Saturday, September 13, 2008

Sampling Methodology

Hi guys,

What type of sampling method works the best? Share with us your "cafeteria experiment." You need to report your findings and comment on someone else's findings. Then your two other posts can be anything you like. But remember, you must post four times to get credit... oh, and those of you who will not be named, it does not count if you post four together. First of all, that is only one post; secondly, it defeats the entire purpose of having a blog, which is to have a threaded discussion on a topic.


Look forward to viewing the results!

53 comments:

Turtle Face said...
This comment has been removed by the author.
Anonymous said...

WOO! Im so excited to ask random people questions.
I know this post doesnt count, i just wanted to share my enthusiasm...

Shannon Elizabeth said...

Okay, here goes nothing...

Cluster (preppy jock crowd)
mean: 16 years, 2 months; college classes; played sports
median: 17 years, 3 months; college classes; played sports
mode: 17 years, 3 months; college classes; played sports

Simple
mean: 16 years, 1 month; college classes; played sports
median: 15 years, 9 months; college classes; played sports
mode: 14 years, 4 months and 16 years, 8 months; college classes; played sports

Convenience (people I saw during my free period)
mean: 17 years, 11 months; honors classes; doesn’t play sports
median: 17 years, 4 months; honors classes; doesn’t play sports
mode: 17 years, 3 months; honors classes; doesn’t play sports

By far, the most accurate portrayal of an average high school student is the results of the simple random sample, for more and many reasons indicated in the text. First of course, there is the absence of bias and pattern as well as the presence of an element of chance which is not found as strongly in the other two methods of sampling. In my cluster sample, I interviewed the jocks; of course they are going to be playing sports! Also, it was a group of mostly junior and senior jocks, keeping the mean age high. Similarly, the people I saw during my free period were all good friends of mine, meaning they were all in my honors classes with me and didn’t really go for the sports scene.

Looking at my data, it is hard for me to determine any correlation. Well, of course obviously there is the fact that jocks play sports. But they wouldn’t be jocks if they didn’t, so, this isn’t a very interesting correlation. I suppose there is the fact that in my convenience sample, all my friends in the honors classes don’t play sports. From this, we could make the inference that kids in honors classes aren’t as likely to play sports. But this would need to be expanded with a specific focus on this one idea rather then the small sampling from this exercise. Though, I don’t think it would be worth the effort in examining this inference any further because I know for a fact that almost the entire starting line up for our girl’s soccer time is in honors classes.

In conclusion, I believe that this is a good illustration of what the textbook took 20 some odd pages to explain; simple random sampling free of bias and pattern with it’s element of chance gives the most representative data for a general population.

Nathan Lessard said...

Shannon, wouldn't the third sampling method be purposive sampling, rather than convenience? We only chose participants that we knew prior to the sampling, so we chose them for a particular reason.

I suppose you could argue that it was convenience sampling as well, becuase you were only able to interview the friends of yours that you saw in a certain timeframe, but I think that Mr. Smith intended for us to do focus on a purposive method of sampling.


Nate

Abby_Beggs said...

I thought it was a conveinience sampling. YOu may have used people you know, but, the basic idea, I thought was that you were just choosing the people that showed up in front of you face, as in the lunch room.

Abby_Beggs said...

Well, here's what I got:
Cluster:
Mean: Born 09/91
A.P. classes
Didn't play sports
Median: Born 06/91
A.P. Classes
Didn't play sports
Mode: Born 07/91
A.P. Classes
Didn't play sports
Simple:
Mean: Born 06/92
College Prep CLasses
Played sports (1)
Median: Born 08/91
College Prep classes
Played sports (1)
Mode: Born 04/92
College Prep Classes
Played sports (2)
Convenience:
Mean: Born 11/91
A.P. CLasses
Played Sports
Median: Born 06/91
A.P. CLasses
Didn't play sports
I felt that the simple expiriment best exeplified people at my school, mostly this is because the cluster expiriment took only one "type" of person and the conveinience one took only the people that I personally knew, making it another "group". The simple expiriment however, deviated and showed people that I didn't ussually talk to and people other than seniors and juniors making the expiriment more broad as oppossed to narrowed down.

After looking at my data I am able to determine (much like shannon)that most people who take A.P. or honor courses are less likely to play sports, and, if they do, they ussually only play one sport. Instead, I found that they are more involved in arts programs or debate as oppossed to the more physical aspects of life. But this is just from my cluster sample.

I found that this expiriment was a good exemplification of what the lectures and nnotes and all those flippin' power points tried to explain to us. Overall I thought it was a good expiriment.

Abby_Beggs said...

