Académique Documents
Professionnel Documents
Culture Documents
Lab No. 3: Developing a Gender Social Interaction Coding System and Assessing Reliability
Authors: Erika Tibaut, Erin Walker, Shannon Wedge
Real Coders: Erika Tibaut Erin Walker
Reliability Coder: Shannon Wedge
University of Guelph
FRHD 3180*0101
conducting through time sample recording on a specifically designed coding system, the results
were analyzed to determine whether there is a pattern between the type of preferred social
interaction, and the gender of the observed child.
Three major hypotheses guided this research. First, it was expected in free play that
gender segregation, that is same-sex interactions, would be more likely observed than opposite or
mixed gender interactions, regardless of gender. This hypothesis is consistent with earlier
findings by researchers who have documented that across cultures, gender segregation remained
stable and more common with children from early life to six years of age (Ellis et al.,1981;
Maccoby & Jacklin, 1974; Powlishta, Serbin & Moller, 1993; Walker, Irving, Berthelsen, 2002).
Gender segregation defined by Staffer et al. (2005) as childrens tendency to associate with
same-sex playmates and to think of the opposite sex as an out-group, has been observed in both
natural and experimental contexts to become progressively stronger with every passing year
(Powlishta et al., 1993). Studies conducted at a nursery school in Sweden and across the United
States of America have shown that 60 percent of preschool girls and 20 percent of preschool
boys prefer to play with, and reliably selected, a same gendered peer (Staffer et al., 2005;
Powlishta et al., 1993). According to related research, this number will dramatically increase
within the years and by five or six years of age, the approximate age of children targeted for the
present study, children will spend ten times as much time with same sex peers than opposite sex.
(Staffer et al, 2005). Although reasons for these findings of gender segregation have not been
clearly established, it was hypothesized that it would be observed and replicated in this study.
A second hypothesis was that when engaged in same-sex social interactions, males would
exhibit more assertive behaviours and females would exhibit more pro-social behaviours. The
third, and final, hypothesis of the present study predicted that when engaged in opposite and
mixed-sex interactions, males and females would both display a tendency toward more prosocial
interaction. In a literature review on psychological sex differences, Maccoby & Jacklin (1974)
concluded that many differences between males and females are overstated, but males definitely
showed more aggressive and assertive behaviours than females (Hay, 2007). In the child care and
learning center that the observations for the current study took place, any obvious aggressive
behaviour is strictly prohibited in the classrooms, and because of this it was expected that males
would display more assertive behaviours as an outlet in their interactions with same gendered
peers. Within the first year of life, aggressive behaviours emerge at similar levels thus creating an
even balance between the two sexes (Hay, 2007), however as early as two years; boys begin to
emerge as more aggressive and assertive than girls (Staffer et al, 2005). Maccoby & Jacklin
(1974) believed this largely reflects differences between boys and girls play and interaction
styles. Boys tend to be more physically active throughout childhood and when engaging in
heightened activity with same gendered peers, boys display more dramatic, rough-and-tumble
behaviours, and teasing than when interacting with girls (Staffer et al., 2005; Rubin & Krasnor,
1983). These types of play most likely reflect pro-social, assertive, and possibly hostile
interactive styles, which would support each of the three hypotheses, for boys, in different ways.
In the same study, preschool and kindergarten girls used more pro-social problem solving
strategies for different given scenarios, regardless whether the children in the story were female,
male or mixed-sexes. (Rubin & Krasnor,1983; Green, Cillessen, Rechis, Patterson, Hughes,
2008). This supports the hypothesis that girls would engage in more pro-social behaviours,
regardless of gender. Although girls often engage in play that involves more turn taking, dual
decision making and sharing with other females, research has shown that girls often display these
behaviours with male peers as well (Green et al, 2008). These findings also support this
hypothesis because these play styles are highly reflective of pro-social interaction, which was
expected to be seen in female children over the other types of interaction being coded.
The observations were be taken during free play time in the kindergarten room which
allowed the children to engage in free social interactions without constraints of predetermined
tasks. Observers used time sampling recording in a private observation booth where children
were be unaware of their presence and observes had no previous relationship to the children, thus
decreasing the occurrence of biases.
