Methods
Subjects
In Study 2, we aimed to conduct a closer psychometric analysis of the Czech SPQ and SBQ. As the previous study showed, the scores on both scales did not differ depending on whether the translated scale was administered before or after the original version (except for SpB at p = 0.031). Therefore, for the following analyses, we decided to use all the data from the Czech SPQ (N = 662) and SBQ (N = 634) no matter the order of administration. Additionally, we completed this sample with unpublished data from our previous research projects focused, in general, on emotions triggered by various animals.
Doing so, we collected a sample of another 3 201 individuals who completed the Czech SPQ. Moreover, 3 147 of them also completed the Disgust Scale – Revised (DS-R; Haidt et al., 1994, modified by Olatunji et al., 2007, Czech translation by Polák et al., 2019), 2 585 subjects completed the Snake Questionnaire (Klorman et al., 1974, Czech translation by Polák et al., 2016), and 399 did the Czech SBQ. In this pooled sample of 3 863 participants aged 15-88 years (mean age 29.9 ± 0.2), there was a considerably higher proportion of women (2 778 vs 1 066 men, 9 subjects did not disclose their gender). Most of the participants have completed high school (N = 2 018) or a university degree (N = 1 489), only a minority have had elementary school as their highest completed education (N = 285). Finally, we also gathered information on their field of study and categorized the subjects as having a biology (N = 975) or non-biology background (N = 2 888) as this has been found by previous studies as an important factor affecting animal fears (Polák et al., 2016, 2020a).
Psychometrics
Snake Questionnaire (SNAQ)
The SNAQ is a 30-item self-report scale to assess the verbal-cognitive component of snake fear. Each item is a fearful or non-fearful statement related to snakes. Participants rate each item as true or false. The instrument is scored by assigning a “1” to each true response and “0” to each false response, items are reversed-scored. A total score (ranging from 0 to30) is calculated by summing all ‘true’ statements and it serves as a measure of the degree of phobic fear (Wright et al.,2002; Wikström et al., 2004).
Disgust Scale – Revised (DS-R)
The DS-R is a self-report personality scale to assess individual differences in propensity to disgust. There are 25 disgust elicitor items loading on one of the three factors (core disgust, animal reminder disgust, contamination-based disgust) and two catch questions (item 12 and 16) to identify those respondents that are not paying attention to the task or do not take it seriously. Each item is rated by the participant on a 5-point Likert scale from 0 (“Strongly disagree/Not disgusting at all“) to 4 (“Strongly agree/Extremely disgusting”). The total score (ranging from 0 to 100) is calculated by summing scores on all the 25 disgust elicitor items but three (item 1, 6, 10) that are reverse scored. Similarly, subscale scores may be calculated. All the participants that do not give valid answers on the catch questions should be dropped. The DS-R demonstrates acceptable Cronbach’s alpha estimates for the overall internal consistency (0.84) and the three subscales (core disgust: 0.74; animal reminder disgust: 0.78; contamination-based disgust: 0.61; Olatunji et al., 2007, which was replicated by van Overveld et al., 2011).
Statistical analysis
First, reliability of the Czech SPQ and SBQ was calculated using the split-half method, internal consistency was expressed as the Cronbach’s alpha. To normalize nonlinear score distributions, we applied the McCall area transformation with data continuity adjustment (McCall, 1922). Using the transformed z-scores we calculated norms for our sample. We also computed a Spearman correlation coefficient between scores on the SPQ, two SBQ subscales (SpB and SrB), and DS-R to demonstrate convergent validity. Discriminant validity was expressed by a Spearman correlation coefficient between the SPQ/SBQ and SNAQ scores. Next, we employed a General Linear Model (GLM) for quasibinomial distribution (log-link function) to analyse the effect of sex, age, highest level of education (categorized as either elementary school, high school, or university), and biology vs. nonbiology background on the SPQ scores. The full model was further reduced according to the Akaike information criterion (AIC).
