The Italian PICCOLO was administered to 152 mother-child dyads. The inclusion criteria were: (a) child’s age between 10 and 47 months; (b) full-term birth (after 37th week of gestation), natural childbirth and no complications in childbirth; and (c) absence of pre/post-natal pathologies and/or developmental difficulties as reported by parent. The sample was 55% male and 45% female, with 8% of the children under 1 year-old (10–11 months), 42% between 12 and 23 months, 28% between 24 and 35 months, and the remaining 22% between 36 and 47 months (M = 25.5, SD = 10.6). Mothers were between 23 and 50 years old (M = 36.4, SD = 4.0) and had high levels of education (M years of study = 17.2). The socio-economic family status (SES) was assessed using Hollingshead (1975) classification and the SES family scores were in the range from 10/100 (low socioeconomic level) and 90/100 (high socioeconomic level) (M = 68.9, SD = 16.0). Parents were informed about the general procedures and aims of the study and were asked to sign a written informed consent form. The study was approved by Ethics Committee of IRCCS “Eugenio Medea” and was conducted according to the World Medical Association’s Code of Ethics (World Medical, and Association., 2013).

Recruitment was completed before pandemic restrictions took hold in March 2020 and was made possible thanks to the collaboration of pediatric centers and kindergartens located in three Italian regions: Lombardia, Emilia-Romagna and Marche. Researchers contacted kindergarten educator teams and pediatricians by letter or telephone and informed them of the study and its aims. In turn, educator teams and pediatricians invited parents who access their service to take part in the study. Parents who expressed interest in participating and who met the eligibility criteria were contacting by the study team. After determining the families’ availability to take part in the research, videotaping sessions were organized at either pediatricians’ offices or at kindergartens. The parents were informed that their participation would be entirely voluntary and anonymous. Information about the study, informed consent, and a demographic questionnaire were delivered to the parents prior the videotaping session. The observational setting consisted of a carpet measuring 2 m × 2 m in the middle of which was placed a box with some toys (rubber cubes, two small books, a puppet, a rattle, an electronic game); the camera was placed on a tripod at one side of the room and the researcher left the room leaving mother and child to interact during the recording. On the day of the appointment, after welcoming the dyads and familiarizing the child with the new environment and strangers, the recording began. Mothers were asked to interact with their son or daughter, for 10 min, according to the following instruction: “interact and play with your children as you typically do.” To assess test-retest reliability, a subsample of 20 dyads (average child’s age = 25.6 ± 9.5 months, nine males) was scheduled to take part in two observations (T1 and T2) carried out approximately 2 weeks apart in the same location.

PICCOLO (Roggman et al., 2013b) is an observational tool designed to assess the quality of parenting interactions of parents of children aged between 10 and 47 months. The original version was developed and validated in USA on a large, diverse, primarily low-income sample (sample of over 2,000 families from different ethnic groups). PICCOLO has strong reliability and validity for typically developing children (Roggman et al., 2013b) and for children with atypical development or disabilities (Innocenti et al., 2013). PICCOLO scores from when children were in the first 3 years of life were found to predict improved academic outcomes for children when 11 years of age (Innocenti et al., 2013). PICCOLO scores were found to be even more predictive for children with atypical development or disabilities than for typically developing children. PICCOLO was reported by users to be an easy-to-use instrument. PICCOLO includes 29 items which are scored on a three-point rating scale, from 0 (absent, no behavior observed) to 1 (some brief or minor behavior) to 2 (clearly, strong or frequent behavior) based on a 10 min recording of parent-child free play interaction. The items are summed up into four domains that measure various aspects of parenting interactions that have been identified in the research literature as promoting early child development:

The sum of the item scores contribute to a score for each domain between 0 and 14 (and 0 to 16 for Teaching domain) and to a total PICCOLO score.

