Participants and procedures
This cross-sectional validation study was performed in the city of Sari, the capital of Mazandaran province, northern Iran, from February to June 2017. In the present study, the sample size was estimated based on the number of items in the scale, multiplying by 10 (24 × 10 = 240). The most commonly used minimum sample size estimation method in structural equation modeling (SEM) is the 10-times rule procedure, which makes on the hypothesis that the sample size should be higher than 10 times the maximum number of internal or external model links implying at any latent variable in the model. The 10‐times rule was preferred due to its simplicity of use (29, 30). In all, 480 pregnant women participated in exploratory factor analysis (EFA) (240 pregnant women) and confirmatory factor analysis (CFA) (240 pregnant women). Data for this study came from pregnant women who attended a Baghban specialist clinic, public health care centers, and private gynecological clinics in Sari. Women were chosen using a multistage random sampling method. The first step of the sampling method was aimed at selecting samples from all regions. To this end, a list of public health care centers and gynecological clinics was provided. Subsequently, in proportion to the number of target groups in each of the public and private service centers, the number of samples required was consecutively entered into the study via a simple sampling method.
The inclusion criteria were to be a pregnant woman with gestational age from the seventh to the ninth month and to be interested in participating in the study. Exclusion criteria were a lack of willingness, having a mental illness, or having a specific physical condition such as complete placenta Previa, which was an absolute indication for cesarean delivery with no option for a vaginal birth, making it impossible to participate and complete the questionnaire. The demographic characteristics of the women included age, level of education, and employment status. Data collection approaches were based on nameless scales that were completed by an expert interviewer for protecting the privacy of women and the confidentiality of the data. The interviewer received guidelines for similarly completing the scales after attending a training session.
Scale development process
This study was performed to develop an instrument to measure the intention of pregnant women to choose the cesarean section delivery method. The scale was developed across several stages. During the first stage, the content domain of the construct was specified. In this stage, interviews were conducted with the experts (gynecologists and midwives) and pregnant women, and a review of the literature relating to the TRA (31-35) was performed to develop an item pool and content domain. The main dependent variable in the present analysis was the cesarean delivery method. In addition, the independent variables included five factors, organized into a logical framework, as follows: (a) behavioral beliefs; (b) evaluation of behavioral outcomes; (c) motivation to comply; (d) normative beliefs; and (e) behavioral intention. The item pool contained 39 items at this point. The principal researcher and other team members then read the items and removed extraneous ones. The first draft of the instrument comprised 27 items. In the second stage, the psychometric properties of the Iranian version of the Theory-Based Intention to Cesarean Section (IR-TBICS) scale were examined to assess its validity and reliability.
Content validity
Content validation requires a wide-ranging review by a panel of experts to determine whether the scale items sufficiently address the subject they aim to assess. It is a crucial phase for developing a tool and a method for linking abstract notions with tangible and measurable indices. The expert panel comprised 10 specialists in health education and promotion, gynecologists, and experts familiar with scale making. Qualitative content validity was assessed in terms of the wording, scaling, grammar, and item allocation indices (36). All items were tested, and the expert panel’s suggestions were added to the scale. We applied the content validity index (CVI) and content validity ratio (CVR) to reach the quantitative content validity of the new scale. To measure CVR, the expert panel was questioned to evaluate each item through a 3-point Likert scale, where 1 = essential, 2 = useful but not essential, and 3 = unessential. The CVR for each item was measured by means of the following formula: CVR = [Ne - (N/2)] ÷ (N/2) (Ne is the number of panelists indicating “essential” for each particular item and N is the total number of the professional panel). The numeric value of CVR is documented by the Lawshe table. Based on Lawshe’s table (37), items with a CVR score of .62 or above were selected (36). For the CVI, consistent with Waltz and Bausell (38), the same panel was questioned to assess the items based on a 4-point Likert scale on “relevancy,” “clarity,” and “simplicity.” The number of those rating the item as relevant or clear (rating 3 or 4) was allocated by the number of a content expert panel. A CVI score of .79 or above was applied acceptable (37, 39).