We have also performed a redundancy analysis (RDA) as implemented in the R package vegan (Oksanen et al., 2020; R Core Team, 2020) to quantify contribution of the explanatory variables (respondent’s gender, age, highest education, biology background, three subscale scores on the DS-R, and the SNAQ score) on all SPQ item scores. The RDA is a multivariate direct gradient method (ter Braak & Šmilauer, 2018), which extracts and summarizes the variation in a set of response variables and permits to plot both the response and explanatory variables to a space defined by the extracted gradients to detect redundancy (i.e. shared variability). Statistical significance of the gradients was confirmed by permutation tests.
To evaluate effects of the above factors on the SBQ scores, we employed linear models with square-root transformation improving normality of the data distribution. The full model was further reduced according to the Akaike information criterion (AIC). Finally, we conducted a receiver operating characteristic (ROC) curve analysis and calculated the Youden index (J = maximum {sensitivity + specificity - 1}; Schisterman et al., 2005) to find an ideal cut-off point on the two SBQ subscales for potential spider phobia as identified by the SPQ score. Calculations were performed in SPSS, version 22 (IBM Corp., 2013) and R, version 3.6.3 (R Core Team, 2020).
Results
SPQ
The distribution of Czech SPQ scores significantly deviated from normality (SW = 0.90, df = 3 863, p < 0.001) with skewness 0.77 ± 0.04 and kurtosis -0.58 ± 0.08 (see Fig. 1 for raw scores distribution). The translated scale demonstrated high internal consistency (Cronbach’s α = 0.94) and split-half reliability (Guttman split-half coefficient λ = 0.92). The mean score was 9.02 ± 0.13 (SD = 8.04) and median 6.00, standardized McCall transformed scores and respective norms can be found in Table 1.
Table 1
Raw and transformed Spider Questionnaire scores with norms; McCall area transformation with a continuity correction (cumulative frequency) was applied.
Raw score
|
Absolute frequency
|
Relative frequency
|
Cumulative frequency
|
Z-score
|
Percentile
|
0
|
327
|
0.0846
|
0.0846
|
-1.724
|
4.2
|
1
|
377
|
0.0976
|
0.1822
|
-1.110
|
13.3
|
2
|
350
|
0.0906
|
0.2728
|
-0.747
|
22.8
|
3
|
299
|
0.0774
|
0.3502
|
-0.491
|
31.2
|
4
|
244
|
0.0632
|
0.4134
|
-0.301
|
38.2
|
5
|
196
|
0.0507
|
0.4641
|
-0.154
|
43.9
|
6
|
185
|
0.0479
|
0.5120
|
-0.030
|
48.8
|
7
|
142
|
0.0368
|
0.5488
|
0.076
|
53.0
|
8
|
129
|
0.0334
|
0.5822
|
0.165
|
56.5
|
9
|
102
|
0.0264
|
0.6086
|
0.241
|
59.5
|
10
|
114
|
0.0295
|
0.6381
|
0.314
|
62.3
|
11
|
105
|
0.0272
|
0.6653
|
0.390
|
65.2
|
12
|
99
|
0.0256
|
0.6909
|
0.462
|
67.8
|
13
|
85
|
0.0220
|
0.7129
|
0.530
|
70.2
|
14
|
78
|
0.0202
|
0.7331
|
0.592
|
72.3
|
15
|
106
|
0.0274
|
0.7605
|
0.665
|
74.7
|
16
|
93
|
0.0241
|
0.7846
|
0.747
|
77.3
|
17
|
102
|
0.0264
|
0.8110
|
0.834
|
79.8
|
18
|
90
|
0.0233
|
0.8343
|
0.926
|
82.3
|
19
|
73
|
0.0189
|
0.8532
|
1.010
|
84.4
|
20
|
85
|
0.0220
|
0.8752
|
1.099
|
86.4
|
21
|
84
|
0.0217
|
0.8970
|
1.206
|
88.6
|
22
|
76
|
0.0197
|
0.9166
|
1.321
|
90.7
|
23
|
62
|
0.0160
|
0.9327
|
1.437
|
92.5
|
24
|
59
|
0.0153
|
0.9480
|
1.558
|
