Hypothesis Analysis
Weighted Calculation Method
Applies to: Developmental Affordance (DA) groups and Maslow levels.
Each image belongs to a high-level search category \((c_i)\).
Let \(G\) be the set of DA groups or the set of Maslow levels.
Definitions
- \(n_{ig}\): object count in image \(i\) for group \(g \in G\)
- \(N_{c}\): number of images in category \(c\)
- \(c_i\): category of image \(i\)
- \(O_i\): total objects in image \(i\) across groups
\[ O_i = \sum_{g \in G} n_{ig} \]
1) Per-image category weight
\[ w_{\mathrm{img}}(i) = \frac{1}{N_{c_i}} \]
2) Per-object weight
\[ w_{\mathrm{obj}}(i) = \frac{w_{\mathrm{img}}(i)}{O_i} \quad \text{(drop image if } O_i = 0 \text{)} \]
3) Weighted value per group (image level)
\[ x_{ig} = n_{ig} \cdot w_{\mathrm{obj}}(i) \]
4) Row-mass check
\[ \sum_{g \in G} x_{ig} = w_{\mathrm{img}}(i) \]
5) Column totals (dataset level)
\[ X_g = \sum_i x_{ig} \]
6) Optional normalisation (per-group sums to 1.0)
\[ \tilde{x}_{ig} = \frac{x_{ig}}{X_g} \qquad \Rightarrow \qquad \sum_i \tilde{x}_{ig} = 1 \]
Outputs
Weighted (stats):
- Weighted + normalised (figures):
- Removed list: images with
- Raw (Unweighted) Calculation Method (DA and Maslow)
Where is the raw count of objects in image for category .
This ignores category size and object counts — any presence is 1, absence is 0.