February 14 Methodological Comparison Quantitative- Quantity: amount of something~Objective~Deductive~Applied~Quicker~Positivism~Questionnaires, secondary data~Dominant Qualitative- Quality: nature of something~Subjective~Inductive~Pure~Longer~Symbolic interactionism ~Ethnography, interviews~Waning CausalityDefinition: the relation between two events or states of affairs in which one brings about or produce the other - Steps to establishing causality o Concomitant variation: covariance or arelationship must exito Temporal sequencing: cause must precede the effecto Elimination of rival factors: rival casual factors must be excluded Four basic models of causalitySingle sequential linear causal chain o A->B->C->DMultiple causalityo A+B+C+DInteractive Modelo ABCD->EFebruary 19Validity and reliability Three primary purposes to research- Exploration (rare)- looking at a new problem or un-researched area - Description (most common)- discovering the state of affairs in some area- Explanation(hardest)- why something occurs. e.g., crime, unemploymentTwo critical issues that must be resolved withany research: - Validityo Accuracy- are you measuring what youwant to measure?- Reliabilityo Repeatability- consistency of a measuring instrument *You can be reliable but you’re not valid. To be valid you have to be reliable Two major types of validity- Internal- what other variables may have accounted for the change?- External- how generalizable your findings areSeven threats to internal validity- History no way to fix it- Maturation- respondents change over the course of a study - Testing- where the respondents know what you’re studying so they change theiranswer to make you mad ask several off-topic questions and throw in the researchquestion to throw them off- Instrumentation- changes in the measuring instrument between applications - Statistical regression- moving closer to the average- Selection bias- nonequivalent groups chosen for comparison. Randomization- Experimental mortality- subjects drop outthe study before it concludes. Over sampleFebruary 21External validity threats- Testing- where the respondents know what you’re studying so they change theiranswer to make you mad- Instrumentation- Selection bias- nonequivalent groups chosen for comparison.- Reactivity- people know they are being study and they are going to behave differently - Multiple treatment interference- Other casual factors that may influence findingso Halo effect- rating a group higher because you know them o Placebo effect- substance that you give somebody experimental drug andthey react different because they think that they got the real thingo Criterion problems- a -poor measure of success or failureSampling- census- when every element of a population is surveyed/ studied- population- the group of the persons or element t0 be studied - sample (of a population)- selected elements from a population that will be observed in order to learn something about the entire populationo you study the sample, and make generalizations from the sample to the entire population - Sampling error- the amount of difference between the characteristics of the population and the sample chosen to studyo Any sample will have some degree of error. The key is to:o Minimize the error o Be able to identify the amount of errorso you can calculate the effect of the error on the findings statistically - Sampling frame- a list of all elements in the population to be studied - Sampling with/without replacement- oddsstay the same because you put the card back. Odds constantly changing without replacement - Two types of sampling techniqueso Probability o Nonprobability sampling Probability sampling- The most powerful form of sampling techniques - Allows you to determine the probability that any given element in the population will be included in the sample - If you can determine this (the probability that..) then you can use inferential statistics (being able to make projects into the future)Simple Random Sampling- The first basic type of probability sample - Each element has an equal chance of being selected - the problem is that you need to know what the sampling frame is- which often you d0 not- steps:1. draw up list of all those in population (sampling frame)2. number all those in population 3. select a number from random number chart for place to start4. move horizontal or vertical on randomnumber chartStratified random sample- use this when you want to make sure the sample representative of the population according to the distribution of an independent variable (e.g., political party influences policy choice)- divide into strata (categories)- randomly sample sub-populationCluster sampling- divide population into clusters based on some characteristics unrelated to theory (e.g. city)- assign each cluster a number- randomly draw number of clusterso often used in public opinion polls Proportionate sample- want same percentages in sample that are in the study population- 1,000 voters: X50% Y40% Z10%-50% libertarian-40% republican-10% Democrat Final probability sample- Multi-stage cluster samplingDifferences between stratified and clustersampling- With stratified sampling, you use independent variable to create strata- Must know something about elements of population to know how to divide for stratified not for cluster- In cluster, you are sampling groups not individuals - In stratified, you have an unequal numberof elements in each strata, whereas in cluster, you want approximately an equal number Comparison chart. Stratified cluster Independent V yes noPopulation yes noSample individuals groupsElements unequal equal #February 25Non-probability sampling - why would we use this type?o Less moneyo Conveniento Type of research - Disadvantageo Cannot estimate amount of sampling error Five types of non-probability sampling- Accidental sample (convenience)-“Person” on the street interviews - Quota sample-Certain %’s of specific groups - Typical-case or purposeful sample - Systematic sample-every nth person -
View Full Document