For the Italian translation of PICCOLO were used International Test Commission (ITC) guidelines for the translation and adaptation of tests (International Test Commission [ITC], 2017). First, following the ITC guidelines, permission was first obtained from the holder of intellectual property rights of the tool. The Italian version of PICCOLO was submitted to Brooks Publishing (Baltimore), which gave permission to proceed with the adaptation. Second, we used appropriate translation procedures to maximize the suitability of the tool adaptation in the Italian population. Here following the main steps performed to obtain the Italian PICCOLO: (1) translation from the original version (English) to the Italian; (2) an independent back-translation from the Italian to the English; and (3) developers (LR and MI) compared the two versions in the original language to determine to what extent both item versions capture the same parenting behaviors. When step 3 indicated a difference or differences between both item versions in the English, then the translation was revised to correct mistakes. Developers provided suggestions which were incorporated into the final Italian version. This process continued until no significant difference in the items meaning was found and the items version in the Italian was accepted. Translation team consisting of a professional translator and two psychologists with specific knowledge in the field of the developmental psychology and mother-child relationships developed the two first phases in the back-translation methods. The translation in Italian covered both the items and coding guidelines as well as the entire PICCOLO manual (Montirosso and Giusti, 2022) which was used for coders training and for coding the Italian sample. Third, we were careful to minimize the influence of potential cultural and linguistic differences that were relevant to the intended uses of the tool in the Italian context. For example, according with developers we have translated the name of first domain “Affection” with “Coinvolgimento emozionale” (“emotional involvement”). Indeed, the Italian term affection may have a slight different meaning than in English, as it also refers to an internal dimension (“to feel affectionate with my own child”) and it not only describes overt behaviors (“to exhibit affection at my own child”). Also cultural differences in parenting behaviors were taken into account by the authors of the Italian PICCOLO manual and further discussed during coders training.

Three developmental psychologists working at the 0-3 Center for the at-Risk Infant of the Scientific Institute IRCCS Medea coded the mother-child interaction videos according to the PICCOLO user’s guide (Roggman et al., 2013a). To reach a high level of reliability, the coders participated in a three-stage training process. First, the coders read the original manual and watched the PICCOLO™ training DVD (5 h). Next, each coder separately scored all videos provided on the DVD. Score differences between coders and the ratings reported in the original manual were discussed and consensus obtained. Lastly, coders scored eight videos of Italian mother-child interactions using the translated material (i.e., scoring sheets and guidelines) and discussed the differences in ratings until they came to an agreement for all items. The training was considered satisfactorily completed when the percentage of inter-rater agreement was equal to or above 80%. These three coders coded parenting behaviors in all video observations of parent–child interaction of the current study, which were used to analyze the psychometric properties of the Italian version of the tool. Each coder coded about one third of sample (Coder#1 = 56 videos; Coder#2 = 50 videos; Coder#3 = 46 videos).

First, descriptive statistics for each domain were calculated and Pearson’s correlations were conducted to verify that the PICCOLO domains met the recommended criterion for discriminant validity, which requires a correlation between two constructs of less than .85 (Kline, 1986).

Eighteen video observations (12%) were independently rated by two of the three coders and used to estimate inter-rater reliability, which reflects the variation between raters who measure the same group of subjects. Researchers have assessed inter-rater reliability with the intra-class correlation coefficient (Bartko, 1966), so here we used two-way random intra-class correlation coefficients (ICCs) at the domain levels. Additionally, we calculated the Cohen’s Kappa as measures of interrater concordance at the item levels.

Internal consistency was assessed through calculating Cronbach’s alpha coefficient for each domain and for the whole instrument.

In order to assess test-retest reliability, which is the stability in measurements taken by an instrument on the same subject under the same conditions, 18 dyads (children’s mean age = 25.11 months, SD = 9.76, 61% female) were recalled 2 weeks after the first observation (T1) and their interaction was again recorded and coded (T2). Student’s t-test for dependent samples was used to compare the domain scores at T1 and T2. Furthermore, to assess the presence of associations between T1 and T2, Pearson’s correlations between the scores obtained in the two observations were also calculated.

The validity of the Italian version of the PICCOLO was examined through confirmatory factor analysis, applying the factorial structure that emerged in the original version of the instrument in which the single dimensionality of each domain was tested. Goodness of fit of each model was assessed using the following criteria: relative chi-square (χ2/df): a good fit is indicated by a value lower than two; (b) comparative fit index (CFI) and Tucker-Lewis index (TLI): for both indices a value >90 indicates an acceptable fit; (c) root mean square error of approximation (RMSEA): a value of ≤0.08 RMSEA is indicative of an acceptable fit, while a value of ≤ 0.05 RMSEA is indicative of a good fit (Hu and Bentler, 1999).

