Summary of Research Methods and Statistical Concepts
Research Methods and Statistical Concepts: Your Comprehensive Guide
Introduction
Research Methods and Statistics provide the tools researchers use to design studies, collect and analyze data, and draw valid conclusions. This material covers sampling techniques, classifications of research, links between theory and evidence, innovation and R&D roles, and core statistical concepts and calculations using a worked dataset. The aim is to make complex ideas digestible and directly applicable to university-level research projects.
Definition: Research methods are systematic approaches used to answer questions, while statistics are mathematical tools used to summarize, interpret, and draw inferences from data.
Section 1: Sampling Techniques
Sampling selects a subset of a population for study. Good sampling reduces bias and increases the validity of inferences.
Stratified sampling vs Quota sampling
Definition: Stratified sampling: the population is divided into homogeneous subgroups (strata) and random samples are drawn from each stratum. Quota sampling: interviewer selects respondents to meet predefined quotas for subgroups, without random selection.
- Purpose
- Stratified: ensure representation and enable precise estimates within strata.
- Quota: ensure sample composition matches population proportions quickly and cheaply.
- Selection method
- Stratified: random selection within each stratum (probability-based).
- Quota: non-random selection until quotas filled (non-probability).
- Bias and generalizability
- Stratified: lower selection bias, supports probability inference to population.
- Quota: higher selection bias; cannot confidently compute sampling error.
- When to use
- Stratified: when you need precise subgroup estimates and can sample randomly.
- Quota: when resources are limited and approximate subgroup representation suffices.
| Feature | Stratified sampling | Quota sampling |
|---|---|---|
| Selection basis | Random within strata | Non-random to meet quotas |
| Probability sampling? | Yes | No |
| Estimation of sampling error | Possible | Not straightforward |
| Typical use | Academic surveys, elections | Market research, quick polls |
Stratified sampling vs Cluster sampling
Definition: Cluster sampling: the population is divided into clusters (usually heterogeneous mini-populations); entire clusters or random clusters with elements inside sampled.
- Structure
- Stratified: strata are homogeneous internally and different across strata.
- Cluster: clusters are mini-representations of population and ideally heterogeneous internally.
- Cost and logistics
- Stratified: can be costlier when strata are widely dispersed.
- Cluster: often cheaper and practical for geographically spread populations.
- Precision
- Stratified: tends to increase precision when strata chosen well.
- Cluster: can reduce precision (higher sampling error) because units within clusters are similar.
- When to use
- Stratified: want accurate subgroup estimates.
- Cluster: large-scale surveys where travel/time costs are important.
| Feature | Stratified sampling | Cluster sampling |
|---|---|---|
| Internal homogeneity | High within strata | Usually low within cluster (clusters heterogeneous) |
| Cost | Moderate to high | Often lower |
| Sampling error | Lower if strata well chosen | Higher unless many clusters sampled |
| Typical use | Social science surveys with known strata | National censuses, school-based surveys |
Section 2: Classifying Research
Clear classification by purpose, method, or design guides choices that affect validity and reliability.
Definition: Research classification organizes studies by their objective (exploratory, descriptive, explanatory), method (qualitative, quantitative, mixed), or design (experimental, correlational, case study).
Why classifi
Sign up for the full summaryAlready have an account? Sign in
Research Methods & Statistics
Klíčové pojmy: Stratified sampling uses random selection within homogeneous strata for precise subgroup estimates, Quota sampling is non-probability, meeting subgroup quotas without random selection, Cluster sampling groups by clusters and is cost-effective but can increase sampling error, Classifying research by purpose/method/design guides valid methodological choices, Quantitative research often tests theory deductively; qualitative research can generate theory inductively, R&D and innovation create competitive advantage when linked to strategy and commercialization, Random assignment enhances internal validity; independent variable is the manipulated predictor, Inferential statistics generalize from sample to population given assumptions, Negatively skewed distributions have a long left tail with mean < median < mode, Standard deviation quantifies average dispersion around the mean, Uniform frequency distributions produce flat histograms indicating even spread across intervals, Compute mean by summing observations and dividing by $n$, median as middle order statistic