RT. Figure 2 shows the RT in categorization answers as a function of stimulus type. A main effect of Stimulus was found, (F1, 57 = 14.02, p < .001, η2 = .19), indicating a fast RT for the categorization of house stimuli (mean RT = 1494.89) compared to face stimuli (mean accuracy = 1564.32). An interaction between Stimulus and the RPBS score was found, (F1, 57 = 8.34, p = .005, η2 = .12) reflecting that the effect of stimulus on RT was modulated by the score on the RPBS. As expected, a main effect of Noise was found, (F3,171 = 86.49, p < .001, η2 = .60), indicating that with increased levels of visual noise the RT increased (Fig. 2). Noise also interacted with the score on the RPBS (F3,171 = 13.97, p < .05, η2 = .19). These main effects were qualified by a significant interaction between Stimulus and Noise, (F3,171 = 26.0, p < .001, η2 = .31) and an interaction between Stimulus, Noise and the RPBS score, (F3,171 = 9.96, p < .001, η2 = .14).
This three-way interaction can be seen in Fig. 2, which shows that whereas skeptics responded differently to face than to house stimuli with higher levels of visual noise, paranormal believers' accuracy for face and house stimuli was unaffected by visual noise. Two different ANOVAs were done for the groups of skeptics and paranormal believers to examine this three-way interaction (for suggestions on the interpretation of three-way interactions in mixed design ANOVAs26). For skeptics we observed a trend of a significant main effect of Stimulus (F1,27 = 3.66, p = .06, η2 = .12), a main effect of Noise, (F3,81 = 170.63, p < .001, η2 = .86) and a significant interaction between Stimulus and Noise, (F3,81 = 28.44, p < .001, η2 = .51). Post-hoc t-tests indicated that there was a significant difference between house and face stimuli on the RT in 40%, 50%, and 60% visual nises. In addition, for believers only a main effect of stimuli (F1,54 = 15.35, p = .001, η2 = .46), and Noise were observed, (F3,54 = 109.80, p < .001, η2 = .85) and the interaction between Stimulus and Noise was not significant (F3,171= 40.907, p > .005, η2 = .69). Post-hoc t-tests indicated that the RT was long for face compared to house stimuli in 50%, and 60% visual noises.
Error rate. Figure 3 illustrates the error rate in categorization answers as a result of stimulus type. A main effect of Stimulus was found (F1,57= 14.02, p < .001, η2 = .19), indicating a more error for the categorization of face stimuli (mean accuracy = .315) compared to house stimuli (mean RT = .182). An interaction between Stimulus and the RPBS score was found (F1,57 = 3.58, p = .05, η2 = .06), reflecting that the effect of stimulus on error was modulated by the score on the RPBS. As expected, a main effect of Noise was found (F3,171 = 32.45, p < .001, η2 = .36), indicating that with increased levels of visual noise the error increased (see Fig. 3). Noise also interacted with the score on the RPBS (F3,171 = 5.37, p < .05, η2 = .08). These main effects were qualified by a significant interaction between Stimulus, Noise and the RPBS score (F3,171= 4.82, p < .003, η2 = .07).
This three-way interaction can be seen in Fig. 3, which shows that whereas skeptics responded differently to face than to house stimuli with higher levels of visual noise, paranormal believers' errors for face and house stimuli were unaffected by visual noise. Two different ANOVAs for the skeptics and paranormal believers groups were carried out to investigate this three-way interaction (for suggestions on the interpretation of three-way interactions in mixed design ANOVAs 26). For skeptics we observed a trend significant main effect of Stimulus (F1,27 = 96.344, p < .006, η2 = .78), a main effect of Noise, (F3,27 = 221.63, p < .001, η2 = .89) and a significant interaction between Stimulus and Noise, (F3,81 = 20.31, p < .001, η2 = .42). Post-hoc t-tests indicated that the error was more for face compared to house stimuli in 50%, 60%, and 70% visual noises. In addition, for believers only a main effect Noise was observed (F3,54 = 155.76, p < .001, η2 = .89), and the interaction between Stimulus and Noise was no significant (F = 2.21, p = .09, η2 = .11). Post-hoc t-tests indicated that the there was no significant difference between house and face stimuli on error in 50% visual noise in believers.
Regression Analysis
The impact of demographic factors (age, gender, and education) and paranormal beliefs on the error rate was assessed using linear regression analysis. The stimulus category for which both groups responded differently to face and home stimuli was the face stimuli with all visual noise (40%, 50%, 60%, and 70%), hence we especially focused on error rate in these stimuli. Age and gender were significant predictors of the error rate, as shown in Table 1. While accounting for the demographic variables, the RPBS was a significant predictor of the error rate as expected (Fig. 4A and Fig. 4B).
Table 1
Hierarchical multiple regression analyses predicting the error rate for face stimuli with all visual noise.
Model | B | SE | b | R2 | t | p |
---|
Step 1 | Intercept | 1.796 | .335 | | .17 | 5.354 | .001 |
Age | − .037 | .017 | − .401 | -2.105 | .040 |
Education | .053 | .132 | .078 | .402 | .689 |
Gender | .202 | .116 | .224 | 1.735 | .089 |
Step 2 | Intercept | 1.246 | .348 | | .31 | 3.576 | .001 |
Age | − .053 | .017 | − .584 | -3.189 | .002 |
Education | .220 | .131 | .322 | 1.677 | .099 |
Gender | .172 | .107 | .191 | 1.611 | .113 |
Paranormal beliefs | .022 | .007 | .421 | 3.344 | .002 |