Convergent interviewing: a technique
for qualitative data collection 1
This is a resource file which supports the regular public program "areol" (action research and evaluation on line) offered twice a year beginning in mid-February and mid-July. For details email Bob Dick firstname.lastname@example.org or email@example.com
... in which a form of interviewing is described in some detail. The interviewing combines some of the features of structured and unstructured interviews, and uses a systematic process to refine the information collected
Convergent interviewing is a technique you can use to gather information. Although it has many uses, it is most valuable when you are in some doubt about the information which is to be collected. Also, if you intend to use surveys to collect information, convergent interviewing can help you to decide what questions to ask in the survey.
You could summarise convergent interviewing as follows, with only a little over-simplification:
First put the person at ease. When you've established rapport, ask a single, broad question. Then keep the person talking for as long as you can, about one hour or a little longer. Then and only then ask any specific questions.
Conduct a pair of interviews (preferably by different interviewers). Compare the themes which emerged from each. If the two informants agreed on a theme, in later interviews probe for disconfirming views. If they disagreed about some topic, in later interviews probe for an explanation.
It may now be apparent that convergent interviewing combines some of the key advantages of both unstructured and structured interviews.
- Unstructured interviews (without specific questions) collect broad information. But they can be hard to interpret.
- Structured interviews (conducted like a face-to-face survey) collect information efficiently. But you may never know if you asked the right questions.
Convergent interviewing achieves its result by leaving much of the content unstructured. You don't ask only a series of pre-determined questions. The information is therefore determined by the person being interviewed.
The process, however, is tightly structured. You analyse the information systematically. You use only relevant information from earlier stages in subsequent stages. The systematic approach extends to sampling, data collection, and particularly interpretation. This helps to improve efficiency and reduce bias.
A typical program of convergent interviewing might move through the following steps. 2
1 Reference group
A reference group is a group of people drawn from the community, and chosen to provide guidance in community matters. In community consultation, you can probably find the people yourself, perhaps with additional people co-opted to fill any gaps in your knowledge. If so, a separate reference group may not be needed.
If you need a reference group, and have not already obtained one, do so.
2 Define the information
Define the nature of the information to be collected, at first in very general terms.
One of the difficulties with almost all data collection is this... sometimes you don't realise the questions you should have asked until after you collect the data. What is worse, sometimes you don't ever realise that you asked the wrong questions. The open-ended nature of convergent interviewing helps to avoid this.
You need ask only the questions that are needed to define the next step. Your initial question may even be as broad as "What is relevant to making this the best possible community to live in?" or "What are the most pressing issues in this community?" Or even "Tell me about this community.
In organisational settings I've had good results from questions such as "Tell me what good and what's not so good about this organisation" or "What's it like working for this organisation?"
Sometimes the study is more focussed. It may then be justifiable to ask something like "What do you like most, and least, about travelling in the western suburbs?" If in doubt, I suggest erring on the side of breadth and open-ended questions.
3 Target population
Define the target population. This may consist of all the stakeholders, or only the local stakeholders. (For the purposes of local consultation, it will most often consist of local stakeholders; you may be able to rely on other contacts to provide you with information about non-local stakeholders.)
In organisations, you may decide to define the stakeholders as everyone within the organisation. Or, preferably, you will include people outside the organisation who also have a stake in how it operates. For example, clients are an important group of stakeholders for most organisations.
4 Inform the stakeholders
Let the target population and other interested parties know what is happening. Be clear and explicit about why you are doing it, as people are otherwise quick to attribute sinister motives to you.
Choose the sample. If you are sure you have enough information about the community to select the sample, you may decide to risk doing it yourself. Perhaps you can co-opt others for the occasion, to help you. Otherwise, use a reference group for the task.
In organisations it is usually easy to find a small group of people who know the organisation well.
Decide the person "most representative" of the population. She will be the first person interviewed. Then nominate the person "next most representative, but in other respects as unlike the first person as possible"; then the person "next most representative, but unlike the first two" ... And so on. This sounds "fuzzy"; but in practice most people use it quite easily.
