Development and validation of paternal breastfeeding attitude scale

Abstract

Objective

The aim of this study is to develop a valid and reliable instrument for the measurement of fathers’ attitudes towards breastfeeding.

Methods

This basic research-designed study included two independent participant groups of fathers who resided with their wives and had children aged 0–24 months who were exclusively breastfed. The final draft of the Paternal Breastfeeding Attitude Scale consisted of 32 items, which were scored on a five-point Likert scale.

Results

Exploratory Factor Analysis was conducted on the scale comprising 14 items. The Kaiser-Meyer Olkin value of the 14-item Paternal Breastfeeding Attitude Scale (0.864). The Paternal Breastfeeding Attitude Scale sub-dimensions Cronbach’s α were (0.85) for “devaluation (six items)”, (0.75) for “caring (five items)”, and (0.68) for “necessity (three items)”. Confirmatory Factor Analysis was conducted on the 14-item scale. The t-tests of all factor loadings obtained from Confirmatory Factor Analysis yielded statistically significant results (p < .05). This model obtained in confirmatory factor analysis was derived through EFA and based on the items belonging to the factors. In this multifactor model, observable variables were grouped under multiple independent dimensions. The model showed acceptable model fit parameters (χ2 = 210.94; RMSEA = 0.08; CFI = 0.94; NNFI = 0.93; SRMR = 0.05; IFI = 0.94).

Conclusions

The study demonstrated that the Paternal Breastfeeding Attitude Scale was a valid and reliable instrument for assessing fathers’ attitudes toward breastfeeding. Fathers’ positive attitudes towards breastfeeding can be promoted through breastfeeding training or counseling provided by women’s health nurses.

Clinical trial number

Not applicable.

Introduction

Breastfeeding plays a pivotal role in the protection and development of maternal and neonatal health. Newborns who do not receive breast milk in the first 24 h after birth have an increased risk of sudden infant death syndrome [1]. Furthermore, those who are unable to receive breast milk exclusively for the first six months are more likely to experience respiratory tract infections [2, 3]intestinal system problems [2] and behavioral disorders [4]. Other studies have demonstrated the positive impact of breastfeeding on the cognitive and social language development of infants in the subsequent period [5, 6]. Mothers who do not breastfeed or who cease breastfeeding prematurely have been found to have an increased risk of developing breast cancer [7]ovarian and endometrial cancer [8]hypertension [9] and myocardial infarction [10]. In light of these benefits, it is recommended that breastfeeding, which is accepted as the international gold standard, should be initiated within the first hour after birth and should continue for the first six months [11, 12]. Globally, the proportion of infants who are breastfed within the first hour after birth is 44%, while 42% are exclusively breastfed for the first six months [13]. In Turkey, the rate of initiation of breastfeeding within the first hour after birth is 71%, while the rate of exclusive breastfeeding for the first six months is 41% [14]. Despite the national and international campaigns to promote breastfeeding, approximately half of postpartum women do not initiate breastfeeding within the first hour and approximately one-fifth of women do not exclusively breastfeed for the first six months.

Although breastfeeding is often perceived as a choice and decision made by the mother after careful consideration, numerous factors influence this choice. In a patriarchal family structure, the father may influence the mother’s decision to initiate and continue breastfeeding. The meta-analysis of Pereira et al. (2024) showed that fathers’ breastfeeding knowledge, attitudes, and participation in the process affected the duration of breastfeeding [15]. Another meta-analysis by Mahesh et al. (2018) underlined the importance of breastfeeding support perceived by mothers from fathers in increasing the duration of exclusive breastfeeding for the first six months [16]. Various studies have demonstrated that fathers’ positive attitudes towards breastfeeding have a positive effect on breastfeeding duration [17–20]. Other studies have underlined the importance of the participation of fathers in the process of breastfeeding in improving the rates of breastfeeding [21]. Therefore, it is crucial to include fathers in the initiation and maintenance of breastfeeding, which has been recognized as a fundamental human right. At this juncture, one of the most pivotal responsibilities falls to nurses and midwives, who are the health professionals tasked with providing breastfeeding education and counseling to mothers [22]. Nurses and midwives should incorporate fathers into breastfeeding education and counsel them to make a conscious choice in feeding their babies. They should evaluate fathers’ attitudes towards breastfeeding, support positive attitudes, and plan and implement interventions to change their negative attitudes. Therefore, they need a valid and reliable measurement tool to evaluate the paternal attitudes towards breastfeeding.

