RT socstat

Slide1: 

Social Statistics Instructor: Tamás Rudas Office: 7.92 Office hours: Tuesdays 1pm-2pm, always by appointment E-mail: dekan@tatk.elte.hu Website: http://statisztika.tatk.elte.hu/tanszeki_honlap/Rudas_Tamas.htm Class meets: Thursdays, 10 – 11:30

Slide2: 

Text: notes to be posted on the website Frankfort-Nachmias, C., Leon-Guerrero, A. Social Statistics for a Diverse Society. Pine Forge Press. Copies are available int he library. Aims: To provide the students with an introduction to using data and statistical methods for the description of the society. Develop basic understanding of the main sources of data, of methods of interpretation of data and of elements of statistical inference.

Slide3: 

Class procedures: Class attendance will be recorded Every student has to make two 20-minutes presentations Topics for presentations: Topic 1: Data and data sources Hungary, economic data Hungary, demographic data Hungary, regional data European Union, economic data European Union, demographic data European Union, regional data The World, economic data The World, demographic data The World, regional data

Slide4: 

Topic 2: Use and misuse of data and statistics Find, present and criticize the use of data or statistics in the media Grading: Each presentation: 30% Final test (last class): 40%

Slide5: 

Bonus example Death penalty data from Florida, 1976-77 (Radelet, 1981, American Sociological Review) How many white/black defendants? How many death/other penalties? Who has a higher chance to receive the death penalty, a white or a black defendant? Has this anything to do with racial discrimination?

Slide6: 

More complete form of the data How are the two tables related? Does the race of the victim has any influence ont he chance of receiving the death penalty?

Slide7: 

Some basic concepts Data Statistics Official statistics Census Surveys Designed experiments Observational studies

Data(plural)Usually numerical informationNumbers are not dataOne needs to know the method of collection: 

Data (plural) Usually numerical information Numbers are not data One needs to know the method of collection

StatisticsThe science of collection, analysis and interpretation of dataOfficial statistics – mathematical (inferential) statisticsAdministration – scienceCentral role in governance – feedbackNew role: evidence based medicine, governance, etcNarrow meaning: data summary (e.g., average) : 

Statistics The science of collection, analysis and interpretation of data Official statistics – mathematical (inferential) statistics Administration – science Central role in governance – feedback New role: evidence based medicine, governance, etc Narrow meaning: data summary (e.g., average)

Official statisticsThe word „statistics” comes form the word „state”Collect, arrange (in tables, lists, etc) knowledge about the population, the economy, etc.Hungarian Central Statistical Office Census Bureau, Bureau of Labor StatisticsNatonal Statistics: 

Official statistics The word „statistics” comes form the word „state” Collect, arrange (in tables, lists, etc) knowledge about the population, the economy, etc. Hungarian Central Statistical Office Census Bureau, Bureau of Labor Statistics Natonal Statistics

CensusComplete enumeration of the populationUsually every 10th yearLegal requirementPolitical decision about the questionsExtensive and expensiveMicrocensusCensus undercountPost-enumeration survey: 

Census Complete enumeration of the population Usually every 10th year Legal requirement Political decision about the questions Extensive and expensive Microcensus Census undercount Post-enumeration survey

SurveysA small fraction of the population is interviewedSample and populationGeneralization to the entire populationSample selection, scientific methodsIntensive and cost effecftive: 

Surveys A small fraction of the population is interviewed Sample and population Generalization to the entire population Sample selection, scientific methods Intensive and cost effecftive

Tuition fees and incomesAbout 45% of university students in Hungary pay tuition feesThe others are in state-supported spacesCurrent legislation makes these to pay about HUF 105 000 per year from next year onTuition fees are about 250 000 – 400000 a yearHow much is this?Average monthly gross income about 220 000Average monthly net income about 130 000The tuition is 2 – 3 month’s average net income: 

Tuition fees and incomes About 45% of university students in Hungary pay tuition fees The others are in state-supported spaces Current legislation makes these to pay about HUF 105 000 per year from next year on Tuition fees are about 250 000 – 400000 a year How much is this? Average monthly gross income about 220 000 Average monthly net income about 130 000 The tuition is 2 – 3 month’s average net income

Who pays tuition?Little is known but it is likely that children of families in less advantageous positions – otherwise these children would do better and would obtain the cheaper placesA central issue – involving lots of statistics – is inequality within the societySocial stratification: crystallization of inequalitiesSocial mobility: moving form one stratum to anotherIntra- or intergenerational mobilityThe school system does not necessarily reduces social ineaqualitiesSome researches claim it may even enlarge the differences : 

Who pays tuition? Little is known but it is likely that children of families in less advantageous positions – otherwise these children would do better and would obtain the cheaper places A central issue – involving lots of statistics – is inequality within the society Social stratification: crystallization of inequalities Social mobility: moving form one stratum to another Intra- or intergenerational mobility The school system does not necessarily reduces social ineaqualities Some researches claim it may even enlarge the differences

Do most of the people earn above or below average?Income distribution : 

Do most of the people earn above or below average? Income distribution

Most people earn below averageMedian is the value that half the peole earn less than this, half the people more than thisTypical incomes over lifetimeVariations depending on educational levelDifference if the graph is made for an individual or for from cross sectional data: 

Most people earn below average Median is the value that half the peole earn less than this, half the people more than this Typical incomes over lifetime Variations depending on educational level Difference if the graph is made for an individual or for from cross sectional data