Method
Development of coding system. The original coding system was developed based on
Olswang, Svensson, Coggins, Beilinson and Donaldsons (2006) coding system used to code
language ability, mainly with children who had speech impairments. This system was found to be
valid and reliable for the purpose of their study (Olswang et al., 2006). This coding system was
adapted for its coding categories: Hostile/Corrosive, Prosocial/Engaged, Assertive,
Passive/Disengaged, Adult Seeking and Irrelevant (Olswang et al., 2006). Each of these
categories was defined to be mutually exclusive and seemed to represent many of the important
categories that would be required to research social interaction in children.
The initial coding system for the current study used four gender categories: interacting
with same sex, interacting with the opposite sex, interacting with a group, and solitary. The initial
coding categories to measure social interaction were hostile, pro-social, assertive, passive, adult
seeking, and irrelevant. After live observation and feedback from peers, it was decided that the
sex categories would be changed to the following: same-sex interaction, opposite sex interaction,
interaction with both sexes, and miscellaneous. Group interaction needed to be altered because it
is possible for a child to be in a group with only girls or only boys. With the initial coding sheets,
a girl interacting with two or more girls would be coded as a group interaction, implying that the
child was interacting with both sexes. This study took into consideration sex differences in the
way children interact, therefore the use of a solitary category, nested in the sex categories, and
the irrelevant and adult seeking categories, nested in the social interaction categories, were not
important to the purpose of the study. In order to simplify coding, the miscellaneous category
was created so that these behaviours could still be recorded; this was important so that if these
behaviours became dominant at all, they could still be easily analyzed.
After live observation at a child care and learning center, it was established that some of
the operational definitions required alteration in order to be more mutually exclusive and
exhaustive. The prominent overlaps in observation occurred between the categories pro-social,
assertive, and passive interaction. Pro-social interaction was initially defined to be behaviour that
helped or benefited another child. Assertive interaction was defined as the child communicating
their needs, wants and feelings in a direct and honest way with respect. These two definitions
were intended to be inclusive of all positive and engaged behaviour; however, after observing the
way children interact, these categories needed to be expanded to be more exhaustive. While
performing initial observations, there was a lot of confusion and disagreement between which
behaviour would be considered pro-social or assertive, based on the initial definitions. It was also
difficult to code using the initial definitions because they relied a lot on spoken communication,
and hearing was sometimes difficult in the observation booth. For example, to code assertive
behaviour because a child was asking a question, the observer would have to hear the words
being spoken, which was not always feasible. To fix this issue, non-verbal interaction was added
to the definitions, and each category was expanded to be more mutually exclusive and
exhaustive.
Passive behaviour was originally defined as being unoccupied and having no focus or
purpose. After doing observations, we found that children who were not speaking, but were still
involved with an activity and peers, were still interacting. We therefore altered this definition to
include interactions that could be considered positive but that were happening alongside peers.
For this reason, it was renamed parallel interaction, and the definition was altered to exclude any
behaviour that could be coded as unoccupied or absent; these behaviours were moved into the
miscellaneous category.
The final coding systems full operational definitions can be found in Appendix A. The
categories that were used to measure social interaction were: same-sex interaction, opposite sex
interaction, interaction with both sexes, and miscellaneous. The first three gender categories were
measured with the following social interaction categories nested within them: pro-social, hostile,
parallel, and assertive. Pro-social interactive behaviour was coded for any behaviour that that
demonstrated positive and cooperative interactive behaviour with peers. Hostile interactive
behaviour was recorded for behaviour that demonstrated negative and aggressive interactive
behaviour with or towards peers. Parallel interactive behaviour was recorded for activity that
occurred alongside a peer, without any verbal or direct communication. The final social category,
assertive interaction, was recorded when the childs behaviour demonstrated respectful but bold
communication of ideas and beliefs in their communication with peers. The miscellaneous
category had three alternate categories nested within it: unoccupied, adult seeking and
uncodeable.
A hierarchy was determined in the event that more than one behaviour occurred. If two
behaviours occurred during the same interval, the behaviour that occurred for the longest amount
of time was recorded. If, during any interval, two behaviours occur in equal proportions, the
hostile or assertive behaviour categories were to be coded over pro-social or parallel interaction.
Reliability assessment.
