Inventory of Interpersonal Ambivalence

How Can One Measure Interpersonal Ambivalence?

What is the Inventory of Interpersonal Ambivalence?

The Inventory of Interpersonal Ambivalence-18 (IIA-18) is an 18-item self-report measure to assess highly mixed feelings about close relationships. We believe interpersonal ambivalence to be the core feature of fearful-avoidant attachment style. While the measure was developed from an attachment framework, there are likely multiple ways and experiences that can lead an individual to develop ambivalent feelings about close relationships.

Clicking on the link below will allow you to download a copy of the 18-item and 8-item version (IIA):

Download the IIA

The sections below are split into two parts. The first part (which you are currently reading) focuses on the technical details for the IIA-18 and IIA-8. The second section below (entitled "What is Interpersonal Ambivalence?") describes the theoretical ideas underlying the construct. How Are the IIA-18 and IIA-8 Scored?

Both measures are scored by treating "false, not at all true" as a one, "slightly true" as a two, "mainly true" as a three, and "very true" as a four. To generate a scale score, simply average responses across all items given. This yields scores ranging from 1 to 4. Higher scores are indicative of higher levels of ambivalence. A benefit of using an average score (vs. a sum scoring method) is that it makes the long-form and short-form versions of the measures more easy comparable.

Can I use a Bipolar Likert Scales with the IIA-18 or IIA-8?

We do not advise using bipolar Likert Scales with either version of the IIA. Many traditional measures of the adult attachment dimensions do use 7-point bipolar Likert scales. However, the nature of the statements rated in the items is quite different. So, we strongly recommend against this practice with the IIA for three reasons: 1) Several items are worded in such a way that it would be difficult for a respondent to determine the difference between "strongly disagreeing," "disagreeing," or "slightly disagreeing." 2) The instructions are worded such that if any part of the item statement is false for that respondent, the respondent should simply rate the entire item as "false, not true." 3) While we fully expect the IIA-18 and IIA-8 to show strong, inverse associations with measures of attachment security (as it has thus far), our intent is to build a unidirectional scale where low scores are not intended to provide a measure of security. Instead, they should be interpreted to reflect an absence of interpersonal ambivalence (which may or may not indicate security).

How Many Items Can a Respondent Fail To Complete?

For the IIA-18, a respondent may miss up to three items. If a respondent fails to endorse four or more items, the respondent should be eliminated from the analysis. A minimum of 15 items completed is recommended to calculate a score for the IIA-18. For the IIA-8, a respondent's data should be considered invalid if they miss more than a single item. A minimum of seven items must be completed to calculate the overall score.

Are There Reference Norms for the IIA?

We have not engaged in a formal, standardized normalization process. Based on a sample of 2,431 adults from the United States (49% of the sample was male and 51% was female) we provide the distribution data below to allow for comparisons. Respondent's ages in the ranged from 18 to 77 with an average age of 29.53 (SD = 12.25). Of the 2,431 adults providing us with this data, 1,611 completed the IIA-18 online through Amazon's Mechanical Turk as a part of the various development studies. An additional 820 college respondents completed the IIA-18 as part of developmental studies.

Who Can Use the IIA-18 and IIA-8?

The IIA is free to use for any one conducting academic or non-profit research. For all other uses, contact Dr. Caleb J. Siefert ( to obtain permission.

How Was the IIA-18 Developed?

For detailed information on how the IIA-18 was developed, download the manuscript by Siefert and Haggerty (in review) in the next section. This paper contains much of the information you are looking for. We'll be adding additional information (e.g., original item pool) to this webpage in the very near future).

How Do I Learn More About What the IIA Relates To?

An early draft of the development article is provided at the bottom of this webpage (i.e., Siefert & Haggerty, in review). You can also view several posters on the IIA that have been presented by student members of the lab by going to the menu and going to the "Posters" page of this website.

I Want to Do a Study on Adult Attachment; Should I just use the IIA?

The IIA can certainly be used on its own, but we'd only recommend doing this if you were really solely focused on interpersonal ambivalence. If, instead, you really want to focus more broadly on "adult attachment status," then it's best to include measures that tap attachment anxiety, attachment avoidance, and attachment ambivalence. To do this, you should administer a version of the IIA and you should use an additional attachment measure. Personally, I prefer either the Experiences in Close Relationships Inventory (ECR, Brennan, Clark, & Shaver, 2000) or the Experiences in Close Relationships Inventory- Revised (Fraley, Waller, & Brennan, 2000). To get more information on the ECR. I advise you to check out Dr. Phil Shaver's wonderful webpage here.

To get access to the ECR, follow this link.

There is also a 12-item short form version of the ECR developed by Wei, Russell, Mallinckrodt, and Vogel (2007). See their article at this link.