There are various tools to measure maternal attitudes towards breastfeeding [26, 27] and infant feeding [23]. However, only one instrument exists to measure paternal attitudes towards breastfeeding [24]. The scale consists of two sections: Fathers’ Breastfeeding Attitude and Fathers’ Participation in Breastfeeding Process. However, as this valuable scale was developed in a different cultural context, cultural differences may affect its validity and reliability. Furthermore, having complementary scales developed in different cultural contexts will enrich the existing literature by increasing the variety and scope of measurement tools in this field. Therefore, given the limitations of the existing scale, there is a need to develop a new scale appropriate to local cultural characteristics. The number of studies on the attitudes of fathers towards breastfeeding is limited. Therefore, this study aims to develop an instrument that measures paternal attitudes towards breastfeeding, which is specific to culture. Such an instrument may contribute to the promotion of breastfeeding and provide empirical evidence for nurses and midwives in planning education and counselling services on breastfeeding.

Materials and methods

Aim

This study aims to develop a valid and reliable instrument to measure the attitudes of fathers towards breastfeeding.

Research design

Basic research encompasses not only the development and testing of theories and hypotheses but also the development, examination, verification, and re-evaluation of methodologies, techniques, and tools that constitute the foundation of research methodology [25, 26].

The summative scale, which is frequently employed in attitude measurement studies within the field of health sciences, is an approach developed by Likert (1932) that permits inferences to be drawn about the attitudes of participants based on their responses to a series of statements [27]. The scales developed based on this approach include both positive and negative items, which are answered using a scale of degrees, such as “strongly agree,” “agree,” “undecided,” “disagree,” and “strongly disagree.” This allows the levels of the participants regarding the attitude of interest to be revealed.

Population and participants

The study was conducted at the Pediatrics polyclinic of Mersin University and six randomly selected highest population density Family Health Centers (FHCs) in the Mersin province of Türkiye.

In studies of measurement tools, the objective is not to generalize findings derived from a sample to the universe. Rather, the aim is to provide evidence of the tool’s reliability and validity through a sample of participants representative of the population to which the tool will be applied in subsequent research [28]. As the objective of this study was to develop an instrument to measure fathers’ attitudes towards breastfeeding, no sampling was conducted. The study included three independent participant groups who met the inclusion criteria. The inclusion criteria were of fathers who resided with their wives, had healthy children aged 0–24 months who were exclusively breastfed and visited places where institutional permission has been obtained (university hospital and FHCs). Apart from the inclusion criteria and the voluntary participation of fathers in responding to the questionnaire, no additional criteria were applied. There were no exclusion criteria in this study. The first group (n = 20) was selected for a semi-structured interview who met inclusion criteria. The second group exploratory factor analysis (EFA) (n = 480), while the third was chosen for confirmatory factor analysis (CFA) (n = 285). In both EFA and CFA analyses, care was taken to ensure that participants were drawn from a diverse range of educational levels and occupational groups. Additionally stop to selection participants on rule of thumb that there should be at least 10–15 participants for each item of the scale. It is a common rule of thumb that there should be at least 10 participants for each item of the scale, making an ideal of 15:1 or 20:1 [29].

The EFA group of participants consisted of 480 fathers between March 6, 2023, and July 3, 2023, and who met inclusion criteria. One participant who did not respond to all questions and 58 participants with response patterns that were deemed to be outliers were excluded from the analysis. Consequently, the study was conducted with the data of 421 participants.

The CFA group of the participants consisted of 285 fathers between July 24, 2023, and February 5, 2024, and who met inclusion criteria. Given that 26 participants had outlier response patterns, the third group was concluded with 259 participants.