Given that there's nothing else to comment on and this is my last chance to pass, I'll start analyzing something mildly related to this topic. I thought the books explanations of correlational studies was confusing. I ended up having to go online to figure out what they meant by it, so, if the correlational study on my expiriment was off, I apologize. Anyway, I thought the expiriment results on Shannon's expiriment were interesting enough. I found that the people I interviewed who were "jocks" actually took A.P. or honors courses, along with a sport. A lot of them held class office and were involved in another after school activity, like NHS. Maybe the people in my school who are "jocks" are more or less overacheivers...

lumpytrousers said...
This comment has been removed by the author.
lumpytrousers said...

Here goes nothing!

Day 1- "Brains" - Cluster Sampling

Mean: 17 years, 4 months; 2.85 classes (on that 1-3 scale), 0.65 sports played
Median: 17 years, 3 months; Honors/AP classes; 0.5 sports played
Mode: 16 years; 11 months; Honors/AP classes; 0 sports played

Day 2- Simple Random Sampling

Mean: 16 years, 0 months; 2.05 classes; 1.5 sports played
Median: 15 years 10 months; College Preparatory (regular) classes; 1.5 sports played
Mode: 15 years, 6 months; College Preparatory (regular) classes; 2 sports played

Day 3- Convenience Sampling

Mean: 17 years, 1 month; 2.25 classes; 0.95 sports
Median: 17 years, 5 months; College Preparatory (regular) classes; 0 sports played
Mode: 17 years, 5 months; College Preparatory (regular) classes; 0 sports played

Predictably, the "Brains" have the highest class mean on the scale from 1-3 at 2.85. All but two of the members of that social strata take AP, Honors, or both. They also shied away from sports (half played none, a quarter played one, and the other quarter played two). Since cliques at my school tend to be formed along graduating class lines, the mean, median, and mode of their ages lie within five months of each other. It would not be too radical to hypothesize that their friendships began in the classes they all took together, but a developmental study would be needed to verify that claim.
As a cluster sample, this social group definitely fails to represent the high school as a whole.
The convenience sampling resulting in a slightly more accurate picture of the typical student,and true to the definition of the method I didn't seek out specific students. However, the lack of age range (I mostly spend time with the people in my own class) leads to a possible source of error. I was also by limited my own biases as far as sports goes (I'm friends with many more arts lovers than sports fanatics). That could explain the lower than average 0.95 sports played mean and the 0 sports played mode and median of the convenience samples.
This means the most accurate portrayal of the average student at my school came in the form of the random sample, though this method was not without its flaws. A major source of error lies in the fact that freshmen have their own lunch period alone (so no freshmen data was collected), and also because a number of the students walked into the cafeteria with other members of their own cliques. Also, students of classrooms closest to the lunch room when the bell rings, as well as students who bring cold lunch were heavily favored by the nature of the sampling (no waiting in lunch lines). Not to mention the fact that applied courses students ("Region 9" at my school), are only at our school lunch every other day. Because of these sources of error, there were patterns, less than equal chances for people in classrooms farther away who ordered hot lunch, and no chance for applied classes students who happened to be at Region 9 that day. Despite these reservations, I believe the 2.05 classes mean and the 1.5 sports played mean portrays a good cross-section of the typical student body, based on my own observations over the years.
I humbly beg your forgiveness for this unforgivably long post. I tend to be a windbag rambler.

Tayla said...

This is what i got..

Day 1:
Cluster (preppy-jocks)
mean- 17 years, 5 months; regular classes; play sports.
median- 16 years, 3 months; honors classes, play sports.
mode- 16 years, 3 months; regualar classes; play sports.

Day 2:
Simple
mean- 14 years, 11 months; regular classes; play sports.
median- 15 years 1 month; regular classes, play sports.
mode- 14 years, 11 months; regular classes, play sports.

Day 3:
Convenience (people I was with during break or study hall)
mean- 16 years, 4 months; honors classes; play sports.
median- 16 years, 3 months; honors classes; play sports.
mode- 16 years, 3 months; honors classes; play sports.

Tayla said...

By looking at everything I think that the simple random sample best explains people in my school. Although the age is a little young because it happened that I went into the freshman/sophomore lunch, it does show that most people in my school play sports. We have a very small school and regardless of which classes are taken most will be involved in sports.

Unlike most, I found that even those taking AP or honors challenge classes in my school play sports. This could have something to do with that our honors challenge is not that much harder or different than the regualar classes taken.

This experiment helped me understand all that was on the powerpoints a little better. Sometimes hands on work and seeing differences in other people schools can change the way you are first looking at something new and give you a much broader explination.

