Bean

Patterns

A catalog of design and cognitive patterns that we identified when working in the spreadsheet environment. These are curated here as a sense-making device to help build a vocabulary and navigate around the spreadsheet design space.

Intrinsic Motivation: High vs. Low

A two part division brought in Advait / Gordon paper on how people's motivation levels determine their behaviours. It is interesting to see how people with low intrinsic motivation can still use the spreadsheet environment without learning much of the formal features of the software. This table nicely captures how these behaviors differ:

	Low Intrinsic Motivation	High Intrinsic Motivation
Feature Discovery	Passive discovers features on being informed by a colleague or received as documentation in a spreadsheet	Actively seeks out information in external resources
Expertise Acquisition	Informal, opportunistic, and social	Trial and error and less likely to be social. Usage creates learning opportunities
Attention Investment	Need strong evidence of reward from using a technique or feature	Bricoleur attitude. Lower threshold for evidence of reward

Attention Investment Tradeoff

The attention one is willing to invest before they reap the rewards for their investment. People with high intrinsic motivation have a high drive to invest their attention even if there's a low chance of getting a reward out of their efforts. This is a tradeoff because one has to draw the limit somewhere before there are diminishing returns for the effort they put in. Advait / Gordon mentions this idea in the context of how users with different intrinsic motivations pick up tooling expertise.

Reinhart-Rogoff Error

The paper was used as an advocacy for austerity in fiscal policy as they found an inverse relationship between gross external debt and GDP. The paper was cited by Paul Ryan in US and Olli Rehn in Europe to push for pro-austerity measures after the 2008 financial crisis.

But it turned out that there was a typo in the results which affected the result stated. The paper originally stated that there was a -0.1% decrease in GDP, but with the rows included it was 2% increase.

There were also other criticisms about how they interpreted the data that lead to their conclusions. Further papers by Reinhart-Rogoff found much lower impact on the GDP.

Reinhart and Rogoff responded by saying that the central point of the paper stands despite the discrepancy in the data. Moreover, they pointed out a growing literature and said that the consensus there is broadly that there is an “overhang threshold” beyond which the correlation between debt and growth becomes negative.

A good chunk of spreadsheet literature has been devised to avoid this kind of errors.The Lish paper points out how their cursor selection allows for preventing “out-by-one” errors.

Data Grain

Data grain is the level of detail at which data is stored in tables. Depending on the problem at hand, the data can be stored at finer levels of detail or could be coarse-grained and stored in a way to give a big-picture view of the information.

Marginalia

When working with spreadsheets, users make use of the adjacent cells to do important cognitive work. They use the cells adjacent to their data organizations to make comments, add descriptions/summaries, or perform experimental calculations. These craft practices help develop control and confidence in the userʼs on the data they are dealing with. It could also add semantic information about the data.

Master Spreadsheet

There occurs a prominent pattern of having an immutable master table and a separate area where derived data is dealt with. The data in the master spreadsheet is kept intact while the derived data is played around with to build confidence and develop one's understanding. This kind of experimentation builds a sense of trust and ownership.

Tidy Data vs. Untidy Data Dichotomy

Tidy data has been given a formal description by Wickham but here we use that in a more general sense of any prescriptive formal definition which requires the data work done in the spreadsheet to be compliant with this definition.

Untidy data is the organization of data in formats other than a neatly structured table as they fit the sense making process of the user. This could include metadata and marginalia that doesn't fit into a tabular structure.

Follow Us for Updates!

We are in the process of rolling out new content to this section over the next few days. Stay tuned.