Construction of the paternal breastfeeding attitude scale

The preliminary version of the Paternal Breastfeeding Attitude Scale was developed by first identifying the items to be included in the scale following a review of pertinent literature [15, 16, 18, 20, 30, 31]. A semi-structured interview was conducted with the participation of 20 fathers on digital platform. These fathers did not participate to nor EFA neither CFA. The interview was designed to elicit fathers’ attitudes towards breastfeeding using the relevant literature [30, 31]. The researchers evaluated the interview data and identified sentences that were thought to be attitude statements. These statements were then reorganized into a draft form with 35 items in accordance with the rules of writing attitude statements.

In the subsequent stage, the strength, comprehensibility, and clarity of each of the 35 items were evaluated by three academicians of nursing and two academicians of measurement and evaluation via an online survey. The experts were asked to score each item on a three-point Likert scale, ranging from 1 (not appropriate at all) to 4 (very appropriate). The expert opinion was evaluated using Davis’s technique, and the item-content validity index (CVI) and the content-validity ratio (CVR) were calculated for each item [32] (Table 1).

Table 1.

CVI scores of the preliminary paternal breastfeeding attiude scale

Items	CVI	Decision	Item	CVI	Decision
1	1.000	A	19	0.798	A
2	1.000	A	20	1.000	A
3	0.798	A	21	1.000	A
4	0.798	A	22	1.000	A
5	1.000	A	23	0.798	A
6	1.00	A	24	0.592	NA
7	0.592	NA	25	0.592	NA
8	0.798	A	26	0.592	NA
9	0.592	NA	27	1.000	A
10	0.798	A	28	1.000	A
11	1.000	A	29	1.000	A
12	0.592	NA	30	0.798	A
13	0.592	NA	31	0.798	A
14	0.798	A	32	0.798	A
15	1.000	A	33	0.592	NA
16	1.000	A	34	0.592	NA
17	0.798	A	35	1.000	A
18	0.798	A

A: A; NA: NA

Table 1 presents the CVI scores of the preliminary Paternal Breastfeeding Attitude Scale (PBAS), which ranged from 0.592 to 1.000. The existing literature suggests that a minimum CVI score of 1.000 is required for expert opinions with five experts [33]. Consequently, the items with a CVI score below 0.80 were revised in accordance with the expert opinion and removed from the draft scale, namely items 12, 13, 26, 33, and 34. In addition, items 7, 9, 24, and 25 were reformulated. Following these revisions, the CVI of the 30-item PBAS increased to 0.867. In addition to the aforementioned 30 items, two further items were added based on expert opinion. The final draft of the PBAS consisted of 32 items, which were scored on a five-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree).

The flow diagram of the Construction of the Paternal Breastfeeding Attitude Scale was illustrated on Fig. 1.

Fig. 1.

Flow diagram

Procedure

The purpose and content of the study were explained to fathers by the researcher (X) at places (university hospital and FHCs) where institutional permission has been obtained. Those who volunteered to participate and gave their informed consent were sent a data collection form via a QR code. This enabled them to access the form on their phones and respond via self-reporting. This approach was also preferred because it is environmentally friendly.

Data analysis

EFA and CFA were performed to examine the construct validity of the PBAS, while Cronbach’s alpha, composite reliability (CR), and McDonald’s ω coefficients were calculated to assess internal consistency reliability. Furthermore, convergent and divergent validity analysis were conducted [33–36].