94.0
|
25
|
53
|
0.0137
|
0.9617
|
1.694
|
95.5
|
26
|
47
|
0.0122
|
0.9739
|
1.849
|
96.8
|
27
|
38
|
0.0098
|
0.9837
|
2.029
|
97.9
|
28
|
29
|
0.0075
|
0.9912
|
2.240
|
98.7
|
29
|
19
|
0.0049
|
0.9961
|
2.493
|
99.4
|
30
|
10
|
0.0026
|
0.9987
|
2.796
|
99.7
|
31
|
5
|
0.0013
|
1.0000
|
3.217
|
99.9
|
The reduced GLM model revealed a significant effect of gender, age, education level, biology background, core and animal reminder disgust score of the DS-R (all ps < 0.001), and finally the SNAQ score (p = 0.007), see Table 2 for SPQ scores according to gender, education level, and biology background. Contrary to that, the effect of contamination disgust score was not significant (p = 0.682). Parameters estimated from the reduced GLM model were as follows: intercept: -1.910, p < 0.001; male gender: -0.476, p < 0.001; age: -0.016, p < 0.001; high school education: 0.175, p < 0.001; elementary school education: 0.239, p = 0.031; biology education: -0.311, p < 0.001, core disgust: 0.047, p < 0.001; animal reminder disgust: 0.023, p < 0.001; and SNAQ score: 0.011, p = 0.007. We also checked for convergent and divergent validity by calculating a Spearman correlation coefficient between the SPQ and other assessments (SBQ, DS-R, and SNAQ), see Table 3 for more details.
Table 2
Descriptive statistics of raw scores obtained from the Czech translation of the Spider Questionnaire categorized according to gender, education level, and biology background.
|
|
N
|
Percent
|
Mean
|
Median
|
95% CI of mean
|
Overall
|
|
3863
|
100%
|
9.02
|
6.00
|
8.77 - 9.28
|
Sex
|
Men
|
1072
|
27.8%
|
5.44
|
3.00
|
5.10 - 5.79
|
Women
|
2791
|
72.2%
|
10.40
|
8.00
|
10.09 - 10.71
|
Education level
|
Basic school
|
285
|
7.4%
|
10.78
|
8.00
|
9.83 - 11.73
|
High school
|
2018
|
52.2%
|
9.46
|
7.00
|
9.10 - 9.81
|
University
|
1489
|
38.5%
|
8.01
|
5.00
|
7.62 - 8.40
|
Biology background
|
No
|
2888
|
74.8%
|
9.39
|
7.00
|
9.09 - 9.69
|
Yes
|
975
|
25.2%
|
7.93
|
5.00
|
7.47 - 8.40
|
Table 3
Spearmann correlation coefficients between the Spider Questionnaire (SPQ), Spider Phobia Beliefs Questionnaire (SBQ), Disgust Scale-Revised (DS-R), and Snake Questionnaire scores (SNAQ). The SBQ has two subscales, the spider-related (SpB) and self-related beliefs (SrB).
|
SBQ
|
DS-R
|
SNAQ
|
|
SpB
|
SrB
|
Total score
|
Core
|
Animal rem.
|
Contam.
|
Total score
|
SPQ
|
0.727
|
0.788
|
0.401
|
0.394
|
0.319
|
0.223
|
0.324
|
SpB
|
|
0.828
|
0.387
|
0.350
|
0.368
|
0.188
|
0.219
|
SrB
|
|
|
0.363
|
0.321
|
0.366
|
0.153
|
0.208
|
All coefficients significant at the α = 0.01 level. |
The RDA model of SPQ item scores generated seven constrained axis which explained only 10.28% of the full variability. We then performed a permutation test (number of permutations = 20,000) to confirm the significance of each of the independent variables (constraints) in a sequential (‘type I’) test: gender, F1,2471 = 76.13, p < 0.001; age, F1,2471 = 31.60, p < 0.001; education level, F1,2471 = 5.76, p < 0.001; biology background, F1,2471 = 9.74, p < 0.001; core disgust, F1,2471 = 91.31, p < 0.001; animal reminder disgust, F1,2471 = 10.67, p < 0.001; and SNAQ, F1,2471 = 58.85, p < 0.001; for visualization of the RDA results see Figure 2).
