Factor analysis methods (exploratory and confirmatory) by the help of SPSSV19 and LISRELV8.80 were utilized to analyze research findings. The overall status of data was examined before performing relevant analyses. The out of range data was initially identified and modified by referring to original questionnaires, and then the outlier data was examined using a boxplot. The results indicated that there was no outlier data among data. The sample group of study was in the age range of 17 to 42 years. The mean±standard deviation of the sample group was 27.91±5.74. Results of analyses are presented below after reviewing the assumptions.
Table 1: Frequency distribution of demographic characteristics(N=240)
Variable
|
Number
|
Percentage
|
Education
|
Under the diploma
|
82
|
34.2
|
diploma
|
98
|
40.8
|
Up the diploma
|
9
|
3.8
|
Bachelor
|
32
|
13.3
|
Master
|
19
|
7.9
|
Pregnancy week
|
First trimesters
|
17
|
7.1
|
Second trimesters
|
91
|
37.9
|
Third trimesters
|
132
|
55.0
|
Parity
|
1
|
122
|
51.3
|
2
|
60
|
25.4
|
3
|
38
|
15.8
|
4
|
16
|
6.7
|
6
|
1
|
.4
|
7
|
1
|
.4
|
Exploratory Factor Analysis
Exploratory factor analysis and varimax rotation were used to determine psychometric properties of the Prenatal Health Behavior Scale. The Kaiser-Meyer-Olkin Measure of Sampling Adequacy (KMO) and value of Bartlett's test of sphericity were calculated before performing the factor analysis. In this analysis, value of KMO was equivalent to 0.92; and value of Bartlett's test of sphericity was 4001.341 that was significant with a degree of freedom of 276 (p≤0.001). Accordingly, two factors were extracted.the health promoting behavior was obtained with an eigenvalue value of 6.60 and explained variance of 59.58. Table 1 summarizes factor loadings of the PHBS.According to the table above, all factor loadings are appropriate (above 0.3). Scree plot is presented as follows(Figure 1).
Table 2: Factor loads of questionnaire of prenatal health behaviors (n = 230)
Item number
|
health-impairing
|
health-promoting
|
24
|
0.79
|
-
|
21
|
0.78
|
-
|
20
|
0.77
|
-
|
16
|
0.77
|
-
|
19
|
0.75
|
-
|
7
|
0.74
|
-
|
9
|
0.73
|
-
|
13
|
0.72
|
-
|
14
|
0.72
|
-
|
5
|
0.71
|
-
|
4
|
0.69
|
-
|
8
|
0.67
|
-
|
17
|
0.64
|
-
|
22
|
0.64
|
-
|
3
|
-
|
0.88
|
18
|
-
|
0.85
|
2
|
-
|
0.82
|
6
|
-
|
0.81
|
23
|
-
|
0.80
|
12
|
-
|
0.76
|
1
|
-
|
0.74
|
10
|
-
|
0.73
|
15
|
-
|
0.72
|
11
|
-
|
0.71
|
Confirmatory Factor Analysis:
The confirmatory factor analysis was also used to fit the 2-factor model of the PHBS. Fit indices of final form of the PHBS were assessed. Findings indicated that most of indices confirmed the optimal data-model fit. In the model, X2=499.11, df=243, and thus x2/df= 2.053 (Figure 2).
Confirmatory factor analysis results indicated that all factor loadings were appropriate and significant in the initial model, but the model did not have good fit indices; hence, covariance paths were established between some items based on LISREL recommendations for model correction until the final model was obtained with optimal fit indices. The following table presents the 2-factor model fit indices for the initial and modified model(Table 3).
Table 3: Fit-fit indices of factor 2 model of pregnancy health behaviors questionnaire
Model
|
*X 2
|
CFI
|
IFI
|
NFI
|
RMSEA
|
RMR
|
SRMR
|
GFI
|
AGFI
|
basic model
|
909.24
|
0.84
|
0.84
|
0.79
|
0.048
|
0.66
|
0.066
|
0.77
|
0.73
|
Modified model
|
499.11
|
0.98
|
0.98
|
0.95
|
0.068
|
0.040
|
0.059
|
0.85
|
0.81
|
* Chi-square by the maximum likelihood estimation (MLE)
|
Reliability analysis: Reliability of the PHBS was calculated using internal consistency and Cronbach's alpha coefficient. The following table presents mean, standard deviation and alpha coefficient for each component of the PHBS and total score.
Table 4: Mean, standard deviation, correlation and alpha coefficient for the prenatal health promoting and impairing behaviors questionnaire.
Variable
|
Mean
|
Standard deviation
|
Alpha coefficient
|
health-promoting
|
26.06
|
6/65
|
0.93
|
health-impairing
|
8.20
|
8/09
|
0.93
|
The table above presents results of mean and standard deviation of each variable. Another column of table reports results of the reliability analysis by Cronbach's alpha method. The results indicated alpha coefficient was 0.93 for the prenatal health promoting behavior, and 0.93 for the adverse prenatal behavior. As shown, the rates were appropriate and above 0.7 indicating high reliability of tool in the Iranian sample (Table 4).