Who Listens To Music?

Why did you pick the topic you chose?

• I enjoy observing people in social situations
• I always listen to music when I walk to class,
• I’ve noticed that not everyone always does,
• I want to see if there was a specific group or demographic that did that more than others.

What were you interested in investigating?

The differences in pre-class habits among different demographics.

What relationships did you hypothesize would exist among your variables?

I hypothesized that women would be more likely to listen to music while walking to class, but I didn’t think there would be a relationship between race/ethnicity and whether or not they listened to music.

1. Randomization - What did you do to get as “random” a sample as possible?
2. Records - How many sample records did you end up with
3. Outliers - Did you remove any outliers?

What did you do to get as “random” a sample as possible?

I observed the students rather than surveyed them in order to prevent volunteer bias. I spread out my observations over multiple days and various locations in order to create a large enough and varied enough sample to control for selection bias.

How many sample records did you end up with? 1875,

• Yes: 782 records
• No: 1093 records

Did you remove any outliers? Yes, we remove a few of the outliers for the black students, because there were so few of them at a couple locations

Report the results of your model in a table.
The table will look different depending on whether you ran an ANOVA or a GLM, but be sure you include the appropriate estimates/sums of squares and test statistics for each explanatory variable and interaction term, along with an appropriate measure of how well the model fits the data.

## General Linear Model  
LinMod <- lm(Yes.percent ~ Race + Sex, 
   weights = Total.Obser, data = Obs_df)
glm(LinMod) %>% summary()

Call:
glm(formula = LinMod)

Deviance Residuals: 
     Min        1Q    Median        3Q       Max  
-0.76636  -0.08820   0.00397   0.13156   0.66312  

Coefficients:
              Estimate Std. Error t value Pr(>|t|)    
(Intercept)   0.412165   0.013477  30.583   <2e-16 ***
RaceBlack    -0.010690   0.029535  -0.362    0.718    
RaceHispanic  0.010609   0.017039   0.623    0.535    
RaceWhite     0.009461   0.014501   0.652    0.516    
SexWomen     -0.005528   0.010937  -0.505    0.615    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

(Dispersion parameter for gaussian family taken to be 0.05520662)

    Null deviance: 4.3203  on 81  degrees of freedom
Residual deviance: 4.2509  on 77  degrees of freedom
AIC: -231.69

Number of Fisher Scoring iterations: 2

qqnorm(LinMod$residuals)
qqline(LinMod$residuals)

plot(LinMod$fitted.values, LinMod$residuals,
     xlab="Fitted Values", ylab="Residuals")
abline(h=0)

1. Model - What software did you use? What function(s)?
2. Response variable - Description, Method of measurement, Units
3. Explanatory variable - Description, Method of measurement, Units

What software did you use? What function(s)?

I used RStudio to in my analysis, utilizing the following packages

magrittr, dplyr, reshape2, ggplot2, RCurl, ggthemes, RColorBrewer

Response variable: The proportion of the population that listens to music on the way to class or not.

Explanatory variables:

1. Race/ethnicity
2. Gender

Description: The response variable is the proportion of students listening to music on the way to class, as well as the proportion of students not listening to music. I made a table that had a column for each potential response of the ethnicities and genders and whether they listened to music.

Method of measurement: It was measured by observing the students at each of the locations and marking down a tally in each of the specific columns.

Units:: Count of people

Briefly describe your explanatory variables, how they were measured, and their units (if applicable).

Race/Ethnicity: The race and/or ethnicity of each student observed. For simplicity, I made the best approximation that I could of their combined race and ethnicity of the student. Students fell into four broad catagories:

1. Asian (hispanic and non-hispanic)
2. Black (hispanic and non-hispanic)
3. Hispanic (and white)
4. White (non-hispanic)

Briefly describe your explanatory variables, how they were measured, and their units (if applicable).

Sex: The biological sex of student observed (male or female).

Units: For each location that I observed, I recorded the count of people observed wearing headphones and not wearing headphones.

Report the appropriate descriptive statistics of your response variable. You can display these in a table like this

Report the appropriate descriptive statistics for two of your main explanatory variables. You can report them in a table like the one above, or in a contingency table.

Interpret the results of your model in context.

What does this analysis suggest about your variables?
Relate it back to the original goal outlined in your introduction!

Were there any limitations with your analysis?

• Was there a known bias in your sample collection?
  
• Did your data fail any assumptions?
  
• Can you think of any confounding variables that you didn’t control for?

Which assumptions did you confirm? Include how you checked each assumption, and if space permits, include any graphs you created to check them.

Plot the interaction of your two explanatory variables (even if it is not significant). Briefly (1-2 sentences or bullet points) interpret the graph.

List at least one specific thing you would change if you were to replicate this research.

• How would that change impact your data or results?
  
• Did you find that the method you used was sufficient in answering your research question?

Source code available on Github

Copyright © 2015 Hunter Ratliff

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.