Thursday, May 13, 2010
Wednesday, May 12, 2010
Variables
Ideally, I would conduct the studies and tests for all the volunteers on the same day at the same time at the same location (I was thinking a park), in case different surroundings, time of day or weather affected their thinking abilities. Actually, if we're talking about ideally, it would have been good to get a little plain room where I could test them one at a time, sort of like the police interrogation rooms on tv.
Sadly, when I perform this experiment it won't be ideal. I won't be able to control all the variables. Originally I thought I could get everyone together in a park near my house, but as I've had to ask friends of friends, I think it would be rather presumptuous of me to ask them all to come out of their ways for my experiment. Instead I'm going to have to carry out the studies and first tests over at least a week, all in different locations. Hopefully I'll be able to have them all either indoors (assuming that wherever I have to go there is some sort of building) or outdoors (which will be rather impractical if it rains).
Sadly, when I perform this experiment it won't be ideal. I won't be able to control all the variables. Originally I thought I could get everyone together in a park near my house, but as I've had to ask friends of friends, I think it would be rather presumptuous of me to ask them all to come out of their ways for my experiment. Instead I'm going to have to carry out the studies and first tests over at least a week, all in different locations. Hopefully I'll be able to have them all either indoors (assuming that wherever I have to go there is some sort of building) or outdoors (which will be rather impractical if it rains).
more method...
So, that's my method. Now there are just a few more things to work out. First of all, I forgot to mention this in my last post, each slide show will have an equal number of Asian models (equal number of males and females) and Anglo-Saxon models (equal number of males and females).
The other thing I have to work out is how long to leave between the study and test. I'm not sure whether to do them on the same day; either leaving an hour in between or performing them right after each other, or to leave a considerable amount of time in between. One experiment I read about left 3 weeks between study and test, but it seems like that would be more memory testing than just own-race bias. Another option is to perform more than one test e.g. one week after the study, then 3 weeks after the study. I suppose then I would have to change the aim a bit to include not only "to determine whether a person from one race is able to distinguish between faces of their own race more accurately than those of another race" but also something like "and whether these results change after a period of time".
My current plan is to leave one hour between the study and first test and then one week between the study and second test.
The other thing I have to work out is how long to leave between the study and test. I'm not sure whether to do them on the same day; either leaving an hour in between or performing them right after each other, or to leave a considerable amount of time in between. One experiment I read about left 3 weeks between study and test, but it seems like that would be more memory testing than just own-race bias. Another option is to perform more than one test e.g. one week after the study, then 3 weeks after the study. I suppose then I would have to change the aim a bit to include not only "to determine whether a person from one race is able to distinguish between faces of their own race more accurately than those of another race" but also something like "and whether these results change after a period of time".
My current plan is to leave one hour between the study and first test and then one week between the study and second test.
Method...in progress
Okay, so last post was about a study I'd found online that's really helpful for my method. This post is me working out my method.
The first stage of my experiment will be the study. I'll arrange pictures of models' faces in a slide show with a timer to regulate how long it remains on each slide. I'll set the timer to 4 seconds. I'll insert a blank slide between each photograph, the slide show will remain on each blank slide for 2 seconds. This slide show will be shown to all my volunteers.
In the test, I will create a different slide show will some of the same pictures from the study and some pictures of different models' faces. This slide show will be shown to all volunteers. They will be asked to give a yes or no answer to the question "Have I shown you this face before?" for each model. While there will be no timer on the slide show, if at 2 minutes they have still not given an answer, there will have to give an answer and the next picture will be displayed.
The first stage of my experiment will be the study. I'll arrange pictures of models' faces in a slide show with a timer to regulate how long it remains on each slide. I'll set the timer to 4 seconds. I'll insert a blank slide between each photograph, the slide show will remain on each blank slide for 2 seconds. This slide show will be shown to all my volunteers.
In the test, I will create a different slide show will some of the same pictures from the study and some pictures of different models' faces. This slide show will be shown to all volunteers. They will be asked to give a yes or no answer to the question "Have I shown you this face before?" for each model. While there will be no timer on the slide show, if at 2 minutes they have still not given an answer, there will have to give an answer and the next picture will be displayed.
