I am going to report on the number of occuranc of a certain interactional strategy used by each teacher-participant (n=4) of my project in diffrent classroom contexts (e.g., meaning and fluency, form and accuracy...). Please note that this certain interactional strategy (SLC) is also comprised of 14 element that should be reported exactly with each teacher-participants in order to examine what happens to this number after recieving instructional mediation (to see either the mediation has positive impact on the occurance of this strategy or not!).
I'v tried to count the strategies (separately for each participant by watching the video-recorded data of typical classroom sessions of each) to see how many times they have been uesed before and after mediation session and i'v report the frequency. yet, it had become a large and messy table! Is there any other more precise and better way to report?
Have you tried using Hattie's meta - analyses to see which of the strategies you are looking at they rank the highest. It may help you prioritise your work.
I'm not familiar with "Hattie's meta-analysis! and unfortunatly i could not find any thing related to it.
Btw, would you plz send me the related papers that have used this tool?!! if it is possible for you
any way thanks alot
It might help to search for ways to quantify and categorize your qualitative data. After, you can of course do some statistical analysis to examine whether there are difference between pre- and postobservations. See for example articles by Mercer et al. (they were looking for difference in so called linguistic key features), Michaels & O'Connor (differences in the use of productive talk moves by teachers), Janneke van de Pol (differences in scaffolding). Another helpful article: http://www.public.asu.edu/~mtchi/papers/Verbaldata.pdf It reports on ways to quantify your qualitative data. This might be a starting point to do some statistical analyses for an easy comparison of the differences in interactional strategies in your pre- and postobservations. Is this in any way helpful?
In problem-based learning skills reported by the student to propose a solution are measured. Besides the solution is feasible, imports solution algorithm used.
This sounds a little like something I've done in the past with a co-teach classroom. I had two cameras and someone else also looked at the video feed and observed along with me so we could compare results. My inspiration for this project was Bill Gates MET Project. You may find it interesting: http://www.metproject.org/
In classroom research, medians tell a better story than averages that can skew the picture because of extreme scores. With medians the distribution can be more finely described at the quartile points Q1 and Q3 that roughly correspond to -.67 and .67 Standard deviation from the mean (yes, report the average/mean). What you are describing, as I understand from your inquiry, is a study of 4 teachers and the data pooled from those teachers. Your study is to explore and report pre-post change effects. Pre-post with the same subjects requires that you use Related t-test for means (parametric), and/or the McNemar test for significance of change in the same subjects (non-parametric) according to your change criteria (how far from the pre-test average or median must the post-test score be in order to qualify as change?). For McNemar SoC, Numerator: subtract those who worsened from those who improved, subtract .5 and square the remainder (ignore the unchanged); Denominator: add worsened and improved. Resulting z-score relates probability table for the normal curve in deviation from the mean of the distribution. Without one of these statistical tests you have nothing definitively to discuss (left with a general impression that may or may not be true--we use statistics for a reason since we have a sign/sign-ificance cut off value). These tests are for the overall significance of the total of your 15 sub skills or clusters of skills. The effect size "e" is reported as "d" ( Hedge's d) when you divide the difference of pre-post means by the standard deviation of the pre-test to find the effect of your intervention. Effect has become the most important statistic that you can report since it is the most practical. Again, we are examining the total of all sub skills combined. The reason for analyzing and reporting the overall/ total scores is that the subskill scores are less reliable and have larger st. dev. You can certainly analyze/inspect the distributions and changes of the sub skills--and discuss them in order to answer questions or to conduct manipulations of the data. For example, if one or more sub skills shows no change then teacher awareness/importance or better training may require addressing and this situation must be reported. Also report any sub skill that substantially accounts for the overall change. Clusters of related skills may be revealed that would be helpful to know in future training or research. Your n is small, and your research may be indicative of a need for further research that could reach the level of "definitive" with much larger n.
All of this discussion assumes that you know the normal curve distribution well, including the various scales and scores commonly used and each representing the same value on the curve expressed by different designations: percentile, standard score (mean 100, SD=15, z, quartile, quintile, decile, stanine, etc that each have their own uses/applications depending upon the audiences you intend to inform. Describe the training given to the teachers and the theoretical base (cited).
Dear professor Palmer,
I'm really grateful to you for taking time and providing such a detailed answer.
- Badges
- Science topic