The Adult Attachment Interview Protocol: THOUGHTS FROM A CODER

Helen Deane Dozier

DeaneInterviewing skills are crucial to proper scoring and classification of the Adult Attachment Interview. 

A good interviewer puts respondents at ease, helping them to understand there are no right or wrong answers on the AAI, but only their answers, and whatever they choose to say in response to AAI questions is important.

With experience, an interviewer learns the importance of not just asking the questions, but really listening to the answers, for it is the subject’s responses that help the interviewer know if further probing is needed.  

An interviewer must think on the fly, and ask for elaboration on some answers without leading the subject. Does the subject need a bit of time to think, reconsider, or maybe just warm up to the topic? Would it help to rephrase a question that seems to be stumping the subject? Wow, the respondent has gotten away from the question – what should be done? A good interviewer allows the subject to digress when appropriate – even acknowledging an interest in what the subject is saying without encouraging the digression, while never making subjects feel their answers were inappropriate.

And all this while being sensitive to the subject’s feelings.  A delicate balance, for sure!

The most authoritative sources of information on how to administer the AAI are the “Adult Attachment Interview Protocol” and “Avoiding Common Errors in Conducting the Adult Attachment Interview,” both by Mary B. Main.  They are full of valuable detailed information that should be read and reread as an interviewer gains practice. These copyrighted documents are available online in the Attachment Theory and Research pages on the website of the State University of New York at Stony Brook or from these links:

Here is an abbreviated version of the AAI Protocol, which contains all the questions and follow-ups, but with abbreviated explanatory notes to make it easier to use in actual interviewing.

Here are some thoughts of my own as I have coded AAIs, which can be viewed as a continued discussion of what is so well-covered in Mary’s Main’s writings.  These are some of the critical areas and key questions that can make a tremendous difference in assisting a coder to score an AAI correctly.

Importance of Good Sound Quality

A good interview starts with good recording equipment.  Perhaps it should go without saying that an interview must be recorded clearly for proper transcription, but many interviews have problems because of extraneous room noise, traffic noise, a crying child on the subject’s lap, a mumbling subject who is too far from the microphone, or a tape that runs out without being noticed.  It’s a good idea to listen to a brief sample recording before starting the interview.

Questions about Adjectives and Supporting Memories.

The AAI has just a few warm-up questions before arriving at what could be considered the heart of the interview – adjectives to describe the relationship with parents and the subjects’ support of the adjectives they choose.  Helping the subject come up with adjectives is a skill that improves with experience.

Always ask for a full set of adjectives and do full follow-up for all primary caregivers, even if there are several (a mother, a father, a grandmother) who took care of subject as a child for significant period of time.  Someone who babysat or who the child stayed with on weekends might not matter so much, but if the child, say, lived with a grandmother for two years, especially at a young age, that might be important.  It’s a judgment call, but err on the side of inclusion – the probe for adjectives and memories regarding caregivers is crucial to accurate scoring.  A transcript that exhibits strong F (secure/autonomous) characteristics even though both parents were highly rejecting or abusive might be less surprising when you discover that the subject’s loving aunt actually did most of the caretaking in the early years.

Occasionally, the importance a particular caregiver played in the life of the subject may not be revealed until an interviewer is well past the subject of who caregivers were, but it’s still important to ask for adjectives and memories after such information comes to light. 

Some respondents will understand the idea of coming up with “adjectives” for a relationship, but you may come across subjects who don’t know what an adjective is, and might not admit it because they think they should know, or with subjects who find it difficult to sum up a relationship in a single word.  Describing a relationship in a word is not necessarily an easy thing to do.  In keeping with AAI protocol and in order to produce information that will help a coder accurately score these AAIs, it’s best to ask for adjectives or words to describe the childhood relationship first, and give increasing assistance as the subject indicates he or she is having a hard time with the question of how to describe a relationship.

You can tell respondents that if they can’t think of single words, they could use phrases, and then probe again by saying something to like, "Okay, go back in your mind to childhood and try to remember what the relationship with your mom was like."  If they are still stymied, you may need to help them understand (but as a last resort), perhaps by asking, “How did you and your mother feel toward each other or act toward each other when you were a child? Can you think of words for that?”  Or, “How did you and your father get along with each other when you were a child?” Further probing (if they still need assistance, or if they just answer that they got along “fine”) might be to ask how the parent acted toward them or how they acted toward their parent.

