Social support was measured using two constructs adapted from Mappen (n.d). Friends’ computer use consisted of three items. Girls were asked to report separately what percentage of their male and female friends are interested in computers. Responses ranged from none (1) to all (5). Also, they were asked how often their closest friends use computers, from everyday (1) to never (5). Students were asked to report who they talk to the most about computers in the last three months, and one option was to mark “There is no one I talk to about computers.”
In addition, quotations from the interviews provide examples of how students changed in their subjective task value, expectations for success, and perceived social support. We highlight responses to two of the interview questions: “Do you think you have changed at all from being in GCG?” and “Did you do anything in the program that surprised you or that you hadn’t done before?”
Satisfaction with the program was measured by responses to prompts in the electronic notebooks. Questions included: “What is one thing you like about [the program]? What is one thing you do not like?” Most responded to the question as a pair, and there were 81 responses to this question over the six different implementations of the program.
Data Analysis
A series of t-tests were used to determine whether there was a significant level of change from pre- to post-test survey. Paired t-tests were used to assess change within the treatment group (GCG program participants) and within the comparison group. Independent group t-tests were used to assess whether the change within each group was significantly different from the other group. Due to the small sample size, we report and interpret changes at or near the p<.05 significance level.
Data from the interviews and electronic notebooks were analyzed using a multi-step process. Following guidelines for coding qualitative data (Auerbach & Silverstein, 2003 ; Miles & Huberman, 1994), we first identified our research questions and expected themes. For the interviews, two researchers separately read through the audiotape transcripts to identify sections where students talked about the variables in which we expected change. In subsequent readings, the researchers sorted excerpts of the text into key themes and entered these segments of text into a spreadsheet. Responses by participants from their electronic notebooks were coded by first reading the entries and identifying repeating themes, and then re-reading the entries to assign them to one or more categories. Each interview and each entry was coded separately by two researchers and discrepancies were resolved, resulting in 100% inter-rater agreement.
Results
Comparability Between Treatment and Comparison Groups
The two groups differed in their race/ethnicity and on their average scores on three variables. The percentage of White students was higher in the treatment group (60%) than the comparison group (36%), because some of the comparison group participants attended a club with larger numbers of African American and mixed race/ethnicity students. Representation of Hispanic/Latina students was similar: 31% in the treatment group, and 27% in the comparison. Three significant group-level differences on scales and individual items at pre-test were identified with t-tests. The treatment group reported lower levels of skill( p< .01), but more positive attitudes toward computers (p < .01), and the comparison group reported more frequently that their teacher thinks they know about computers ( p < .05).
Attrition
A total of 214 participants completed pre-tests (126 in the treatment group; 88 in the comparison group), and attrition varied between groups: 28% of the treatment group and 17% of the comparison group did not complete the post-test. T-tests determined that age, confidence, skills, and attitudes toward computers at pre-test did not explain this attrition. The salient factors contributing to attrition included computer accessibility and knowledge about computers. In the comparison group only, those with no computer at home were less likely to have taken a post-test (64%) than those who had a computer at home (84%), and in the treatment group, those who dropped out reported higher levels of knowledge about computers (p < .05) at pre-test.
Did Participants Increase Their Computer Technology Capacity?
The survey data indicate how program participants changed over time and if that change is significantly different from that in the comparison group. The analyses include students who completed a pre- and post-test (90 in GCG and 71 in the comparison group). The survey findings are described below and excerpts from the interviews are used to illustrate what that change looked like from the girls’ perspectives.
Table 1: Means (Standard Deviations) on Pre- and Post-test Surveys
| |
GCG Program |
Comparison |
| |
Pretest |
Posttest |
Pretest |
Posttest |
| Stereotypes about computer workers+ |
2.08 (.55) |
2.07 (.59) |
2.21 (.54) |
2.36 (.59) |
| Intentions to study computers |
3.41 (.78) |
3.26 (.92) |
3.36 (.88) |
3.16 (.96) |
| Attitudes toward computers |
4.26 (.59) |
4.14 (.67) |
3.99 (.66) |
3.94 (.58) |
| Confidence to use computers |
3.86 (.62) |
3.98 (.65) |
3.78 (.65) |
3.87 (.66) |
| Computer skills*** |
2.77 (.90) |
3.59 (.91) |
3.13 (.96) |
3.34 (.95) |
| Knowledge about computers* |
3.13 (.59) |
3.39 (.59) |
3.27 (.64) |
3.31 (.67) |
| Gender stereotypes** |
2.52 (.85) |
2.23 (.90) |
2.20 (.97) |
2.48 (.94) |
***p<.001 **p<.01 *p<.05 +p<.10 |
|
|
|
|
Only one of the three measures of subjective task value was significant. As shown in Table 1, the intervention group reported virtually no change in their stereotypes about computer workers, while the comparison group reported an increase in negative stereotypes. An independent group’s t-test approached significance (p = .05). The following quotations illustrate the participants’ perspectives of the program’s role in reducing their stereotypes.
