Designing a semi-structured interview guide for qualitative interviews

clipboard by wikidave https://www.flickr.com/photos/wikidave/7386337594

 

Interviews are a frequently used research method in qualitative studies. You will see dozens of papers that state something like “We conducted n in-depth semi-structured interviews with key informants”. But what exactly does this mean? What exactly counts as in-depth? How structured are semi-structured interviews?

 

The term “in-depth” is defined fairly vaguely in the literature: it generally means a one-to-one interview on one general topic, which is covered in detail. Usually these qualitative interviews last about an hour, although sometimes much longer. It sounds like two people having a discussion, but there are differences in the power dynamics, and end goal: for the classic sociologist Burgess (2002) these are “conversations with a purpose”.

 

Qualitative interviews generally differ from quantitative survey based questions in that they are looking for a more detailed and nuanced response. They also acknowledge there is no ‘one-size fits all’, especially when asking someone to recall a personal narrative about their experiences. Instead of a fixed “research protocol” that asks the same question to each respondent, most interviewees adopt a more flexible approach. However there is still a need “...to ensure that the same general areas of information are collected from each interviewee; this provides more focus than the conversational approach, but still allows a degree of freedom and adaptability in getting information from the interviewee” –MacNamara (2009).

 

Turner (2010) (who coincidentally shares the same name as me) describes three different types of qualitative interview; Informal Conversation, General Interview Guide, and Standardised Open-Ended. These can be seen as a scale from least to most structured, and we are going to focus on the ‘interview guide’ approach, which takes a middle ground.

 

An interview guide is like a cheat-sheet for the interviewer – it contains a list of questions and topic areas that should be covered in the interview. However, these are not to be read verbatim and in order, in fact they are more like an aide-mémoire. “Usually the interviewer will have a prepared set of questions but these are only used as a guide, and departures from the guidelines are not seen as a problem but are often encouraged” – Silverman (2013). That way, the interviewer can add extra questions about an unexpected but relevant area that emerges, and sections that don’t apply to the participant can be negated.

 

So what do these look like, and how does one go about writing a suitable semi-structured interview guide? Unfortunately, it is rare in journal articles for researchers to share the interview guide, and it’s difficult to find good examples on the internet. Basically they look like a list of short questions and follow-on prompts, grouped by topic. There will generally be about a dozen. I’ve written my fair share of interview guides for qualitative research projects over the years, either on my own or with the collaboration of colleagues, so I’m happy to share some tips.

 


Questions should answer your research questions
Your research project should have one or several main research questions, and these should be used to guide the topics covered in the interviews, and hopefully answer the research questions. However, you can’t just ask your respondents “Can the experience of male My Little Pony fans be described through the lens of Derridean deconstruction?”. You will need to break down your research into questions that have meaning for the participant and that they can engage with. The questions should be fairly informal and jargon free (unless that person is an expert in that field of jargon), open ended - so they can’t be easily answered with a yes or no, and non-leading so that respondents aren’t pushed down a certain interpretation.

 

 

Link to your proposed analytical approach
The questions on your guide should also be constructed in such a way that they will work well for your proposed method of analysis – which again you should already have decided. If you are doing narrative analysis, questions should be encouraging respondents to tell their story and history. In Interpretative Phenomenological Analysis you may want to ask more detail about people’s interpretations of their experiences. Think how you will want to analyse, compare and write up your research, and make sure that the questioning style fits your own approach.

 

 

Specific ‘Why’ and prompt questions
It is very rare in semi-structured interviews that you will ask one question, get a response, and then move on to the next topic. Firstly you will need to provide some structure for the participant, so they are not expected (or encouraged) to recite their whole life story. But on the other level, you will usually want to probe more about specific issues or conditions. That is where the flexible approach comes in. Someone might reveal something that you are interested in, and is relevant to the research project. So ask more! It’s often useful in the guide to list a series of prompt words that remind you of more areas of detail that might be covered. For example, the question “When did you first visit the doctor?” might be annotated with optional prompts such as “Why did you go then?”, “Were you afraid?” or “Did anyone go with you?”. Prompt words might reduce this to ‘Why THEN / afraid / with someone’.

