1. Ethical Design for individual raw data or the 4.

Organizing a Thesis
Vulnerable Populations names of participants. Proposal vs. Full Thesis
Anonymity vs.
Principles: Beneficence, Confidentiality: * A thesis must follow a logical
Justice, and Respect for Anonymity: Best if possible flow to prove the research is
Persons. (e.g., no names or ID feasible (proposal) and valid
Respect for Persons: The numbers collected at all). (full thesis).
researcher must ensure Confidentiality: If names Front Matter: Title, Abstract,
informed consent is truly are needed for tracking, use Table of Contents.
voluntary and that pseudonyms and store the Chapter 1: Introduction:
participants understand they "key" (the link between Problem statement, research
can withdraw at any time names and codes) in a questions, and significance.
without penalty. For separate, encrypted file. Chapter 2: Literature
vulnerable groups, use Reporting: Present findings Review: Current state of
simplified language and in aggregate form (e.g., knowledge and the "gap"
ensure no coercion is "60% of staff felt...") rather your research fills.
present. than using direct quotes that Chapter 3: Methodology:
Beneficence: This is the "do might reveal a specific Design, participants,
no harm" principle. Since person's identity based on instruments, and data
participants are at risk of their unique way of speaking analysis plan. (In a proposal,
distress, the researcher must or specific job role. this is the main focus).
have a distress protocol in 3. Data Integrity in Chapter 4 & 5 (Full Thesis
place (e.g., providing contact Longitudinal Studies only): Results (presentation
info for mental health of data) and
services). The benefits of the Inconsistencies are common Discussion/Conclusion
research must outweigh the in long-term data; the ethical (interpretation and
potential psychological risks. path is transparency, not implications).
Justice: Ensure that the deletion. 5. Integrating Mixed
vulnerable population isn't Data Cleaning & Methods (Survey +
being targeted simply Documentation: Maintain a Interviews)
because they are "easy to codebook or a "change log"
recruit," but because the that records every edit made Integration is what turns two
research specifically to the dataset. Never delete separate studies into one
benefits them. The burdens the original raw data file. cohesive mixed-methods
and benefits of the research Handling Inconsistencies: study.
should be distributed fairly. If data is missing or Triangulation: Compare
2. Workplace Satisfaction: contradictory, use survey results with interview
Consent & Anonymity established statistical themes. Do they agree?
methods (like imputation) (e.g., If 80% of survey takers
When employees fear and disclose these methods say they are happy, do the
repercussions, the in the final report. interviewees explain why?)
researcher must build a Storage: Use secure, Sequential Explanation:
"firewall" between the data password-protected servers Use the quantitative survey
and management. with version control to to find broad trends, then
Informed Consent: prevent accidental data loss use the qualitative interviews
Explicitly state in the consent or unauthorized access. to "deep dive" into outliers or
form that management will specific patterns found in
never have access to the numbers.
Joint Display: Use a table to text that represent a single Themes: Refine the
show a quantitative result in idea, such as "lack of specifics of each theme and
one column and a matching resources" or "positive the overall story the analysis
qualitative quote in the next. student engagement". tells.
6. Challenges and Benefits Axial Coding Producing the Report:
of Mixed Methods (Categorization): Group the Select vivid extract examples
initial codes into broader and relate the analysis back
Challenges: categories by identifying to the research question and
Time & Resources: It takes relationships between them. literature.
double the effort to collect For example, codes like 9. Content Analysis vs.
and analyze two types of "software bugs" and "slow Thematic Analysis (Social
data. internet" could be grouped Media)
Skill Set: The team must be under the category
proficient in both statistical "Technical Barriers". In the context of public
software (like SPSS) and Thematic Development: sentiment on government
thematic coding (like NVivo). Refine these categories into policy:
Integration Conflict: What overarching themes that Content Analysis Process:
happens if the survey says answer the research The researcher defines a
one thing and the interviews question regarding teachers' unit of analysis (e.g., a
say the opposite? (The experiences. tweet), creates a coding
"divergence" problem). Rigorous Analysis: Use a scheme (e.g., positive,
Benefits (Justification): "constant comparative" negative, neutral), and
Completeness: Numbers method, checking new quantifies the frequency of
provide the "what," while codes against existing ones these codes to find
stories provide the "why." to ensure consistency statistical trends.
