Step 1- Item analysis:
The item analysis aimed to identify the item of SBSRI that had poor psychometric properties, which would be excluded from the further analysis. For each item, we determined the following criteria for poorly performing items: distribution characterized by substantial departure from normality Kurtosis (-7, +7), skewness (-2, +2), floor or ceiling effects to be present if > 15% of the participants scored the lowest or highest possible score on the scales, as well as an ITC of 0.3 ≥ formed the basis for the elimination of an item. There are no definite cutoffs for item variance; nevertheless, one suggestion is to choose items with higher variance. However, if an observed value deviated from these standards and the item was regarded as critical to the domain measured by the instrument, such an item was retained(38,39) Table2 shows the 38 items that were preserved.
Step 2- Exploratory factor analysis
The KMO value was.925, and there was a statistically significance of Bartlett’s sphericity (χ2 = 4549.882, df=703, p < 0.000), indicating that the samples met the criteria for factor analysis. Communalities of the variables ranged between 0.52 and 0.75. Factor analysis yielded a 7-factor solution with an explained loading of 64.058% and eigenvalues >1. The Eigenvalues for the seven factors were 14.403, 3.086, 1.965, 1.478, 1.211, 1.154 and 1.045 respectively. According to the screen plot (Fig. 1), the slope of the curve became smooth at the seventh point, and factor 1 contributed 37.902% of the accumulated variance. The majority of items held their highest factor loading on the scale to which they theoretically belonged. Several items were also loaded on other factors, but never higher than.44, and the loading was 0.40 or above for all items in Table 3. We then sorted the items by factor loadings and removed an additional 2 items (15-22) from Factor 1, 2 items (21-19) from Factor 2 and 1 item (26) from Factor 5, given that their factor loadings were <.42. As a result, thus, 33 items were retained. We had ten items in Factor 1 (Cronbach’s alpha =.93), six items in Factor 2 (Cronbach’s alpha =.879), four items in Factor 3 (Cronbach’s alpha =.76), five items in Factor 4 (Cronbach’s alpha =.782), three items in Factor 5 (Cronbach’s alpha =.765) and five items in Factor 6 (Cronbach’s alpha =.74). The 7th factor has been removed, but its belonged question (se2) will be analyzed in factor 6.
More details of the factors are described below:
Factor 1-The first factor that explains the most variance in behavior is related to the goal-setting questions and motivational beliefs that are properly loaded into this factor, except for question 28, which was designed as a reward itself but loaded into this factor and we accepted it because we identified self-satisfaction as a motivational belief.
Factor 2-The second factor is related to the self-control part of the performance phase and includes questions about controlling the risk factors for delaying the sleep-wake program.
Factor 3-The third factor is related to some of the self-efficacy questions that deal with the ability to perform behaviors that make sleep time not be delayed.
Factor 4-The fourth factor of expectations is the positive result of having enough and regular sleep, which includes a positive effect on education, mood and relationships with others.
Factor 5-The fifth factor consists of three questions, each of which is a kind of self-reaction to the waking sleep program, and we considered it to be related to the self-reactive part.
Factor6-The second part of the questions is self-efficacy and includes the ability to perform behaviors that are required to have a regular sleep schedule and are related to the individual, such as the ability to control negative emotions.
Step 3-Confirmatory Factory Analysis (CFA):
Following the above EFA results, we constructed a model Using the two halves of the sample (n= 201) that included the 6 factors measured by the 33 items, these factors included 3 to 9 items, with the factor loadings ranging between 0.51 and 0.88, with the exception of two items in factor 1, which were deleted. The correlations among factors ranged from 0.31 to .78. Model fit results showed that Chi-square was significant (X2 = 729.199, df = 458, p = 0.00), and the values of CFI (.920), GFI (.826), RMSEA (.053), and PCFI (.71). were all within the proper range Overall, we deemed the model fit acceptable. Figure 2 displays the path model and the factor loadings within each latent factor. Note that the factor loadings of the items are majority greater than 0.50 and can be considered important for the associated items.
Step 4 - tests of reliability:
Step 5-Concurrent Validity
Concurrent validity was explored by correlating the SBSRI with the Adolescent Sleep Hygiene Scale-Revised (ASHS-r) self-reported was used to measure how often sleep hygiene behaviors or events occurred with each item rated on a 6-point scale (never, 0% to always, 100%). Scores are reversed, so higher scores indicate better sleep hygiene. This scale has reliability with Cronbach's alpha of internal consistency of 0.71 in the Iranian version (40).
The findings indicated a positive and significant correlation between the SBSRI and ASHS (r = .58, p < .001). demonstrates good concurrent validity (Table 4).
Finally, the predictive validity of the SBSRI was examined using correlational analysis and level of sleepiness. a negative linear relationship between Increasing sleepiness scores and decreasing the sleep behavior self-regulation was found. The results of the correlations between the SBSRI and Cleveland Adolescent Sleepiness displayed in Table 7, showed that Predictive validity was also found to be Medium