In the initial phase of data analysis, EFA was employed to examine the data obtained from the first group with the objective of elucidating the latent structure of the scale to be developed, determining the number of dimensions to which the items can be grouped, and elucidating the relationships between the items. During the EFA, the data obtained from the application of the PBAS preliminary form (n = 480) were transferred and transcribed, and one participant’s responses that did not answer all questions were excluded. Items 4, 8, 9, 12, 13, 14, 19, 21, 23, 27, 28 and 30 were reverse scored. The data set was found to be normally distributed [36]. To ascertain whether the data exhibited single and multiple outliers, standard-z and Mahalanobis distances were examined. The results indicated that the standard-z values ranged between 3.66 and − 2.64 for the EFA data and that there were no single outliers in the observation set, in accordance with the criteria suggested by Tabachnick (-4 ≤ z ≤ + 4) [41]. In order to identify multiple outliers, Mahalanobis distance measures were compared with the chi-square value at 32 degrees of freedom at the 0.05 significance level. The results indicated that the data of 58 participants exhibited multiple outliers (, p < 0,05). Consequently, the study on the first group of participants was concluded with the participation of 421 fathers on a 32-item instrument.

During the EFA, the Kaiser-Meyer-Olkin (KMO) and Barlett’s test of sphericity were employed to assess the factorability of the data set. In order to determine the adequacy of the sample size, the criteria of KMO value being at least 0.70 and Barlett’s test of sphericity being significant were taken into consideration [36]. The results of the Barlett’s test of sphericity indicated that the chi-square statistic was significant , and that the KMO value (0.892) was relatively high. These findings indicate that the data set was suitable for multivariate normal distribution [36]. The linear relationship among the items may lead to a problem of multicollinearity. In this regard, multicollinearity coefficients were considered and the reference values of the Variance Inflation Factor (VIF < 5) and Tolerance (> 0.10) were examined. The calculated tolerance intervals for the EFA data ranged between 0.215 and 0.777, while the VIF values ranged between 1.287 and 4.64. These findings indicate that there was no multicollinearity problem among the items. Finally, the Durbin-Watson (DW) statistic, which was used to examine whether the errors have autocorrelation, was 2.01, indicating no autocorrelation. In the social sciences, it is expected that there will be a relationship between factors. In light of the interrelationships between the factors and the recommendations in the literature, the direct oblimin rotation method, one of the oblique rotation methods, was employed.

In determining the number of factors, the following criteria were considered: common variance explained (> 0.30), factor loadings, KMO eigenvalues (> 1.00), and total variance explained (>%50). Additionally, scree plot and Horn’s parallel analysis were taken into consideration [36, 37].

During the second stage of the data analysis, a CFA was conducted on data obtained from the participants in order to evaluate the model-data fit of the structure revealed with EFA and to provide additional evidence of construct validity. Initially, negative items (items 2, 4, 6, 7, 10, 12, 13, and 14) were identified and subsequently reverse-scored. The data set (n = 285) was found to be normally distributed. The standard z-values calculated to determine the univariate outliers in the data set were found to vary between 2.2958 and − 4.0952. According to, Tabachnick’s criteria (-4 ≤ z ≤ + 4), there was a univariate outlier in the data set (-4.0952) [36]. The Mahalanobis distance measures were compared with the chi-square value at 24 degrees of freedom at the 0.05 significance level and a margin of error of 0.001. The results indicated that the data of 25 participants should be excluded from the analysis, as they were identified as outliers (, p < 0,05). Multivariate outliers were removed from the data set, and subsequent analyses were conducted on the data of 259 participants. All tolerance intervals calculated for the CFA data (n = 259) ranged between 0.218 and 0.708, VIF values ranged between 1.412 and 4.592, and no multicollinearity problem was found among the items. The DW statistic was calculated to be 1.828. A DW value approaching 2.0 indicates that the errors are independent [38].

The findings obtained from CFA were evaluated by using a path diagram and some Goodness of Fit model criteria, including the chi-square fit test, root mean square error (RMSE), root mean square error of approximation (RMSEA), non-normed fit index (NNFI), and comparative fit index (CFI) [39, 40]. The test-retest reliability of the scale was calculated using Pearson’s moment correlation coefficient.