Offsetting Weaknesses: throughout all transcripts. Interpretation: While it can
Quantitative data lacks 8. Step-by-Step Thematic be qualitative, content
depth; qualitative data lacks Analysis (Patient analysis often aims to
generalizability. Together, Experiences) provide a descriptive
they provide a more robust account of the data's
conclusion for community Thematic analysis is a characteristics.
health interventions. systematic six-step Key Differences:
7. Qualitative Coding and approach often defined by Focus: Content analysis
Categorization (Teachers' Braun and Clarke: focuses on the frequency of
Curriculum) Familiarization: Immerse in specific words or concepts;
the data by transcribing and thematic analysis focuses on
To identify patterns in reading. identifying underlying ideas
interview transcripts, a Generating Initial Codes: or patterns of meaning.
researcher should follow a Code the entire data set Outcomes: Content analysis
systematic coding process: systematically. often results in counts and
Familiarization: Read Searching for Themes: percentages; thematic
through all 15 transcripts Collate codes into potential analysis results in a rich,
multiple times to get a deep themes. descriptive narrative of
sense of the overall narrative Reviewing Themes: Check themes.
before any formal coding if the themes work in relation
begins. to the coded extracts and
Open Coding: Assign initial the entire data set.
labels or tags to segments of Defining and Naming
10. Descriptive Statistics Alternative Hypothesis 13. Interview Types
(Exam Scores) ($H_a$): The new teaching (Remote Workers)
method significantly
With 100 students, improves performance. The choice of interview
descriptive statistics Analysis: The researcher depends on the research
summarize the "typical" might use an Independent goal:
score and the "spread" of Samples T-Test to compare Intervi Advan Limita Impac
performance: the mean scores of two ew tage tion t on
Measures of Central groups. Type Data
Tendency: Drawing Conclusions: If the Struct High Low Standa
Mean: The average score, p-value is less than the ured reliabili depth; rdized
providing a general sense of significance level (usually ty; cannot but
class performance. 0.05), the researcher rejects easy follow rigid.
Median: The middle score, the null hypothesis and to up on
useful if there are extreme concludes the new method compa interes
outliers (very high or low is likely effective for the re ting
scores). broader population. across points.
Mode: The most frequent 12. Statistical Software: partici
score, showing where the SPSS and R (Survey Data) pants.
majority of students are Semi- Balanc Harder High
clustered. For 300 survey participants, struct e of to depth
Measures of Dispersion: software simplifies complex ured consist analyz and
Standard Deviation: Shows calculations: ency e than reliabili
how much scores vary from Functionalities: and struct ty.
the mean. A low SD means Data Cleaning: Handling flexibili ured;
students performed similarly; missing values and recoding ty; time-c
a high SD indicates a wide variables. allows onsum
gap in achievement. Descriptive Analysis: for ing.
Range: The difference Generating frequency tables, deep
between the highest and means, and standard probin
lowest scores. deviations. g.
Interpretation: These Inferential Analysis: Unstr Maxim Low Explor
measures help identify if the Running regressions, ucture um reliabili atory
exam was too hard, too easy, ANOVAs, or Chi-square tests d depth; ty; and
or if the student body has a to find relationships between partici very rich.
diverse range of abilities. variables. pant difficul
11. Inferential Statistics and Visualization: Creating leads t to
Hypothesis Testing histograms, scatter plots, the replica
(Teaching Methods) and box plots to see data narrati te or
trends. ve. compa
To determine if a new SPSS vs. R: SPSS is re.
method is better than a user-friendly with a 14. Focus Groups
traditional one: "point-and-click" interface; R (Customer Perceptions)
Formulating Hypotheses: is a coding-based language
Null Hypothesis ($H_0$): that offers more flexibility Focus groups leverage the
There is no significant and advanced visualization interaction between
difference in performance through packages. participants to uncover
between the two methods. shared meanings:
Collective Opinions: as a "fly on the wall" to 17. Reliability and Validity in
Participants can agree, record behaviors without Surveys
disagree, or build on each intervening.
