Research Questions and Proposals

Research Questions and Proposals

• A research question is what a researcher aims to answer through their study
• Good research questions should be specific, researchable, informed by literature, and value-neutral
• Research questions can be descriptive, exploratory, explanatory, about the meaning of social phenomena, for theory testing/refinement, comparative, or for the assessment of interventions

Types of Research Proposals

• Quantitative research proposals involve deductive reasoning, hypothesis testing, survey/experimental design, and examine relationships/strength between variables
• Qualitative research proposals often involve inductive reasoning, explore/meaning-based questions, use interviews/focus groups/ ethnographic observation/multimodal analysis
• Concepts are more general at the proposal stage and research questions are refined as analysis proceeds

Parts of a Research Proposal

• Key components include the title, abstract, introduction, literature review, research question/aims/objectives, theoretical framework, methods, data analysis plan, ethical considerations, dissemination/policy relevance, references/appendices, resources, and a schedule/timetable

Quasi-Experimental Designs

• Quasi-experimental designs use statistical operations at the analysis stage to control for spurious variables
• Used to analyze survey data and in situations where strict experimental designs are impractical or unethical
• Evaluation research is a form of research used to assess the effectiveness/value of social interventions or programs
• Evaluation research can be quantitative (trial), but is usually qualitative (investigators) using qualitative or mixed methods, and investigators remain outside intervention development/implementation
• Process evaluation focuses on how a policy/intervention unfolds
• Formative/action research involves real-time information about a policy/intervention as it unfolds, in order to provide feedback that can be used to modify it over time

Natural Experiments

• A natural experiment is conducted without creating an artificial environment or manipulating variables, and are rare and opportunistic

Field Experiments

• Field experiments take place in a natural setting (not a lab), based on an occurrence that has already happened
• The researcher randomly assigns subjects to experimental/control groups, then exposes the experimental group to an intervention in a real-world setting
• Groups may be aware of manipulation
• Porousness among groups may occur

Case-Control Studies

• Case-control studies explore multiple exposures
• Researchers select a group with an event/characteristic of interest (cases) and a similar group without it (controls)
• Comparison of past exposures between cases/controls indicates an association
• Useful for rare outcomes
• Limitations include recall bias, subjective assessment of exposure, and being unable to measure incidence.

Randomized Controlled Trials (RCTs)

• Randomized Controlled Trials (RCTs) are effective in ruling out spurious/false causation
• Participants are randomly assigned to receive an experimental treatment, a control, or another treatment.
• Experimental groups and control groups are randomly allocated
• Experimental group receives exposure/treatment and the control group does not
• Single blinding is restricting knowledge about which treatment arm participants are receiving after allocation
• Double blinding is when both participants and the intervention providers are unaware of treatment allocations
• Causality occurs when the treatment (independent variable) causes a statistically significant change in the outcome (dependent variable) in the experimental group compared to the control group
• Ethical considerations must be considered
• Issues of validity and generalizability should be addressed
• Difficulty in blinding and reduced social complexity/artificiality can be limitations
• RCTs may not be applicable to sociological questions
• The Hawthorne effect can occur

Sampling: Probability and Non-Probability

• Probability sampling gives everyone an equal chance of being selected, and non-probability sampling does not
• Representative sampling aims to provide a good estimate about the whole population and focuses on statistical generalization
• Purposive sampling allows researchers to selectively choose participants based on research objectives

Historical Context of Probability Sampling

• Probability sampling's development is credited to Arthur Bowley in the early 20th century., which is an efficient alternative
• Bowley introduced sampling theory, which allows researchers to select a representative subset rather than conduct full censuses

Sampling Frame

• The sampling frame is a list of people who could participate in the study
• Examples include the Postcode Address File, phone directories, electoral register, schools, and organizations

Types of Probability Sampling

• Three main types of probability sampling: simple random, stratified, and cluster.

Simple Random Sampling

• A simple random sample can be obtained by using the lottery method, a table of random numbers, or a systematic approach
• The sampling fraction is calculated, and then every nth person from the list selected
• The higher the sample size, the lower the margin of error

Stratified Sampling

• Individuals are first divided into groups (strata), and then a random sample is selected from each stratum
• Stratification improves accuracy and representativeness
• Disproportionate stratification samples different subgroups at different rates than their actual size, in order to over-sample smaller or underrepresented groups

Cluster Sampling

• Cluster sampling is useful when a population is widely dispersed, as it economizes on time and travel
• Cluster sampling is a probability sampling strategy involving successive sampling of units/clusters
• Multi-stage cluster sampling involves more than one stage

Response Rate

• The response rate is the number of people participating in a research study divided by the total number asked to participate
• The higher the response rate, the more representative the sample is likely to be
• Factors like refusal to participate and non-contact can lower response rates

Non-Probability Sampling

• Non-probability techniques are often used to access groups whose activities are normally 'hidden' from public or official view
• Volunteer samples involves a magazine or advertisement where interested readers a asked to respond
• Snowball sampling involves obtaining respondents through referrals of people who share the same characteristics

Maximum Variation Sample

• Maximum variation sampling is a purposive sampling technique used in qualitative research to capture a wide range of perspectives/experiences/characteristics
• Researchers seek participants who represent the full spectrum of variation in the population instead of selecting a typical group

Purposive Sampling

• Purposive sampling is a non-probability sampling strategy where the researcher selects participants who are considered typical of the population or have significant relation to the research topic.
• Purposive sampling is used by qualitative researchers to explore research questions and seeks depth and targeted understanding

Theoretical vs Statistical Generalization

• Theoretical generalization applies findings to broader theoretical concepts
• Statistical generalization applies findings from a sample to a larger population using probability-based sampling methods

Quota Sampling

• Quota sampling is a purposive sampling strategy where the researcher identifies the proportions in various strata and ensures they are proportionately represented in order to increase representativeness

Purposive Sample Sizes

• Sample size is determined by reaching a point where new insights are no longer emerging
• Sample size depends on the study's goals and the depth of information needed

Analyzing a Single Variable

• A frequency distribution involves counting how often each value of a single variable occurs
• Can be presented by raw count, percentages, or a pie chart

Measures of Central Tendencies

• Mean is the average value of a variable
• Median is the middle value when data is arranged in order from smallest to largest
• When there is an odd number, it is the middle value and with an even number, it is the average of the two middle numbers

Mode

• Mode is the most frequently occurring variable
• It is the value that occurs most frequently in a distribution of measurements
• The distance from the mode can indicate deviance or resistance
• Change in the mode can signal a shift of social norms
• Multiple modes indicate social polarization, subcultures, or class stratification

Children of the Prison Boom

• For a factor (X) to influence structural inequality, it has to be common, unequally distributed among segments of the population, and result in negative outcomes
• The 2008 economic crisis was common, with millions of Americans losing jobs.
• Black and Latino families were disproportionately affected by discriminatory lending practices
• The negative consequence was that the Black-white wealth disparity in median household income increased