Question 1
Most people don't succeed in losing weight by dieting, and on average people don't lose any weight. Proponents of the quantized-self movement believe that by tracking eating behavior, using apps like MyFitnessPal, people will lose more weight. Pedro randomly selected 25 people from his university who were interested in dieting. He then had them use MyFitnessPal to track their eating behavior. He wants to understand whether using the app leads to greater weight loss.
1a) (Must answer all correct for points on question 1a.)
What is Pedro’s research question?
What is the predictor (independent variable)?
What is the outcome (dependent variable)?
What kind of dependent (outcome) measure is this?
1b) What should his null and alternate hypotheses be in words and in symbols?
1c) Would he be more likely to conclude that there is a significant effect is he uses a one -tailed or two-tailed test?
1d) Pedro found that after 6 weeks his sample of 25 had lost an average of 2 pounds. He set an alpha level of 0.10 at the start of the study and his p-value from the analysis is 0.12.
Are Pedro’s results statistically significant?
Is his group different from the rest of the population?
What can you conclude about MyFitnessPal and weight loss?
Do you believe Pedro’s findings to be substantively significant? (This question is not graded.)
Question 2
Professional athletes are often judged in terms of performance on a wide variety of metrics, which feed into a wide array of uses from fantasy leagues and gambling to AI development in video games. We will use the Fifa 2017 video game data as a proxy for actual soccer (football) player performance. Import the Fifa CSV file.
Mean composite scales are variables that represent a mean score built from other variables. Use R to construct a mean composite scale that represents a player’s kicking abilities. Use the following variables: Shot_Power, Finishing, Long_Shots, Curve, Freekick_Accuracy, Short_Pass, and Long_Pass. Attach your new variable to the Fifa 2017 dataset. See the R setup file for more on how to construct a mean composite scale with an example.
The following questions will have you evaluate the new kicking ability scores of all Fifa 2017 players, with those of the top and bottom ranked football clubs: FC Barcelona and Longford Town respectively. [ As reported using the top and bottom non-zero average rating on https://www.futhead.com/17/clubs/?sort=-average_rating. ] The club names are the same in the R data. You will need to use subsetting techniques to isolate the scores for each team. You may use whichever subsetting approach you prefer.
2a) Report summary statistics below for this new kick scale variable for all Fifa players, FC Barcelona, and Longford Town clubs (PQ table with n, min, max, 1st & 3rd quartiles, median, mean, and sd).
2b) Calculate a 95% confidence interval for the mean for all Fifa players, FC Barcelona, and Longford Town. Do this by hand below and confirm your work using R. Round your answer to 2 decimal places.
What can you conclude from your confidence intervals? Not graded, be descriptive.
Question 3
The General Social Survey has 6 questions that pertain to optimism. The variables are named LOTR1, LOTR2, LOTR3, LOTR4, LOTR5, and LOTR6.
3a) What type of variables are LOTR1 through LOTR6? (scale of measurement). Look up the full survey question asked for one of the six LOTR variables and enter it below. Hint: This will be easier if you use the GSS online data explorer rather than the pdf codebook.
3b) Use the R code from the setup file to create an item scale, which allows us to do quantitative analysis on it. The new variable will be called LOTR.scale. Examine the setup code to learn more about how we constructed this new variable.
Using a PQ table, report the n, min, max, 1st & 3rd quartiles, median, mean, and sd.
3c) Use R to produce a histogram of lotr.scale with PQ labels and title and insert it below this question. Describe if you believe the data to be relatively normally distributed.
3d) Use R to calculate a confidence interval for LOTR.scale. Use a = 0.01. Round your answer to 2 decimal places. You may find the confidence interval by hand or R, but do not need to do both.