Findings

Characteristics of fathers

The EFA group of participants

The ages of 421 participant fathers in the first group ranged from 28.73 to 41.2 years, while that of their spouses ranged from 26.34 to 37.21. They had infants whose age ranged from 0.19 to 13.81 months. With regard to their sociodemographic variables, 42% of the fathers had obtained a university education, 97.4% were employed, 52.7% were self-employed, 80.5% worked daytime shifts, 82.4% had a nuclear family structure, and 72.2% resided in central districts or provinces. Conversely, 41.8% of the mothers had obtained a university education, while 62.2% were unemployed. Pregnancy was planned for 76.2% of the fathers, and 58% had received education on breastfeeding, including 88.1% who stated that the education was sufficient. The primary source of information on breastfeeding was health professionals for 92.2% of the participants in the first group, while the least common source of information was books, newspapers, or magazines, with a percentage of 29.1%. Finally, 54.4% of the participants in the first group stated that they provided support to their spouses together with their wives’ mothers, while 27.8% of the fathers were alone in providing support.

The CFA group of participants

The age of these participants ranged from 29.24 to 42.84, while that of their spouses ranged from 27.41 to 39.70. They had infants, whose age ranged from 3.35 to 18.61 months. With regard to their sociodemographic variables, 49.6% of the fathers had obtained a university degree, 94.6% were employed, 39.2% were civil servants, 85% had a day shift, 78.5% had a nuclear family structure, and 58.5% resided in central districts or provinces. Conversely, 39.6% of the spouses had a university degree and 42.72% were unemployed. Pregnancy was planned for 62.7% of the fathers, and 27.3% had prior education on breastfeeding, including 50.8%, who believed that their education was sufficient. The primary source of information on breastfeeding was health professionals for 48.1% of the fathers, while the least common source of information was books, newspapers, or magazines, with a percentage of 12.7%. Finally, 66.5% of the participants in the first group stated that they provided support to their spouses together with their wives’ mothers, while 21.5% of the fathers were alone in providing support.

Findings of exploratory factor analysis

Exploratory Factor Analysis is shown in (Table 2).

Table 2.

Exploratory factor analysis scores of the paternal breastfeeding attitude scale

Item	Common variance	Factor
		1	2	3
Eşimin bebeğimizi ev ortamı dışında emzirmesindenrahatsız olurum	.508	.697
Eşimi emzirirken görmeye tahammül edemem	.646	.811
Bebeğimizin emzirilmesinde bana sorumluluk düşmez	.478	.709
Bence eşimin bebeğimizi emzirmesine gerek yok	.723	.804
Eşimin emzirmesi ondan uzaklaşmama neden olur	.638	.750
Bebeğimizin doğumdan itibaren mamayla beslenmesiyeterlidir	.472	.650
Eşimin bebeğimizi emzirmesini desteklerim	.704		.772
Eşimin bebeğimizi emzirmesini bebeğimizin sağlığıaçısından gerekli bulurum	.632		.726
Eşimin bebeğimizi emzirmesini önemserim	.648		.765
Bebeğimizi emzirmek eşimin psikolojisine iyi gelir	.545		.641
Emzirmenin eşim ile bebeğimiz arasındaki bağıgüçlendirdiğini düşünürüm	.543		.681
Eşimin emzirmesi bebeğimiz için yaşamsal birzorunluluktur	.509			.642
Bebeğimizin doğar doğmaz emzirilmesi gerektiğinidüşünüyorum	.630			.587
Ağız sütü, bebeğimiz için önemlidir	.561			.492
Eigenvalues		4.913	2.289	1.039
Explained variance		35.092	16.350	7.423
Cronbach’s α	.842	.855	.750	.683
Mc-Donald’s ω	.851	.862	.785	.698

The analysis of the data from the EFA of the 32-item preliminary form conducted on the data of 421 fathers in the first participant group was repeated by removing the items with factor loadings less than 0.45 (items 2, 13, 16, 17, and 25) and the items with a factor overlap less than 0.10 (items 3, 5, 7, 10, 11, 29, and 31). The resulting structure revealed the presence of two items under the same factor (items 22 and 32). Furthermore, four items (items 15, 19, 21, and 24) were identified as superfluous and were thus excluded from the analysis. As a consequence of these revisions, an additional EFA was conducted on the scale comprising 14 items. The KMO value of the 14-item PBAS (0.864) was greater than 0.80, and the Bartlett’s test of sphericity results yielded statistically significant results (χ² = 2243,423; p < .001). These two findings indicate that the data can be factorized effectively, resulting in a meaningful correlation matrix [36, 38].

