Commit fce6450e authored by Morgane Flores's avatar Morgane Flores
Browse files

Update fitts_law_study.md

parent 66eb3994
......@@ -40,7 +40,8 @@ We can see the results are very different :
- maybe she forgot to put it in her results ?
- maybe she made a mistake in the parameters she gave us ?
- the slope of the regression lines are not at all the same
- for my trial (blue), I am faster at the begining but I also take more time at the end
- for my trial (blue), I'm faster for easy tasks but I also take more time than Cecile
as the difficulty increases
- Cecile takes more time but seems more consistant in her results
From that experiment, I must say that it was really hard for me to do the experiment.
......@@ -147,3 +148,18 @@ Here I sum up all the conclusion I can get from my experiments :
- it can also vary on more "human" parameters, such as the bihavior of the tester (see Test 5)
- Moreover, it can vary on the device we use to perform the task (see Test 4)
## Notes about advices
After sending this project, I received some very interesting advices from the professors.
Here are few complementary comments :
-
-
| Professors comments | Additionnal information |
| ------ | ------ |
| "As this exercise is not about programming, I simply used Google Sheet to generate graphs." Argh. Come on! :) [Céline: +1. The fact that you are not programming the experimental software does not mean that you should not script your analysis] Btw, you should apply the checklist for good graphics to https://gitlab.ensimag.fr/floresmo/Mosig-SMPE-2021/-/blob/master/Exercices/Fitt's%20Law/result_test_4.png which one is which ? [Céline: +1 and R would make this easy because a default graph is better ;)] | I didn't coded a real script because I am lacking time recently. I hope I'll be able to have more time for the next exercise. I prefered to focus on the real challenge of this exercise, the analysis of the data and the understanding of the Fitt's Law. Moreover, the study of the (ggplot flipbook)[https://evamaerey.github.io/ggplot_flipbook/ggplot_flipbook_xaringan.html#1] really motivates me to practice R when I'll have the occasion. About the color of the lines on my graphs, I is not written on the graph itself (and I agree it's a pity, so I updated it !) but the information was written just above the graph ;) |
| "I'm faster at the begining but I also take more time at the end". What's beginning and end ? | Updated, I actualy meant "I'm faster for easy tasks but I also take more time as the difficulty increases" |
| "Cecile takes more time but seems more consistant in her results". So information on the confidence of the estimates would be interesting, right ? It looks like difficulty has no significant effect in her case. | I agree, I did not see that ! I should conduct a real study on the confidence, but from what I can easily say : (1) She have only 4 dots to draw the regression line, it would have been interesting to have a value for ID = 6 ; (2) The experiment was made with 6 trials, It would have been better to have more trials (at least 10) |
| [Céline: "I have a terrible vision, then clicking on a 1px line was hell. My screen may have been to bright, so it was very exhausting for my eyes" + "different width an[d] distance can make the experiment harder / easier, even with the same IDs" -> These are very interesting comments: Vision seems to be the bottleneck for pointing at very small targets. Is this related to you only, or is it the same for everyone?] | We actually saw in the recent Human in the Loop class that human have a mean precision of 0.002mm with a mouse... So I guess the real problem was about vision. Indeed this doesn't relate to me only, it would be interesting to have more data with that parameters and from a lot of different people to understand better the effect of human vision. |
| "? Hypothesis : I may take a little more time, and that time will the proportionnal to the distance." Good! I like it when people state their hypothesis before making the experiment. :) But then, when analyzing test3, it's good that you checked that it was coherent with test2 but you did not really check whether the model still applies or not and whether the time had increased proportionaly to the distance or not. [Celine: if the model applies, the movement time should increase proportionaly to ID, i.e. log2(D/W+1)] | I should have really tested that ! I'll do it if I have time, but I understand that I forgot something essential. |
| "My conclusion to this experiment is that doubling the distance is totally adding a level of difficulty in the task, and that I should be more careful on the data I work on." 1) You verified that doubling the distance adds a new index of difficulty in the experimental task (you can as well computer it from ID = log2(D/W+1). In addition, you experimentally (started to) verify that this new ID has an impact on the movement time as predicted by Fitts' law. 2) I do not understand why this makes you say that you need to be more careful on the data you work on. | My mistake was to intentionally remove data to plot... If I plotted all the data I had, I would have understood directly the incrementation of the ID. As I removed that crutial information from my graph, I took me more time to understand x) |
| "We can very easily see that the slopes of the two tests is are very different. I took more time with the tactile pad, and the more the difficulty increases, the more I take time comparing to the mouse." -> that fact that you took more time with the tactile pad is not supported by the fact that the slopes of the two tests is are very different, but rather by the fact that the intercept for the pad is higher than for the mouse. | I misunderstood, thank you for the clarification ! |
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment