**3.3 Special single items**

Single items have the advantages of having good content validity and decreasing subjects' confusion [47], and only content validity is the real validity and others can be considered as a validation process [48]. In addition to the three domains of parenting, we used one special item in each of the three domains in parenting styles as they have some special properties, as follows. Traditional Chinese parent-child interaction emphasizes parents' "training" of children [49], which implies high maternal involvement for promoting children's success. The first item was "it is for my children's own good to require them to do what I think is right, even if they don't agree" in the authoritarian domain. This was the only item in the domain that emphasized doing something entirely *for the sake of children's good*. All others emphasized ordering without mentioning children's good. We name it FCG (for children's good), hereafter. The second item was "I always encourage discussion when my children feel family rules and restrictions are unfair" in the authoritative domain. This was the only item in the domain that *encourages, by words*, parent-children interactions. Others do encourage interactions but not by words and are less explicit. We shall name it EBW (encourage by words). The third item was "I usually don't set firm guidelines for my children's behavior" in the permissive domain. This was the only item in the domain where parents set *no guidelines at all*. We name it NGA (no guidelines at all). According to Chao [49]'s discussion on Chinese parenting, FCG, EIW, and NGA items represent descending order for "training." We shall see later that NGA has more indirect effects on children's creativity.

### **3.4 Mediation methods**

The traditional approach to handle mediation analysis was proposed by Baron and Kenny [50]. It requires statistically significant conditions for: independent and dependent variables; independent variables and moderators; and mediators to predict dependent variables when controlling for independent variables. But it has been found that this procedure assumes no measurement error for mediators, which is impractical [51]. More importantly, an indirect effect can exist even though there is no direct effect [52, 53]. In our case, it may be possible that there is an indirect link between maternal personality and children's creativity without a direct link between the two variables. We will return to this in the Section 5. Hence, we used a more recent approach that uses the PROCESS (version 2.16.3) [54] procedure in SPSS with bootstrapping. Bootstrapping takes random samples with replacement of the original data. It has the advantage of not assuming normality and is particularly useful in small samples such as this case. If the "zero" point is not included in the confident interval as outputted, the indirect effect is said to be statistically significant [55]. Technically, if we denote X, Y, and M as independent variable, dependent variable, and mediator. The indirect effects are denoted by *c*, which equals *a*\**b*, where *a* is regressing M on X, and *b* is regressing Y on M controlling for X.

*Indirect Effects of Parenting Style on the Relationship between Maternal Personality… DOI: http://dx.doi.org/10.5772/intechopen.102378*