Don’t give leading ideas, such as saying, “Some people, for instance, say their mother was ‘loving’ or ‘always there’,” because some subjects will latch onto suggestions even if they don’t really fit, and you may even find your own suggested adjectives then popping up to answer questions during the rest of the interview.  In a case where the subject is completely stumped, you could ask if they have any memories of the parent from childhood.  If the response is no, that is an answer of significance.  In time, you will develop a line of questioning that works to assist the subject open their mind to thinking about childhood relationships without leading them down a path that is not their own.  Of course, there will be those who are unable or unwilling to think about childhood, but you’ve done your job when you’ve uncovered that strategy of dealing with the past.

So not only are the narratives from childhood important, but it’s also vital to discover whether or not memories even exist and if they do, whether or not the subject is willing to talk about them.  A properly executed AAI is designed to uncover this information, and the importance of making sure the respondent understands that the questions regarding adjectives relates to childhood memories cannot be overstated. 

Sometimes it’s hard to tell if the subject is giving a childhood memory or not.  If you are unclear whether or not the memory being recounted is from childhood (say, for instance, the subject describes “caring” by saying they remember a time they brought a friend home and the mother was nice toward the friend), you can ask, “About how old were you when that happened?”  In scoring the AAI, it’s important to see what memories exist during the formative years of childhood, so a memory about “caring,” for instance, when the subject was 16 years old is not near as important as a memory from age 7 or age 10.

If I fill a lot of space talking about the importance of adjectives and asking for childhood memories related to those adjectives, it’s because this part of the AAI is so critical to scoring and classifying, yet far too often is rushed through by interviewers who do not understand that the task they have asked the subject to undertake may require considerable thought.  In scoring hundreds of AAIs from various research projects and private practice, I have encountered countless transcripts where the interviewer was far too quick to continue with the interview with just one or two adjectives supplied by the respondent, despite Mary Main’s admonition to “be patient” and “be encouraging” and despite her instructions on p. 2 of the AAI Protocol referenced above, that “the interviewer must not begin to probe until the full set of adjectives has been given” (italics hers) or only after the interviewer is fairly certain that more adjectives cannot be given.

Also, I have encountered just as many interviews where the subject is not reminded during the probe for support of each of their adjectives that they are being asked for childhood memories.  Here is an example of a probe for support of an adjective according to protocol:  “You described your childhood relationship with your mother as “loving.”  Can you think of a memory or an incident that would illustrate why your chose “loving” to describe the relationship?”  

You are gently reminding the subject throughout the request for support of adjectives that the subject picked the words to describe a childhood relationship.  After such reminders, if the subject reverts to present tense or to the more recent past, such moves away from the questioning has far more significance to the coder than when the questioning is ambiguous.

Having said that, if a subject does digress or revert to present tense, it’s important to let them keep talking off-subject until they come to a pause, and never to make them feel their answers were inappropriate.  Again, Mary Main’s documents have excellent suggestions about how to deal with a respondent who wanders off topic.

Examples of importance of adjectives:  A transcript that is clearly dismissive (Ds) in tone but has too little information can be difficult to classify.  Sometimes there’s the feeling that derogation or idealization might be evident if the interviewer had just probed all adjectives and given the subject a chance to reveal what is going on. Or if the interviewer doesn’t make it clear that they want specific examples to illustrate the adjectives a subject chose, the coder may be left wondering if the subject has memories but just didn’t understand what the interviewer was asking.  It’s understandable and instinctive in everyday conversations to back off when someone seems reluctant to talk about something, but in the AAI, questions– and especially the questions about adjectives -- need to be probed unless the subject indicates they would rather not discuss something.  On sensitive matters, if the subject seems to be having a hard time, they can always be asked if it’s okay to discuss.