The influence of child’s age and sex on the PICCOLO’s scores was then explored. In line with the Spanish validation (Vilaseca et al., 2019a), the sample was divided into three age groups: 12–23 months (M = 17.3, D.S. = 3.5); 24–35 months (M = 29.7, D.S. = 3.3); 36–47 months (M = 41.0, D.S. = 3.1). Preliminarily, differences between these age groups and between male and female participants in maternal variables (age, SES) were assessed through one-way Analysis of Variance (ANOVA) models and Student’s t-test, respectively. Thus, the scores obtained at the four PICCOLO domains were entered as dependent variables into a multivariate analysis of variance (MANOVA) with child’s age group and sex as between-subject factors. This approach was chosen since it combines multiple dependent variables into a single value, thus maximizing differences across groups, but also controls for the number of inter-correlations among them (Bray et al., 1985). Post-hoc comparisons were eventually carried out adopting the Bonferroni correction of the p-value.

Confirmatory factor analysis was performed using lavaan, an R package for Structural Equation Modeling, version 0.5–12 (Rosseel, 2012). The other statistical analyses were executed using IBM SPSS Statistics, version 22.0, with p ≤ 0.05.

All Pearson coefficients between PICCOLO domain scores were statistically significant, with the only exception of the correlation between Affection and Teaching (p = 0.09). The lowest correlation coefficient was found between Responsiveness and Teaching, and the highest between Affection and both Responsiveness and Encouragement (see Table 1). Therefore, the PICCOLO domains were moderately correlated with one another, although not at a level that would suggest discriminant validity issues.

The ICCs were .98 for Affection, 0.99 for responsiveness, 0.97 for Encouragement, and 0.96 for Teaching, thus indicating an excellent reliability for all the domains. Similarly, Cohen’s Kappa values indicated a high inter-rater agreement at the item level within each domain (Affection: 0.82; Responsiveness: 0.83; Encouragement: 0.84; Teaching: 0.82).

The alpha coefficient was satisfactory for the total score (0.68), while non-optimal values were reliable for the single domains (Affection: 0.45; Responsiveness: 0.53; Encouragement: 0.42; Teaching: 0.40).

Test-retest reliability was based on observations at one time point (T1) compared with scores based on the observation of the same dyad 15 days later (T2). No difference emerged from t-test comparisons of domain scores between T1 and T2 (all p > 0.05), suggesting no statistically significant changes in PICCOLO scores over time. T1–T2 correlations were all significant (p < 0.05), with the exception of the Teaching domain (Affection: 0.83; Responsiveness: 0.55; Encouragement: 0.57; Teaching: 0.29), indicating that domain scores from the two observations were similar over the 2 weeks time period.

The results of the confirmatory factor analysis confirmed the one-dimensionality of the four domains, even though the saturation values were not optimal (0.29 for Affection, 0.39 for Responsiveness, 0.31 for Encouragement and 0.33 for Teaching). However, the model’s overall goodness of fit indices were acceptable: all chi-square values were not statistically significant and the association effect size indices were less than 2; all CFI and TLI values were greater than 0.90 and all RMSEA index values were less than 0.05.

Preliminary analyses did not reveal any statistically significant differences for age and gender (all F < 2.92, all t < 1.74, all p > 0.06) (Table 2). The MANOVA showed that neither age nor sex of the child, nor their interaction significantly affected the scores obtained across the four domains (all Wilks’s λ < 0.99, all F < 1.70, all p > 0.99). Nevertheless, between-subject effects revealed a significant influence of age group on the Affection domain (F_2,134 = 4.42, p = 0.014, η²_p = 0.06). Post-hoc comparisons further clarified this effect showing that mothers of the younger group (12–23 months) obtained higher scores than mothers of the older group (36–47 months) (p = 0.008). All other between-groups effects were non-significant (F < 2.14, all p > 0.120). PICCOLO domain scores for the three age groups are showed in Figure 1.