2.1 Design and setting
This methodological research was conducted in two phases between May 2019 and January 2020. A qualitative content analysis and item generation were performed in the first phase. The second phase involved a psychometric evaluation of the tool and the assessment of its validity and reliability.
This project was conducted in prenatal clinics of health care centers affiliated to Mazandaran University of Medical Sciences (MAZUMS), Sari, Iran. Sari is the capital of Mazandaran province- northern Iran. Expectant fathers, who attended in the prenatal visits with their spouses in second or third trimester of pregnancy, were employed. In Iran, expectant fathers can attend pre-arranged prenatal classes or prenatal care visits with their spouses upon their will.
2.2. Qualitative phase (Item generation)
At first for designing the FFCS, the necessary data to generate items were collected by semi-structured interviews with expectant fathers who were selected purposefully. During face to face interviews, fathers were held to explore their experiences related to fear of childbirth. Sampling was performed with maximum variation in terms of age, education and reproductive history. The interviews lasted 60–90 minutes and were recorded by the researcher. All interviews were held in Persian and by the same interviewer (A Master of Science in Midwifery on the supervision of research team including a psychiatrist and a reproductive health specialist). Data saturation was reached after 20 interviews. After transcribing the interviews, they were analyzed through the conventional four-step content analysis approach proposed by Graneheim and Lundman. Accordingly, each interview was divided into meaningful units, which were condensed and coded. Then, the resulting codes were grouped into categories and subcategories (22). Data was managed using the MAXQDA 10 software, and trustworthiness was ensured via Guba and Lincoln’s criteria, including credibility, dependability, confirmability, and transferability (23).
Also, in this phase, a comprehensive search in the relevant databases such as: Scopus, Science Direct, Psych Info, PubMed, and Cochrane were conducted to finalize the initial item pools which generated based on the results of the interviews. Keywords and syntaxes were as follows: ["Fear" OR "Tocophobia"] AND ["Childbirth" OR "Delivery" OR "Parturition" OR "Birth"] AND ["Pregnancy" OR " Gestation"] AND ["Father" OR "Men" OR "Couple" OR "Paternal"] AND ["Related Factors" OR "Influence Factors" OR "Contributed Factors"] AND ["Psychometry" OR "Psychometric"] AND ["Scale" OR "Questionnaire" OR "Tool" OR " instrument"]. The literature review was carried out until saturation was reached for the items of the FFCS. Finally, based on the results of the interviews and literature review, an item pool was generated. Items were scored on a five-point Likert scale ranging from 1 (Not important at all) to 5 (Completely important).
2.3. Quantitative phase (Psychometric evaluation)
In the quantitative phase, reliability as well as face, content, and construct validity of the FFCS were established.
2.3.1. Face validity
The face validity of the FFCS was evaluated both qualitatively and quantitatively. For qualitative face validity, 10 expectant fathers were invited to comment on the difficulty, appropriateness, clarity, and essentiality of the items. The items were then modified based on their comments. For quantitative face validity, 20 expectant fathers were asked to rate the importance of each item on a 5-point scale ranging from 1 (Not important) to 5 (Very important). The impact score of each item was calculated by multiplying its importance score by the number of fathers who had rated it 4 or 5. An impact score greater than 1.5 was considered appropriate (24).
2.3.2. Content validity
Content validity was also evaluated using both qualitative and quantitative methods. For qualitative content validity, 11 experts (i.e., two gynecologists, six reproductive health and midwifery specialists, and three psychiatrists) who were experienced in instrument development were asked to comment on the structure, wording, item allocation, and scoring of the FFCS items. The scale was amended based on their comments. Next, content validity ratio (CVR) and content validity index (CVI) of the primary version of the scale were evaluated. Accordingly, for CVR evaluation the expert group was asked to rate the essentiality of each FFCS item as being “Essential” (score of 1), “Useful but not essential” (score of 2), or “Not essential” (score of 3). Based on the Lawshe Table, items with CVR values less than 0.59 were excluded (25). For CVI calculation, the same expert group was invited to rate the relevance of each item. Item CVI (I-CVI) with values more than 0.79 were considered appropriate, between 0.79 and 0.70 were revised, and scores below 0.70 were considered unacceptable (24). In addition, an average scale-level CVI (S-CVI/Ave) was evaluated. An S-CVI/Ave of greater than 0.80 was considered acceptable.
2.3.3. Item analysis
Prior to construct validity evaluation, 30 expectant fathers were asked to complete the FFCS. Their responses were used for internal consistency evaluation. Items with an inter-item correlation coefficient of less than 0.30 were omitted.
2.3.4. Construct validity
Comrey and Lee (1992) offered a rough rating scale for adequate sample sizes in factor analysis as follows: 100 = poor, 200 = fair, 300 = good, 500 = very good, 1,000 or more = excellent (26). Therefore, 433 eligible fathers were recruited to complete the 23-items FFCS and socio-demographic checklist for exploratory (200 fathers) and confirmatory factor analyses (233 fathers). The sociodemographic checklist included items on age, level of education, occupation, number of children, and having a wanted pregnancy. Convenience sampling was used to choose the participants from the health care centers affiliated to MAZUMS. Inclusion criteria were basic literacy, no history of hospitalization in psychiatric hospitals, and consent to participate in the study. High-risk pregnancies, any history of a child with physical or mental abnormalities in the family, and chronic maternal illnesses that endanger the mother's life were exclusion criteria.
For exploratory factor analysis (EFA), sampling adequacy was assessed via the Kaiser-Meyer-Olkin and Bartlett tests. Then, the latent factors of the FFCS were extracted via the maximum-likelihood EFA with Promax rotation. The number of extractable factors was determined via parallel analysis. The minimum acceptable factor loading for the presence of an item in a factor was 0.3, which was calculated using the equation below:
𝐶𝑉=5.152÷√ (n-2). Based on the three-indicator rule, each factor had to have at least three items (27). Items with communality values less than 0.2 were excluded (28).
Using confirmatory factor analysis (CFA), the extracted factor model was evaluated via maximum likelihood estimation by using the following model fit indices: incremental fit index (IFI), comparative fit index (CFI), adjusted goodness of fit index (AGFI), parsimony normed fit index (PNFI), parsimony comparative fit index (PCFI), root mean score error of approximation (RMSEA), and minimum discrepancy function divided by degrees of freedom (CMIN/DF).
2.3.5. Normal distribution, outliers, and missing data
Univariate normality was evaluated using skewness (±3) and kurtosis (±8). Multivariate outliers were assessed via the Mahalanobis D squared test (P< 0.001). Moreover, multivariate normality was assessed via Mardia coefficient of multivariate kurtosis (< 20) (23). Missing data was assessed via multiple imputations, and it was replaced with the mean of participants’ scores.
2.3.6. Reliability
Cronbach’s alpha, McDonald's omega, and AIC were calculated to evaluate internal consistency (29). An acceptable internal consistency involved a coefficient greater than 0.70 and an AIC between 0.20 and 0.40 (27). The data was analyzed using SPSS-AMOS24 and SPSS R-menu2.0.