Created: 2026-01-07 07:56:45 - Touched: 2026-06-13 09:46:58 - Status: Stable

Mongeon, P. (2025, January 3). Research methods. https://pmongeon.github.io/info5520/

Typical class is 75 minutes of lecture and 75 minutes of group work on assignments.

Methods broadly split between quantitative and qualitative. Common qualitative approaches:

Techniques:

Common quantitative approaches:

Also, mixed methods.

Lecture 2026-01-07, chapters 1 and 2

Research Process/Stages

Fortin & Gagnon (2016) describe a 5-phase research process:

Terminology

Ontology: what exists, what can be known Epistemology: how do we know it Axiology: values and beliefs around knowledge

Positivism: Reality is objective and can be known with certainty, knowledge is logic applied to observation, quantitative methodology, neutral. Postpositivism: Reality can only be partially known, knowledge is accumulation of evidence that justifies a "true" belief, both quant and qual, values are inevitable but should be mitigated. Interpretivism: Reality is subjective, knowledge is interpretation of reality, qualitative, values are integral to knowledge and should focus on rigour instead of elimination of bias.

Positionality: Me and my values and beliefs within the context of the research I'm doing and how they can influence things Reflexivity: self-examination (how is this different from criticality?)

Methodology: high level description of the approach Method: specific step-by-step Instruments: survey questions, interview guide, coding guide

Literature review

Pick an initial idea to explore, ask an initial question, review existing literature to see if it's answered/how thoroughly it's answered, identify a gap, repeat. Inspiration comes from your own experiences, past research, social problems, or given to you by your boss. Examine a problem from a new (disciplinary) perspective, identify an affected population, try to identify core concepts to explore.

Fortin and Gagnon 2016, 69–70 question vs research type. Descriptive (what is, explore and classify, not fully understood), explanatory (identify process, factors, influences, understand phenomena and identify relationships, correlational studies), predictive (what effects, differences between groups, how effective, causality, requires solid existing research, experimental studies).

Good questions should be relevant (of interest to audience), significant (advances understanding), feasible (can be done), related to theory (I argue this one is less important).

Literature reviews establish state of knowledge, set boundaries, identify key concepts and methods, build on existing research, position this research in context, give credibility (don't roll your eyes, people are human), identify areas of consensus and disagreement, identify gaps. Empirical, theoretical, primary vs secondary sources (meta-analyses).

Define your search strategy (including terms) and databases/collections to search, perform the search, find the relevant results in the search results, critically examine them, and synthesize the results to accomplish the goals in the prior paragraph. LitReview spreadsheet: title, year, source (which journal), abstract, research objectives.

Source evaluation: see Literature reviews criteria, and possibly Collection development process.

When you finish the first round of literature review: identify gaps, use citation chaining, look for other works from prominent authors, refine search strategy for next round.

Lecture 2026-01-14, chapter 3

Theoretical frameworks, research problem, purpose, and questions.

Intro, intro but in more depth, empirical context, theoretical context (why this framework is a good fit, explain it).

Objective:

Statement that indicates the key concepts, the target population, and the action verb (e.g., explore, describe, verify, predict) that matches the type of study. (see week 2 slides for more details and examples).

Questions: list of questions you need to answer in order to fulfill the objective. Answers will allow for declarative statements of knowledge in the conclusion.

Hypotheses can be non-directional/directional, associative/causal, existence/absence (null). State as future positive, "there will exist a causal relationship between X and Y." Derived from the theoretical framework (how you think something works).

Empirical context: summarize what's known but it's not a literature review, just a summary of the results.

Theoretical context: list possible framework choices and why your choice is the most appropriate.

Lecture 2026-01-21, methodology

Research design. Qualitative, quantitative, and mixed methods.

Comparisons, interventions, location/context, collection methods and instruments, chronology, communication with participants, biases, exogenous and endogenous factors.

Big list of research types, goals, and methodologies that fit them. I guess my 6680 study is phenomenology or case study.

Lecture 2026-01-28, sampling, operationalizing and measuring

Population: the group of people you're studying. Accessible vs representative.

Probability based sampling methods

Simple random sampling: pick people from the population at random.

Systematic random sampling: order the population, pick a random starting position, use a fixed interval to achieve desired sample size.

Stratified random sampling: stratify population (homogenous grouping). Select samplesize/number of groups from each group to get a non-proportional sampling, or adjust the sample size of each group based on the group's proportion of the total population for a proportional sample.

Random cluster sampling: create groups (eg by geography), assign number to each group, generate a random set of numbers within the range of your list of groups. Repeat those 3 steps for each level of grouping if your grouping strategy is hierarchical. Can apply above sampling strategies at each level of the hierarchy.