Finally, to set a cut-off point for spider phobia and thus be able to estimate its prevalence in the Czech population, we adopted the mean SPQ score 23.76 (SD = 3.80) of 17 spider phobics reported by Fredrikson (1983) and calculated the 95% confidence interval (CI) as 23.76 ± 1.81 [21.95, 25.57]. We took its lower bound, i.e. SPQ score 22 as a cut-off point for potential spider phobia, which was reached by 398 subjects in our sample representing 10.3% that could be preliminary classified as spider phobics.
SBQ
Both the spider-related (SpB) and self-related beliefs score (SrB) on the Czech SBQ significantly deviated from normality (SpB: SW = 0.92, df = 1 086, p < 0.001, skewness = 0.93 ± 0.07, kurtosis = 0.34 ± 0.15; SrB: SW = 0.79, df = 1 086, p < 0.001, skewness = 1.48 ± 0.07, kurtosis = 1.57 ± 0.15 (for raw score distributions of SpB and SrB subscale, please see Additional file 1 and 2). The translated scale showed an excellent reliability, expressed either through internal consistency (Cronbach’s α = 0.98) or Guttman split-half coefficient (λ = 0.91). The mean SpB score was 24.65 ± 0.55 (SD = 18.16), mean SrB score 15.68 ± 0.61 (SD = 19.96).
Based on the reduced linear models, the SpB score was significantly affected only by the biology background (F(1, 392) = 10.62, p = 0.001), and DS-R subscale scores on core (F(1, 392) = 9.29, p = 0.002) and animal reminder (F(1, 392) = 55.84, p < 0.001); the model explained 16.02% of total variability. Parameters estimated from the reduced linear model were as follows: intercept: 2.955, p < 0.001; biology education: -0.373, p = 0.038; core disgust: 0.048, p < 0.001; animal reminder disgust: 0.067, p < 0.001. The SrB score was only affected by the gender (F(1, 392) = 33.15, p < 0.001) and score on animal reminder disgust (F(1, 392) = 43.95, p < 0.001); the model explained 16.96% of total variability Parameters estimated from the reduced linear model were as follows: intercept: 1.230, p = 0.004; male gender: -0.887, p = 0.002; animal reminder disgust: 0.101, p < 0.001. For SpB and SrB scores according to gender, education level, and biology background, see Table 4.
Finally, based on the previously found threshold of SPQ score 22 and higher signifying spider phobia, we identified 98 potential phobics in the subsample of subjects who also completed the SBQ. By calculating the Youden index for each coordinate of the ROC curve, we have found a cut-off point of 32.64 on the SpB subscale, which corresponds to sensitivity 0.867 and specificity 0.754; J = 0.621. The area under curve (AUC) was 0.871. For the SrB subscale, a cut-off point of 25.79 has been identified, which corresponds to sensitivity 0.847 and specificity 0.814; J = 0.661. The AUC for this ROC curve was 0.866.
Table 4
Descriptive statistics of raw scores obtained from the Czech translation of the Spider Phobia Beliefs Questionnaire categorized according to gender, education level, and biology background. The assessment is divided into two subscales, the spider-related (SpB) and self-related beliefs (SrB).
|
|
N
|
Percent
|
SpB
|
SrB
|
|
|
|
|
Mean
|
95% CI
|
Mean
|
95% CI
|
Overall
|
|
1086
|
100%
|
24.65
|
23.57 - 25.73
|
15.68
|
14.49 - 16.87
|
Sex
|
Men
|
291
|
26.8%
|
17.60
|
16.08 - 19.12
|
7.20
|
5.75 - 8.65
|
Women
|
795
|
73.2%
|
27.23
|
25.91 - 28.56
|
18.78
|
17.30 - 20.26
|
Education level
|
Basic school
|
120
|
11.3%
|
27.23
|
23.59 - 30.87
|
17.55
|
13.81 - 21.30
|
High school
|
446
|
42.0%
|
25.30
|
23.53 - 27.07
|
16.22
|
14.25 - 18.20
|
University
|
497
|
46.7%
|
23.34
|
21.87 - 24.81
|
14.55
|
12.94 - 16.17
|
Biology background
|
No
|
481
|
44.3%
|
28.47
|
26.67 - 30.27
|
19.06
|
17.13 - 20.99
|
Yes
|
605
|
55.7%
|
21.62
|
20.35 - 22.88
|
12.99
|
11.54 - 14.44
|