other's ideas, revealing the Implications: Reliability (Consistency):
"socially constructed" view Ethical: Participant Use Cronbach’s Alpha to
of the product. observation risks check if multiple questions
Group Dynamics: The "deception" if students don't about "satisfaction" yield
researcher can observe know the TA is a researcher; similar scores. Use the
body language and power non-participant observation test-retest method to see if
dynamics (e.g., one must manage the results are stable over time.
dominant voice vs. quiet Hawthorne Effect (students Validity (Accuracy): Ensure
participants). behaving differently because content validity (do
Methodological they know they are being questions cover all aspects
Considerations: watched). of satisfaction?) and
Moderation: The facilitator Practical: Participant criterion validity (do survey
must ensure everyone observation is scores correlate with actual
speaks and the discussion time-consuming and hard to turnover rates?).
stays on track. document in real-time; Challenges: Social
Composition: Groups non-participant is easier to desirability bias (employees
should be diverse enough to record systematically. answering what they think
cover different views but Analytical: Participant data management wants to hear).
homogenous enough for is subjective and rich; 18. Probability vs.
people to feel comfortable non-participant data is more Non-Probability Sampling
talking. objective but may miss the
Challenges: "Groupthink" "why" behind behaviors. Probability Sampling (e.g.,
(where people agree just to 16. Survey Design for Simple Random): Every
fit in) and the difficulty of Employee Satisfaction student has an equal chance
transcribing overlapping of being picked.
voices. Question Types: Use a mix Impact: High
15. Participant vs. of Likert scales (1–5) for generalizability; the results
Non-Participant quantitative analysis and can be statistically applied to
Observation open-ended questions for the whole university.
qualitative insights. Non-Probability Sampling
In a classroom setting to Clarity and Relevance: (e.g., Convenience):
understand student Conduct a pilot test with a Picking students in the
engagement: small group of employees to library or cafeteria.
Implementation: identify confusing jargon. Impact: Low generalizability
Participant Observation: Ensure questions are but easier and cheaper to
The researcher joins the "single-barreled" (avoiding execute. It provides a
class activities, perhaps asking two things at once). "snapshot" rather than a
acting as a teaching Maximizing Response representative map.
assistant. This allows for an Rates: Ensure anonymity, 19. Determining Sample
"insider" perspective on keep the survey short, and Size (Mixed Methods)
student dynamics. have leadership
Non-Participant communicate the value of Quantitative (Survey): Size
Observation: The the results to the staff. is determined by power
researcher sits at the back analysis to ensure statistical
or watches via video, acting significance. Larger samples
increase reliability and 21. Grounded Theory vs. 22. Internal and External
reduce the margin of error. Case Study Validity
Qualitative (Interviews):
Size is determined by data When exploring how small Testing a new teaching
saturation—the point where business owners adapt to method in a single school
no new information or digital marketing, the choice raises concerns about the
themes are being between these two accuracy and generalizability
discovered. qualitative methodologies of the results.
Impact: Too small a sample depends on the researcher's Internal Validity (Causal
in quantitative leads to "Type goal. Accuracy):
II errors" (missing a real Grounded Theory: Threats: Confounding
effect); too small in Application: The researcher variables, such as student
qualitative leads to shallow, would use theoretical socio-economic status or
non-credible findings. sampling, where they teacher experience, may
20. Sampling Bias in collect data, analyze it, and influence scores rather than
Dietary Surveys then decide what data to the teaching method itself.
collect next to develop a Strategies: Use random
When a survey on dietary theory. assignment to ensure
habits only attracts urban Strengths: It is excellent for groups are equal at the start
respondents: generating new theories or utilize a control group to
Sampling Bias: The sample directly from the data when isolate the effects of the
is not representative of the no existing theory explains a teaching method.
whole population (missing phenomenon. External Validity
rural perspectives). Limitations: It is highly (Generalizability):
Impact: The conclusions time-consuming and Threats: Because the study
might wrongly suggest a requires the researcher to is limited to one school, the
national trend that is actually remain open-ended without results might not apply to
only true for city-dwellers a pre-defined framework. schools in different
(e.g., higher access to Case Study: geographic or economic
organic stores). Application: The researcher settings.