On some occasions, especially in urban settings, you may lack the information to select a sample. If so, you can choose people at random from the electoral roll. Better, choose half a sample in this way; then ask each informant to nominate someone different to themselves who it would be good for you to include in the sample.
6 Select and train interviewers
You may decide to do the interviewing yourself. Alternatively, and in keeping with the emphasis on participation, it is appropriate if local community members are used. If so, choose and train the interviewers.
The technique works best if interviewers work in pairs. It can be used cautiously by single interviewers, at the cost of a possible increase in interviewer bias.
7 Plan the interview
There are two parts to this...
- Decide the opening question. Define the approximate topic without leading the person being interviewed in any particular direction.
Think of it as almost a "content-free" question. It does not so much define the content of the answer as start the interview. In this way, you don't predetermine the answers by the questions you ask.
You may occasionally decide to save time by asking more precise questions, though this may bias the information you get. Even where I have specific questions to ask, however, my preference is to start in a very open-ended way, and ask the specific questions later.
Some typical opening questions were given earlier.
- Determine any probe questions for specific information. Probe questions may be used when you already have some specific questions to which you want answers. The probe questions for later interviews are used to clarify uncertainties arising from earlier interviews -- see below.
Step 7 is the beginning of a recurring cycle which is the heart of the technique.
8 Conduct the interviews
This description assumes that you are working in pairs for the interviewing.
Each of you conducts an interview with a different member of the sample. The sample is drawn on in the same order as it was compiled. The early interviews are therefore between the people who are least alike. Agreement which arises under these conditions is agreement indeed.
The ideal is a sample which, at each stage, is the best sample for the sample size at that point.
I work without taking notes. The greater eye contact and attention build earlier and better rapport. I also use a mnemonic system to help me remember the themes as they arise.
In contrast, some interviewers find it hard to recall the information unless they take notes. So if you are a relative novice, you may feel more confident if you have notes to aid your recall. If so, I suggest "key word" notes which you can take without losing much eye contact.
A compromise is to use a tape recorder. I prefer not to, as it doubles the time taken -- once to conduct the interview, and once to play it back. Some people, too, may be less frank with a recorder present (though most seem to adjust reasonably quickly to its presence).
The interview follows a number of more or less distinct stages.
- Set the person being interviewed at ease; try to give the person all of your attention. Introduce yourself, explain the purpose of the interview, and generally try to establish a person-to-person relationship. Explain (though only if it is true) that the information will be anonymous, and will be reported in ways which conceal the identity of individual people. Indicate clearly what use will be made of the information, and who will have access to it.
- Ask the opening question.
- Keep the person talking, typically for about an hour or so. The length of the interview is an important feature of the technique. Until you develop some experience you may find it hard to maintain an interview for this time; at first, 45 minutes may be more realistic.
The questions during this part of the interview are best kept as content-free as possible. You can then be assured that the content was determined by the person interviewed, not by the questions you asked. If you maintain rapport, you will find that few questions may be needed for most informants.
- Invite a summary. If you have been keeping a mental note of the key issues, this gives you a chance to check your own impressions.
- Follow up on doubtful or ambiguous issues.
- Ask any probe questions not already answered by what the person has said.
Probes are not always used in the first interviews, though they are always included in later interviews. The number of probe questions typically increases from interview to interview. You can also include probe questions on any doubtful part of the design, for example the sampling.
9 Individually interpret your interview
Working alone, interpret the information so far collected. Record this interpretation in writing. In the early interviews, the report might consist of between half a page and a page of handwriting. Note form is all that you need.
Your report can take into account both the interview just completed, and any previous interviews. It thus becomes more elaborate over the course of the interview program. By the end of the series of interviews it may consist of two or three pages, or perhaps more.
10 Compare interviews
Compare notes with the other interviewer. Give particular attention only to information which occurs in more than one report. Such information must logically be one or the other of two types: in agreement or in disagreement.
Devise a probe question which identifies how widely the phenomenon occurs. Test apparent agreements; seek explanations of apparent disagreements...
