SOCIAL PERCEPTIONS
Social Perception Survey Do people make prejudices based on appearance/stereotypes? We used photos as a bias to test this.
Randomization Using the master schedule, our group immediately disregarded all classes without a second period Next, we used the random number table (line 125) to produce three random numbers that were associated with the teachers alphabetically Our random number table gave us three teachers (two first choice, one back-up): Krogh, Lockheart, and Leal
The Survey Biased: Had only pictures, there was no way for the sample to tell what the people were like, their personalities, or behavior outside of what they saw and what their first impression was Unbiased: Given short descriptions and no pictures, made more logical and accurate inferences on behavior and personality
The Survey The sample size for the unbiased survey was 38 and the sample size for the biased survey was 33. Since both of these samples are greater than 30, the CLT applies here and we can assume normality for all calculations.
Question 1 Who would you trust with wiping the tears of your 4-year-old child: Man or Woman? Photos to provide bias:
Data for Question 1 Unbiased Biased Man Woman Man Woman n=38 Man: 2.0526 Woman: 36.9474 n=33 Man: 25.7576 Woman: 8.2424 Here it is clear that the photo created a bias. The unbiased survey showed that woman was the most frequently chosen response, but when we added photos, the responses changed completely, with an over-whelming support of the answer man.
Statistical Inference, Question 1 (unbiased) P: We want to estimate the actual proportion of people in CCA who would rather have a man wipe their child s tears, than a woman. A: SRS-We took a random sample of CCA students Normality- n(p) and n(1-p) 10? 38(.0526)=1.999 38(.9474)=36.0012 But, sample 30, so normality should apply. Independent- Population of CCA 10(38) T: pˆ±z* [ {pˆ(1-pˆ)/n}].0526±1.96 {(.0526*.9474)/38}.0526±1.96(.0362).7105±.0709 (-.0183,.1235) (0,.1235) C: We are 95% confident that between 0% and 12.35% of the students at CCA would rather have man wipe their child s tears instead of a woman.
Question 2 Who would you rather have cook you a 3-course meal? Contestant on America s Next Food Network Star or a first year student in culinary school? Photos to provide bias:
Data for Question 2 Unbiased Biased Middle-Ager 6-year-old Food Net. Star Student n=38 Food Net. Star: 27.7105 Culinary Student: 11.2895 n=33 Food Net. Star: 10.303 Culinary Student: 23.697 When no photo was provided, most people said that they would rather have a Food Network Star cook their food as opposed to a culinary school student. But when shown pictures, most people ended up choosing the culinary student. This shows that the bias was effective in changing people s opinions.
Statistical Inference, Question 2 (unbiased) P: We want to estimate the actual proportion of people in CCA who would rather have a contestant on America s Next Food Network Star cook them a meal rather than a student in culinary school. A: SRS-We took a random sample of CCA students Normality- Normal because n(p) and n(1-p) are 10 38(.7105)=26.999 38(.2895)=11.001 Independent- Population of CCA 10(38) T: pˆ±z* [ {pˆ(1-pˆ)/n}].7105±1.96 {(.7105*.2895)/38}.7105±1.96(.0736).7105±.1442 (.5663,.8547) C: We are 95% confident that between 56.63% and 85.47% of the students at CCA would rather have a contestant on America s Next Food Network Star cook them a 3-course meal.
Question 3 Who would you allow to use your bathroom? Middle-Ager or 6-year old? Photos to provide bias:
Data for Question 3 Unbiased Biased Middle-Ager 6-year-old Middle-Ager 6-year-old n=38 Middle-Ager: 27.7105 6-year-old: 11.2895 n=33 Middle-Ager: 28.8485 6-year-old: 5.1515 We hypothesized that people would be more likely to allow a 6-yearold to use their bathroom as opposed to a middle-ager just because they re typically seen as more innocent. Therefore, our bias wasn t much of a bias at all and only shifted the results slightly.
