Table 4.
Linear mixed model (Table 2) was fitted to the average dwell times on Animate and Inanimate foveal interest areas (FIAs) with and without audio markers under the two viewing conditions that involved audio stimuli that referenced visual elements in the artworks and complex images: Contextual information and External perspective.
| Model 2: Effects of audio stimuli and social cues on gaze patterns across visual artworks and complex images in healthy adults | ||||||
|---|---|---|---|---|---|---|
| Effect | df | ChiSq | p | VS-MPR * | ||
| Viewing condition | 1, 17.96 | 1.911 | 0.184 | 1.182 | ||
| Foveal interest area (FIA) subcategory | 3, 10114.39 | 467.507 | < 0.001 | 3.236 ×10+280 | ||
| Viewing condition * FIA subcategory | 3, 10114.39 | 63.797 | < 0.001 | 5.382 ×10+37 | ||
| Fixed effects estimates | ||||||
| Term | Estimate | SE | df | t | p | VS-MPR * |
| (Intercept) | 1426.346 | 146.740 | 17.960 | 9.720 | < 0.001 | 1.445 × 10+6 |
| Reference viewing condition: External perspective | ||||||
| Viewing condition: Contextual information | −202.873 | 146.740 | 17.960 | −1.383 | 0.184 | 1.182 |
| Reference FIA subcategory: inanimate | ||||||
| FIA subcategory: animate | −112.519 | 31.893 | 10112.425 | −3.528 | < 0.001 | 112.542 |
| FIA subcategory: animate audio marker | 1090.315 | 39.376 | 10113.190 | 27.690 | < 0.001 | 1.013 × 10 +159 |
| FIA subcategory: inanimate audio marker | −195.003 | 48.323 | 10103.286 | −4.035 | < 0.001 | 682.866 |
| Interaction effects | ||||||
| Animate FIAs * Contextual information viewing condition | −148.690 | 31.893 | 10112.425 | −4.662 | < 0.001 | 9168.838 |
| Animate audio marker FIAs * Contextual information viewing condition | 265.278 | 39.376 | 10113.190 | 6.737 | < 0.001 | 8.710 ×10+8 |
| Inanimate audio marker FIAs * Contextual information viewing condition | −148.690 | 31.893 | 10112.425 | −4.662 | < .001 | 9168.838 |
| Estimated marginal means | ||||||
| 95% CI | ||||||
| Viewing condition | Foveal interest area subcategory | Estimate | SE | Lower | Upper | |
| Contextual information | Animate | 962.264 | 212.045 | 546.662 | 1377.865 | |
| External perspective | Animate | 1665.390 | 209.406 | 1254.962 | 2075.818 | |
| Contextual information | Inanimate | 694.569 | 207.897 | 287.099 | 1102.038 | |
| External perspective | Inanimate | 592.538 | 207.821 | 185.216 | 999.860 | |
| Contextual information | Animate audio marker | 2579.066 | 211.888 | 2163.774 | 2994.359 | |
| External perspective | Animate audio marker | 2454.256 | 219.165 | 2024.700 | 2883.813 | |
| Contextual information | Inanimate audio marker | 657.992 | 225.433 | 216.152 | 1099.833 | |
| External perspective | Inanimate audio marker | 1804.694 | 218.711 | 1376.029 | 2233.359 | |
Model terms tested with the Satterthwaite test method.
The following variable is used as a random effects grouping factor: “Image.”
Type III sum of squares.
The intercept corresponds to the (unweighted) grand mean; for each factor with k levels, k - 1 parameters are estimated with sum contrast coding. Consequently, the estimates cannot be directly mapped to factor levels. The estimated marginal means below provide estimates for each factor level/design cell and their differences.
*Vovk-Sellke maximum p-ratio: Based on a two-sided p-value, the maximum possible odds in favor of H1 over H0 equal 1/(-e p log(p)) for p ≤ 0.37 (Sellke et al., 2001).
Fixed effects were calculated for “viewing condition” with 2 levels (Contextual information and External perspective) and “FIA subcategory” with 4 levels (Animate, Inanimate, Animate audio marker, and Inanimate audio marker), with “Image” (N = 20) selected as a random effects grouping factor. The LMM random-effect structures for “Image” included only the intercept as there was not enough data to estimate the random slopes for the FIA subcategories with audio markers. All random slopes involving “Viewing condition” were automatically removed from the model by JASP as the factor “Viewing condition” did not vary within the levels of random effects grouping factor “Image.” The dwell-time data that were included in the analysis were recorded between 750 and 20,000 ms after image presentation. Estimated marginal means were calculated for the four FIA subcategories across the two Viewing conditions. A control analysis found no main effects for age or sex, so the presented results were averaged across all research participants in the study (N = 37). Total number of observations: 10,138. The bold values indicate a statistically significant result, defined as a p value smaller than 0.05.