To elaborate, if an interviewer asks the subject why he or she chose “caring” to describe the mother, and the subject says, “That’s just how she always was, she took care of me and was a very good mother.”  The interviewer should explain to the subject, “That gives me a good idea in general.  Can you think of a specific time in your childhood that your mother did something that made you feel she was caring?”  The subject may or may not have specific memories, and may or may not want to discuss them, but that can only be determined if the subject understands the nature of the question.  Notice in that follow-up question that the interviewer is reminding the subject that the question pertains to childhood.  Again, if the subject responds to questions with more recent memories, that “off topic digression” takes on far more significance when the coder is certain that the interviewer was clearly asking for information related to childhood.

Another example of the importance of discussion around these adjectives for childhood, but on the other end of the state-of-mind spectrum, would be a transcript that has passages that hint at underlying anger or preoccupation with a rejecting stepparent.  The interviewer may have spent considerable time quizzing the subject about the mother and father, so not to prolong the interview, decides to just ask for two adjectives about the stepparent, perhaps not realizing that the stepparent may have had more impact on the subject than either of the parents.  The coder may be left wondering how much passivity or anger might have been left unexpressed because the interviewer did not obtain a full set of adjectives to probe.

Questions on Loss

After identifying that the subject lost someone who was important to them (and not just a relative who lived elsewhere and didn’t figure prominently in their life, for instance), it’s important to ask the follow-up probes regarding important losses according to protocol.  The questions as prescribed by Mary Main are designed to get the subject discussing their feelings about the loss as opposed to just answering “yes” or “no” questions.  The discussion allows the coder to analyze the way the subject uses language, giving clues that help in deciding whether the loss has been resolved.

Questions on Abuse

Knowing how and when to probe regarding possible abuse while being sensitive to the subject’s feelings is one of the biggest challenges to the interviewer.  Sometimes there will be no prior clues, but an alert interviewer will pick up on mentions of strict parenting or being afraid of a stern parent, for instance, suggesting a further probe to discover the nature of the parent’s discipline, even if the subject has said “no” when asked if they remember any abusive behavior. A subject’s own opinion of whether or not they were abused may or may not be “abuse” by AAI standards, so it’s important to have a description or examples.  If subject indicates they received spankings or beatings, or if they were subject to yelling and screaming, it’s important to know the nature of the beatings, or how frightening the yelling or the beatings were, for instance.  Also, just as in the questions about loss, it’s important to ask some questions that encourage the subject to discuss the nature of the abuse rather than just answering with a yes or no, because how the subject uses language when discussing loss and abuse is critical to accurate scoring.  Again, the discussion of how to probe possible abuse is explained in the AAI Protocol by Mary Main.  For instance (Mary Main’s example), if a subject says he or she “would get the belt,” the interviewer could ask what “getting the belt” meant.  Or a subject who has said their mother hit them a lot could be asked, “Could you tell me a little more about that?” followed with, “Can you give me an example of when she would hit you?”

Here’s an example from an interview where a subject had mentioned early in the interview that she had received frequent beatings from her father.  When the interview progresses to the question about abusive behavior, notice that the respondent seems to be saying “no,” that she was not abused, but because of the way the question was phrased, she may actually be responding to the last part of the question, which has to do with sexual abuse.   

Example from an interview: 
Interviewer:  “And some people have memories of threats or behavior that was abusive.  Did you have anything like that in your family, to you or in your family?
Subject:  “I really – like what?”
Interviewer:  “Like physical abuse, where somebody would get beat and it would leave marks, or sexual abuse.”
Subject:  “No.”

Example of follow up question from an alert interviewer would be, “You mentioned earlier that your father sometimes beat you. Did you ever have welts or bruises from that?” Or the interviewer might ask if the subject could elaborate on the beatings, describe how painful or frightening they were, or describe what happened on a particular occasion that stands out in the subject’s memory. If a subject had mentioned earlier in the interview that a parent was strict without elaborating at the time, the interviewer could ask the respondentfor an example of a time the parent was “strict,” or what the subject meant by choosing that word.

