I've finally sorted out and organised my raw data! However, there were some unexpected results which I have to ponder over and do further research on.
Baking soda emerged tops for 3 stains of the 4 stains tested - protein, oil and dye, instead of hydrogen peroxide, which I had expected to be the most effective, but which came in a close second instead. On the other hand, hydrogen peroxide was the best cleaner for tannin stains, with baking soda being the worst. Combining ranking scores revealed a tie between the two ... which means my hypothesis was both right and wrong! ><
It appears that I will have to reject my original hypothesis in favour of another, and also delve deeper into the possible reasons for this. Where were the likely areas where errors could have occurred and how could I improve my experimental techniques next time, in order to reduce the margin of error?
Although I was a little uncomfortable at first (I usually get my hypothesis 'right'!), I realised that the findings discovered in just one small-scale study like this, certainly does not earn me the right to claim that something is or is not 'true' for sure. Besides, my hypothesis was NOT something I randomly plucked out from the sky but was thoughtfully formulated after extensive research! I was actually deliberating between hydrogen peroxide and baking soda, but settled for the former as several all-purpose, general stain removers invented within the past 20 years or so, used hydrogen peroxide as their main 'ingredient'. There was no literature that I could find that utilized baking soda in the same way.
This led me to think about how statistically significant my results were, and the methods I could employ to support or disprove the results obtained. I'm still trying to figure out statistical analysis (very new and foreign to me) and am presently still unsure if there are any statistical tools I can apply to my experiment (HELP MISS TAN!), which is not exactly very conventional in that it deals with multiple variables and is more like 2 experiments rolled in one (so theoretically I should have double the page-limit)! This makes reporting especially tricky and challenging. I am currently considering using one of the statistical ranking coefficients - Kendall's Tau or Spearman Rho, probably the latter as it is easier to compute.
I really need to discuss several issues with Ms Tan in greater depth and I hope to finish my report soon. :D
Baking soda emerged tops for 3 stains of the 4 stains tested - protein, oil and dye, instead of hydrogen peroxide, which I had expected to be the most effective, but which came in a close second instead. On the other hand, hydrogen peroxide was the best cleaner for tannin stains, with baking soda being the worst. Combining ranking scores revealed a tie between the two ... which means my hypothesis was both right and wrong! ><
It appears that I will have to reject my original hypothesis in favour of another, and also delve deeper into the possible reasons for this. Where were the likely areas where errors could have occurred and how could I improve my experimental techniques next time, in order to reduce the margin of error?
Although I was a little uncomfortable at first (I usually get my hypothesis 'right'!), I realised that the findings discovered in just one small-scale study like this, certainly does not earn me the right to claim that something is or is not 'true' for sure. Besides, my hypothesis was NOT something I randomly plucked out from the sky but was thoughtfully formulated after extensive research! I was actually deliberating between hydrogen peroxide and baking soda, but settled for the former as several all-purpose, general stain removers invented within the past 20 years or so, used hydrogen peroxide as their main 'ingredient'. There was no literature that I could find that utilized baking soda in the same way.
This led me to think about how statistically significant my results were, and the methods I could employ to support or disprove the results obtained. I'm still trying to figure out statistical analysis (very new and foreign to me) and am presently still unsure if there are any statistical tools I can apply to my experiment (HELP MISS TAN!), which is not exactly very conventional in that it deals with multiple variables and is more like 2 experiments rolled in one (so theoretically I should have double the page-limit)! This makes reporting especially tricky and challenging. I am currently considering using one of the statistical ranking coefficients - Kendall's Tau or Spearman Rho, probably the latter as it is easier to compute.
I really need to discuss several issues with Ms Tan in greater depth and I hope to finish my report soon. :D
No comments:
Post a Comment