Best Practices for Researchers

Make the most of your EA knowledge on your research


Good to have native coders

Among these important lessons include that having natives to a particular country code their own tapes is very important, and the decision to train professionals to do their own coding has been a good one. If you are working in an international context, ideally, your coders should be native.  Native coders will understand the language of course and will be able to decipher subtle cues.  This ability to decipher subtle cues of a particular culture, combined with EA Training (that emphasizes that one does not "excuse" behaviors because they are culturally common) is ideal.  For example, hitting or harsh language is considered acceptable behavior toward children in some cultures and one does not "excuse" such because they are common.  On the other hand, less obvious types of cultural cues can be helpful to a coder.

Translations optimal if native coders evaluating immigrant groups

On some occasions, the native investigators are coding immigrant samples. To the extent possible, having translation is important, because without that translation, you are mainly coding nonverbal communication.  While EA in the early years is predominantly guided by nonverbal signs, nonetheless, many statements can convey endearments as well as hostility, and sometimes there is mismatch between what a parent says and how they are saying it.  With that said, parents and children in refugee camps (where it is so loud that it is not possible to understand what they are saying to each other) have been successfully coded.

Related to this is that there is no such thing as an “expert scorer”, as sometimes professionals refer to themselves in publications if they have been doing coding for a long time. There is no such thing as an expert coder and certainly no external expert coding service–you would need to join a research team at an intellectual level (and hopefully, through co-authorship) and be committed to doing inter-rater reliability with another partner, refresher training, and abide by all other aspects of the guidelines outlined here. On occasion, my students, associates, and I have done coding for other research labs, but this practice has been discontinued. This modification is in line with no one being an expert coder, including the method developer.

Copyrighted and trademarked; therefore, use as designed with no adaptations without prior permission

Research publications

  • One of the authors should be EA certified;
  • Try and use the term emotional availability or EA, preferably in the title and hopefully throughout (so that it can easily be located and contribute to the coherence of the EA field);
  • Try and describe the system in narrative format, rather than with scale points, which can mislead a quick reader/scanner, despite the authors’ best efforts to be clear that they are not depicting the “real deal”;

A minimalist example of what can be written is as follows, and helps the reader identify key aspects, such as the edition used, duration of observation, context of filming, authorized training, and reliability of said coders:

A 15-minute parent-child free play with knights and princesses was videorecorded. Instructions were to “be together” as you normally would. Two raters attended an authorized online training using the 4.1 edition of the EA System (EA® System,, accessed on 25 November 2022) (Biringen, 2022; Biringen, 2008; Biringen et al., 2014). After certification of each coder through authorized training, inter-rater reliability of the first 10 cases was achieved, and then 20% of the whole sample was coded by these coders. Intraclass correlations were conducted for each of the scales and the EA Clinical Screener exceeded .75 in each case, and for the 4-classification EA zones, kappas ranged between .80-.85. At least one of the authors (Dr. Jones) in this publication is a trained/certified EA coder.

Moving from our standard cases on which you obtained reliability needs to be followed up by formal inter-rater reliability on your own cases. This should be done early in the data evaluation. See the manual for more detail on how to obtain reliability, typically with a member of your ‘group’, or on some occasion with a previously certified professional in your facility.

To “translate” your training into your own sample, Level II, as described on this site, is recommended. This is when we (or an authorized supervisor) scores 2-3 of your own cases. Simply having scored for a long time is not a way to become an authorized supervisor. We have seen those who were certified a long time ago providing inputs that are not accurate. EA Basic certification provides only a minimum level of competence, and without additional recalibration, this competence can become diluted over time (also known as ‘drift in reliability’).