Sensitive Subjects

Sexual abuse, of course, can be a very sensitive subject, and questions may need to be more general.  For instance, a subject could be questioned about the frequency of the abuse, how the subject reacted to it, and whether the abuser was frightening or threatening.  If the subject appears willing to discuss the abuse, please ask them to tell as much as they are willing to share about a specific incident, because it is in the reliving of traumatic experiences that the coder is often able to discover lapses in reasoning or discourse. The subject’s answers can help the coder determine the general nature of the abuse, and might even reveal that the activity was limited to conversations rather than actual sexual contact. Some subjects will talk freely about sexual abuse, but questions of a general nature will allow subjects to choose how much they want to say.

There will be times when the subject gives an answer that is too brief to be very informative, but you will be hesitant to probe further because of the possible sensitive nature of the subject, such as when they say a simple “yes” when you ask if anything abusive or traumatic happened with someone outside the family.  Again, be ready with questions of a general nature that allow the subject to choose how much they tell you, rather than allowing a brief answer like that to be a conversation stopper unnecessarily.  For instance, you can ask the subject how old he or she was (providing easy questions to prolong the discussion) and you could inquire about the general nature of the experience and how it affected the subject, and if it still affects their personality today, and if so, how.  Was it overwhelmingly terrifying?  How did they deal with it at the time?  Again, try to be prepared with questions that go beyond yes-or-no answers and promote discussion, while being sensitive to a subject who seems distressed and may not want to discuss the matter further.  Again, it’s always okay to ask the subject if they mind talking about a sensitive subject if you feel they might be reluctant.

The Subject Gets Off Topic

It’s natural for an interviewer to try to keep a subject on topic, answering the questions that are being asked.  In the case of some questions, digressions into relationship discussions, indeed, need to be curtailed (e.g., the first question for demographics and the request for five adjectives in Question 3).  But in the AAI, a subject should be allowed to digress and ramble on certain other questions without being interrupted by the interviewer (within reason), especially on questions about their relationships with important caregivers and when discussing Loss and Abuse.  From a coding/scoring perspective, the digressions are important to note, and might indicate a strategy for dealing with attachment issues.  The fine line for the interviewer to walk is to indicate interest in the subject’s response (with “mm-hmmm” for instance) while not encouraging the digression with off-topic questions or comments.  After the subject has completely finished his or her thought, maybe even wandering from topic to topic, the interviewer can go on to the next question, or a follow-up to the original question that was asked.  It’s important not to make the subject feel uncomfortable about getting off the subject, because this might hinder further ramblings and again, these wanderings off topic can provide important clues to states of mind for scoring.  So after the subject runs out of steam and ends their thought, the interviewer can be conversationally polite, saying something such as, “That’s interesting, and we may talk more about that later.  Regarding your adjective that your mother was ‘loving,’ can you think of a specific memory from childhood that would illustrate why you chose that word?” 

My addendum to this document, “Maintaining Gentle Control,” contains a further discussion on this subject. 

Fear of Loss

The question on how the subject feels when separating from a child (or an imagined child if they don’t have one) is an important one and is often the only chance to uncover the rare Ds4 classification.  Mary Main places importance on asking the question just as it is, without elaboration, allowing the subject to respond.  But the follow-up question asking if the subject feels worried about the child can provide invaluable information for the coder, too.  If the subject reveals what sort of things he or she worries about, it can help the coder understand if the fears make sense in the subject’s life. For instance, if a mother says she is scared to death that something might happen when her child goes to school or goes to stay with a relative, it can be helpful to the coder to know what the mother fears could happen and perhaps how she responds to that fear. If the subject grew up in a dangerous neighborhood, such fears would have a basis in reality, but if there is no apparent reason for fear that “something might happen,” it can be a significant discovery for the coder.  If the subject acts on that fear (such as keeping the child at home), that can also be important to know.

A tip regarding follow-up probes

You may have noticed a recurring theme throughout this paper on the importance of questions that do not lead the respondent down particular paths, and which require a discussion type of answer instead of a “yes” or “no.”  This is a generality, and there are certain parts of the AAI where “yes” or “no” is important, but in general, scoring the AAI is dependent on discovering what memories and feelings the subject has regarding attachment, what they are willing to tell and how they use language in recounting their narrative.


In summary, these are thoughts from a coder’s perspective about interviewing procedures that can make or break a transcript for scoring and classification purposes. This paper is a work in progress and will be updated and expanded on from time to time.


Back to Clients Page

Back to AAI Coding Home Page