A sample is biased if it produces results that are systematically different from those of the population in a specific direction.
*EX: an election sample that included only republicans
*EX: an unbiased sample would include democrats, republicans and independents in proportion to their numbers in the population.
Reliability & Validity
-Variables can be judged by how validly and reliably they measure a concept.
-Validity: does is measure what we think it measures?
-Reliability: is the variable consistent?
-To study a theory, at least some of its concepts must be measured by variables.
-A variable is a measurable trait or characteristic which can vary and which is used to measure a concept.
*EX: age may be measure with a specific Q on a questionnaire: :how old are you (in years)?" The Q measures a variable and the answer to the Q can take on a range of possible values.
-Concepts are connected to specific variables by operational definitions.
Operational definition: a description of procedures used to measure a concept in sufficient detail so that someone else could peform the same procedure and get a similar result.
In social surveys the questions asked of respondents tend to be answered with greater reliability and validity when they:
* ask things respondents could reasonable be expected to know
* ask things respondents want to tell you correctly
EX: avoid a social desirability bias--respondents tend to answer Qs in ways that make them appear to have socially desirable traits as being truthful,smart, and fair.
* ask things that are neither too difficult to answer nor consume too much time
-Most sociological reseach examines a sample
-A sample is a subset of members of the population rather than the entire populations
-Even though we study samples we usually do so with the intention of gerneralizing the reults to the broader population.
-A sample of people who are selected because they are easy to find.
*EX: the next 20 ppl who enter through a door.
-A sample including specific numbers (quotas) of cases falling in various subcategories.
*EX: the first 10 men and the first 10 women you meet on the street.
A probability or "random" sample is a sample in which
-each case in the populatiun has some known probability of being included
-all segments of the population are represented.
Statistics are mathematical measures summarizing important characteristics about a sample.
summarize the distribution of a variable
3 measures of 'central tendency are often used
Measures of Association
Examine the relationship between variables.
the most commonly occuring category or value in a sampling of cases
Tests of significance
Ask whether a result could have occured by chance
Types of statistics
measures of associations
tests of significance
the midpoint for a series of cases for which half the cases are below and half are above.
the average or central tendency of scores computed by totaling the scores and dividing by the number of cases.
Rates: Proportions and Percentages
Donald Rumsfield at one point said that being a soldier in Iraq was no more dangerous than bein a citizen in CA because roughly the same number of soldier met a violent death in Iraq as the number of ppl dying violently in CA during the same period.
-100 soldiers killed in Iraq
-100 violent deaths in CA
However there were 30 mil ppl in CA at the time and only 1 mil US troops in Iraq
100/1,000,000= 1 death per 10,000
100/30,000,000= 1 death per 300,000
30 times the death rate
Association vs. Causality
2 variables are associated when the values of one variable depend on or can be predicted from the values of the other variable.
just because 2 variables are associated does not mean that one causes the other. (not causal)
Types of research
emphasizes numbers, things you can cunt, and statistical analysis
permits the use of statistics to deudce the implications of theories and test hypotheses
ofte uses social surveys, experiments, and systematic observation.
emphasizes more verbal, descriptive info; downplays numbers, captures the richness of soc life
often exploratory rather than confirmatory, developing new theories rather than testing hypotheses
often uses participant observation and historical/comparative methods
Types of studies
in observational studies researchers watch subjs to see how they behave in various circumstances
2 common methods of observation are:
a formal, quantitative method of observation in which researchers typically develop a systematic set of codes, use those to code each event observed, and analyze the results
reactivity must be considered
often use audiotapes or video-cameras to document behavior
*Bales (1950) stages of group problem soling in lab
* Sykes and Brent (1983) studying police-civilian interaction in field.