 

 

Be flexible with order
Generally, an interview guide will be grouped into several topics, each with a few questions. One of the most difficult skills is how to segue from one topic or question to the next, while still seeming like a normal conversation. The best way to manage this is to make sure that you are always listening to the interviewee, and thinking at the same time about how what they are saying links to other discussion topics. If someone starts talking about how they felt isolated visiting the doctor, and one of your topics is about their experience with their doctor, you can ask ‘Did you doctor make you feel less isolated?’. You might then be asking about topic 4, when you are only on topic 1, but you now have a logical link to ask the more general written question ‘Did you feel the doctor supported you?’. The ability to flow from topic to topic as the conversation evolves (while still covering everything on the interview guide) is tricky, and requires you to:

 

 

Know your guide backwards - literally
I almost never went into an interview without a printed copy of the interview guide in front of me, but it was kind of like Dumbo’s magic feather: it made me feel safe, but I didn’t really need it. You should know everything on your interview guide off by heart, and in any sequence. Since things will crop up in unpredictable ways, you should be comfortable asking questions in different orders to help the conversational flow. Still, it’s always good to have the interview guide in front of you; it lets you tick off questions as they are asked (so you can see what hasn’t been covered), is space to write notes, and also can be less intimidating for the interviewee, as you can look at your notes occasionally rather than staring them in the eye all the time.

 


Try for natural conversation
Legard, Keegan and Ward (2003) note that “Although a good in-depth interview will appear naturalistic, it will bear little resemblance to an everyday conversation”. You will usually find that the most honest and rich responses come from relaxed, non-combative discussions. Make the first question easy, to ease the participant into the interview, and get them used to the question-answer format. But don’t let it feel like a tennis match, where you are always asking the questions. If they ask something of you, reply! Don’t sit in silence: nod, say ‘Yes’, or ‘Of course’ every now and then, to show you are listening and empathising like a normal human being. Yet do be careful about sharing your own potentially leading opinions, and making the discussion about yourself.

 

 

Discuss with your research team / supervisors
You should take the time to get feedback and suggestions from peers, be they other people on your research project, or your PhD supervisors. This means preparing the interview guide well in advance of your first interview, leaving time for discussion and revisions. Seasoned interviewers will have tips about wording and structuring questions, and even the most experienced researcher can benefit from a second opinion. Getting it right at this stage is very important, it’s no good discovering after you’ve done all your interviews that you didn’t ask about something important.

 

 

Adapting the guide
While these are semi-structured interviews, in general you will usually want to cover the same general areas every time you do an interview, no least so that there is some point of comparison. It’s also common to do a first few interviews and realise that you are not asking about a critical area, or that some new potential insight is emerging (especially if you are taking a grounded theory approach). In qualitative research, this need not be a disaster (if this flexibility is methodologically appropriate), and it is possible to revise your interview guide. However, if you do end up making significant revisions, make sure you keep both versions, and a note of which respondents were interviewed with each version of the guide.

 

 

Test the timing
Inevitably, you will not have exactly the same amount of time for each interview, and respondents will differ in how fast they talk and how often they go off-topic! Make sure you have enough questions to get the detail you need, but also have ‘lower priority’ questions you can drop if things are taking too long. Test the timing of your interview guide with a few participants, or even friends before you settle on it, and revise as necessary. Try and get your interview guide down to one side of paper at the most: it is a prompt, not an encyclopaedia!