Research
I found a different study online, this one was actually free to access. In the introduction it says that although several independant studies have confirmed the cross-race effect, other studies have lead to different results. They claim that there are a number of possible reasons for this including the type of stimulus used, the encoding time and the delay between study and test. This lead to a number of questions for me, the most important being that I had thought that the study and test were the same thing. But obviously they're not. So, what's the difference?
Well apparently this sort of experiment is made up of two main stages, not one. I'd intended to just show the pictures of models to my volunteers and ask "have I shown you this face before?" all in one go, but this isn't how it's done. Instead the first stage is the study where faces are shown to the volunteers for a set amount of time (in this study it was 3 seconds with an interstimulus time of 1.5 seconds). During the study the volunteers aren't asked any questions about the faces, just told to remember the faces. The next stage is the actual test. The volunteers are shown faces and asked questions (in this study they were asked if they had been shown the faces before and just had to give a yes/no answer. There was no time limit, the picture remained until they had answered.) This method of performing the experiment, with both a study and test, makes a lot more sense. It's rather exciting, I feel like I'm a considerable way closer to performing my SRP.
The report also confused me with mention of an "encoding time". When I first read this phrase I had no idea what it meant, however I think, it refers to how long each image is shown in the study.
Another question the introduction raised for me was "is there more than one type of possible stimulus?" I'd thought not, you could only show pictures (or maybe computer-generated faces). However after reading their method, I think what they mean is the type/quality of picture instead of completely different types of stimulus. For example they go into a lot a detail about the picture: the background was digitally removed, the colour resolution and pixels of the images remained the same the light in the image was fixed etc.
During the discussion of this report, which was titled "Recognizing faces across continents: The effect of within-race variations on the own-race bias in face recognition" the authors suggested another reason that different studies have found varying results. It is because the definition of race in these cross-race effect experiments is rarely given. For example if the races being studied were termed black and white, it would be assumed that white South Africans would be able to distinguish the faces of white Americans as well as other white South African faces, and that they would find black American faces just as hard to distinguish as black South African faces. But this is not always the case. Their conclusion was that own-race bias (another term for cross-race effect) should be "retired from literature" as race cannot be defined with precision and this can result in false conclusions. Basically, they've said me whole experiment is pointless. Ah well, I still find it interesting.
NB. In case anyone's interested, the website where I found all this is http://works.bepress.com/cgi/viewcontent.cgi?article=1039&context=christian=meissner
Well apparently this sort of experiment is made up of two main stages, not one. I'd intended to just show the pictures of models to my volunteers and ask "have I shown you this face before?" all in one go, but this isn't how it's done. Instead the first stage is the study where faces are shown to the volunteers for a set amount of time (in this study it was 3 seconds with an interstimulus time of 1.5 seconds). During the study the volunteers aren't asked any questions about the faces, just told to remember the faces. The next stage is the actual test. The volunteers are shown faces and asked questions (in this study they were asked if they had been shown the faces before and just had to give a yes/no answer. There was no time limit, the picture remained until they had answered.) This method of performing the experiment, with both a study and test, makes a lot more sense. It's rather exciting, I feel like I'm a considerable way closer to performing my SRP.
The report also confused me with mention of an "encoding time". When I first read this phrase I had no idea what it meant, however I think, it refers to how long each image is shown in the study.
Another question the introduction raised for me was "is there more than one type of possible stimulus?" I'd thought not, you could only show pictures (or maybe computer-generated faces). However after reading their method, I think what they mean is the type/quality of picture instead of completely different types of stimulus. For example they go into a lot a detail about the picture: the background was digitally removed, the colour resolution and pixels of the images remained the same the light in the image was fixed etc.
During the discussion of this report, which was titled "Recognizing faces across continents: The effect of within-race variations on the own-race bias in face recognition" the authors suggested another reason that different studies have found varying results. It is because the definition of race in these cross-race effect experiments is rarely given. For example if the races being studied were termed black and white, it would be assumed that white South Africans would be able to distinguish the faces of white Americans as well as other white South African faces, and that they would find black American faces just as hard to distinguish as black South African faces. But this is not always the case. Their conclusion was that own-race bias (another term for cross-race effect) should be "retired from literature" as race cannot be defined with precision and this can result in false conclusions. Basically, they've said me whole experiment is pointless. Ah well, I still find it interesting.