Amanda said...

Here is my data!

-Cluster Sampling- Band Members

Mean: 16 years, 1 month; 2.5 classes (using the scale given); played sports
Median: 15 years, 9 months; 3 classes; played sports
Mode: 14 years, 3 months; 3 classes; played sports


-Simple Random Sampling

Mean: 15 years, 3 months; 1.5 classes; played sports
Median: 15 years, 9 months; 1 classes; played sports
Mode: 15 years, 3 months; 2 classes; played sports


-Convenience Sampling

Mean: 17 years, 2 months; 2.9 classes; played sports
Median: 17 years, 2 months; 3 classes; played sports
Mode: 17 years, 2 months; 3 classes; played sports


Both the text books and the powerpoints say that simple random sampling is the best way to gather generalizable data about a large group. This experiment helped to show that first hand. I do believe that simple random sampling is the best method. In my simple random sample, my school is more accurately portrayed than in the other samples. The age is a little off because I happened to see a lot of young people when I was gathering the data. However, the other data is right on. A majority of students in my school take college prep or a level below that; a majority also plays sports.

In the group sample, the only data that does not accurately portray the school is the types of classes. Most of the band members took honors classes. This is not typical of the whole school and therefore, cannot be generalized to the population. Only a few band members did not participate in a sport.

In the last group, the convenience sample, the age and the class types are a little off and do not accurately portray the school either. The people I know are older and this shows with the age being around 17 years. Also, most of the people I interviewed were in my classes, which are honors or AP.

In conclusion, the random sample most accurately portrays the school that I attend. My data confirms what the text book and the powerpoints say. And this is good!

Amanda said...

I think it is fun to look at the other data being posted! I find it interesting that my convenience sample looks different from some of my classmate's samples. In Chelsea's and Shannon's convience samples not many people played sports. They were mostly in honors or college prep classes, but it looks like only a few played sports. I guess the only reason this surprised me was because all of my samples had a majority that played sports. The people that did not play sports maybe had an extracurricular activity not associated with the school or something.

lumpytrousers said...

After looking at Amanda's data, I found it interesting that sports seem to be a common thread throughout her entire school, dominating the mean, median, and mode of every sampling method. This trend continued even in the "Band Member" group, which usually doesn't conjure up images of athleticism in the common stereotype (ex:"One time, at band camp...", not to say anything against band, I've been a member myself). Go disproving of stereotyping! But since sports seemed to be the common denominator in her data, I was wondering if she gathered any more specific answers concerning the level of involvement (how many sports did each person play?) and also what constituted as a sport in her sampling (did she count dance, for instance? or swimming? or biking?). Just a thought.

samchasse said...

Day#1 Cluster (Preps/Jocks)

Mean- 16 years, 1 month; college prep; played sports
Median- 16 years, 4 months; college prep; played sports
Mode- 16 years; 4 months; college prep; played sports

Day #2 Simple
Mean- 16 years, 5 months; college prep; played sports
Median- 16 years, 1 month; college prep; played sports
Mode- 16 years, 2 months; college prep; played sports

Day #3 Convenience

Mean- 16 years, 7 months; AP/Honors; played sports
Median- 16 years, 2 months; AP/Honors; played sports
Mode- 16 years, 11 months; AP/Honors; played sports

Each sampling method provided valid research given the target group. Students at Fort Kent Community High School are not very diverse, thus most of the data was similar. In this test, the simple random sample most accurately represented the average student at FKCHS. Most students are 16 years old and some-odd months, taking college prep courses, and play sports.
The only outlier of the data involved the convenience sample. I gathered data during my first two period classes, which are both AP classes, so most of the students were taking AP/Honors courses. This is an example of how convenience sampling can misrepresent a population. If one were to look at the data from the convenience sample without knowing the circumstances, they would assume that most students at FKCHS are taking Advanced Placement classes, which is not true. Only a small percentage of students are taking AP classes whereas the rest of the school takes applied or college prep courses.
One result you could gather from this data is that all students avidly participate in sports, irregardless of the classes they are taking. Actually, the conveience sample had the highest number of students playing sports and the higest number of AP courses. Most students would find this odd (geeks don't play sports) but this is not true in Fort Kent.

This could also be used to describe a correlation. The higher percentage of students taking AP courses, the higher amount of students participating in sporting events.
Another correlation could be made claiming that if students attend FKCHS they will participate in at least one sport.

This project used data to point out what everyone already knew. That is basically the point of research in Psychology. Psychologist gather research to give valid proof on what they already know to be true, or feel to be true. Research just puts the nail in the coffin.