- When your reports agree, test the agreement by attempting to find out if it is ever untrue, and if so, under what conditions. For example, suppose in a program two people have both reported that traffic volume at a certain intersection is a problem. A probe question might ask "On which occasions is traffic volume not a problem?"
- When your reports disagree, seek to explain the discrepancy. Suppose that one person has said that the community wants greater use of traffic calming, and another has said they do not. A probe question would seek to determine the conditions under people desire it. "To what extent do local people favour traffic calming? ... What distinguishes those who want it from those who do not? In your opinion, when is traffic calming desirable, and when isn't it."
11 Review the process
If necessary in the light of the interviews so far, modify the approach used. For instance, change the sample, or provide follow up training for interviewers, or change the interview design and the probes.
Return to Step 8 to recycle. This is the central cycle of the technique. Continue to do so until two succeeding interviews have added no significant information. Then move on to the next step.
Compile a combined report and decide the next thing to be done.
As you can see, each interview starts broadly and generally. It becomes more specific over its duration. You keep the most specific probes until the very end. Similarly, you develop the information base and the interpretation of it gradually, from interview to interview. At each cycle, the only essential information is that required to decide the next cycle.
Keeping the interviewee talking
An important part of the technique is being able to keep someone talking for an extended period of time. Remember that the purpose is to keep them talking without leading them. If all information is freely volunteered you have some assurance that it isn't determined by the questions you asked. The effect to aim for is a "content-free" question or something similar. Some methods include the following.
- The so-called pregnant pause, where you smile sweetly and say nothing. This is obviously content-free. (It can be quite effective provided the only thing pregnant about it is your own expectant expression. Overdoing it may succeed only in making you and the other person uncomfortable.)
- Friendly, encouraging, but otherwise non-committal noises such as "mmm?", "uh-huh?" and the like. Similar gestures -- smiles, nods, and the like. This too can provoke discomfort if overused.
- Requests for more specific information: "Could you give me an example?".
- Repeating back a key word or phrase, or the last word or phrase, with a questioning intonation: "Accident?". This is very effective if not overused.
- Returning to earlier business which was passed over at the time: "You mentioned public transport...".
All of these techniques work better if you have been able to develop and maintain good rapport.
The opening question is also important to the success of the technique. Some care in choosing it is warranted.
Suppose the end goal of the program were to improve community satisfaction with road planning. A suitable opening question is one to get people talking about the strengths and weaknesses of current design; preferably, you won't wish to define the particular topic more closely than that. Examples: "What do you like most and what do you like least about travelling in these suburbs?" "What are the particular strengths and weaknesses of road planning in these suburbs?"_____
Convergent interviewing can be described as an interviewing technique which defines, quite efficiently, the most salient issues in a community or organisation or the like. I find it yields good results from surprisingly small samples if they are compiled with some care to be as diverse as possible.
It is suited to the early stages of an action research project, especially if you can find ways of involving the stakeholders other than just as informants.
There are some features of convergent interviewing which can be applied fruitfully to other data collection techniques. I draw your attention to the way in which it is driven by the data collected. Most questions, and for that matter the sample size, are determined on the run, from interview to interview.
- The following description is suited especially to community consultation or citizen participation. The technique can be modified easily to suit other applications, including qualitative research generally.
Many people, too many to name, have contributed to the development of convergent interviewing. (You know who you are. I thank you all.) However, I would like to mention, in particular, Tony Pascarelli who helped to start it all, and Karyn Healy for introducing some recent innovations. [ back ]
- This description is loosely based on one modified from Tim Dalmau, Bob Dick and Phill Boas, Getting to change, Canberra: Work Organisation Branch, Department of Industrial Relations. That in turn was condensed from the document Convergent interviewing, Chapel Hill: Interchange, 1990, which provides much more detail. [ back ]
Copyright (c) Bob Dick 1995-1999. This document may be copied if it is not included in documents sold at a profit, and this and the following notice are included.
This document can be cited as follows:
Dick, B. (1998) Convergent interviewing: a technique for qualitative data collection [On line]. Available at
Maintained by Bob Dick; this version 2.04w last revised 20000103
A text version is available at URL ftp://ftp.scu.edu.au/www/arr/iview.txt