 


Hopefully these points will help demystify qualitative interview guides, and help you craft a useful tool to shape your semi-structured interviews. I’d also caution that semi-structured interviewing is a very difficult process, and benefits majorly from practice. I have been with many new researchers who tend to fall back on the interview guide too much, and read it verbatim. This generally leads to closed-off responses, and missed opportunities to further explore interesting revelations. Treat your interview guide as a guide, not a gospel, and be flexible. It’s extra hard, because you have to juggle asking questions, listening, choosing the next question, keeping the research topic in your head and making sure everything is covered – but when you do it right, you’ll get rich research data that you will actually be excited to go home and analyse.

 

 

Don’t forget to check out some of the references above, as well as the myriad of excellent articles and textbooks on qualitative interviews. There’s also Quirkos itself, software to help you make the research process engaging and visual, with a free trial to download of this innovative tool. We also have a rapidly growing series of blog post articles on qualitative interviews. These now include 10 tips for qualitative interviewing, transcribing qualitative interviews and focus groups, and how to make sure you get good recordings. Our blog is updated with articles like this every week, and you can hear about it first by following our Twitter feed @quirkossoftware.

 

 

Starting out in Qualitative Analysis

Qualitative analysis 101

 

When people are doing their first qualitative analysis project using software, it’s difficult to know where to begin. I get a lot of e-mails from people who want some advice in planning out what they will actually DO in the software, and how that will help them. I am happy to help out individually, because everyone’s project is different. However, here are a few pointers which cover the basics and can help demystify the process. These should actually apply to any software, not just Quirkos!

 

First off: what are you going to be able to do? In a nutshell, you will read through the sources, and for each section that is interesting to you and about a certain topic, you will ‘code’ or ‘tag’ that section of text to that topic. By doing this, the software lets you quickly see all the sections of text, the ‘quotes’ about that topic, across all of your sources. So you can see everything everyone said about ‘Politics’ or ‘Negative’ – or both.

 

You can then look for trends or outliers in the project, by looking at just responses with a particular characteristic like gender. You’ll also be able to search for a keyword, and generate a report with all your coded sections brought together. When you come to write up your qualitative project, the software can help you find quotes on a particular topic, visualise the data or show sub-section analysis.  

 

So here are the basic steps:

 

1.       Bring in your sources.
I’m assuming at this stage that you have the qualitative data you want to work with already. This could be any source of text on your computer. If you can copy and paste it, you can bring it into Quirkos. For this example let’s assume that you have transcripts from interviews: this means that you have already done a series of interviews, transcribed them, and have them in a file (say a Word document or raw text file). I’d suggest that before you bring them in, just have a quick look through and correct them in a Word Processor for typos and misheard words. While you can edit the text in Quirkos later, while using a Word or equivalent you have the advantage of spell checkers and grammar checkers.

 

Now, create a new, unstructured project in Quirkos, and save it somewhere locally on your computer. We don’t recommend you save directly to a network location, or USB stick, as if either of these go down, you will have a problem! Next, bring in the sources using the (+) Add Source button on the bottom right. You can bring in each file one at a time, or a whole folder of files in one go, in which case the file name will become the default source name. Don’t forget, you can always add more sources later, there is no need to bring in everything before you start coding. Now your project file (a little .qrk file you named) will contain all the text sources in one place. With Quirkos files, just backing up and copying this file saves the whole project.

 


2.       Describe your sources
It’s usually a good idea to describe some characteristics of your qualitative sources that you might use later to look for differences or similarities in the data. Often these are basic demographic characteristics like age or gender, but can also be things about the interview, such as the location, or your own notes.

 

To do this in Quirkos, click on the little grid button on the top right of the screen, and use the source properties. The first thing you can do here is change the name of the sources from the default (either a sequential number like ‘Source 7’ or the file name. You can create a property with the square [+] ‘Quickly add a new property’ button. The property (eg Gender) and a single value (eg Male) can be added here. The drop down arrow next to that property can be used later to add extra values.

 

The reason for doing this is that you can later run ‘queries’ which show results from just certain sources that have properties you defined. So you can do a side-by-side comparison of coded responses from men next to women. Don’t forget, you can add properties at any time, so you can even create a variable for ‘these people don’t fit the theory’ after you’ve coded, and try and see what they are saying that makes them different.

