RESULTS
The sociodemographic and obstetric characteristics of the participants are presented in Table 1. The mean age of the participants was 28.00±4.96 (min: 18, max: 43), their mean number of pregnancies was 2.01±1.18 (min: 1, max: 9), their mean number of childbirths was 1.77±0.98 (min: 1, max: 8), their mean number of miscarriages/curettages was 1.19±0.55 (min: 1, max: 4), and their mean number of living children was 1.77±0.97 (min: 1, max: 8). It was determined that 6.5% of the participants had chronic diseases (30.8% hypothyroidism, 30.8% diabetes mellitus, 19.2% asthma, 11.6% hypertension, 3.8% rheumatism, and 3.8% Crohn’s disease). While 18.8% of the participants stated that they experienced pregnancy complications (54.6% abortus imminens, 25.3% bleeding, 18.7% premature rupture of membranes, and 1.3% diabetes mellitus), 6% said they experienced complications during labor (45.9% difficult labor, 37.5% preterm labor, 12.5% cord entanglement, and 4.1% cord prolapse), and 6.5% reported that the health status of their babies was not good (73.1% respiratory distress and 26.9% preterm birth).
In the Kaiser-Meyer-Olkin (KMO) test that was conducted to decide whether the sample size was sufficient, the KMO value was found to be 0.806. KMO values greater than 0.7 are generally considered satisfactory.20,24,26,35,36 Accordingly, it was determined that the sample size of this study was “highly sufficient” to conduct factor analyses. The result of the Bartlett’s test of sphericity revealed an acceptable chi-squared value (χ2(276)=2571.919; p<0.01) (Table 2).
The 39-item Maternal Emotional Well-Being Scale was developed in this study based on three dimensions in line with Jean Ball’s theory. These dimensions were “social support, healthcare and education, and mother’s personality and experiences”. Accordingly, to identify the factor structure of the scale, an EFA was performed. Principal component analysis and varimax rotation were used in the EFA. The minimum factor load value to be met was determined as 0.40. Fifteen items that were found to have factor load values smaller than 0.40 in the EFA (items 1, 2, 3, 4, 10, 11, 12, 13, 14, 22, 23, 24, 26, 38, and 39) were removed from the scale (Table 2).
According to the varimax rotation results, it was observed that a four-factor structure was applicable. These factors explained 51.824% of the total variance in the measured characteristic (Table 2). In multi-factor constructs, variance explanation rates between 40% and 60% are considered adequate.20,21,22,24,26 This shows that a defined factor has a sufficient contribution to the total variance.
As the factor loads of items 15, 27, 28, 34, and 35 were below 0.40 according to the CFA results, these items were removed from the scale (before the items were removed: CMIN=555.970, CMIN/df=3.731, RMSEA=0.118, CFI= 0.818, GFI= 0.775). Modifications were made for items 30, 33, and 37 under the Healthcare and Education factor, item 15 under the Social Support factor, and item 5 under Mother’s Personality and Experiences factor to improve the model (for the improvements, covariances were drawn between error terms with high modification index (MI) values), and the resulting goodness-of-fit index values were CMIN=180.685, df=83, CMIN/df=2.177, RMSEA=0.077, CFI=0.937, and GFI=0.896, which showed an excellent fit.20,27-30According to the results of the CFA, in total, 9 items were removed from the scale. The path coefficients of all items under Factors 1, 2, and 3 were found to be statistically significant. Based on the standardized path coefficients, the item with the greatest effect on F1 was item 36 (β0: 0.718), the item with the greatest effect on F2 was item 17 (β0: 0.887), and the item with the greatest effect on F3 was item 8 (β0: 0.881) (Table 3).
The first-level CFA results of the scale are shown in Figure 2. Accordingly, in the scale that consisted of 15 items, the lowest factor load value was 0.43, while the highest was 0.89.
In reliability analyses, it is accepted that Cronbach’s alpha (0≤α≤1) coefficients in the range of 0.61-0.80 indicate high reliability, while those in the range of 0.81-1.00 indicate excellent reliability.21,34 It was determined that the Maternal Emotional Well-Being Scale, its social support dimension, and its mother’s personality and experiences dimension had excellent levels of reliability, whereas the healthcare and education dimension had high reliability. The overall reliability coefficient of the scale was calculated as 0.860 (Table 4).
In the reliability analyses of the scale, it was determined that all corrected item-total score correlation coefficients were greater than 0.20 and adequate (Table 5).
According to the results of the Hotelling’s T² analysis conducted to test the presence of response bias in the scale, it was found that the participants responded to the items based on their own opinions, and there was no response bias (Hotelling’s T2=397.260, p=0.000). Tukey’s test of additivity was also conducted to determine the additivity of the scale. As the significance level found in the test was p=0.000, it was determined that the factors of the scale were additive (non-additivity: 0.000). That is, a single total scale score could be obtained by the addition of the scores of all scale items.