Temporary Page
Ray Poynter’s NTU Lecture
28 November 2018


I have pasted the notes from the online discussion platform at the bottom of this page. At the moment they are quite rough and I need to check my spelling and typing.

Welcome to this virtual lesson with Ray Poynter and NTU. The format is experimental, so we welcome any feedback. The flow of the lesson is.

  1. Take two short surveys, to get a sense of where the lesson is heading
  2. Watch the videos listed below
  3. Complete the practical task (creating your own survey, collecting data, and exporting it to Excel and/or SPSS)
  4. Log in to the online discussion about the lesson to ask questions and hear feedback from other students.

You will need a Qualtrics account, so please set one up, you can download information about how to create an account by clicking here.

Take these two surveys

Watch these videos

You can download a PDF copy of the “Introduction to the Lesson” PowerPoint deck by clicking here.

Practical Task
Using the Qualtrics account:

  1. Design your own survey – you can base it on one of mine
  2. Share the links with classmates to gather some data
  3. Check and review your data in Qualtrics
  4. Export the data to Excel or SPSS

Discussion and Q&A
Register for the Discussion and Q&A by clicking here. The Q&A session will be hosted on the Zoom video-conferencing platform.

Once you register you can join the session any time between 9am and 1pm on Wednesday 28 November. You can watch the videos and do the practical task before the 28 November, or you can do them during the session.

If you have any questions or comments, email Ray.Poynter@NewMR.org

Here the questions and answers from the virtual session.

Question Answer
How are the discursive survey dissimilar to qualitative survey? Both require in-depth responses which are quite subjective as well. Good question. “Real” qualitative is never coducted with a survey. For a focus group, or a depth interview, or an online discussion, we would have a discussion guide, but that is only a list of topics, initital questions, and probes. A discursive or semi-structured survey is half-way between a quant survey and “Real qualitative”
That is, to say, quali surveys have prompts to encourage the “right” answers from respondents? Yes, you put in open-ended questions and follow up questions.

The next tool that is being developed are Chatbots which will use artificial intelligence to know what question to ask next, and when to move on to the next project.

Thank you! I’m assuming that one isn’t better than the other, seeing as both have their advantages.

Say that our respondents will consist mostly of students, and we know many are quite opinionated when it comes to surveys, would you recommend discursive over quali?

For your project, a discusrive survey will probably be easier to manage. But you are correct, they all have their advantages. Qual where people can interact verbally with you can get deeper, but it can produce a large amount of material that needs to be analysed.
I’m aware that different research topics are better suited to different research methods, however does would discurvsive surveys work when trying to guage participants feelings towards a brand’s identity? Or would more in depth qualititative research be better suited? If you have the time and resources, conventional qual will be better than a discursive survey. The key benefits of the discursive survey is speed and convenience. If you have 10 conversations with people, you will have different amounts of information form each person, they may follow different paths through the conversations. From an insight point of view this is good.

However, with a discursive survey your data is already organised into chunks, every person follwowed a similar route through the conversation. So processing is faster/easier.

We are doing qualitative research about how consumer actually perceive the brand image, by indepth interviewing technique. Should we have a specific target customer like age group? Yes, you need to define who you should be interviewing. This starts with the business question. If you want to understand how loyal users of the brand can be persuaded to use it more often, you need to speak to loyal uses. If you want to understand how to persuade lapsed users to start using the brand again you need to speak to lapsed users.
I understand that quantitative data will be easier to manage but due to how the respondents will mainly be students, would it be okay to include possibly one open ended qualitative question to broaden our results and give greater quality of analysis later on in the review? Definitely! I would always include at the end of a survey “Is there anything else you would like to tell us about this topic or this survey?” and it is a good idea to ask a question such as “Why did you give it that score?” after a numeric evaluation. Or, “What is the main thing you like about this brand?” or something similar.
Can you give me some samples of conceptual framework in qualitative research? It is best to look these up in the literatue. One exmaple would includ “Jobs to be done” this is quite popular at the moment, it assumes that everything we buy/choose is there to do a job. So, if somebody buy a bright red sports car, what is the job to be done? the job is not just to get to and from work, it is to create an image, what is the image, how does that car help meet that image. Jobs to be done can be applied to small things like cup of coffee, or big things cars, holidays, houses etc.

