7

As part of some research I am doing on measuring documentation quality, I have come across the terms "accurate" and "believable" as two separate dimensions of information quality.

But the differences between them are not clear at all to me: they define accurate as "correct, reliable, and certified free of error" and believable as "true, real, and credible".

Looking at the dictionary didn't help me much either.

So I'm putting this out here - what is the difference between these two dimension as they relate to documentation quality?

The original source is here: Wang R. & Strong, D. (1996). Beyond accuracy: what data quality means to data consumers. J. of Man. Info. Sys., 12 (4), p.5-34)

migrated from techcomm.stackexchange.com Feb 20 '18 at 15:14

4

If one assesses this not in terms of data capture, quantitative analysis and other hard metrics, but from the user experience and sociological/psychological/behavioural direction, the implications are different.

For the reader (the end-user whose tasks, experience, and schematic understanding we're hoping to improve) believability is a matrix of authorial voice / tone, the perceived level of authority (based not on structural hierarchy but on skill or knowledge-based merit) of the writer and their data source(s), the relevance to the topic and / or meta topic being researched, and the clarity, structure and utility of presentation.

The term trustworthiness is germane to this discussion. Can I trust this source? Do they reliably present things in a way I can readily assess, apprehend and take action upon? Have they misled me previously? Does this information "feel" fast-but-loose, hastily-thrown-together, or worse yet apocryphal? Do I feel not only that my point of pain or confusion was clarified, but also reassured that my instructional source is of high value, that my time here was not squandered? do I feel my intelligence and competences have been understood and accounted for, or discounted out of hand?

I think this particular aspect of technical communications is now in a continually-escalating importance spiral due to the vast toxic seas of inaccurate, non-authoritative but easy-to-find rubbish purporting to be instructional content: especially in the peer-to-peer specific technique learning milieu.

The danger to user's perceptions of technical communications' value and believability is not necessarily in the direct instruction included in the specific skills tutorial or video, but rather the often deeply mistaken field-overview or meta comments thrown in by those peer coaches in passing: many users have taken something they "learned" from a trusted technique source and have constructed false understandings of far more critical and overarching ideas, only to later realise that these were deeply flawed. This leads to a certain degree of cynicism around instructional content.

How one goes about differentiating one's content as being of high trustworthiness seems critical to me; not only for actual acceptance of methods, modes and concepts, but to save our readers time and frustration, and to engage them further to improve both content and process.

I'm glad Yoel is putting time into researching this, and look forward to hearing about those results.

6

These terms come from the world of management information, data warehousing and data quality.

Accuracy is easy:

First, it has to be or the right value. Second, it has to precisely represent the value in consistent form in accordance with the business data model and architecture.

Believability is harder. It consists of the quality of the source and the quality of the processing, also called (data) provenance or lineage.

As with accuracy definitions in the field differ, a nice one is:

“the extent to which data are accepted or regarded as true, real and credible”

Be assured that data quality has lots more dimensions. Some references:
https://en.wikipedia.org/wiki/Data_quality
http://www.learn.geekinterview.com/data-warehouse/data-quality/what-is-data-accuracy.html
http://web.mit.edu/smadnick/www/wp/2007-11.pdf

  • I'm quite familiar with the categories and dimensions of information quality (especially those of Wang et al. in the literature). My research is based on (attempting to) apply them to the idea of documentation quality. You can see slides from my STC 2017 presentation here: sched.co/8thX – Yoel Feb 15 '18 at 6:56
2

The main difference is that when something is believable, it is or can be accepted as true without proof. To determine if something is accurate, however, implies that the something is or can be compared to an observation.

Believable can also be used to state an opinion, as in "I believe this piece of documentation is accurate". You can opt to believe me without further proof, or you can determine for yourself whether or not the documentation I'm talking about is accurate by comparing it to what it documents.

Related to your research on documentation quality, it may signify whether or not the respondent(s) have prior experience with the documentation in question. If a reader has experienced several issues of inaccurate descriptions within a piece of documentation that spans multiple versions, for example, the content may still be believable - although not accurate.

If I may offer an opinion: Technical communication should strive to be accurate, not believable. Inaccurate technical documents undermine their own very existence.

1

I don't agree with the answer of Paul.

Technical communication should be accurate, without doubts. If it is not accurate, people tend not to use it, so it will annihilate the purpose of technical documentation.

But technical documentation should be believable, for the same reason, else the people will start not to read it.

To be believable, the difficult points should be explained (maybe just in a footnote, or as reference). Readers must be sure that their (previously held) believes were wrong, not the technical documentation.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.