Shannon Elizabeth said...

I just thought of something really cool. What if this sampling was expanded to a sampling of schools as a whole and not just of the students within a specific school? We already have a sampling of different schools from across the state since we all go to different schools, so what if we gathered all our data to see if the trends and correlations we noticed in our own schools is also visible schools across the state. Do the majority of kids in schools in Maine play sports like most of us got in our samplings? It would be sort of interesting to figure out. Now, correct me if I'm wrong, but this new sampling would be a purposive sampling because it's not really random; it's the schools that we all attend.

lumpytrousers said...

Hey, Shannon that's pretty sweet idea--to take what was a specific school survey and make it statewide. I think it's convenience sampling in that context, not purposive...we didn't specifically choose our schools based on their relevance to the state, we just happen to be going there and so its convenient for us to sample kids from those schools. If we were to do that, we'd need to operationally define what playing sports means (some people consider cheering a sport, some don't, etc.) and such, so that the criteria extended evenly across all schools. Like I said in response to Amanda's data, it'd be nice to know how many sports each student played. Was I the only one that did do that? I feel so alone...

Abby_Beggs said...

I don't know Shannon, I think it would still be a random sampling because it is the schools that we all attend but we're not picking out individuals. It's like if a psychologist went to Boston to do a random sample study it wouldn't be TRULY random because it's just the people in Boston, but it would be pretty random. I don't know, I could be wrong, but there's my guess.

Amanda said...

Looking at Chelsea's comment, I also found it interesting that most of the people I sampled played sports. There were only a handful that did not. Even more surprising, at least a quarter of all three samples played more than one sport. And I did count dance and biking. The reason I counted them was because they show more about the person and just add to their profile. It would significantly change the data if I counted only a certain number of sports. Overall though, sports were common in my school, as Chelsea noticed!

Amanda said...

I agree with Abby on the subject of Shannon's comment about broadening the survey. I think it would still be random sampling. Yes we are just interviewing students but it would be on the subject of sports. We would just be choosing random people from the schools. Like Mr. Smith posted, in purposive sampling you must choose the group for a specific reason. The group chosen would have to have certain qualities that go along with the experiment or survey. That's just what I think though! :-P

Unknown said...

Ok...Lake Region is the most average high school out there.

Random sample
Age: Mean = 16 years and 3 months
Median = 16
Mode = 16
Classes: Mean = 1.85
Median = 2
Mode = 2
Sports = 17/20 played sports

Convenience sample
Age: Mean = 16 years and 4 months
Median: 17
Mode = 16
Classes: Mean = 1.91
Median = 2
Mode = 2
Sports = 20/22 played sports

Music Geek Group (band + chorus)
Age: Mean = 16 years and 3.5 months
Median = 16
Mode = 16
Classes: Mean = 2.25
Median = 2
Mode = 2
Sports = 12/20 played sports


I think that the random sample is the best type of random sample because it does not discriminate and it is a fair way to find out information from a wide group of people like at school. The other sampling methods were not as good because they allow discrimination and when asking a group, the people in the group are more likely to be like the others in that group just because they are friends.

samchasse said...

I find it interesting that most students in Maine do play sports. I wonder how this correlates with the amount of overweight students in our state. I assume there would probably be a correlation between the amount of students in a state play sports and the amount of overweight students in that state.

Carly said...

Cluster Sample: Jocks (I walked around the cafeteria asking everybody in a school team jersey)
Mean: 16 years, 9 months; 2.6 classes; played sports (obviously).
Median: 16 years, 11 months; 3 classes; played sports.
Mode: 17 years, 10 months; 3 classes; played sports.

Random Sample
Mean: 16 years, 2 months; 1.9 classes; 10 played sports, 10 didn't.
Median: 16 years, 6 months; 2 classes.
Mode: 15 years, 8 months; 1 classes.

Convenience Sample
Mean: 16 years, 6 months; 2.4 classes; 13 played sports, 7 didn't.
Median: 17 years, 0 months; 2 classes.
Mode: 17 years, 4 months; 2 classes.

Based on the data I collected, it appears that the Random Sample method is the most accurate when surveying students at my high school. The Cluster Sample portrayed 100% of the student body as playing sports, which is obviously not accurate, which is why that form of sampling doesn’t work. Plus all of the athletes are fairly good students, which means higher level classes, which doesn’t represent the whole student body either.
The Convenience Sample was a little closer to accurately portraying the students at the high school, except that they were mostly 17 year olds, because they were all my friends who are seniors.
The Random Sample portrayed a 50/50 split on sports teams which might not be quite accurate, but it’s closer than the other methods. The ages are about 16 which seems to be the average age at the high school, and the classes are right in the middle. This was the most difficult sample, but it is definitely the most accurate.

samchasse said...