Statistical Inference, Question 3 (unbiased) P: We want to estimate the actual proportion of people in CCA who would rather allow a middle-ager to use their bathroom than a 6-year-old. A: SRS-We took a random sample of CCA students Normality- Normal because n(p) and n(1-p) are 10 38(.7105)=26.999 38(.2895)=11.001 Independent- Population of CCA 10(38) T: pˆ±z* [ {pˆ(1-pˆ)/n}].7105±1.96 {(.7105*.2895)/38}.7105±1.96(.0736).7105±.1442 (.5663,.8547) C: We are 95% confident that between 56.63% and 85.47% of the students at CCA would rather allow a middle-ager to use their bathroom than a 6-year-old.
Question 4 Who would write a more beautiful poem? Old man or young girl? Photos to provide bias:
Data for Question 4 Unbiased Biased Old Man Young Woman Old Man Young Woman n=38 Old Man: 25.6579 Young Girl: 13.3421 n=33 Old Man: 22.6667 Young Girl: 11.3333 Again, there was not a large shift in results after adding a bias. That means the general population of CCA thinks an old man would write a better poem, and images of the writers would not change their opinions.
Statistical Inference, Question 4 (unbiased) P: We want to measure the actual proportion of people in CCA who think that an old man would write a better poem than a young girl. A: SRS-We took a random sample of CCA students Normality- Normal because n(p) and n(1-p) are 10 38(.6579)=26.999 38(.3421)=11.001 Independent- Population of CCA 10(38) T: pˆ±z* [ {pˆ(1-pˆ)/n}].6579±1.96 [ (.6579*.3421)/38].6579±1.96(.0769).6579±.1508 (.5071,.8087) C: We are 95% confident that between 50.71% and 80.87% of the students at CCA believe an old man would write a better poem than a young girl.
Question 5 Megan is a 16-year-old girl who is hard working and pays attention in class. She likes to knit, ice-skate, and play soccer. What do you think her GPA is (4.0 maximum)? Photo to provide bias:
Data for Question 5 Unbiased (no photo) Mean GPA: 3.64 Median: 3.7 St. dev:.3107 Variance:.0965 Min: 2.9 Max: 4.0 n=32 95% CI for the mean: (3.5374,3.7416) Biased (photo) Mean GPA: 3.71 Median: 3.75 St. dev:.2884 Variance:.0832 Min: 3.0 Max: 4.0 n=38 95% CI for the mean: (3.6048,3.8127)
Question 5 Unbiased Biased
Overall-Unbiased All unbiased responses were as predicted, apart from the public s apparent preference for allowing middle-aged strangers to use their bathrooms over 6-year-old strangers. We had anticipated the results might have been opposite. Reason for this was probably a flaw in our own hypothesis. Perhaps we were wrong to think people would feel sensitive about what kinds of strangers use their bathrooms. Perhaps these results were more indicative of this uselessness of a question, or apathy on our surveyees parts.
Overall-Biased Our use of visual bias seems to have been effective. For example, both tearwiping and meal-cooking responses completely changed. Bathroom-using slightly increased as a result of our bias, though we feel it should have been more dramatic, if only its unbiased responses were not so strange in favoring middle-aged strangers. Last, the question of poem-writing changed in favor of our bias as well, though not as obviously as we would have hoped. The concepts tested in these last two questions may not have been significant enough to show strong results in such a small sample size. We might also just have very strange ideas about social perception that would not have been statistically supported in any survey, no matter the sample size.
More There are not really statistics on these topics, but research on national opinion. Studies say most people feel more comfortable with a female babysitter, and our unbiased survey supports that. Our surveys did not suffer from undercoverage or nonresponse. Bias naturally occurred from people s prejudices and long-standing opinions.
In conclusion We found that by changing the survey in most cases changed the outcome of how people answered. In almost all cases when we made the question bias, the response followed and people answered exactly as we expected them to do. One big problem we had with our bias is in some cases it is hard to find the exact visual aid we needed in order to introduce the exact bias we were looking for, so we had to use pictures slightly different then we intended. While conducting this survey, we were surprised to find that the mean GPA of the student with the picture showing was.07 higher, as we expected it to be significantly lower with the bias of the picture introduced. If we were to repeat the survey, we may consider removing the description of the girl in the biased survey and simply show a picture of her and ask what the survey taker suspects is her GPA.
Any Questions?