Other frameworks will look at things like needs, transactional analysis etc. A nice quote about conceptual frameworks is “the system of concepts, assumptions, expectations, beliefs, and theories that supports and informs your research”- if you search the library you should find the source of this quote and a discussion of the topic.

how can we decide on the wording and the scale for measuring our research questions. In general it is best to use scales and questions that other people have usd. For example the five point agree-disagree Likert scale. This way you reduce the chance of mistakes. You can search published sutdies for scales. I will post the name of a good book on questionnaire design, it should be in the library I hope. Questionnaire Design : How to Plan, Structure and Write Survey Material for Effective Market Research
by Ian Brance is a good book for questionnaire design.
how many questions should I have in the questionnaire? There is no correct number of questions. In general, there is a feeling that most surveys are too long, which is reducing the cooperation of people with surveys, and data from later in the interview is often less reliable. But, you need to ask enough questions.

A good way to check a question is to make two lists. List A, the research questions. List B, the survey questions. You want to make sure that there are no items in List B that do not link to List A. You need to make sure that there are no items in List A that do not have something in List B. A more important point is probably “How long does the survey take to complete?” Think about how long you would be prepared to spend taking somebody else’s survey? 5 to 10 minutes is good, 12 to 18 is possible, but 20 or more minutes is likely to cause problems with people not wanting to complete it – especially if you are not paying them.

What are some of the best practices to distribute survey online? Would you recommend purchasing respondents? If you have budget, then the most common way to find sample is from a market research sample provider, for example YouGov, Lightspeed or ResearchNow. However, for a university project, it is normally necessary to find free sources, such as friends, family or fellow students etc (i.e. a convenience sample).

I would not advice buying sample from a list broker, these lists are used for marketing and tend to get a very negaitve response.

It is possible to find sample by buying advertising, but that is likely to expensive.

One option that is popular in academic circles is to use Amazon’s Mechanical Turk (low paid workers who do a range of tasks online). I have seen examples of very bad samples being collected this way – because nobody with sample management skills is looking after the project. So I would not recommend Mechnanical Turk.

I am wondering about the survey’s starting question, according to my project research, should we suppost to start from specific question. For example, I want to investingate about the most way peoeple know Virgin Media, thus, I have set the first question to be “How do you know about Virgin Media?” and the option of my question is “Official Site/ Banner/ Offline Advertising/ Friends/ Social Media” Is this way applicable to our researching? Great question. One of the problems with quantitative research is that people will often answer your questions, even if it does not make sense.

In this case, imagine that you have had a virgin media account for 10 years. In all hoensty, you probably do not know how you first became aware of virgin media – but you will try to help by answering the question.

Also, in most research situations you do not ask the key question as the first question, you try to warm the participant up first.

So maybe we do the following:
1) Which of the following media companies are you aware of (add a type in other), inlcude a “None of these” 2) Which of these are you currently using – only ask from the list of aware of
3) If Virgin was on the list from question 1, How did you first become aware of Virgin Media (use a select all that apply question), include a type in other, and include a “Can’t remember / Don’t know” option. Also, if somebody is not aware of Virgin media at the first question, either a) politely terminate the interview or b) ask the demographics, but not the media questions.

According to the target audience of our survey, could we ask international students to answer our questionaires or just NTU student able to be included in the survey? You are always trying to balance you sample with your target group. Also, for you project you want to make life as straighforward for yourself as possible. Feel free to choose eitehr group, whichever will be the easiest to collect, but modify the definition of your target group to match you decision.