Subjects. There were four children observed in this study. Two were female and two were
male. The children were chosen for observation because they were completely visible to the
observers and because, at the time that they were chosen, they were not demonstrating any
unoccupied behaviour. Child 1, the boy in the white striped shirt, was observed for the majority
of time playing with the water sensory bucket with another boy and girl. There was a transitional
period before he went to play in the block area with a girl and a female teacher. The boy was last
observed in the drawing and colouring area. Child 2, a boy wearing a burgundy striped shirt, was
observed playing with a toy house in the middle of the block area. He was playing alongside a
second boy for the majority of the time. The children were instructed to clean up because free
play was over, at which time the boy was also observed putting away blocks and ended the
observation reading a book with a young girl, while the other children in the room cleaned. Child
3, a girl wearing a green shirt, was observed playing with a miniature house set in the carpeted
area. She spent most of her time playing beside two other boys. Child 4, a girl wearing a pink
shirt and black vest, spent approximately equal amounts of time at the craft table and the play
dough table (located on the opposite side of the room). The girl played alongside girls and boys
at the craft table, but spent the majority of time using the play dough alongside a girl and a
female teacher. A boy was seen at the table towards the end of the observation, but he was
ducking under the table for many of the intervals (out of sight).
Setting. The observations were conducted in a kindergarten room of a child care and
learning center. The observations took place over two days; approximately 15-18 children were
in the kindergarten room during each observation. There were 3-4 adults in the room, including
student helpers, who would interact with the children casually as they engaged in their activities.
The room was divided into several play areas including crafting, storybook, and block area
which were located towards the front of the room. A dramatic play and sensory area were located
towards the back of the room, and a separate drawing area was located at a table at the right side
of the room. The room was very open and each area was easily accessible to the children and
adults in the room. The room was appropriately sized for the amount of children it held; no one
seemed to be cramped in their working/play environment. In the play area, four microphones
hung from the ceiling to project the noise into the observation booth. Observers were located in
the observation booth, out of sight from the children in the room being observed.
The socio-emotional climate of the room seemed to be positive and balanced during both
observations. The children in the room, particularly the four observed children, were observed
for the majority of the time occupied in an activity; there were not many transitions taking place.
The interactions between the children and the adults in the room seemed to be helpful and
friendly; the children appeared to be enjoying their free play time.
Procedure. Data was collected using time interval sampling. Observations were
completed by marking down the predominant behaviour that was viewed within each fifteen
second interval, on a coding sheet. Each time an interval passed, the behaviour was to be
recorded immediately. This permitted each observer to be recording the behaviours at the same
time which allowed for increased intercoder reliability. Each child was coded for 40 intervals,
totaling 10 minutes of observation per child.
The observers for this study were three university students with some introductory
observation experience using coding systems to assess behaviour. The students practiced
Table 1: Total and relative frequencies of gender, social interaction, and miscellaneous
categories.
Table 2: Mean relative frequency of the combined gender and interaction categories, by
observed.
Figure 1: Mean relative frequencies of social interaction categories, by sex of child
observed.
Table 4: Mean relative frequencies of peer gender categories, by gender of child
observed.
Figure 2: Mean relative frequencies of peer gender categories, by gender of child
observed.
Table 5: Confusion matrix.
Same Sex:
Hostile
Prosocial
Assertive
Parallel
Opposite Sex:
Hostile
Prosocial
Assertive
Parallel
Both Sexes:
Hostile
Prosocial
Assertive
Parallel
Miscellaneous:
Solitary
Adult Seeking
Uncodeable
0
0.1
0.062
5
0.087
5
0
0.1125
0
0.012
5
Females
0
0
0.05
0.2125
0
0.1
0.025
0.4375
0
0.275
0.012
5
0.1125
0
0
0.175
0.0125
0.0375
0.0625
0.0375
0.075
0
0
Gender of Observed
Males
Female
s
0.1
0.075
0.65
0.0625
0.0375
0.3
Males
0.2
Females
0.1
0
Gender of Observed
Gender of Peers
Males
Female
s
Same
Oppisite Both
Sex
Sex
Sexes
Solitary Adult
0.25
0.125
0.4
0.175
0.0125
0.2625
0.5625
0.0625
0.0375
0.2
Males
Females
0.1
0
Gender of Peers