stats

met with the stats consultant today. good. we went over every question and how i'd been entering the answers and how we could code the answers if necessary, for the analysis. the only big problems were with the birth dates. considering that 0.5% of the dates are written according to the arabic calendar, it's kind of hard to write a formula that would automatically calculate the participants' ages! in addition, we often have incomplete dates, such as 1984, or sept. 3, only. the differences in format (86 or 1986) is also problematic. which means two things: i'm going to have to create a second column where i will have to enter every date again by using the full year, and if i don't have a month and day i'll simply make it january 1, AND i'll have to find a calculator of arabic dates to convert them to the western calendar and verify every single date i've entered from arabic questionnaires. fun fun.

i need to codify question marks, which i used when the students wrote "i don't know," and i also need to code the two open-ended questions for the students and all those for teachers and administrators. i have 74 teacher answers so far. finally, we talked about missing data and how we could manage the matching of questionnaires. that'll be hell. one thing that i've always wanted to do was to look at responses given about individual teachers. however, with the way things work now, it's impossible, because at one given school, there might be more than one advanced grammar teacher from the US, so even in the improbable case of students actually giving us all this info (they do, sometimes, but don't, about 30% of the time), it would be impossible to know which teacher they actually have. the stats consultant said i could collect this info on the final questionnaire only, since i then would match pre and post questionnaires. SO! i talked about that with margie, who says that asking students to write their teachers' names on the questionnaire would violate IRB agreements so it'd be hell to go over that again! however, three of the largest and most helpful programs did not have very difficult irb protocols. i might simply ask these three programs to do it, and make sure that the teachers agree... and that would be maybe only 10-20 different teachers, but having such data would be priceless!

i'm wondering how much i need to change the final questionnaires. there were tons of problems (arabic going the wrong way, korean being threatening, spanish missing the first few questions, etc.) and i could change all this... but that would mean working on those translations again, which was such a nightmare... + sending new questionnaires to schools, which is super expensive... and time consuming... and hair graying... so... maybe not. after all, i could make the case that any change would influence students' responses in ways that i didn't influence them the first time so that's bad for the validity of my results.

i need to start coding open-ended questions and send new letters to new program aministrators to ask them if they want to participate in the teacher/admin study. could use a few more admininstrators, really.