To get access to the ECR-R, you should visit Dr. Chris Fraley's website (where there is also a good amount of excellent information there on adult attachment). You can visit that site by following this link.

View a Seven Minute Talk about the IIA

What is Interpersonal Ambivalence?

Interpersonal ambivalence refers to very mixed feelings about close relationships. Ambivalence, as a concept, means that one is pulled in two opposing directions at the exact same time. It doesn't mean one isn't sure of what one wants. It means one wants two conflicting things at the same time. So, interpersonal ambivalence is the simultaneous desire to develop and avoid close interpersonal relationships.

For individuals high in interpersonal ambivalence there is a battle between impulses. On the one hand, they do experience desires for connection and support from others. This gives rise to wishes for to develop close relationships with others. On the other hand, these individuals have also had experiences that have left them skeptical in terms of the benefits of closeness. Being close to others isn't just disliked, it something that triggers worries, anxiety, or fear (e.g., The other person will let me down if I stay close; If I'm close, I'll end up being hurt). These feelings give rise to wishes to separate and keep distance from others. This conceptualization of interpersonal ambivalence as having wishes for connection in conflict with fears regarding the risks of connection is in keeping with the core feature of the fearful-avoidance attachment status in adulthood. Though there are other features that distinguish the fearful-avoidant status from other forms of attachment, interpersonal ambivalence is among the most salient.

What is the Fearful-Avoidant Attachment?

The fearful-avoidant style of attachment was first discussed by Bartholomew (1990). Bartholomew sees several factors playing a role in the ambivalence experienced by those with fearful-avoidant attachments. These individuals have had experiences that have led them to view themselves in more negative terms and as less capable of managing stress or coping with challenges on their own. Typically, such feelings would likely trigger support seeking behaviors. However, these individuals also tend to view others in negative terms. They don't anticipate that seeking help will prove fruitful. In fact, they may even believe that others will actively heighten their distress. This conflict sets the groundwork for ambivalence about close relationships. Other features of fearful-avoidance have been identified, such as unstable ways of viewing others, vulnerability to stress, and difficulty interacting with others generally. These individuals often possess a wide range of positive features and character strengths, but their attachment style can sometimes interfere with their ability to make full use of these.

It's common for folks with this attachment status to manage in one of the following ways. First, the individual longs for relationships, but simply avoids them. This avoidance leaves them feeling safe, but also isolated, lonely, and unsupported. The second way is by people beginning relationships with others, but becoming increasingly anxious and nervous as the relationship deepens. In many cases, these fears lead them to prematurely leave a budding relationship in order to reduce their anxiety and feel relief. Third, a pattern in which the individual enters relationships with others who treat them poorly. Often, these individuals do leave these relationships; however, their core belief that relationships are dangerous is further confirmed by their lived experience. If you'd like more information, check out this an excellent blog article on Fearful-Avoidant Attachment by my colleague Hal Shorey (link). Of the various features of fearful-avoidant attachment, deep interpersonal ambivalence is at the core. The person is constantly feeling uneasy about closeness. Behaviorally, this causes them to be more likely to answer the question, "should I stay or should I go?" with withdrawal. From a motivational point of view, they feel badly about this. They do desire relationships. They're just afraid of it as well.

Why Should We Care About Interpersonal Ambivalence?

There continues to be growing research support that close relationships provide us with a number of benefits for psychological and physical health. Close relationships seem particularly important and the ability to forge and maintain relationships is consistently linked to well-being (Mikulincer & Shaver, 2007). Sadly, not all adults are able to pursue these aims. Further, some find close relationships to actually be a source of distress. A number of research studies have found that those with fearful-avoidant attachment styles are at increased risk for things like depression, stress, and interpersonal challenges (Cyranowski, et al., 2002; Karreman & Vingerhoets, 2012; Pickard, Caputi, & Grenyer, 2016; Woodhouse, Ayers, & Field, 2015). More specifically, our research focusing specifically on the interpersonal ambivalence component indicates that people with higher levels of interpersonal ambivalence are at greater risk for stress, difficulty understanding their own and other's feelings and motives, and interpersonal challenges. Further, it's common for them to report less access to social support to help them manage their distress. Even when support is available, they find it harder to make use of (likely due to fears of intimacy or challenges with interpersonal problems). As such, their lives and their relationships are less satisfactory and more inconsistent (for more information on these specific studies; download the paper currently in review or check out the student posters on this topic).

How is Interpersonal Ambivalence Related to Adult Attachment?

