Haphazard (convenience/accidental) sampling - Get any cases in any manner that is convenient
Quota sampling - Get a preset number of cases in several predetermined categories that reflect the diversity of the population.
Purposive sampling - Get all cases that fit a criteria
Snowball sampling - Get cases using referrals. Creates networks that are represented by sociograms
Sequentialsampling - Get cases until there is no new information or diversity from the cases. Similar to purposive sampling.
Probability Sampling - Type of sampling that gives every subject in thepopulation the same likelihood to be selected
Sampling element: Unit of analysis or - case in a population
Population/universe: the large pool from which a case is drawn.
Samplingratio = sample size / population size x 100
Parameter: Characteristic of a population estimated from a sample(e.g. average age of students with GPA 4.0)
Statistic: Characteristic of a sample that is used to estimate the population parameter
Random sample: each case in the population has an equal chance tobe selected
Sampling error: deviation of the sample from being representative ofthe population
Margin of error: An estimate of the amount of sampling error thatexists in a survey result.
Systematic sampling - A type of random sample in which a researcher selects every k th (e.g. 12 th) case in the sampling frame using a sampling interval.
Samplinginterval -Sampling interval = inverse of sampling ratio Population = 900.Sample= 300 Sampling ratio = 300/900• Sampling interval = 900/300
Cluster samplingA type of random sample that uses multiple stages and is often used to cover wide geographic areas in which aggregated units (clusters) are randomly selected. Samples are then drawn from the sampled aggregated units
RandomDigit Dialing (RDD) - A way of sampling used for phone interviews where the sampling frame is all phone number
Inferential statistics Branch of statistics based on a random sample that permits to draw conclusions with confidence from the sample to the population
Principles of good question writing -
Avoid jargon, slang and abbreviations.
Avoid ambiguity, confusion, and vagueness
Avoid emotional language
Avoid prestige bias
Avoid double-barrelled questions (two questions in one)
Avoid asking about intentions in the distant future
Avoid double negatives
Avoid overlapping or unbalanced response categories
Good question writing
Threatening questions
Socially desirable questions
Knowledge questions
Skip/screen or contingency questions
Questionnaire design issues
Length of survey or questionnaire
Question order or sequence
Format and Layout
Probing -A follow up question or action in survey research used by aninterviewer to have a respondent clarify or elaborate on an incomplete or inappropriate answer
Ethical issues related to surveys• Keep privacy and confidentiality• Must get voluntary participation of respondents• Avoid pseudosurveys (made to mislead others and use survey formatto persuade others to do something, e.g. “suppression polls” or “pushpolls”• Misuse of survey results or purposely rigged surveys
Secondary data analysis Advantages:• Focus only on analysis rather than collection of data• Inexpensive• Permits comparisons across countries and longitudinal analysis• Answers new issues not thought of at the original time of datacollection• Facilitates replication
Secondary data analysis - The researcher accesses data that isalready available in reports, books, libraries, digital archives and reassembles the information in new ways to answer acertain research question
The fallacy of misplaced correctness: When a person uses too manydigits while quoting statistics reports in an attempt to create theimpression that the data are accurate or the researcher is highlycapable. (e.g. population of Canada= 38,654,738, better say 38.7million)
Validity - When the secondary data is a proxy for whatthe researcher is interested in (e.g. the researcher is interested in the extent of robbery and uses police records as a proxy, problem:some robbery cases are not reported)
Reliability - Changes in definition of terms over time and changes in methods ofdata collection affect reliability (e.g. police using computerizedmethods will increase crime records, not because crime actuallyincreased).
Codebook: A document that describes the procedure for codingvariables and their location in a format for computers
Code sheet: A printed grid on which the researcherrecords the information
Directentry: Enter the information as the respondentanswers (through CATI, CAPI, online survey)
Types of data cleaning:Possible code cleaning (wild code checking) e.g. an error if someone’s height = 0 or age=30 02- Contingency cleaning (consistency checking) e.g. Look at combination of impossible variables (e.g. gender= male,breast cancer= 1)
Descriptive statistics: A type of simple statistics used to describe patterns in the data (e.g. age distribution, average age, etc. )
Univariate analysis: one variable (e.g. age distribution)
Bivariate analysis: two variables (e.g. relationship between income and years of experience)
Multivariateanalysis: three or more variables (e.g. how do years ofexperience and education affect income)
Mode: most common or frequently occurring 6, 5, 7, 20, 9, 5There can be more than one mode 5,6,1,2,5,7,4,7 (bimodal if two modes)Any distribution with more than one mode is multimodal