The EFA yielded a three-factor structure with eigenvalues exceeding 1.00, collectively explaining 58.87% of the total variables. As illustrated in Fig. 2, the scree plot demonstrated that the graph lines reached a plateau from the fourth point onwards, indicating that the construct exhibited a three-factor structure. In addition to the aforementioned indicators, Horn’s Parallel Analysis was also considered in determining the number of factors [37]. Table 3 presented a comparison of the eigenvalues obtained from the principal components analysis and the simulative data obtained by Horn’s Parallel Analysis, which demonstrated that the scale had a three-factor structure.

Fig. 2.

Scree plot

Table 3.

Number of factors according to Horn’s parallel analysis

No	Real eigenvalues	Random eigenvalues		No	Real eigenvalues	Random eigenvalues
		Mean	%95			Mean	%95
1	4.913	1.314109	1.378901	8	0.535	0.974299	1.006742
2	2.289	1.239884	1.283233	9	0.473	0.935495	0.963251
3	1.039	1.185785	1.228238	10	0.466	0.894468	0.922581
4	0.965	1.139071	1.179361	11	0.411	0.855514	0.891750
5	0.715	1.095882	1.129257	12	0.377	0.816069	0.851177
6	0.679	1.051767	1.084643	13	0.325	0.771849	0.809876
7	0.574	1.010521	1.036473	14	0.239	0.715287	0.756420

A three-factor structure was determined for the final scale when the eigenvalues, scree graph, and Horn’s parallel analysis findings were taken together. The results of the EFA analysis for the 14-item construct were presented in Table 3. The table shows that the explained common variances of the 14-item instrument exhibited a range between 0.472 and 0.704. The factor loadings of the six items assigned to the first factor ranged between 0.811 and 0.650, while the factor loadings of the five items assigned to the second factor ranged between 0.772 and 0.641. The factor loadings of the six items assigned to the first factor ranged between 0.811 and 0.650, while the factor loadings of the five items assigned to the second factor ranged between 0.772 and 0.641. The eigenvalues of the factors exhibited a range of 4.913 to 1.039 accounting for a total of 58.865% of the total variance, with the first, second, and third factors accounting for 35.092%, 16.350% and 7.423% of the variance, respectively. The Cronbach’s alpha (0.842) and McDonald’s omega coefficients (0.851) for the entire scale were notably high.

The Cronbach’s alpha of each subscale ranged from 0.683 to 0.8555. Given that a Cronbach’s alpha value exceeding 0.70 is considered as A, only the Cronbach’s alpha of the third factor of the PBAS remained relatively low. This finding may be related with the limited number of items in the relevant factor. When all the findings are taken together, it can be stated that the scale is a reliable measurement.

The final stage of the EFA entailed the assignment of labels to the three factors in accordance with the extant literature, expert opinion, and theoretical findings. As illustrated in Table 5, the labels were devaluation (six items), caring (five items), and necessity (three items).

Findings of confirmatory factor analysis

The standardized values, estimated error variances, and t-values associated with the PBAS, are shown below (Fig. 3).

Fig. 3.

a. Standardized values. b. t values. Standardized values, estimated error variances, and t-values associated with the instrument

The confirmatory factor analysis was conducted to ascertain the suitability of the three-factor measurement model derived from the EFA. Figure 3 illustrated the path diagrams of the standardized values, estimated error variances, and t-values for the measurement model resulting from the CFA conducted using the data of 259 participants on the 14-item scale. The figure illustrated that the standardized values of the instrument ranged from 0.38 to 0.91. The t-tests of all factor loadings of the instrument yielded statistically significant results (Fig. 3b) (p < .05). The multifactor model depicted in Fig. 3 was designed to aggregate observable variables under multiple independent dimensions. This model was derived through EFA and was based on the items belonging to the factors. The fit indices for the measurement model are presented in Table 4.