Historically, research into individual differences in attachment has generally differentiated people in terms of their focus on attachment relationships. Specifically, some people are overly-focused on attachments, more dependent on others, and experience chronic fears that they will be abandoned. For adults, this prototypical style of attachment is known as "preoccupied." At the opposite end of the spectrum are individuals who are under-focused on close relationships. They care little about close relationships, desire high levels of independence, and dislike emotional closeness and intimacy. These adults are often said to have a "dismissive" style of attachment. The preoccupied and dismissive styles are both known as "insecure" patterns of attachment.

The majority of adults, however, strike a balance between these two styles. They are invested in cultivating close relationships, but also focused on exploring and developing themselves. In other words, they are more flexible. They turn to others when they need comfort and they provide support for others when others need comfort; however, when they are not distressed, they are focused on things other than their relationships. They are thus able to balance needs for autonomy and needs for connection. These adults are typically said to have a "secure" style of attachment.

During the 1990s, social psychologists and adult attachment researchers began examining how to best measure attachment using self-report questionnaires. Bartholomew (1990) was among the first to note that differences in attachment appeared to correspond to two dimensions based on the person's view of others (i.e., model of others) and the person's view of the self (i.e., model of self). Secure individuals were said to have positive models of self and others. Thus, they were happy to turn to others and confident that others would be helpful. At the same time, they were also viewed themselves as capable and able to handle challenges as they arose.

Preoccupied individuals are said to have positive models of others, but negative models for the self. Thus, they tend to view themselves as incapable of coping or managing challenges. Because they view others positively, many adopt the solution of being highly dependent on others (e.g., Because I can't manage things on my own, I need to keep someone close who can take care of everything). However, because they view others positively and see themselves negatively, they chronically worry about abandonment (e.g., Why would that wonderful person stay with a loser like me?).

Dismissive Individuals are said to have a negative model of others, but a positive model of self. Thus, they view themselves as capable of coping and managing challenges, but don't see others as helpful or supportive in these efforts. In fact, they may even feel that others will "just get in the way,", "will slow me down," or "will just make things worse." As such, they strive to be highly independent and are skeptical of emotional intimacy and closeness. Because they view others in negative terms, it's typically hard for them to be open or turn to others for support (e.g., Why would I share my distress, the other person won't understand and will just make things worse; I'll handle it on my own).

Because Bartholomew's model was dimensional, she bisected the dimensions and postulated that there was likely a fourth status. These individuals, which she described as "fearful," would have both a negative model for self and a negative model for others. Thus, like the preoccupied individuals, they would view themselves as less capable and long for supportive others to assist them. However, like dismissive individuals, they would also fear that others would hurt them, let them down, or misunderstand them. As such, they were pulled in two directions, never quite feeling satisfied. When they let others close, they began to feel anxious that things would go wrong; however, when they distanced themselves from others they begin to feel isolated, lonely, and overwhelmed. In short, they experience deep interpersonal ambivalence. Later work on adult attachment measures led to the development of excellent self-report measures for assessing adult attachment. Today, the Experiences in Close Relationships Inventory (ECR; Brennan, Clark, & Shaver, 1998) and the Experiences in Close Relationships Inventory-Revised (ECR-R; Fraley, Wallner, & Brennan, 2000) are among the most widely used research measures in psychology. Both of these measures were developed by collecting all of the available self-report attachment measures and analyzing responses to individual items. The two measures selected different items because they used different methods. The ECR was created using exploratory factor analysis to select items. The ECR-R was created using item-response theory (IRT). Despite their differences, both measures found that individual differences in attachment were best explained along two orthogonal dimensions: attachment avoidance and attachment anxiety. Attachment avoidance involves discomfort with intimacy and dependence on others, difficulties trusting others, desires for independence, and reluctance to share in close relationships. Attachment anxiety involves preoccupation with close relationships, anxiety about being abandoned, concerns one is not worthy of love, and desires for excessive levels of closeness. Thus, these two dimensions were quite similar to Bartholomew's model of self and model of others; and bisecting them yielded the same four-quadrant attachment model (see image below).

If You Say Existing Attachment Measures are Great, Why Make another One?

This is the million dollar question. It's true that most dimensional adult attachment measures, such as the ECR and ECR-R, are quite strong. Nonetheless, all measures are to an extent dependent on the method utilized to create them and the initial item pool. When we went back through the major adult attachment measures, including those that contained a fearful scale, we found that very few of them contained items worded to assess ambivalence. Instead, items tended to be worded to assess either attachment anxiety or attachment avoidance (with a handful of exceptions). A dearth of items tapping ambivalence is likely to render it challenging for factor analysis to detect an ambivalence dimension. We have found that when you include an equal number of ambivalence items, a third factor emerges (that is not orthogonal, but is instead related equally to both attachment anxiety and attachment avoidance).