NB. In case anyone's interested, the website where I found all this is http://works.bepress.com/cgi/viewcontent.cgi?article=1039&context=christian=meissner
Sunday, May 9, 2010
Good news! I have another two Asian girls since I last updated (bringing my current total up to 4 Asian girls) and maybe, just maybe, a few more Asian guys; a friend said she'd ask around at her youth group and another friend said they'd ask around at school. On the negative side, I think quite a few people have gotten the wrong impression of me and think that I'm crazed and obsessive, understandable really as I have been going up to people and saying "What country were you born in?". Ah, the sacrifices one must make for science.
Monday, May 3, 2010
Umm..., I've come across a few bumps in the road. Basically all the little but vital details that, when I decided to go with the facial recognition experiment, I thought "I'll work it out later". Well, that later has come, but unfortunately the answers to all my problems haven't.
First of all, I need volunteers. I wanted at least 32 between the ages of 14 and 17: 16 anglo-saxons who were born and raised in Australia (8 boys, 8 girls) and 16 Asians who were born in Asia and have been in Australia for less than 10 years (8 boys, 8 girls). It turns out that it's harder than I thought to find 32 people who fit my specifications. I've got enough anglo girls and I think enough guys (I just haven't asked them all yet). I have 2 asian girls and 1, only 1, asian guy. So I've got a few solutions, sort of:
1. Go round to every person in my grade, then perhaps in other grades until I find enough girls, and hopefully some of them will have brothers.
2. Maybe increase my age restrictions if I still don't have enough.
3. Decrease the number of volunteers (but I really don't want to as I think I've got the minimum as it is).
4. If worse comes to worst, don't have the separate genders; either blend them of cut the guys completely.
Now, to the other problems I'm facing.
1. I was going to use pictures of models, but I need more than one picture of each model and how do i get those?
2. I need to work out the questions to ask in my experiment. At the moment I don't have much of an idea.
Solutions:
1. don't know
2. Research. Try and get a copy of that report I've mentioned in previos blogs and see how they carried out their (more advanced, more acurate) experiment.
NOTE: I also have to get the permission notes written.
Aaaaaaaaaaaaarrrrrrrrrrrrrr!!!!!!!!!!!!!!! There's so much to do!
Also, I apologise for the way this blog rambles on. Well, my username isn't jabbering jess for nothing.
First of all, I need volunteers. I wanted at least 32 between the ages of 14 and 17: 16 anglo-saxons who were born and raised in Australia (8 boys, 8 girls) and 16 Asians who were born in Asia and have been in Australia for less than 10 years (8 boys, 8 girls). It turns out that it's harder than I thought to find 32 people who fit my specifications. I've got enough anglo girls and I think enough guys (I just haven't asked them all yet). I have 2 asian girls and 1, only 1, asian guy. So I've got a few solutions, sort of:
1. Go round to every person in my grade, then perhaps in other grades until I find enough girls, and hopefully some of them will have brothers.
2. Maybe increase my age restrictions if I still don't have enough.
3. Decrease the number of volunteers (but I really don't want to as I think I've got the minimum as it is).
4. If worse comes to worst, don't have the separate genders; either blend them of cut the guys completely.
Now, to the other problems I'm facing.
1. I was going to use pictures of models, but I need more than one picture of each model and how do i get those?
2. I need to work out the questions to ask in my experiment. At the moment I don't have much of an idea.
Solutions:
1. don't know
2. Research. Try and get a copy of that report I've mentioned in previos blogs and see how they carried out their (more advanced, more acurate) experiment.
NOTE: I also have to get the permission notes written.
Aaaaaaaaaaaaarrrrrrrrrrrrrr!!!!!!!!!!!!!!! There's so much to do!
Also, I apologise for the way this blog rambles on. Well, my username isn't jabbering jess for nothing.
Subscribe to:
Posts (Atom)