Empirical Conduct for Social Scientists

This post is about empirical research in social science, data quality and data management.  It is suited for trainees in UX, traffic, crowd, social science and behavioural analysis (ancillary to some stakeholders in HR, L&D, PM)

Data is based on the doctoral research “The composite semiotics of interactional repair” by Albanese Claudia, Ph.D. (University of Luxembourg)


“Hello. Some updates… But I’ll frame them first. If you are a social scientist, and/or run UX, traffic, crowd and behavioural studies, you are probably a very lucky scientist because working with people is absolutely fantastic, but sometimes it can also be … [be diplomatic!!].. a painful process!! you know?!? all that collecting and mining human data is really hard work!

Hard scientists often criticise social scientists for lacking empirical precision. I found that is unfortunately often the case. I try to do things in the best of my possibilities to avoid that in every research project I approach (Even if Agile! Agile needs quality too! “How to produce quality in Agile environments” is for another post ).

In my doctoral work, I produced micro-analytic and multi-layered transcripts of interactional sequences and organised them in a database. I analysed intonational aspects of talk such as pitch, frequency, energy and some muscular movements. I went through several cycles of segmentation, transcription, data mining, reliability checking, proofreading etcetera. A few people got involved, some to a greater, other to a lesser extent, but there was a several-eyes check on the database and it is still not perfect. It will never be (because of the margin of error, [and also for uncertainty, incompleteness, relativity, etcetera] but not for this post).

In an attempt to improve data management quality, I am now in the process of creating logs to navigate the research. For example, this here is Repair.Database_AlbaneseV16Log. I can only show raw data and graphics and I can’t word it properly if I wish to publish in peer-reviewed journals… but in few, rather inelegant words, by performing this log, one can get a “bi-demsional representation of an interactional plane inhabited by 90 (human) speakers in 3 hours of interaction, whereby the circles represent muscular activation and how muscular movements are distributed in 409 interactional events” (Albanese, 2016).

Albanese, C (2016) Muscular Activation in Repair Sequences [Author’s pre-print, in peer review, forthcoming]

Software credits: R Core Team (2016)

Albanese, C (2016) Muscular Activation in Repair Sequences [Author’s pre-print, in peer review, forthcoming]

for more info please get in touch at info.ca@claudia-albanese.com

Leave a Reply