|
|
Post by melihv on Apr 23, 2020 16:52:46 GMT
I have finished coding my experiment and succesfully implemented it in the survey. However I have issues when it comes to carrying the answers from the experiment over to excel succesfully.
My experimental design: present a random animal picture for 2 seconds then ask two questions about it back to back: do you like this animal, do you know this animal? Then repeat the process for all 90 pictures. I did experiment A and experiment B to mix up the order of the questions asked.
What I want: For example we have 30 participants per experiment. I want to note their answers to each photo and each question seperately:
-For example participant A rates the 65th picture between 1-9 based on how much they know the animal AND how much they like the animal. Lets say they rate 3 on likability and 8 on how much they know about it.
I want to have mean likability/ familiarity scores separately for each photograph so I can analyse them accordingly. I just dont know how to do this.
Best regards,
Melih.
|
|
|
|
Post by PsyToolkit on Apr 23, 2020 21:16:24 GMT
Okay, this should not be difficult, there are actually different solutions. I know you put this into an experiment, which is fine. Personally, I would have done it in a survey, which is easier to code for this type of study. Now, in your case, what you need is not only to save the score and the reaction time, but also the number of the bitmap. If you have that, you will get for each participant a data file and that contains for each participant for each picture the scores. Given that you would do it back to back, I would try to put it in one line. So I would have $score1 and $score2, and I would save the RT in a temporary variable and at the end of your trial just have one "save" line. after first question...
set $score1 .....
set $firstRT RT
after second question have your save line:
set $score2 ...
save @1 $score1 $score2 $firstRT RT
|
|
|
|
Post by melihv on Apr 30, 2020 8:23:28 GMT
Thank you for the advice, I managed to run the experiment with real participants. However there is a problem with the data I got. For example in this data there are minus scores for some pictures when they should be scoring between 0-8:
6 6 sivrisinek 8 -2 8449 7132
45 45 zebra 2 8 3166 5146
66 66 saigaantilopu 1 5 2303 2649
11 11 blobfish 0 0 2405 2447
60 60 lelwelhartebeest 4 7 2308 1264
36 36 lama 3 8 1877 1453
72 72 inek 8 8 1353 1623
38 38 palamut 3 5 3114 1956
15 15 proboscismonkey -1 6 1786 1351
69 69 yelkenkuyruklumoli 3 6 3066 3277
62 62 pandaant -3 4 3386 2009
75 75 kuzu 8 8 1240 1338
58 58 hummingbirdhawkmoth 1 7 983 1852
26 26 cita 5 8 1499 1147
88 88 siyahayakliferret 1 8 1040 1297
43 43 tilki 7 8 1062 1043
47 47 lowlandstreakedtenrec -3 5 1465 1562
79 79 tavsan,79 7 8 4878 1390
44 44 vasak 7 8 1754 1170
10 10 yilan 5 2 2320 2111
33 33 kertenkele 2 7 1446 1635
85 85 meksikasalamanderi -3 -1 1195 1012
4 4 orumcek 2 0 2040 1904
55 55 fossa 2 8 1088 1107
12 12 ciplakkorfare 0 0 838 1575
51 51 addax 0 8 2294 959
78 78 papagan 6 8 2332 805
65 65 platypus 1 7 1920 1977
34 34 kirpi 5 8 1112 827
17 17 seapod 0 3 1662 1780
14 14 monkfish -3 -2 1677 1430
1 1 akrep 2 0 1874 2442
35 35 kunduz 3 8 2721 863
8 8 yarasa 2 6 1733 2166
70 70 zebraduiker 0 5 1038 1236
46 46 glaucusatlanticus 0 5 1203 1475
22 22 at 6 8 861 869
86 86 mirket 6 8 870 938
68 68 sparklemuffin 0 3 910 2303
48 48 narwhal,48 6 5 1035 2289
Most of the data files have no problem and they are very similar with the expected scores. But small number of data in some data files have minus scores. What does it mean?
|
|
|
|
Post by PsyToolkit on Apr 30, 2020 9:54:58 GMT
Sometimes things can go unexpectedly. I recommend you to try your code yourself and try to find out exactly why and where the problem occurs.
Identifying the source of such problems can be quite time consuming. Here is some advice (for all users):
1) Make a copy of your experiment and call it "test" or something like that.
2) Remove all the code that is not really necessary for replicating the problem, so that you do not need to understand everything else, just try to strip everything down to the problem and then just work with that.
3) Try to find out exactly when the problem occurs
4) Once you know, let me know via psytoolkit@gmx.com
|
|
|
|
Post by melihv on May 2, 2020 13:03:55 GMT
I think the issue that I am currently having might have started after the latest update. The first few data sets came perfectly without any problems. But lately, I have been getting minus scores and some participants reported me that some questions pass without them being able to click on them. For example, a participant told me she had the issue 3 times and when I checked the data I saw 3 minus scores. I assumed the problematic questions scored these minus data.
I recorded someone doing the experiment and 179 data out of 180 came out perfectly. The system skipped a question that was recorded as 0 but the reaction time was around 100, well under the time a participant needs in order to understand the question and answer it.
For a few reasons I don't think the issue is with coding:
1) In the beginning, I did not have issues with the data files when I had used the same coding as I have used after the problem occurred.
2) Problems happen at random times, with random pictures, random number of times (between nothing and 12 out of 180, usually around 1-2), and in either one or both questions. I think if coding was imperfect then I would've seen the problems at every level consistently.
3) At first, we saw the problem happening with safari users so we added criteria where we wanted the users to use google chrome in order to continue. The problems continued to occur not only with different types of browsers, but it also happened with different types of devices (computer and phones)
I don't think our problems are big enough to entirely scrap the data we have, the number of spoiled data is well under %10 threshold (at most %5 but mostly around %1) and since the problem occurs at random points we will have an accurate amount of data for all questions and all photos. But we would like to have the most accurate data as possible, so I do not want to continue with the experiment until I find the source of the problem.
Other groups in our research class are planning to use psytoolkit as well so the results I am getting now will be the model for other students to follow through.
Best regards,
Melih.
Edit: I worked on some coding and i will be trying out tomorrow if there is an issue i will let you know.
|
|
