messy data is at risk of, but isn’t inherently, poor quality data

in clinical data, other ‘mess’ examples given: ‘incomplete or duplicated patient records; omission of expected clinical assessment information; incorrect variable classification; inconsistent formatting; insufficient identification labels; inconsistencies between variables; incorrect coding, data entry errors; and/or multiple variables presenting in a single column.’

poor quality data, less useful (less ‘explanatory power’) - ‘inefficient, inaccurate, invalid, unreliable, incomplete, irrelevant, too broad/ detailed, and/or out-of-date’

embracing messy data

transfer of memory from the physical/mental to the digital LOSES information, a lossy process - this should be embraced and worked around rather than treating partial data as full data when this would be inaccurate, non-representational, harmful omission

reduction of the qualitative to the quantitative

Jones (2022) discusses the analysis of ‘messy’ data of vulnerable patient groups in naturalistic (NHS) settings and emphasises the importance of ‘good quality’ data over complete or ‘clean’ data in making clinical decisions with reasonable certainty (Jones, 2022)

Formalised records have been found to be often misleading, ineffective and harmful for TGDI patients (Singer, 2015), (Alpert et al., 2023), (Kirkland, 2021)

an external record as somewhere between an index and direct recall

not an experience in totality but still useful alone. partial info, re-enlivened by proper context/surroundings/etc, experiential knowledge of specifics partially transferred when you have the requisite lived experience
“the records require an active input of my internal memory in order to be more precisely deciphered.” (Smolicki, 2017)

References

Alpert, A.B., Mehringer, J.E., Orta, S.J., Redwood, E., Hernandez, T., Rivers, L., Manzano, C., Ruddick, R., Adams, S., Cerulli, C., Operario, D. and Griggs, J.J. (2023a) ‘Experiences of Transgender People Reviewing Their Electronic Health Records, a Qualitative Study’, Journal of General Internal Medicine, 38(4), pp. 970–977. doi: 10.1007/s11606-022-07671-6. ^1cc529

Jones, A.M. (2022) ‘Improving Healthcare Access and Engagement with Vulnerable Groups Through the Transformation of Complex and Messy Naturalistic Clinical Data’ University of Brighton. ^ef569c

Kirkland, A. (2021) ‘Dropdown rights: Categorizing transgender discrimination in healthcare technologies’, Social science & medicine, 289, pp. 114348. doi: 10.1016/j.socscimed.2021.114348. ^e95683

Singer (2015) ^e0fcea

redlinks are important

wiki redlink - the absent link, the lacuna, just as important Ted Nelson on hypertext and (Rettburg, 2005)
highlighting that which is absent
linkages we don’t yet see, information we don’t yet have - adjacent stuff may be relevant - others’ annotations and marginalia can be useful - “Locating Digitised Marginalia”
seeds for future work
see in (Smolicki, 2017), ‘what we might say’

messy data is at risk of, but isn’t inherently, poor quality data

embracing messy data

Mentions

References

Resources

redlinks are important