Article Method:

Participants Participants included 31 preschool girls as well as 30 of their parents. The girls were between ages 3 (36 months) and 5 (64 months) ( M = 53.48 months, SD = 7.13 months). The child participants also represented a variety of racial/ethnic backgrounds: 13 were Hispanic, 10 Caucasian, 1 Asian, and 7 mixed race. Parents signed informed consent forms for their children and themselves, and child participants gave their assent. Due to real world constraints, including limited time available in the preschool classes, not all of the participants could complete all components of the procedure. Participants were drawn from four classes at two U.S. preschools. Of the four classes, one came from Campus Children's Center (a pseudonym used for confidentiality purposes), which is a member of the organization, Bright Horizons Family Solutions. Campus Children's Center is located in a suburban neighborhood and is the lab school for a well-known college. Monthly tuition rates for participants involved in this study ranged between $495 and $1139 as specified by the preschool's tuition schedule. Of the 31 children involved in this study, 13 of the child participants and 13 of their parents came from Campus Children's Center. The remaining three classes came from Head Start, a government-funded organization, which provides free preschool programs for children whose families live below the poverty line. The Head Start in which we observed is located in an urban area, and the majority of the students come from Spanish-speaking homes (Jacoby and Lesaux 2014). Eighteen of the child participants were drawn from the Head Start center along with 17 of their parents. These two preschools were selected because they provided a range of socioeconomic and racial/ethnic backgrounds. Pretend Play Observation The pretend play observations, conducted by the first author and a trained research assistant, involved three hour-long observation periods of each class's free playtime on three separate days. Throughout the observations, the researchers tracked each child individually, assigning every participant an identification code and keeping her data separate from the others. To acquaint themselves with all of the participants, the researchers referenced classroom photo rosters supplied by the teachers. During the first hour of observations, the researchers recorded notes about Bbusiness as usual,^ that is, how the class functioned on a normal basis without the introduction of external stimuli (i.e., costumes). Of the 31 child participants, 23 attended school on the day of their class's Bbusiness as usual^ observation. The following two observation hours, conducted on separate days, involved observations when costumes, provided by the researchers, were present. On these days, the classes received a box of 12 mass-produced dressing-up costumes for children. Six of the costumes represented Disney Princesses and served as prompts to Disney Princess media for the participants. These included the Disney-produced costumes for Cinderella, Rapunzel, Tiana, Merida, Jasmine, and Mulan, all of which were emblazoned with an image of the character. We selected these six princesses from the eleven available because they represented a variety of ethnicities and races and because Disney released films about these princesses both in present day and in past decades (e.g., Disney first released Cinderella in 1950 while they released Brave in 2012). By representing this range of princesses, we sought to allow for variation in the children's play. The box also contained six non-princess costumes, advertised as Bgirl^ costumes by their marketers. These non-princess costumes were added so as to disentangle the relationship between princess costumes and novel stimuli; for the purposes of our study, play behaviors in these costumes were not analyzed. The non-princess costumes included Wonder Woman, a pink Power Ranger, Transfomers: Optimus Prime for girls, Violet from The Incredibles, a bumblebee, and a ladybug. All 12 of the costumes were of similar value and quality. Before the start of the class free-play period, the teacher or a researcher would place the opened box of dress-up costumes in the dramatic play section of the room where the participants would have the opportunity to play with them. Twenty-seven of the participants were present for at least one of the two days of the observations when costumes were available. In total, for the combined group of four classes, the researchers performed 12 h of play observations, four of which were observations of Bbusiness as usual^ and eight of which were observations when costumes were present. All observations followed a similar protocol. Approximately every 7 min, the researchers filled out a snapshot observation form, which took between 3 and 4 min to complete. The snapshot observation technique served to monitor the percentage of time participants played with the costumes and to track their locations within the classroom. Just as a camera captures a specific moment, the snapshot observation did the same, although with quantitative information. Snapshot observations have been used frequently in other studies to collect quantitative data within the classroom (e.g., Jacoby and Lesaux 2014). This information provided a foundation on which to further interpret and understand girls' play behaviors. In the remaining 3-4 min between snapshot forms, the researchers recorded thick qualitative descriptions about the participants' behaviors. These notes addressed the types of roles that the participants chose to enact as well as their actions while dressing in the costumes, when costumes were available. Semi-Structured Interviews Between 3 and 14 days following each class's final hour of observation, the first author returned to the classrooms to conduct individual, semi-structured interviews with 27 of the child participants. Due to class absences and unwillingness to participate, three participants did not participate in the interviews, and one interview was later removed from analysis due to language barriers between the interviewer and child, resulting in 23 complete interviews. The interviews lasted between 5 and 15 min and included approximately 40 questions from a pre-prepared list. (This full list is available as an online supplement.) A pilot test of the interview questions, which involved two 4-year-old girls who were not part of the sample, determined that the questions were easily comprehensible for preschoolers. The first author asked all of the participants the same questions in the same order; however, occasionally she added other relevant questions. Standard questions, for example, included: BDo you like to pretend to be a princess?^ and BDo you like to watch princess shows and movies?^ In addition, she presented girls with pictures of the 11 official Disney Princesses as they appear on the Disney website. Although Disney does portray other princesses, such as Ana and Elsa from Frozen, we elected, as have other studies (e.g., England et al. 2011; Wohlwend 2009) to include only the official Disney Princesses (Disney 2015). Using the pictures, the first author asked questions such as BWho is your favorite princess?^ and BWould you want to be (fill in princess name)?^ During the interviews, she also asked girls yes or no questions regarding gender stereotypes, on which they had the opportunity to elaborate. For example, she asked: BAre princesses strong?^ and BDo girls have to be pretty to be princesses?^ Interviews took place in a corner of the classroom separate from the other students and were recorded by a digital audio recorder. Participants received a sticker for their participation at the conclusion of the interview. Parent Questionnaire Parents of the child participants received questionnaires regarding their child's media use and, specifically, her access to Disney Princess media. This questionnaire included five questions and provided baseline data about each child's exposure to Disney Princess media and other related topics. Two of these five questions were analyzed for the purposes of the current study. These two questions were: BHow many Disney Princess items (clothing, toys, or any other product with the Disney Princess image on it) does your child have?^ and BHow often does your child play video games and/or watch movies and/or television shows that portray Disney Princesses?^ Pilot testing of this questionnaire with two parents of preschool children determined that the questions were appropriately worded. Parent questionnaires have been used to collect data on children's media use by other social cognitive studies as well (e.g., Coyne et al. 2014; Starr and Ferguson 2012). Indeed, the question regarding Disney Princess screen media exposure was adapted from a parent survey used by Coyne et al. (2014), which also examined the influence of the mass media on children's play behaviors. Parents at Head Start received both an English and Spanish copy of the questionnaire.

Describe thedata collection methodsthat the researchers used to answer their research question - for example, did they use standardised surveys, in-depth interviews, online questionnaires etc? Did thedata collection methodsmatch the methodologyof the study? What is it about the data collection methods that show that they match the methodology?