Why Not Just Use the Interaction Term?

This is a question I sometimes get from reviewers who are indeed familiar with the literature. The notion is that one could simply convert the attachment dimensions to z-scores, multiply them together, and then use the interaction term to determine when a specific quadrant is more predictive of a given variable or outcome. There are three reasons why this is not an optimal approach:

1) In theory, highly ambivalent people would rate all of these items highly. However, in practice, highly ambivalent people are more likely to select ratings near the middle of the scale for items that tap only one dimension at a time.

2) It requires significantly more power to detect interaction effects relative to main effects. One need only compare the adult attachment literature in the 1990s to that since 2000 to see that the number of studies focusing on specific styles of attachment have declined notably in favor of the dimensions. This is not a critique on the dimensions; it is simply noting that people will publish that which is easier to find.

3) Few actually consider what creating an interaction term does. When we create a multiplicative interaction term for attachment we are essentially putting the secure and fearful people together and the preoccupied and dismissive people together. For example, someone who scores two standard deviations below the sample mean on both dimensions (i.e., very secure) would have a multiplicative score of four and someone scoring two standard deviations above the sample mean for both dimensions (i.e., very fearful) would also be given a score of four. This means that to find an effect it needs to be very specific and very large, or else the effect will get washed out.

The most important argument I can give you for not using an interaction effect is empirical evidence. For example, across multiple studies in which we looked at how attachment status predicts self-esteem, interpersonal problems, subjective well-being, interpersonal functioning, relationship status, relationship satisfaction, fear of intimacy, mentalization, smoking motivation, substance use, and social support, we've found that the interaction term is almost never significant. Alone, that finding wouldn't be important, but when you pair that finding with the fact for every variable just mentioned that IIA-18 incrementally increased the amount of variance explained by at least 5%, then we clearly see an issue. If the two dimensions (plus the interaction term) were fully capturing the four-quadrant space, then inclusion of the IIA-18 wouldn't add significant variance (and certainly it wouldn't do it so consistently).

Might Other Vectors Be Directly Measurable?

Yes. This is quite possible. In other words, someone might be able to develop a measure that more precisely assesses the diagonal vector running into the preoccupied quadrant. That may even be a good thing to do. In fact, one of my favorite measures in all of psychology, The Inventory of Interpersonal Problems, uses this precise approach. Like attachment measures it generates two orthogonal dimensions; however, it also produces eight octant scores for specific vectors. This allows researchers (and clinicians) to further breakdown and better understand an individual's overarching dimensional scores. This precision allows researchers to ask more specific questions and helps clinicians develop more specific treatment plans. While I can see wisdom in such endeavors, I myself am not personally working on this issue, but I'd be happy to try and help anyone who wanted to have a go at it.


Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective. Journal of Social and Personal Relationships, 7, 147-178.

Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four- category model. Journal of Personality and Social Psychology, 61, 226-244

Brennan, K., A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of adult attachment: An integrative overview. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp.46-76). New York: Guilford Press.

Cyranowski, J. M., Bookwala, J., Feske, U., Houck, P., Pilkonis, P., Kostelnik, B., & Frank, E. (2002). Adult attachment profiles, interpersonal difficulties, and response to interpersonal psychotherapy in women with recurrent major depression. Journal of Social and Clinical Psychology, 21, 191-217.

Fraley, R.C., Waller, N.G., & Brennan, K.A. (2000) An item response theory analysis of self-report measures of adult attachment. Journal of Personality & Social Psychology, 78, 350-65.

Karreman, A. Vingerhoets, Ad J. J. M. (2012). Attachment and well-being: The mediating role of emotion regulation and resilience. Personality and Individual Differences, 53, 821-826.

Mikulincer, M., & Shaver, P. R. (2007). Attachment in adulthood: Structure, dynamics, and change. New York, NY: Guilford Press.

Pickard, J. A., Caputi, P., & Grenyer, B. F. A. (2016). Mindfulness and emotional regulation as sequential mediators in the relationships between attachment security and depression. Personality and Individual Differences, 99, 179-183.

Ravitz, P., Maunder, R., Hunter, J., Sthankiya, B., & Lancee, W. (2010). Adult attachment measures: A 25-year review. Journal of Psychosomatic Research, 69, 419-432.

Wei, M., Russel, D., W., Mallinckrodt, B., & Vogel, D. L. (2007). The Experiences in Close Relationship Scale (ECR)-Short Form: Reliability, validity, and factor structure. Journal of Personality Assessment, 88, 187-204.

Woodhouse, S. Ayers, A., & Field, A. P. (2015). The relationship between adult attachment style and post-traumatic stress symptoms: A meta-analysis. Journal of Anxiety Disorders, 35, 103-117.