To select the pupils in my stratified sample from the school population I will use random numbers. The random numbers I use will be from a calculator and will be completely random and will not be biased in any way. The pupils have allocated numbers in the Microsoft excel file and so random numbers will be used to select the pupils I need for my stratified sample.
Once I have taken my sample, I will look at the data to analyse it. I will look for anomalies. If I find any then I will replace them with data from the database. I will randomly collect data to replace anomalous data if I find any. If I can see what is wrong with the anomalous data I will replace it myself for example there may be a typing error of 170m instead of 1.70m. I could easily change this to make the data more appropriate. I will do this until I do not have any more anomalous data in my sample so that I can analyse it for patterns that give information to test my hypothesis. With the sample I will draw graphs and look for trends. I will look at trends and patterns from different key stage groups and from girls and boys and compare the results I get. I will plot histograms to see where the most information lies and compare them with other histograms to see differences in different ages and genders of pupils. I will plot box and whisker diagrams to look at the data in a way that will show how spread out it is and to compare different key stages and genders. I will plot cumulative frequency curves to see how the cumulative frequency changes as the data values increase and to see a comparison between heights and weights on the same graph to get a good comparison. I expect to get a normal cumulative frequency curve from both heights and weights. From these results I will look at my hypothesis and see how the results have tested it.
This graph shows roughly that as height increases so does weight. This shows my hypothesis to be correct. From this graph I can say that the sample I have taken has positive correlation. The correlation is positive but not very strong. A line of best fit can still be drawn from the points on the graph though so it tests my hypothesis showing that in general it is correct.
This cumulative frequency graph shows that as a pupil’s height increases so does their weight. This also shows my hypothesis to be correct. The graph is not efficiently labelled as I would prefer to have an extra axis on the top of the graph so that I could label height and weight properly instead of having to label it in groups but I could not do this on excel so the graph does not show its best representation of the data. The graph still shows the curve I was expecting from both height and weight so I think the graph is still appropriate and informative.
After looking at my results, I have seen that my hypothesis that the taller the pupil is, the heavier they will weigh, that the pupils from KS4 will be taller and heavier than the pupils in KS3, also that boys in KS4 will be taller than the girls in KS4
has generally been proven to be correct. My diagrams and graphs show clearly that there is a trend between older boys being taller and heavier and younger girls being shorter and lighter.
During my collection of information I had to adjust some of my tables and graphs due to anomalous data. As I planned, I was able to replace anomalous data and change data to make it more appropriate. My box and whisker diagrams showed clear easy comparison between age groups and genders and so were easy to compare results. I also realised once I had plotted the box and whisker diagrams that the data had a big range but almost all of the data was inside the upper and lower quartile ranges. I could have improved my sample by further by replacing some of the data that I considered appropriate as anomalous but also keeping a large range of data so that my investigation was not limited by not enough data. Also, after looking at my cumulative frequency graph I saw that the cumulative frequency curve of pupil’s weight was not as defined as the curve for pupil’s heights. To overcome this, I could have used a larger sample to give me more data as to plot my graphs and diagrams from.