 

 

3.       Create your themes
Whatever you call them: themes, nodes, bubbles, topics or Quirks, these are the categories of interest you want to collect quotes about from the text. There are two approaches here, you can try and create all the categories you will use before you start reading and coding the text (this is sometimes called a framework approach), or you can add themes as you go (grounded theory). (For much much more on these approaches, look here and here.)

 

In Quirkos, you create themes as coloured bubbles, which grow in size the more text is added. Just click on the grey (+) button on the top right of the canvas view to add a new theme. You can also change the name, colour, level in this dialogue, or right click on the bubble and select ‘Quirk Properties’ at any time. To group, just drag and drop bubbles on top of each other.

 

 

4.       Do your coding
Essentially, the coding process involves finding every time someone said something about ‘Dieting’ and adding that sentence or paragraph to the ‘Dieting’ bubble or node. This is what is going to take the most time in your analysis (days or weeks) and is still a manual process. It’s best to read through each source in turn, and code it as you go.

 

However, you can also use the keyword search to look for words like ‘Diet’ or ‘eating’ and code from the results. This makes it quicker, but there is the risk of missing segments that use a keyword you didn’t think to search for like ‘cut-down’. The keywords search can help when you (inevitably) decide to add a new topic halfway through, and the first few interviews haven’t been coded for the new themes. You can use the search to look for related terms and find those new segments without having to go over the whole text again.

 

 

5.       Be iterative
Even if you are not using a grounded theory approach, going back over the data a second time, and rethinking codes and how you have categorised things can be really useful. Trust me: even if you know the data pretty well, after reading it all again, you will see some topics in a slightly different light, or will find interesting things you never thought would be there.

 

You may also want to rearrange your codes, especially if you have grouped them. Maybe the name you gave a theme isn’t quite right now: it’s grown or got more specific. Some vague codes like ‘Angry’ might need to be split out into ‘Irate’ and ‘Annoyed’. Depending on your approach, you  will probably constantly tweak and adjust the themes and coding so they best represent the intersection of your research questions and data.

 

 

6.       Explore the data.
Once your qualitative data is all coded, the big advantages of using CAQDAS software come into play. Using the database of your tagged text, you can choose to look at it in anyway: using any of the source properties, who did the coding or when, or whether a result comes from any particular group of codes. This is done using the 'Query' views in Quirkos.

 

In Quirkos there are also a lot of visualisation options that can show you the overall shape and structure of the project, the amount of coding, and connections that are emerging between the sources. You can then use these to help write your outputs, be they journal articles, evaluations or a thesis. Software will generate reports that let you share summaries of the coded data, and include key statistics and overviews of the project.


While it does seem like a lot of work to get to this stage, it can save so much time at the final stages of writing up your project, when you can call up a useful quote quickly. It also can help in the future to have this structured repository of qualitative data, so that secondary analysis or adding to the dataset does not involve re-inventing the wheel!

 

Finally, there is no one-size-fits-all approach, and it's important to find a strategy that fits with your way of working. Before you set out, talk to peers and supervisors, read guides and textbooks, and even go on training courses. While the software can help, it's not a replacement for considered thinking, and you should always have a good idea about what you want to do with the data in the end.

 

 

How to set up a free online mixed methods survey

It’s quick and easy to set up an on-line survey to collect feedback or research data in a digital format that means you can quickly get straight to analysing the data. Unfortunately, most packages like SurveyMonkey, SurveyGizmo and Kwiksurveys, while all compatible with Quirkos, require a paying subscription before you can actually export any of your data and analyse it.

 

However, there are two great free platforms we recommend that allow you to run a mixed-method survey, and easily bring all your data into Quirkos to explore and analyse. In this article, we'll go through a step by step guide to setting up a survey in eSurv, and exporting the data to Quirkos

 

eSurv.org

This is a completely free platform, funded by contributions from Universities, but is available for any use. There are no locked features, or restrictions on responses, and it has an easy to use on-line survey design. There are customisable templates, and you can have custom exit pages too.