Face validity
Face validity is a calculation of laywomen (pregnant women) in understanding and knowing an instrument. In this step, both quantitative and qualitative methods were used. For the quantitative step, ten pregnant women were questioned to assess the instrument and the degree of importance of each item on a 5-point Likert scale to evaluate the “Item Impact Score” (Impact Score = Frequency (%) × Importance). An impact score equal to 1.5 or more was considered acceptable, as declared (40). For the qualitative step, the same pregnant women were questioned about the ‘relevancy,’ ‘ambiguity,’ and ‘difficulty’ of each item, and some minor modifications were performed on the primary instrument.
A preliminary version of the instrument: Reflecting the above methods, a preliminary version of the instrument containing 24 items was created for the next phases (construct validity and reliability of the IR-TBICS scale).
Statistical analysis
Construct validity
The construct validity of the IR-TBICS scale was assessed using both exploratory (EFA) and confirmatory factor analyses (CFA).
Main study and data collection
A cross-sectional study was planned to assess the psychometric properties of the IR-TBICS scale. A consecutive sample of pregnant women was recruited from the Baghban specialist clinic, public health care centers, and private clinics of gynecologists affiliated with Mazandaran University of Medical Sciences.
a) Exploratory factor analysis (EFA)
A sample of 240 pregnant women completed the IR-TBICS scale, and its factor structure was extracted by principal component analysis with varimax rotation. Bartlett’s test of sphericity and the Kaiser-Meyer-Olkin (KMO) test were applied to assess the suitability of the sample for the factor analysis. Eigenvalues above one and scree plot were conducted to identify the number of factors. Factor loadings equal to or greater than 0.4 were considered appropriate (41).
b) Confirmatory factor analysis (CFA)
A separate sample of 240 pregnant women completed the IR-TBICS scale, and factor analysis was conducted to measure the model fitness. As suggested, several fit indices counting relative chi-square (χ2/df), goodness of fit index, normed fit index, non-normal fit index, standardized root mean square residual, comparative fit index, and root mean square error of approximation were accompanied (42, 43). Relative chi-square is the ratio of the chi-square to degrees of freedom, and its suggested reference value is less than three for accepting the fitness of the model. The values for GFI, CFI, NFI, and NNFI could range between 0 and 1; values closer to 1 reveal data fitness (44, 45). An RMSEA ranging from .08 to .10 displays an average fit; less than .08 identifies a good fit (43). The satisfactory value for SRMR is below .10; values under .08 display satisfactory fit, and values less than .05 show good fit (46).
Reliability
Cronbach’s α coefficient assessed the internal consistency of the IR-TBICS scale. A Cronbach’s α coefficient equal to 0.7 or more was identified as acceptable (47). Floor and ceiling effects were determined as present if more than 15% of the responders attained the lowest or highest possible total score on the IR-TBICS scale (43). Furthermore, a subsample of pregnant women (n = 25) completed the IR-TBICS scale twice with a 2-week interval to test the stability of the IR-TBICS scale by computing the intraclass correlation coefficient (ICC); an ICC of 0.4 or more was deemed acceptable (48). All statistical analyses, except CFA, were done using SPSS v22.0 (49). The CFA was done using the AMOS software v22.0 for Windows (50).
Scoring
In the final version of the IR-TBICS, a minimum of three and a maximum of seven items were generated for each construct. In the present study, behavioral beliefs and outcome evaluation toward cesarean section were measured with seven and five items, respectively. The items were rated on a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicate a more positive attitude toward cesarean section. Normative beliefs were assessed concerning other important factors. In the present study, normative beliefs toward cesarean section were measured with six items. The items were rated on a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). In the present study, the motivation to comply with cesarean section was measured with three items. The items were rated on a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicate more subjective norms persuasive to cesarean section. The intention was assessed using three items. The items were rated on a 5-point scale ranging from 1 (very unlikely) to 5 (very likely). Higher scores indicate more intention to have a cesarean section.
Ethics
The ethics committee of Mazandaran University of Medical Sciences approved the study. All pregnant participants gave their written informed consent.