I find it funny how basically everyone's convenience sampling had students taking AP/Honors courses. I found it weird in my data also because I don't hang out with the predominately "smart kids". Most of my friends are actually not that smart (haha don't mean that badly...they'll admit it!). I wonder if anyone else is similar

Unknown said...

Simple
Age-Mean-16 years 5 months
Mode- 16 years 10 months
median- 16 years 3 months
classes: Mean- 1.95 (scale of 3)
Mode- 2
Median- 2
Sports: ( 0 is no sports, 1 is sports)
Mean-0.8
Mode- 1
Median- 1

Convenience

Age:
Mean- 15 years 10 months
Mode- 14 years 7 months
Median- 15 years 4 months
Classes:
Mean- 1.85
Mode- 2
Median- 2
Sports: (Same scale as above)
Mean- 0.7
Mode- 1
Median- 1

Cluster
Age:
Mean- 16 years 4 months
Mode- 16 years 5 months
Median- 16 years 2 months
Classes:
Mean- 2.25
Mode- 2
Median- 2
Sports:
Mean- 0.65
Mode- 1
Median- 1

The convenience sample was my study hall, the cluster was the brainier kids at my school, and the simple is self explanatory. At a small school like mine (70 students) its hard to find a true group of 20 so my brainy group might have different data from someone else's at a larger school. Some correlation i see is that the closer to average classes you get, the more students play sports. This makes sense because people taking lots of classes are probably less likely to do sports, and complete slackers aren't likely candidates for sports either. The only other thing of note for now is that my mean, mode and median for age are all similar in every group. I attribute that to how difficult it is to find 20 people in my high school without asking people of all ages.

I think the simple sample is the best representative of my school, but i think to get a true sample of the school, you would need to get a purposeful sample. High school students in general are likely to hang around people who are like them, so interviewing the first people you see means you are bound to get similar types of people.

Unknown said...

Ok, so here's my data:

Day 1 - Clique: "Preps"

mean: 17 yrs, 1 month; 2 classes; played sports
median: 17 yrs, 6 months; 2.5 classes; played sports
mode: 17 yrs, 1 month; 2 classes; didn't play sports

Day 2 - Random

mean: 16 yrs; 2.2 classes; played sports
median: 16 yrs, 7 months; 2 classes; played sports
mode: 17 yrs, 9 months; 2 classes; played sports

Day 3 - Convenience

mean: 16 yrs, 9 months; 2.3 classes; played sports
median: 17 yrs, 5 months; 2 classes; played sports
mode: 17 yrs, 9 months; 3 classes; played sports

Based on this experiment I would say that the simple random sampling best portrays the average student in my school. According to the books and multiple power points, simple random sampling is the best form of generalizing a large population. This assignment re-enforced that point. This type of sampling doesn't allow bias to exist towards any group and that's important when you are tryingt o generalize an entire school population.

My cluster sample is not very accurate at representing the student body because the age is a bit high considering most of these "Preps" were in my grade. I do feel as though the classes were represented well in this sampling because many students in my school take a 200 or 300 level class. My convenience sample is also not a very accurate representation because the people I know are around my age and are in my classes which are AP/Honors. I do think that the sports sampling is right on because most of our student body is involved in some type of sport. I did not limit the list of sports because I don't think that it is right for me to tell someone that what they do is not a sport if they think it might be.

Overall, I feel more comfortable knowing abou the different types of sampling there are. It is clearly evident, throughout all the data, that the simple random sampling is the best method to use.

Unknown said...

This comment is regarding one left near the beginning of this blog which dealt with the difference between purposive and convenience sampling. In my opinion, the third sample would be considered a convenience sample and not a purposive one. It's a convenience sample because we knew these people so we could easily obtain data from them. A purposive sampling is designed to target people for specific reasons. For example, if you were conducting research to see the effects of the new football rule you would only ask those who are involved with the sport. You may not ask someone you know because they might not play football. That's my take on the differences, but I may be a bit off.

Unknown said...

I liked the way other people reported their data better than my way. It was clearer, easier to read, and probably faster. I'll apologize for not checking the blog earlier, i don't know what i was thinking, but i only discovered we also need to post and respond yesterday, so my posts will be in quick order. Anyways, I was surprised how many different ways we interpreted the little things in the directions, like do we ask if they play any sports at all, or how many do they play. Some people used the date and year of birth, instead of asking people exactly how old they are. I think its more accurate, because a lot of time, people didn't feel like calculating exactly how many months since their birthday, so they rounded or something. And I let people interpret the difficulty of their schedule, and had no real rule for it. Which probably meant my data was more of how difficult do you think you're schedule is, because often times people i knew had easy classes were giving me twos and threes.

samchasse said...