Comparison of incomes for high school and university graduatesUniversity graduate: starts to earn laterhigher income (the advantage is the highest among the OECD countries)earns longersmaller decrease before retirement age(Cross sectional data. Individual data are different) : 

Comparison of incomes for high school and university graduates University graduate: starts to earn later higher income (the advantage is the highest among the OECD countries) earns longer smaller decrease before retirement age (Cross sectional data. Individual data are different)

Comparison of data sourcesReliability and validityParticulars of data collection about human populationsThe respondent’s attitudeSensitive questions: 

Comparison of data sources Reliability and validity Particulars of data collection about human populations The respondent’s attitude Sensitive questions

Data and causalityDecison makingIntervention researchCause – effect relationships Association and causation: 

Data and causality Decison making Intervention research Cause – effect relationships Association and causation

Data and inference Is smoking bad for your health?How do you know?Medical evidence?Statistical evidence?What is statistical evidence? Data or their interpretation?What is scientific evidence? How is it based on data?Will giving up smoking make you healthier?Will you live longer if you give up smoking?: 

Data and inference Is smoking bad for your health? How do you know? Medical evidence? Statistical evidence? What is statistical evidence? Data or their interpretation? What is scientific evidence? How is it based on data? Will giving up smoking make you healthier? Will you live longer if you give up smoking?

Association is found easilyCausation is rarely found„happens after” is different from „happens because of”Association is not causation.Having long hair is associated with having babies.Having long hair does not cause having babies : 

Association is found easily Causation is rarely found „happens after” is different from „happens because of” Association is not causation. Having long hair is associated with having babies. Having long hair does not cause having babies

How does one see association from data?: 

How does one see association from data?

Those having long vs short hair may also differ in other aspectsOne cannot conclude that the difference in the the chances of having children is caused by differences in hair length.Perhaps some of the other differences may be more relevant„Construct” a sex x hair x baby table : 

Those having long vs short hair may also differ in other aspects One cannot conclude that the difference in the the chances of having children is caused by differences in hair length. Perhaps some of the other differences may be more relevant „Construct” a sex x hair x baby table

Slide25: 

Men Women All Long hair 0 50 350 150 350 200 Short hair 0 400 150 50 150 450 baby no baby baby no baby baby no baby

What kind of data?quantitative – qualitativediscrete – continuousLevels of measurement1. Nominal or categoricalcategories are different – no specification of the differencedisjoint and exhaustive categoriesusually few categoriesgender, hair colour: 

What kind of data? quantitative – qualitative discrete – continuous Levels of measurement 1. Nominal or categorical categories are different – no specification of the difference disjoint and exhaustive categories usually few categories gender, hair colour

2. Ordinal scaleordering among the categories(larger, more beautiful, …)educational level3. Interval scaledifference between categoriesno „zero-point” on the scale??: 

2. Ordinal scale ordering among the categories (larger, more beautiful, …) educational level 3. Interval scale difference between categories no „zero-point” on the scale ??

4. Ratio scalealso ratios are meaningful„zero-point” is impliedincomeLevels of measurement:objective or subjectiveit may depend on the use for proxy variables: 

4. Ratio scale also ratios are meaningful „zero-point” is implied income Levels of measurement: objective or subjective it may depend on the use for proxy variables

Data summaries aka descriptive statistics: 

Data summaries aka descriptive statistics Mode: typical observation – most frequent category or value Median: middle observation – only if ranking is meaningful Mean: average observation – interval or ratio scales Parameters of location

Slide30: 

The mean may not be typical: 3,3,3,3,3, 11,11,11,11,11 Mean: 7 2,2,2,2,2, 92 Mean: 17 2,2,2,2,2,212 Mean: 37 The median is less sensitive: 2,2,2,2,2,4,10,50,100 Mode: 2, Median: 3 2,2,2,2,2,4,10,500,1000 Mode: 2, Median:3

Slide31: 

The mean does not exist for nominal variables: Faculties of Elte Number of students who applied Science 2500 Arts 7000 Law 2500 Social Science 2000 Informatics 2200 Special education 1500 Teacher training 2000 What is the variable? What is its level of meaasurement? What are the observations? Where did the ‘average student’ apply?

Slide32: 

Data sets with the same location may be very different 1,1,1,1, 19,19,19,19 9,9,9,9,11,11,11,11 The observations differ Variation / dispersion Measures of variation

Slide33: 

Number of different categories or values Range – interval scale (largest minus smallest) Mean absolute deviation – ratio scale 2, 3, 4, 5, 11 Mean: 5 Absolute deviations: 3, 2, 1, 0, 6 Mean absolute deviation: 12/5=2.4

Slide34: 

Standard deviation (SD) Variance = square of SD SD = square root of variance Variance: mean squared deviation from the average 2, 3, 4, 5, 11 Mean: 5 Squared deviations: 9, 4, 1, 0, 36 Variance: 50/5=10

About the final test: 

About the final test In class, about 100 minutes Closed book, also no personal notes No calculators are needed Topics: Sources of data (census, survey) Quality of data (validity, reliability, interview situation) Inference based on data (association, causation) Basic numerical facts about Hungary

Possible types of questions: 

Possible types of questions Definitions of basic concepts Assessment of data quality Interpretation and evaluation of analyses based on data Logical relationships between basic economic and demographic statistics

Grading: 

Grading Your final grade will be calculated form the two presentations (30% each) and the final test (40%) Grades will be available from the 2nd week of January by sending an e-mail to me. Those unhappy with their grades will be offered an oral examination on January 24. This exam will override your test grade but not the presentation grades. If you wish to take the exam, you have to sign up using ETR