Once you have signed up for an account, you will be presented with the screen above, and will be able to get going with your first survey. The first page allows you to name the survey, and set up the title and page description, all have options for changing the text formatting. Just make sure you click on the verification link in the e-mail sent to you, which will allow you to access all the features.

 

The next screen shows a series of templates you can use to set the style of your survey. Choose one that you like the look of, and you have the option of customising it further with your logo or other colour schemes. Click next.

Now you are ready to start adding questions.

 

The options box on the right shows all the different types of questions available, and each one has many customisation options at the bottom of the screen. For example, the single text box option can be made to accept only numerical answers, and you can change the maximum length and display size of the box. All questions can be made mandatory, with a custom 'warning' if someone does not fill in that dialogue.

 

The drag and drop ranking feature is a nice option, and pretty much all the multiple-choice and closed question formats you might want are represented.

 

When you have chosen the title and settings for each question, you can click on the 'Save & Add Next' button on the top right to quickly add a series of questions, or 'Save & Close' if you are done.

 

There are also Logic options to add certain questions only in response to certain answers (for example, Please tell us why you didn't like this product). It is of course possible to edit the questions and rearrange them using the drag icon in the main questionnaire overview.

 

You can test the survey to see how it looks, and when happy click the launch button to make it available to respondents. This also gives you a QR code linking to the survey, allowing smartphone users to complete the survey from a link on posters or printed documents. While you can customise the link title, the web address is always in the format of "https://eSurv.org?u=survey_title".

 

You can have a large number of surveys on the go at once, and manage them all from the 'Home' screen, which also shows you how many responses you have had.

 

Once you are ready to analyse your data, open the survey and click on the export button. This gives the options above to select which questions and respondents you want to export, and a date range (useful if you only want to put in new responses). For best use in Quirkos, select the Compact and .csv File format options, and then click download.

 

exported csv file in excel

The only step you probably want to take before bringing the data into Quirkos is to remove the first row (highlighted above). By default eSurv creates a row which numbers the questions, but it’s usually easier to have the questions themselves as the title, not just the number. Just delete the first row starting with ‘Question’ and this will remove the question numbers, and Quirkos will see the first row with the actual question names. Just save any changes in Excel/LibreOffice making sure you save using the CSV (Comma delimited) format, and ignore the warning that ‘some features may be lost’ and choose ‘Yes’ to keep using that format. You can also remove any columns here that you don’t want (for example e-mail address if it was not provided) but you can also do this in Quirkos.

 

In Quirkos, start a new Structured Questions project, and select the Import from CSV option from the bottom right 'Add Source' (+) button. Select the file you saved in the previous step, and you will get a preview of the data looking like the screenshot above. Here you have the option to choose which question you want to use for the Source Title (say a name, or respondent ID) and any you might want to ignore, such as IP address. Then make sure that open ended questions are selected as Question, and Property is associated with any discrete or numerical categories. Click import, and voilà!

 

Should you get new responses, you can add them in the same way to an existing project with the same structure, just make sure when exporting from eSurv that you select the newest responses to export, and don't duplicate older ones.

 

Now you can use Quirkos to go through and code any of the qualitative text elements, while using the properties and quantitative data  to compare respondents and generate summaries. So for example, you can see the comments that people with negative ratings made side by side by comments from positive feedback, or compare people from different age ranges.

 

If you need even more customisation of your survey, the open-source platform LimeSurvey, while not as easy to use as eSurv, gives you a vast array of customisability options. LimeService.com allows 25 responses a month for free, but we have our own unrestricted installation available free of charge to our customers – just ask if you need it!

 

p.s  I've also done a video tutorial covering setting up and using eSurv, and exporting the results into Quirkos.