I think it can be a convenience sample or a purposive sample, depending on what your intentions were with coming up with results. WIth a convenience sample you were probably just interviewing the people easiest to interview, people in your classes, neighborhood, etc. It could have been a purposive sample if they were interviewing a group for a certain purpose. Interviewing the kids in most of your classes (if you take AP/Honors classes) than you have are surveying them for a certain purpose, to get students taking AP/Honors courses.

michelle renee chasse said...

Alright so here are my results...

Cluster (The Smart People)
Mean: 17 years, 3 months; 2.8 classes; played sports
Median: 17 years, 7 months; 3 classes; played sports
Mode: 17 years, 7 months; 3 classes; played sports

Random
Mean: 16 years, 3 months; 2 classes; played sports
Median; 16 years, 6 months; 2 classes; played sports
Mode: 16 years, 6 months; 2 classes; played sports

Convenience
Mean: 17 years, 6 months; 2.9 classes; don't play sports
Median: 18 years; 3 classes; don't play sports
Mode: 17 years, 9 months; 3 classes; don't play sports

I would have to agree that the simple random sampling gives the most accurate representation of a general population. The simple random sample allows random people to be picked rather than a select group, therefore giving a better picture of what the entire student body is made up of. Interviewing only a select group of kids will leave room for error because only one specific group is being interviewed. That select group of kids does not represent the student body as a whole, just a part of the student body. Therefore, in order to get the best results, the simple random sampling is most effective in this situation.

lumpytrousers said...

This is in response to Abby and Amanda concerning the statewide survey: Would it be considered random sampling if we all pooled our data to compile the statewide mean, median, and mode? I don't think so because there is a pattern to the data: we're all AP Psychology students that, according to our conveniences samples, tend to associate with others who do AP classes. That skews the data a bit. Also, how geographically diverse (as far as Maine goes, anyway) is our AP psych class? All that considered, I think that kind of sampling would be convenience, since it was convenient for us to gather data from our own schools--we didn't have to seek anyone out, they were just there.

Jourdan said...

Cluster (The Smart People)
Mean: 17 years, 3 months; 2.8 classes; played sports
Median: 17 years, 7 months; 3 classes; played sports
Mode: 17 years, 7 months; 3 classes; played sports

Random
Mean: 16 years, 3 months; 2 classes; played sports
Median; 16 years, 6 months; 2 classes; played sports
Mode: 16 years, 6 months; 2 classes; played sports

Convenience
Mean: 17 years, 6 months; 2.9 classes; don't play sports
Median: 18 years; 3 classes; don't play sports
Mode: 17 years, 9 months; 3 classes; don't play sports


Michelle and I worked on this project together, hence why our results are the same. Before beginning this project, I anticipated that the random sample would clearly have more variety. The results were not very different from how I thought they would be. The random sample clearly represents the school better because it helps to gain the perspective of all students and not jsut select groups of similar people.

michelle renee chasse said...

I have to agree with what Chelsea mentioned earlier; the people I interviewed for the convenience sample were mostly my friends and only a couple do sports. This does cause a bias, reinforcing the idea that simple random sampling is the more effective method to use. AP classes are more important to most of my friends than sports. However, this is not true for all of the students in the school. This leaves the convenience sample with a bias not only on the sports aspect, but also the AP class aspect. Also, most of my friends are the same age as me, giving the results little diversity. Without a range of data from different groups of people, the results will not accurately represent the school as a whole.

Unknown said...

After reading people answer which sample they thought best represented their school, and answering the question myself, I asked a freshman at my school which sample he thought was best representative of the high school. He chose the cluster sample of brainer kids. He was blind, all I showed him was 3 sets of results, unlabled. And he, ironically, is not a brainy kid, who plays multiple sports, so I'm not quite sure how he decided the cluster sample, with harder classes, and less sports, fit best. I think the reason we all chose the simple sample as best representative of our schools is because the simple is the book's favorite method. We knew, coming in, that the simple would have the best data. If you were just choosing the best sample results blind, i wonder if the simple sample would still be the favorite. I don't doubt that the simple method gave the best results, I just think that its very hard to accurately define a group you are a part of.

Shannon Elizabeth said...

Okay, so yeah, reading everyone's comments about my last post about the statewide survey, I think it is convenience sample too. It is all the most easily accessable data to us.

