What visual statistic can I use ?

[ttl64]

I'm in high school and I'm raising mealworms for my experiment. I have a total of 5 treatments and 5 containers/treatment. Each container contains 10 worms. I measured the length & weight of the worms every three days for a period of 1 month. They didn't start out with the same length and weight. I calculated the differences between the length of the initial day and the final day of the experiment for each container. I did the same thing for the weight. What kind of visual can I use to represent that data ? Can I run an ANOVA on it to determine the significance ? By the way, is the P-value of ANOVA the same as the T-value ?
I would really appreciate if you have any other statistic recommendation.
Thanks so much for your help!

[MelissaB]

Hi,

This sounds like a great experiment! What I suggest doing is taking an average of the worms in each container (so you have one number per container) and then plot the data with length on the y-axis and treatment on the x-axis. You could either just plot the change in length, or plot before and after for each treatment--you might want to plot the change simply so you don't overwhelm the people reading your board. Then do the same with weight.

You should be able to do an ANOVA on your data, though you need to take the average of the worms in each container. Keep in mind that an ANOVA will only tell you if there are significant differences among your treatments and not which treatments are significantly different from which.

A p-value is not the same thing as a t-value. A p-value is a number that indicates how likely you are to get the results you observed in your experiment if there really aren't any differences between your treatments. By convention, p-values less than 0.05 in science are considered to be 'significant', that is, we can say that we did not obtain our results by chance alone and (for example) our treatments did affect the growth of mealworms differently. A t-value is a number you calculate in some statistical tests which you can then compare to published distributions of that number to obtain your p-value; nowadays most computer programs automatically do the comparison and calculate your p-value from the value of t.

Hope this helps!

[ttl64]

Can you please tell me how to calculate the t-value ? Isn't it supposed to tell you that the effect resulted from the treatment and not by any outside causes ? Do I need both t-value and p-value ?

How can I plot the changes ? I think I can't use the line graph because it's not a continuous data. (I did it, and it looked really confusing). If I have to do a bar chart, I'm afraid it would make the data insignificant because I would have to take average across the 5 containers/treatment.

[MelissaB]

To calculate t (actually, the test statistic of an ANOVA is called F), you'll need to use an online calculator. This one: http://www.physics.csbsju.edu/stats/anova.html is really easy to use. When it gives you a 'probability', that is your p-value. The p-value is what most people will be looking for, but typically you report both the t or the F as well as the p-value.

If these 'outside causes' affect your sample randomly, then yes, this is what the p-value tells you.

I was envisioning you plotting the value for each container as a dot. Thus, you would have five columns (representing your five treatments) with five dots each. You could technically make a boxplot with mean and error bars, but with a small sample size like five I would just stick with the dots. Does this make sense?

Let's say that the average changes in worm length for containers in treatment A were 5, 6, 7, 5, and 6 mm and the average changes for treatment B were 10, 9, 7, 8 and 9 mm. You would put 'mm' on your y-axis and have two columns, one labeled 'Treatment A' and one labeled 'Treatment B' on your x-axis. In the A column, you would have two dots at 5 and 6 mm and one at 7; in the B column you would have two dots at 9 mm and one each at 7, 8 and 10 mm.

[staryl13]

Hi!
Just to give you some more background information on statistics, check out these links-
http://en.wikipedia.org/wiki/Student's_t-test
http://www.danielsoper.com/statcalc/calc08.aspx
Good luck!