week 12 - practice final exam questions

35
Crm204 Final Exam 15 April – 9:00-11:00am – TRS 2166 Out of 80 / Worth 40% of final grade 40 Multiple choice 40 Marks for the short answer questions Extra Office Hours: 14 April – 1:00 to 3:00pm Please bring pencils to the final exam for the multiple choice scan cards AND bring your student ID. NOTE: you do not need a calculator for the exam. If you still wish to bring one, it must be a simple adding/subtracting calculator – phones, etc are not allowed.

Upload: ivan-bin

Post on 25-Sep-2015

232 views

Category:

Documents


0 download

DESCRIPTION

Exam

TRANSCRIPT

  • Crm204 Final Exam 15 April 9:00-11:00am TRS 2166

    Out of 80 / Worth 40% of final grade40 Multiple choice40 Marks for the short answer questions

    Extra Office Hours: 14 April 1:00 to 3:00pm

    Please bring pencils to the final exam for the multiple choice scan cards AND bring your student ID.

    NOTE: you do not need a calculator for the exam. If you still wish to bring one, it must be a simple adding/subtracting calculator phones, etc are not allowed.

  • Practice Final Exam QuestionsTake up the three take-home questions assigned last week

    In-Class Group workThe normal distributionT-test CorrelationCross-tabs (contingency tables)

    Controlling for a 3rd variableDoes Location Matter for Victimization?

  • Take home questions from last week:Take home question #1 (pulling percentages out of cross-tab)

    Take home question #2 (reading information from a t-test)

    Take home question #3 (which statistical test would be most appropriate?)

  • Take Home Question #1Look at the cross-tab on the next slide and answer the following questions:Are people who are very satisfied with their personal safety more or less likely than those who are somewhat/very dissatisfied with their safety to believe that sentences are not severe enough?What proportion of those who are very satisfied with their personal safety believe that sentences are not severe enough?What proportion of those who are somewhat/very dissatisfied with their personal safety believe that sentences are not severe enough?What proportion of those who feel that sentences are too severe/ about right feel somewhat/very dissatisfied with their personal safety?What proportion of those who feel that sentences are not severe enough feel somewhat/very dissatisfied with their personal safety?If Im interested in how those who believe sentencing is about right/not severe enough vary with respect to their views on personal safety, would I want to look at row or column percents? Is the relationship between perceptions of sentencing and safety from crime significant?

  • Perceptions of sentence severity and personal safety

  • Take Home Question #1: AnswersPeople who are very satisfied with their personal safety are less likely than those who are somewhat/very dissatisfied with their safety to believe that sentences are not severe enough.62.8% of those who are very satisfied with their personal safety believe that sentences are not severe enough.85.5% of those who are somewhat/very dissatisfied with their personal safety believe that sentences are not severe enough.2.4% of those who feel that sentences are too severe/ about right feel somewhat/very dissatisfied with their personal safety.6.4% of those who feel that sentences are not severe enough feel somewhat/very dissatisfied with their personal safety?If Im interested in how those who believe sentencing is about right/not severe enough vary with respect to their views on personal safety, I would look at column percents.Yes, p
  • Take Home Question #2People were asked to rate, on a 10 point scale, how good of a job they thought the police and courts are doing (1=poor job, 10 = great job). Given the output below, what can you say about how people view the police vs. courts?

  • Take Home Question #2: AnswerPeople's ratings of the police were significantly higher than their ratings of the courts. The average rating for the police was 7.76 while the average rating for the courts was 4.26. This was found to be highly significant and thus very unlikely to be random variation (p
  • Take Home Question #3Which Statistical Test Would be Most Appropriate?1. Looking at kids in youth court, are those who have more 2 or more court appearances more negative (on a scale from 1 to 100) in their ratings of the courts than those who only have one court appearance? T-test, independent groups2. Is there a linear relationship between number of hours of free time in a month and number of crimes committed during that month?Correlation3. Is there a relationship between believing that crime has increased (yes/no) and whether one feels sentences are about right or too lenient? Chi-square4. Which do people feel more confident will reduce crime increasing sentences or providing more social supports for people. (Confidence is rated on a scale that runs from 1/not at all confident to 10/very confident). T-test, paired samples if one group was asked both questions; independent groups if two different samples were each asked one question

  • In-Class Group WorkThe normal distributionT-testCorrelationCross-tabs (contingency tables)

  • a) The Normal DistributionTwo students are talking about how poorly they did when they wrote Sprotts final exam. One student (from Mondays class), Stan, says that he did worse on the final exam because he got 50% when the average was 70% with a standard deviation 10%. Another student, Romeo, (from Tuesdays class), said that he did worse because he got 45 (out of 70) when the class average was 60 (out of 70) with a standard deviation of 5.

    Which student, Stan or Romeo, did worse on Sprotts final exam, relative to the rest of the class?

  • a) Normal Distribution Answer Romeo did worse, relative to the rest of the class. His mark on the final exam was 3 standard deviations below the mean, so only about 1% of the class did worse. Stan, on the other hand, was 2 standard deviations below the mean, so about 3% did worse.