And jacob, I think that's so cool that you asked some random freshmen about your data. Him being blind and everything. That's awesome. And I definitely agree with your whole "the book said simple was best" thought. Would we have thought differently if we hadn't known which one was better? More of us would have to do the same thing that you did in order to see if more people disagree with the book. But I think it would be kind of cool to see. Neat idea :)

Unknown said...

This goes along with Sam's comment on 9/23. It is interesting to note that a lot of kids seem to play sports in Maine. I wonder who's collecting the data for the obesity stats and I wonder what method they are using. It would be interesting to delve into the relationship between obesity and sports in Maine schools. Maybe the obesity data is only taken from certain grade levels or specific age groups. I'd like to learn a bit more about that!

Unknown said...

Wow Jacob! I couldn't imagine going to a school with only 70 students in it. We have around 1,000 students! It's understandable that finding 20 different students would be a bit difficult.

Anonymous said...

Cluster- Mean: 17 years, 2 months; college classes; played sports
Median: 17 years, 3 months; college classes; played sports
Mode: 17 years, 3 months; applied classes; played sports

SRS- Mean: 16 years, 5 months; college classes; played sports
Median: 16 years, 9 months; college classes; played sports
Mode: 17 years, 4 months; college classes; played sports

Convenience: Mean: 17 years, 6 months; honors classes; played sports
Median: 17 years, 5 months; honors classes; played sports
Mode: 17 years, 5 months; honors classes; played sports

As has been noted by almost everyone before me, the simple random sample gave the most accurate picture of the student. This did not come as a surprise to me after reading about sampling and also having taken a course in statistics. There was an obvious bias in the convenience sampling, because I was surveying friends who were in the same classes as me or who ran cross country with me. The cluster sampling was also from surveying a crowd of "jocks", so therefore, the result of lower course levels and 100% of the sample playing sports did not come as a shock.

All in all, this survey just gave more support to the idea of staying away from convenience sampling at all costs, and to only use cluster sampling when I'm looking to study a certain variable. In any other case, the simple random sample will give you the most variation and therefore the most accurate results.

Anonymous said...

Also, I was expecting to see an obvious correlation between playing sports and having an average course level, but this the link wasn't actually distinct in my survey. In the cluster sampling, it appeared that playing sports was closely related to being in a lower course level. However, in my simple random sample, there were many people in honors courses that also played sports. So, perhaps we shouldn't be too quick to judge the geeks and the jocks!

Unknown said...

It's interesting that you chose to ask a freshman to decide which set of data represented your school Jacob! I never really thought about it before but I also wonder if we'd think differently if we hadn't already been told that the simple random sampling would represent the population the best. I think that if we didn't know then some of us may say that the convenience sample best represented our school because many of us have a mind-set that "our school" consists mainly of the people we know. This experiment has definitly given me some new insight into the general population of my school!

Anonymous said...

That is a good point Cory.. many people probably would think that just asking friends would give a good sense of the entire population. I notice convenience sampling all the time now in every day life.. for example, listening to the radio. A station will have someone call in with thoughts on a certain topic. By having people simply call in on their own, you are only getting information from the part of a population that has a strong view on that topic. Anyone else who doesn't have the time to call in, or is just too lazy, won't have their voice heard.

Anonymous said...

Although it did not give a very accurate picture of the entire school, I think the different cluster samples are interesting to look at it. It's funny to see the differences between the "preppy jock crowd" and the "nerdy" crowd. While mine did not follow the average pattern, other people had very stereotypical results for this survey. It would be interesting to use the cluster sample for other surveys, such as finding each "groups" opinion on certain topics regarding things in the school.

Anonymous said...

Clustered Sampling (only 15, Sophomore Girls)
Median: 15 years, 7 months; honors classes; played sports
median: 15 years, 6 months; honors classes; played sports
mode: 16 years, 4 months; honors classes; played sports

Simple Random Sampling (only 15)
mean: 16 years, 1 month; regular classes; doesnt played sports
median: 16 years, 9 months; honors classes; doesnt played sports
mode: 17 years, 4 months ; regular classes; doesn't play sports

As far as I can tell the most accurate depiction of an average high school student came out of the Simple Random Sampling. There was no way for this to be biased, and they were not picked out of one group but of several. In the cluster sample I interviewed a group of 15 Sophomore girls and they all seemed to have the same thing to say about every question I asked. They all seemed to play sports, and were all generally smart so they took honors classes. Where they were all Sophomores this kept the age pretty much the same. The results from the Simple Random Sampling were significantly different. I found that not all of the students were taking honors classes, and obviously not all were playing sports. I also had the opportunity to explore a larger age group.