researcher participates in and is directly involved in the lives of those he/she is studying
typically involves both observation and interviews with participants or informants
Examples of Participation Observation
used frequently for the interactionist perspective
*Julius Roth (1963) conducted participant observation as a patient in a tuberculosis clinic
*Erving Goffman (1967) acted as a card dealer to study casinos
*Michael Burawoy (1979) worked as a machinist in a machine shop in the 1970s
a typically descriptive account summarizing and interpreting a culture or a collection of people studied
*usually richly detailed, descriptive accounts of what went on and what the researcher experienced or observed
*often read like a novel or diary and give the reader a sense of experience the events themselves
researchers may ake diff roles in participant observation
a true insider: someone already participating in the context in a non-research role who chooses to study that setting
*Julius Roth as a tuberculosis patient
a researcher acting as an insider: pretend to be an insider
*Erving Goffman (1967) as a dealer to study gambling
*Donald Roy a machinist in a machine shop in Chicago
an outsider: someone who does not disguise their role as a researcher
*Michael Burawoy had explicit knowledge and consent of manage and told fellow machinists he was doing research
Consequences of Researcher roles
Researcher roles can pose ethical problems when subjects do not know they are being studied.
keeping the researcher role secret also constrains data collection
*Roy, for example, could never explicitly interview coworkers about their lives outside work and had no access to an company records to place his study in a broader context
on the other hand, known observers may have trouble getting cooperation from all participants.
gather info by asking people Qs
*the Qs may be objective factual info(EX: behaviors)
*or subjective info (EX: attitudes and beliefs.
EX: election polls, the census, the Gallup poll, the General Social Survey
common types of surveys
*face to face interviews
About social surveys
A respondent is someone who answers Qs in a social survey
*EX: political election polls, pub opinion polls, the General Social Survey, and the US census
*least expensive research procedures
*more representative sample
*larger sample than other research methods
*researcher has little or no control over extraneous variables
Evaluating social surveys
how were variables measured?
do respondents appear to be telling the truth?
what is the response rate?
was the sample biased?
Sex in America: The Natl Health & Soc Life Survey
first lg scale systematic social survey of sexual practices & attitudes in the US in several decades
How were variables measured?
*the Qs were developed using standard procedures in the social science and subjected to scrutiny by other researchers and pretested before being administered.
Was the sample biased?
*the sample for this survey was selected using procedures commonly used by large natl surveys. they selected a probability sample of respondents using a multi-stage sampling process, first selecting geographic areas: the cities, towns, and rural areas within those areas: then neighborhoods within the chosen cities: then addresses in the selected neighborhoods: then residences with the addresses: and finally selecting random adult members of the household. this sampling process assures a representative unbiased sample of respondents, but is very expensive.
*the resulting sample was compared to the US population on various characteristics and found to be quite similar to the entire population.
Do respondents appear to be telling the truth?
*redundant Qs were included to check the truthfulness of responses
*results were compared to those of similar high-quality rigorous scientific studies such as the General Social Survey and similar studies in other countries, and the results were found to be remarkably similar.
What is the response rate?
*nearly 80% of ppl contacted actually participated in the study (an 80% response rate), increasing our faith in the representativeness of the sample since so many people who were approached agreed to participate
All of these steps taken together help assure the validity of the results
Findings: Sex in America
The national Health and Social Life Survey found a number of results that challenge conventional wisdom about sexuality
*people are far more likely to choose sex partners similar to them in race, ethnicity, religion, age, and social class.
*marital infidelity is less common than thought
*AIDS is much more common for some segments of the population
Can research be value free?
Sex in Amr
*conceived in 1987 to gather data t help ctrl AIDS
*identifying grps at risk due to sexual behaviors, could help target sub populations
*only useful if it asked explicit Qs
supported by scientists and administrators at federal agencies incl
The politics of Social research
pressure t narrow the survey, and quit asking Qs once it was est the couple was monogamous
Senator Jesse helms introduced an amendment to a bill funding the National Institutes of Health specifically prohibiting the Fed gov from paying for such a study. it passed by a vote of 66 to 34 and federal funding became impossible for the study.
Private foundations funded the study but with a much smaller sample making it hard to say much about subcategories.