Table 4.

Goodness of fit values of the instrument

Model	Δχ 2	Δχ /df	RMSEA	CFI	NNFI	SRMR	IFI
DFA	210.94	2.85	0.08	0.94	0.93	0.05	0.94
Normal	-	< 2	< 0.05	> 0.95	> 0.95	< 0.05	> 0.95
Acceptible (A)	-	< 5	≤ 0.10	> 0.90	> 0.90	≤ 0.08	> 0.90

Δχ 2: Minimum weighted least chi-square; Δχ /df: Chi-square fit test; RMSEA: Root Mean Square Error of Approximation; CFI: Comparative Fit Index; NNFI: Non-Normed Fit Index; IFI: Incremental Fit Index; SRMR: Standardized Root Mean Square Error

As illustrated in Table 4, the 14-item PBAS exhibited satisfactory alignment with the model fit indices. The ratio of the chi-square value to the degrees of freedom was 2.85, which may be considered as an A fit. The RMSEA (0.008), CFI (0.94), NNFI (0.93), SRMR (0.005), and IFI (0.94) values also indicated that the factor structure of the scale exhibited an A fit [46]. This suggests that the instrument demonstrated robust fit indices with regard to the RMSEA and near-perfect fit indices with respect to the IFI and CFI values [41].

Table 5 presented the findings regarding the convergent and divergent validity, as well as the composite reliability. Convergent validity refers to the degree of correlation between the statements and the factors derived from them. In contrast, divergent validity defines the relationship between factors other than the factor to which the items belong [42]. As demonstrated in the table, the requisite convergent validity (CR > AVE; AVE > 0.5) and divergent validity (MSV < AVE; ASV < AVE) were achieved. In conclusion, composite reliability coefficients (CR) were calculated to obtain internal consistency estimates of multiple factors and reliability evidence for the relationships between items under these factors [43]. The CR values of the scale subscales ranged between 0.78 and 0.91. As all calculated CR values exceeded 0.70, it can be concluded that the measurements to be made with the items and the factors to which they belong are reliable.

Table 5.

Convergent, divergent and composite reliability coefficients

Factors	AVE	MSV	ASV	CR
Devaluation	0.63	0.20	0.08	0.91
Caring	0.40			0.75
Necessity	0.56			0.78
Criteria	AVE > 0.50 CR > AVE	MSV < AVE	ASV < AVE	CR > 0.70

One month after completing the EFA and CFA studies of the scale, the scale was reapplied to test its test-retest reliability. For this purpose, the scale was applied to a group of 53 fathers, and the same participants responded to the scale again three weeks after the first application. During the control of the scale data, outliers were identified in the data of five participants, and the relevant data were excluded from the analysis. The test-retest reliability of the Paternal Breastfeeding Attitude Scale, calculated based on data from 48 participants, was found to be 0.96. This value indicates that the scale has a reliable structure capable of producing consistent data over time regarding fathers’ attitudes toward paternal breastfeeding.

Discussion

The development of the PBAS commenced with an examination of analogous studies in the extant literature. In this review, the Fathers’ Attitude Towards and Participation in Breastfeeding Scale, developed by Abu Abbas et al. (2016), and Turkish validity and reliability studies conducted by Hazar and Gültekin (2023) were identified [24, 44]. It is anticipated that the PBAS will contribute to the literature as a scale developed specifically for the Turkish cultural context.