From this data I cannot find any specific thing that catches my eye. If I were to have made a hypothesis I would have said that the Sophomores that took honors classes would not have been so active in sports, and that the students that took regular classes would have been. I guess at my school it is different because we have about 140 students in the entire school so therefore everybody knows everybody so it is hard to distinguish "cliques" because we do not have any.

This made the idea in our textbook become a part of real life and we got a chance to experiment a little with interviews and collecting data. We also got a chance to see the difference between data that has been collected with bias, and then some without.

Unknown said...

Has anyone else noticed things in the news semi-recently about how teenagers need more sleep, less stress, ect. because our lives are too busy? Well I think the correlation between class difficulty and sports supports the idea that teenagers are just fine. The data suggests that kids who have more classes, and theoretically less time, are less likely to play sports, which takes up time. I know correlation doesn't prove causation, and my idea has a lot of flaws, but the correlation alone suggests teenagers are aware of time constraints, and acting upon them. I know some kids will fill their time with other things, but sports are universal enough that i would predict you would see the same trend between class load and other extra curriculars. Why 'slackers' don't play as many sports is beyond me, and i frankly don't feel like taking a stab at that one, because its late, and I'm proving myself wrong by staying up late working on homework.

http://stophomework.com/teenagers-drastically-need-more-downtime/71

Chelsea said...

Here is my data:

Cluster: The Skateboard Crowd
mean: 16 years, 7 months; regular classes; did not play a sport
median: 17 years, 2 months; regular classes; did not play a sport
mode: 16 years, 9 months; regular classes; did not play a sport

Simple
mean: 17 years, 4 months; honors classes; played sports
median: 16 years, 10 months; honors classes; played sports
mode: 16 years,7 months; honors classes, played a sport

Convenience
mean: 17 years 3 months; ap classes; played sports
median: 17 years, 7 months; ap classes; played sports
mode: 17 years, 9 months; ap classes, played sports

I think that the best sampling method for my school is the simple random sample, which seems to be the consensus. My school is filled with very ambitious people, so almost everybody takes honors classes and plays sports. I went into this assignment knowing this, so I decided to interview my skateboarding friends because I know that most of that crowd doesn't "fit into the poland mold" if you will. I wanted to bring a little variety into the mix.

Jourdan said...

I would like to add to the conversation about our own personal bias in this. Having read the book and knowing that the simple sample is highly endorsed we became biased and went into the assignment beliveing that would be the outcome. I know i personally read the assignment and thought that I knew the outcome already.

The second topic I wanted to mention was Jacob's last comment about whether kids generally realize their limitations and that we can see this a little through our data. I think that can be true in a lot of ways and there are a lot of trends that show this of course there are many exceptions. It might be said that the students who are not doing sports take many more AP's and honors classes and therefore they become overworked and stressed and possibly those that play sports are more involved in co-curriculars and using their time that way. I definately think the theory makes sense generally, but I think there are several exceptions.

michelle renee chasse said...

Like some others have already pointed out, it's interesting to see the various data patterns that have surfaced. "Jocks" don't take as many AP classes as do the "smart people," nor do the "smart people" participate in as many sports. My data would most definitely fit that hypothesis, however, not everyone does fit that mold. There are some people I know who can juggle AP classes and sports. But a lot of people do tend to focus their time on either sports or academics.

michelle renee chasse said...

Looking at the results of not only my experiment, but everyone else's results as well allowed me to realize how stereotypical the results were. It also proved to me just how diverse the different groups of students are. I think it would be interesting to collect data and see if we could label the type of group it came from just based on the results.

Chelsea said...

I agree with Michelle on this one. Looking at everyone's data, it seems like Maine students are very stereotypical in these areas. It would be interesting to see whether or not the same experiment would have the same results in different states. Another idea is to compare Maine's data to the national averages.

Chelsea said...
This comment has been removed by the author.
Chelsea said...

Jacob,

I would challenge your hypothesis that the correlation between class diffictuly and sports supports the idea that teenagers are doing just fine. I'm not saying it's wrong, I think it would be interesting to look at private schools such as Exeter in NH and Hebron. I think the variety of sports and classes offered at any given school has a lot to do with it.

Chelsea

Chelsea said...

For everyone who posted numbers as classes in thier data:

Is this based on a scale? Or do you carry two to three classes a grading period? At Poland HS, to be eligable you have to carry/pass six classes a semester.

Mr. Smith said...

Great work guys!! I am giving you the next topic off, but your level of discussion is really picking up, I am impressed!