Visual Argument Structure Tool (VAST) by Leising, Grenke & Cramer
2024-10-25
Conjecture: We have a refined methodology¹ to test theories (e.g., experimental designs, statistical methods, preregistration, …). But we had (so far) no good methodology for constructing theories.
¹ “A scientific methodology is an ordered series of steps that assist a researcher in reaching a desired end state from a specified starting point.” (Borsboom et al., 2021)
Leising, D., Grenke, O., & Cramer, M. (2023). Visual Argument Structure Tool (VAST) Version 1.0. Meta-Psychology, 7. https://doi.org/10.15626/MP.2021.2911
This is a brief introduction into some of the core features of VAST. In doing this, I will simplify a few features and ignore others completely.
Six types of relationships:
I
in the box symbolizes a concept
I
) are abstract and basically arbitraryn
stands for a naming relationship""
in VAST displays.A concept may have any number of names. Here, all three names are synonyms regarding concept I
.
This expresses the idea that the same name is used for conditions that may not be the same. This name might thus be a homonym.
F
and T
F
Note: In the original VAST paper, the Data concept has a different shape (but that shape is not available in the Graphviz system used to create the graphs in this presentation):
r
denotes reasoning relationshipsp
stands for a predictionc
stands for causal effectsThis display is supposed to capture the idea that …
Readability may often be promoted by setting naming relationships aside.
X
→ Y
: this relationship is considered relevant and positive (i.e., the more X the more Y)p
: “wearing glasses” makes it 70 percent likely for a person to also be “smart”r
: It is 90 percent reasonable to assume someone “is in love with you” when that person “giggles a lot while talking to you”c
: being “obese” makes it 50 percent likely for someone to develop “Diabetes Type II” as a consequencec
)p
)p
-path is the next author’s theoretical c
-path(A): “Mira is assumes that she has presented Robert’s and Ayse’s views correctly (with strength > 0.5, which is the default for IS and OUGHT elements). This implies that these two have different and incompatible statements about the single causal effect on I
.
(B): Mira says (normatively) that Robert and Ayse should get along well.
(C): Mira is certain (relationship strength = -1
) that the big box below (i.e., the incompatible views) causally leads to Robert and Ayse not getting along well.
These slides are part of the course Formal modeling in psychology at LMU Munich