Non-probability based sampling methods

Convenience sample: whatever happens to be available.

Purposive sampling: cherry pick your sample, eg extreme cases, minimum variation, maximum variation.

Quota sampling: like probability based sampling but without the random.

Snowball sampling: participants refer future participants.

Sample size

Rules of thumb: 10% of accessible population or 25-30 participants in stratified or quota samples. Sample size is driven by type of study, desired statistical significance for the full population, number of studied variables, amount of diversity in the population, and how much each variable is expected to vary.

There are checklists you can use (like Fortin and Gagnon 2016) to evaluate a sampling criteria.

Fortin, Marie-Fabienne, and Johanne Gagnon. 2016. Fondements Et Étapes Du Processus de Recherche: Méthodes Quantitatives Et Qualitatives. 3e ed. Montréal, QC: Chenelière Éducation.

Operationalizing

Indicators: proxies that allow us to measure things that aren't directly measurable. For instance, measuring research impact via citation counts.

Indices: composite measures of multiple indicators. For instance, health as a composite of energy level and ability to perform everyday activities.

Measuring

Actual observations. Different scales available:

Negative/positive observation vs negative/positive reality => true/false negative/positive. Sensitivity and specificity?

False positive: type 1 error False negative: type 2 error (I hate these, it's like diabetes)

https://www.geeksforgeeks.org/machine-learning/false-positives-and-false-negatives/

Precision: how many observed positive samples are actually positive (important when false positives are costly)

Sensitivity: how many of the actually positive were observed positive (important when missed positives are costly)

F1-score: 2*(precision*sensitivity)/(precision+sensitivity) (useful when positive vs negative rate are highly disproportional)

Random errors are noise in the signal/observations that happen by chance, affecting precision. Increase number of observations so that noise averages out. Systematic errors affect accuracy, not random, more data won't help.

Lecture 2026-02-04 Quantitative research

Quantitative is about numbers, counting things, statistical analysis. Quantitative is mostly putting the headers into your spreadsheet and then filling in the rows using the measuring options above. Data usually comes from surveys, structured observation, or secondary data use.

Surveys are easy and cheap, can be easily anonymized and standardized. But can be hard to get depth and nuance, account for bias, may be misinterpreted, low participation. Questions can be leading, ambiguous, biased, too many options, lack of neutral options, overlapping options. Too long, inconsistent, poorly structured, doesn't address all research questions. No follow up. Demographic questions to help identify bias in respondent population.

Structured observation starts with a coding criteria and then making observations. Kind of like filling out your own survey on behalf of what you're observing. "Objective," contextual, real-time. Time-intensive, limited, observer effect.

Secondary data can be qualitative or quantitative, what makes this quantitative is what you do with it. Easy way to get longitudinal data without having to wait a long time. Issues of quality, relevance, formatting.

Lecture 2026-02-11 Qualitative research

N/A

Lecture 2026-02-25 Qualitative analysis

Code books.

Create preliminary codebook, segment the data (split it into codable pieces), first cycle of coding includes updating/expanding/finalizing code book, second cycle of coding, inter/intra coder reliability testing (do people code the same passages the same way?), repeat this over and over until there's good reliability.

Reliability has some options: Cohen's Kappa (stats for determining how consistent different coders are). Krippendorff's Alpha useful for likert scale/numeric coding/degrees or magnitudes (nominal/unordered, ordinal/ordered, interval/integers, ratio). Percentage agreement (tolerance can allow for some number of dissent, just pick a value).

Miles, Huberman, and Saldaña (2020) suggest a list of tactics for generating meaning after coding is completed:

What's LibraQDA? "Using this tool, you can upload a collection of documents, create a hierarchy of codes, and annotate portions of documents with codes and notes that you can recall and organize later."

Lecture 2026-03-04 Quantitative analysis

This is a very big chapter, and yet I don't have any notes about it. It all seems pretty obvious.

Lecture 2026-03-11 Knowledge mobilization

Plan: outputs (what you're producing, eg a paper, blog posts, news letter), events (discussions, go on tv), audiences (which people need to hear about it), needs/interests (why will they care), channels (how are you going to reach those audiences), timeline, budget, evaluation (reach, usefulness, influence). Explain why. How is this going to change the world, why is it important? Honestly a journal article is not likely the most important way to share results. Describe the potential benefits and outcomes (e.g., evolution, effects, potential learning and implications) that could emerge from the proposed project as a result of knowledge mobilization activities.

Good thing to include in your Research Proposal

Advance knowledge in information science.

15 minute time limit.

Pages that link here: