the theoretical pursuit of linguistic accuracy is appreciated, at the same time, the researchers have to make a focused decision whether he/she is actually measuring what he/she is intending to measure. We oftentimes get occupied with numbers on papers and process through software so that the software tells us whether the parameter is within acceptable limit based on pre-defined criteria or  not. How about face validity that is more of a judgement by an informed and rational researcher even before starting data collection?
In fact, these issues will be taken care of during steps 1,2 and 5 of your answer. Again, thanks for sharing.
3 Recommendations
Can you help by adding an answer?