The PBAS is a 14-item scale comprising three subscales: devaluation (six items), caring (five items), and necessity (three items). The range of possible scores for the devaluation, caring, and necessity subscales ranged between 6 and 30, 5–25, and 3–15, respectively. Higher scores on each subscale indicated that the father held the belief that the mother can breastfeed outside the home (i.e., devaluation), that the mother’s breastfeeding is essential (i.e., caring), and that baby formula is sufficient for infant care (i.e., necessity). Possible scores to be obtained from the PBAS may range from 14 to 70, with higher scores indicating higher positive attitudes of paternal attitudes towards breastfeeding. The Cronbach’s alpha values were higher than 0.70, and indicating that the PBSA was a reliable instrument [34]. The developed scale items enable the multifaceted evaluation of fathers’ attitudes towards and participation in breastfeeding. These cover various topics, including the impact of breastfeeding on maternal psychology, its role in fostering the mother-baby bond, paternal involvement in breastfeeding and perceptions of responsibility, knowledge of breastfeeding, and social/cultural sensitivities. The balanced inclusion of statements measuring both positive and negative attitudes also contributes to the scale’s ability to accurately reflect respondents’ true views. The items are written in clear, easy-to-understand language, increasing the scale’s applicability. These features demonstrate that the scale can be used as a valid and reliable measurement tool.

To assess the construct validity of the scale, the convergent and divergent validity calculations proposed by Fornell and Larcker (1981) were conducted. The results demonstrated that the scale exhibited both convergent and divergent validity [42]. The AVE values for convergent validity were 0.63 for the first factor, 0.40 for the second factor, and 0.56 for the third factor, respectively. Although the second factor did not meet the criterion of 0.50 for the average variance explained, this may be associated with the high amount of error involved in the respondent responses for each item in this factor. Furthermore, as a secondary criterion for convergent validity, “construct reliability being greater than the calculated AVE values” was met in all factors. Based on pairwise evaluations and other validity evidence, it was decided that this could be ignored.

Conclusions

The study demonstrated that the PBAS was a valid and reliable instrument for assessing fathers’ attitudes toward breastfeeding. Further studies may utilize this scale to assess fathers’ attitudes towards breastfeeding. Additionally, fathers’ positive attitudes towards breastfeeding can be promoted through breastfeeding training or counseling provided by healthcare providers.

Researchers may employ this scale to investigate the factors influencing attitudes towards breastfeeding or the factors affecting breastfeeding attitudes in disparate sample groups. Subsequent studies may be conducted to assess the validity and reliability of the instrument in different populations.

Implications

When breastfeeding is mentioned, a mother and baby schema comes to mind. In this scheme, fathers can often be ignored by nurses, who are the health professionals with whom the family communicates most frequently during the breastfeeding process. This can be prevented by developing, using and disseminating scales within this framework. By using PBAS, fathers’ attitudes towards breastfeeding can be identified as positive or negative, and their needs for breastfeeding education and counselling can be identified. In this context, increasing the involvement of fathers by using this scale in breastfeeding practice may enable an empathetic approach to fathers. As the involvement of fathers in breastfeeding is directly related to their attitudes towards breastfeeding, it is very important to obtain data through the use of this scale by health professionals (clinicians and academics). It is very important that the data obtained are presented to policy makers and reflect the attitudes of fathers in society. It can lead to the creation and dissemination of breastfeeding education programmes that will improve fathers’ positive attitudes towards breastfeeding. In addition, health policies can pave the way for the active participation of fathers in breastfeeding education programmes and the follow-up of their participation.

Author contributions

All authors participated in the conception and design. A.B. participated in the data collection. A.B., and D.A. participated in the data analysis and interpretation. All authors participated in drafting the text, and approved the final manuscript.

Funding

This study is a “2021-1-TP3-4359” numbered project which is funded by the Mersin University Department of Scientific Research Projects Coordination.

Data availability

Data from this study are available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

Ethical approval for the study was obtained from Mersin University Clinical Research Ethics Committee (February 17, 2021/number: 04). After obtaining ethical approval, institutional approvals were obtained from the Mersin University Pediatrics Polyclinics (E-32705888-774.01.01-1614243) and the Family Health Centres affiliated to the Toroslar and Yenişehir District Health Directorates of the Mersin Provincial Health Directorate (June 14, 2022/number:16). In the face-to-face interviews conducted by the researcher with the fathers who constituted the participant group, the content and purpose of the study were explained. The study was conducted in accordance with the Helsinki Declaration, and written and verbal informed consent to participate was obtained. Also, the study used online data collection forms designed to be environmentally sensitive.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.