Research Methods in Politics 16

11
Research Methods in Politics Research Methods in Politics Chapter 16 Chapter 16 1 Research Methods in Research Methods in Politics Politics 16 16 Factor Analysis and other Factor Analysis and other Advanced Techniques Advanced Techniques

description

Research Methods in Politics 16. Factor Analysis and other Advanced Techniques. Teaching and Learning Objectives. to understand the meaning and application of factor analysis and cluster analysis to learn how to carry out factor analysis - PowerPoint PPT Presentation

Transcript of Research Methods in Politics 16

Page 1: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

11

Research Methods in Research Methods in PoliticsPolitics

1616

Factor Analysis and other Factor Analysis and other Advanced TechniquesAdvanced Techniques

Page 2: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

22

Teaching and Learning ObjectivesTeaching and Learning Objectives

1.1. to understand the meaning and to understand the meaning and application of factor analysis and cluster application of factor analysis and cluster analysisanalysis

2.2. to learn how to carry out factor analysisto learn how to carry out factor analysis3.3. to introduce other advanced functions to introduce other advanced functions

and techniques including the binomial and techniques including the binomial and Poisson distributions and time series and Poisson distributions and time series analysisanalysis

Page 3: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

33

Factor and Cluster AnalysisFactor and Cluster Analysis

– characteristics (variables) occur together characteristics (variables) occur together frequently in social world, e.g. poor health, bad frequently in social world, e.g. poor health, bad housing, povertyhousing, poverty

– they may be observable symptoms of underlying they may be observable symptoms of underlying common factorscommon factors – latent variables – latent variables

poor healthpoor health bad housingbad housing povertypoverty

distribution ?distribution ? high unemploymenthigh unemployment of powerof power Low skillsLow skills

factor/factor/latent variablelatent variable

clustercluster

Page 4: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

44

R-matrixR-matrix

   XX11 XX22 XX33 XX44 XX55

XX11 11

XX22 0.50.5 11

XX33 0.40.4 0.30.3 11

XX44 0.10.1 0.10.1 0.10.1 11

XX55 0.10.1 0.10.1 0.20.2 0.70.7 11

Page 5: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

55

R-matrixR-matrix

  XX11 XX22 XX33 XX44 XX55

XX11 11

XX22 0.50.5 11

XX33 0.40.4 0.30.3 11

XX44 0.10.1 0.10.1 0.10.1 11

XX55 0.10.1 0.10.1 0.20.2 0.70.7 11

Principal factors

Page 6: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

66

Rotation- VarimaxRotation- Varimax

Principal Factors

00.10.20.30.40.50.60.70.8

0 0.1 0.2 0.3 0.4 0.5 0.6

Factor A

Fact

or B

Factor B

Factor chartFactor scores

Page 7: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

77

Advanced Statistics SoftwareAdvanced Statistics Software

SPSS, Minitab, STATA etcSPSS, Minitab, STATA etc available on 30-day trialavailable on 30-day trial maintained by universitiesmaintained by universities SPSS SPSS Statistical Package for the Social ServicesStatistical Package for the Social Services Statistical Product and Service SolutionsStatistical Product and Service Solutions DataData

– Data EditorData Editor Data ViewData View Variable ViewVariable View

Page 8: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

88

Time Series AnalysisTime Series Analysis

UnemploymentUnemployment

YearsYears

Underlying trend

Seasonal component

Irregular component

Cyclical component

Page 9: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

99

Method of Moving AveragesMethod of Moving Averages

Example: Iraqi civilian casualties 2003/06Example: Iraqi civilian casualties 2003/06

NumberNumber

MonthsMonths

Trend LineTrend Line

Moving Average

0500

1000150020002500300035004000

1 6 11 16 21 26 31 36 41

Data Point

Valu

e

Actual

Forecast

Page 10: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

1010

DistributionsDistributions

Normal DistributionNormal Distribution Binomial DistributionBinomial Distribution

only two outcomes, e.g., yes/noonly two outcomes, e.g., yes/no success, p . . . failure qsuccess, p . . . failure q BernoulliBernoulli probability of event 0.5probability of event 0.5

Poisson DistributionPoisson Distribution probability of rare eventprobability of rare event used to identify whether larger incidence of used to identify whether larger incidence of

exceptional event, e.g., clusters of rare illnesses exceptional event, e.g., clusters of rare illnesses can be attributed to chance or local can be attributed to chance or local circumstances which might involve local causecircumstances which might involve local cause

Page 11: Research Methods in Politics 16

Research Methods in Politics ChaptResearch Methods in Politics Chapter 16er 16

1111

Questions for Discussion, Workshops or Questions for Discussion, Workshops or AssignmentsAssignments

1.1. Work through each of the examples given in this chapter. Seek help Work through each of the examples given in this chapter. Seek help when you encounter difficultieswhen you encounter difficulties

2.2. Collect new data from the class population of their assessment of Collect new data from the class population of their assessment of candidates for UK prime ministers using the criteria listed in the candidates for UK prime ministers using the criteria listed in the chapter. Add four additional criteria of your own choice. Carry out a chapter. Add four additional criteria of your own choice. Carry out a factor analysis and comment on the resultsfactor analysis and comment on the results

3.3. Obtain unadjusted data for registered unemployment in the UK since Obtain unadjusted data for registered unemployment in the UK since May 1979. Apply time series analysis. Identify the size of irregular May 1979. Apply time series analysis. Identify the size of irregular movements and offer cogent potential explanationsmovements and offer cogent potential explanations

4.4. Two Departments of Politics, having equal and good RAE and TQA Two Departments of Politics, having equal and good RAE and TQA ratings in ‘new universities’ A and B both have 30 places available ratings in ‘new universities’ A and B both have 30 places available for similar Politics degrees. A attracts 56 applicants whilst B attracts for similar Politics degrees. A attracts 56 applicants whilst B attracts 73. Could the variation have arisen entirely by chance or does it 73. Could the variation have arisen entirely by chance or does it indicate that B’s department or university are significantly more indicate that B’s department or university are significantly more attractive than A? If so, what should A do?attractive than A? If so, what should A do?

5.5. The Conservative party has a working majority of 11 over all other The Conservative party has a working majority of 11 over all other parties combined in the House of Commons following the general parties combined in the House of Commons following the general election. On average, five MPs die each year causing by-elections to election. On average, five MPs die each year causing by-elections to be held. The new prime minister believes that harsh economic be held. The new prime minister believes that harsh economic policies are essential for the first three years. He assumes that half policies are essential for the first three years. He assumes that half the Tory-held seats will be lost in by-elections. Calculate the the Tory-held seats will be lost in by-elections. Calculate the probability that the party can survive in office for three years. probability that the party can survive in office for three years. (Remember that each time a government loses one seat, their (Remember that each time a government loses one seat, their majority reduces by two).majority reduces by two).