  • Stan: 40%50%60%70%80%Romeo: 4550556065

    Chart1

    0.010.01

    0.010.01

    0.1350.135

    0.340.34

    0.40.4

    0.340.34

    0.1350.135

    0.010.01

    0.010.01

    percentage

    percentage

    .1%

    .1%

    2%

    13.5%

    34%

    34%

    13.5%

    2%

    Sheet1

    -4-3-2-101234

    percentage1%1%14%34%40%34%14%1%1%

    Sheet1

    Megan

    percentage

    Normal Distribution

    2%

    13.5%

    34%

    34%

    13.5%

    2%

    Sheet2

    Sheet3

  • b) T-test

    People were asked what proportion of convicted offenders (youths or adults) reoffend within 5 years.

    What can you say about how the public views recidivism among youths vs. adults

  • b) T-test AnswerPeople believed that slightly more adult offenders would recidivate compared to young offenders. On average, people thought 54.9% of adult offenders would recidivate within 5yrs compared to 51.8% of youth offenders.

    That difference was significant (p=.037) which means that there is only a 3.7% chance that the difference is simply a bit of random variation.

  • c) CorrelationWhat, in normal words does it mean if you hear that the correlation between criminal record and length of prison sentence is +.783?

    c) Correlation Answer A correlation of +.783 between criminal record and length of prison sentence means that there is a relatively strong linear relationship between criminal record and length of prison sentence. Specifically, as criminal record increases, so does the length of a prison sentence.

  • d) Cross-tabs (contingency tables)

    What proportion of people from Quebec believe that sentences are not severe enough?

    What proportion of people from Ontario believe that sentences are not severe enough?

    Which region has the largest proportion of people saying that sentences are not severe enough?

    Is the relationship significant?

  • d) Cross-tabs (contingency tables) Answers65.9% of people from Quebec believe that sentences are not severe enough

    84.1% of people from Ontario believe that sentences are not severe enough

    The prairies have the largest proportion of people saying that sentences are not severe enough (88.7%)

    Yes, the relationship is significant (p

  • Controlling for a 3rd VariableDoes Location Matter for Victimization?Dependent VariableExperiencing a Victimization (yes/no)

    Independent VariablesLocation where you live (urban/rural)How often you go out (a little, half the month, every night)

    Where you live (urban/rural) relates to whether or not you experience a victimizationGoing out in the evening relates to whether or not youre victimizedWhere you live also relates to the number of times you go out in the evening

    So if you control for going out, does where you live still relate to victimization?

  • Group WorkControlling for a third variableIndependent variable: GenderIndependent variable: Fear of victimizationDependent variable: View of sentences

  • Controlling for a 3rd VariableDoes Gender Relate to Sentencing Views?Dependent VariableViews of Sentences (about right/too lenient)

    Independent VariablesGender (male / female)Fear of victimization (very worried/somewhat worried/not worried)

    Women are more likely than men to say that sentences are too lenientThose who are worried about being a victim also believe sentences are too lenientBut women are more likely than men to be worried about being a victim of crime

    1) So, if you control for being worried about being a victim of crime, are women still significantly more likely then men to want sentences increased? 2) More generally, what does the following cross-tab tell you about gender, being worried and wanting sentences increased?

  • Does Gender Relate to Sentencing Views?Women are no longer significantly more likely then men to want sentences increased once you control for being worried about being a victim of crime.Of those who were very worried, 75.3% of men and 84.6% of women wanted sentences increased. Although a 11% difference, this was not significant (p=.055).Of those who were somewhat worried, 75.4% of men and 74.2% of women wanted sentences increased. And of those who were not at all worried, 66.5% of men and 67.8% of women wanted sentences increased. Those differences were also not significant (p=.438 and p=.073, respectively).

  • Does Gender Relate to Sentencing Views?There may be an overall effect of being worried about being a victim of crime. For example, of all of those who were very worried 82.4% wanted sentenced increased, compared to only 67.1% of those who were not at all worried.The effect of fear may be stronger for women. For example, 84.6% of women who were very worried wanted sentences increased while only 67.8% of women who were not at all worried wanted sentences increased. That difference is larger than what is seen for men. 75.3% of men who were very worried wanted sentences increased while 66.5% of men who were not at all worried wanted sentences increased.

  • Group WorkDependent VariableViews of Police (good job/poor job)

    Independent VariablesContact with police (yes / no)Visible minority status (Visible Minority / Non-Visible Minority)

    Controlling for visible minority status, are those who had contact with police more or less likely than those who didnt have contact with police to think the police are doing a poor job of treating people fairly?

  • Does Contact with Police Relate to Views of the Police?Controlling for visible minority status, those who had contact with police are more likely than those who didnt have contact with police to think the police are doing a poor job of treating people fairly.Within Visible Minorities, 58.9% of those who had contact with the police said they were doing a poor job compared to only 42% who didnt have contact. This was a significant difference (p=.02)Within Non-Visible Minorities, 42.9% of those who had contact with the police said they were doing a poor job compared to only 33.8% who didnt have contact. This was a significant difference (p=.008)

  • *****