I used to, like, whenever I would think of a computer I would think of a nerd with glasses or whatever sitting in front of it, but I guess since I’ve done this, there are all different types of girls here and none of them were nerdy or anything. So like, I like sports and I like the computer too, so that’s completely going against the stereotype of what people think
(13-year-old, White participant).
Now like I've got some feedback how you make the game and that people that
make them aren't just lazy and … it's hard work and fun too
(11-year-old, Latina participant).
There was no significant change within or across groups in students’ intentions to take computer courses in the future or their attitudes toward computers. However, in response to the interview question, “Do you have any plans to take any more classes?” GCG participants said:
Yes. This man that was with my dad, right now he's going to [local community
college] and he's going to look for classes there for me, so I can do it with him,
he said
(11-year-old, Latina participant).
I really now understand how important it is because I didn't know any of that stuff and now I know what I want to be; I want to be a computer animator
(12-year-old, White participant)
Three of the five measures of expectations for success were significant across groups. As shown in Table 1, there was an increase in the GCG participants’ confidence in using computers, and it approached significance (p = .05), but was not significantly different from non-participants. Comments from the interviews illustrate what their confidence:
…thing I was most proud of learning about was how to do like buttons and stuff because we have the longest game on the whole thing and because we have two different story paths and I didn't understand how the whole button thing worked at all. I played the games that people had done before and I was like, how do we get the buttons to go to each click? But then I learned how to do the buttons
(14-year-old, White participant).
Additionally, computer skill level increased among those in the GCG program (p < .001), and an independent groups t-test showed that the increase was significantly greater for the treatment than the comparison group, (p < .001). One girl described her skills:
I know how to program now and I know how to go on the internet like Google
and get the graphics and computer animation, like the time frame. And I also
know how to put action scripting on a button (12-year old, White participant).
Also, there was a significant change in participants’ perception of their knowledge about computers. The treatment group reported significant increases in what their family thinks they know (p < .01), what they think they know (p < .001), and what other kids think they know (p < .05). As shown in Table 1, the change in self-knowledge for the treatment group was significantly different from that reported by the control group (p < .05). The result of this change is described in the following quote. In response to whether she changed as a result of participating, one girl responded:
Yes, a little because I could teach other people about computers and my mom and my dad and my brother and my cousin and my whole entire family want to learn more about the computer, but I want to learn too
(12-year-old, Latina participant).
A significant difference emerged between the two groups in their view that boys usually do better than girls when using computers. Overall, the treatment group decreased their gender stereotypes (p < .01), while the comparison group increased theirs (p < .05). As shown in Table 1, the difference in change for treatment versus comparison group was significant (p < .01). None of the qualitative data specifically illustrated the students’ views on this change, and there was no significant change in the other three items that measured gender stereotypes.
Problem solving was measured by asking students to respond to the following question: "If I don't know how to do something on the computer at school, the first thing I do is…" Students in GCG reported greater increases in first trying to figure it out themselves (from 38% to 61%) compared to those in the comparison group (from 39% to 50%). However, logistic regression shows that the change is not significantly different for the treatment versus the comparison group. The following two quotes illustrate independent problem solving:
I like… now when I'm on the computer even if I'm not on the Flash program. I kind of can figure out things better, like I was doing my homework and it just all disappeared and I tried everything to put it on and I got so mad at myself that I didn't save it, and I checked a file thing and checked this one little skill in Flash where you check every place, you check on the desktop or document, because I did that a lot in Flash – and I found out I saved it to the other file
(14-year old, Latina participant).
Another student also described how she changed:
Well, a little bit. I never really used to like [sic] computers and now I don't like, really, hit the computer when I'm mad - I sort of try and find out what's wrong. So I guess that's something
(12-year old, White participant).
One of the two measures of social support was significant. There was a decline in the numbers of girls in the treatment group who reported, “There is no one I talk to about computers,” from 10% at pre-test to 2% at post-test. Among those in the comparison group, the change was from 11% at pre-test to 10% at post-test. A logistic regression showed that the difference at post-test between treatment and comparison groups was significant (p < .05), controlling for pre-test ratings. There was no significant change in students’ report of their friends’ computer use. This may be due in part to the high numbers who reported they do not know if their friends are interested in computers or how often they use them (15% to 38%).