Strategies to Minimize focuses on a "bounded Strategies: The researcher
Bias: system," such as one can strengthen this by
Stratified Sampling: specific business or a few replicating the study in
Specifically target and businesses, using multiple various school types or using
"over-sample" rural areas to data sources like interviews a more diverse sample.
ensure their voices are and marketing analytics. 23. Research Designs for
represented. Strengths: Provides a "thick Social Media & Academic
Weighting: Statistically description" and deep Performance
adjust the data to give rural contextual understanding of
responses more "weight" in the specific challenges A graduate student has four
the final analysis. faced by those businesses. primary designs to consider
Diversified Recruitment: Limitations: The findings for this study:
Use physical flyers in rural are often difficult to Experimental: Involves
community centers rather generalize to the broader manipulating social media
than just online ads. population of all small use (e.g., a "digital fast") to
businesses. prove causation, but it is
often difficult to implement
ethically or practically.
Correlational: Measures the effectively. interest in the research.
relationship between usage 25. Strategies for a 27. Synthesizing and
hours and GPA. While easy Comprehensive Literature Organizing Research
to conduct, it cannot prove Search Findings
that social media caused a
change in performance. A master’s student should Synthesis is about moving
Longitudinal: Tracks use the following strategies beyond a "summary" to a
students over a long period to ensure depth and cohesive narrative:
(e.g., four years of high breadth: Thematic Organization:
school) to see how habits Database Search: Use Grouping findings by themes
affect performance over academic databases like (e.g., "Technological
time. JSTOR or Google Scholar Barriers" or "Psychological
Cross-sectional: A with specific keywords and Impact") rather than by
"snapshot" of data collected Boolean operators (e.g., author.
at one point in time. It is fast "Remote work" AND Highlighting Patterns:
and inexpensive but cannot "Productivity"). Identifying where different
show how usage habits or Snowballing: Checking the studies agree on the impact
grades change over time. reference lists of key articles of remote work.
24. Qualitative, to find earlier foundational Contrasting Perspectives:
Quantitative, and Mixed works. Explicitly discussing areas
Methods Inclusion of Gray where researchers disagree,
Literature: Looking at which helps identify "gaps"
For exploring a new health industry reports and for the student’s own thesis
app, each approach offers government white papers to to fill.
different insights: capture the most current 28. Identifying Research
Quantitative: Focuses on trends not yet published in Gaps and Justifying a
measurable outcomes like journals. Study
weight loss, heart rate, or 26. Evaluating Source
frequency of app logins. It Credibility and Quality To ensure a study makes a
provides broad, meaningful contribution, a
generalizable data. To ensure the review is researcher must move
Qualitative: Focuses on robust, the student should beyond what is already
personal experiences, such evaluate sources based on: known to find "gaps" in
as why a patient felt Peer Review: Prioritizing existing literature.
motivated or what they articles that have undergone Methodological Gaps:
found difficult about the academic scrutiny over blog Identifying if previous
interface. posts or opinion pieces. studies relied solely on one
Mixed Methods: Combines Currency: Checking if the method (e.g., only surveys)
both to provide a data is up-to-date, and justifying the need for a
comprehensive view. especially in fast-moving different approach (e.g.,
Benefit: It allows the fields like technology and qualitative interviews) to
researcher to see both what remote work. provide deeper insight.
happened (quant) and why it Authority: Investigating the Population Gaps: Finding
happened (qual). credentials of the authors that existing research
Challenge: It requires more and the reputation of the focuses on a specific
time, resources, and publishing journal. demographic (e.g., students
expertise to integrate two Objectivity: Identifying any in the US) and justifying a
very different types of data potential bias or conflict of study on an
under-researched group correlate with reported
(e.g., students in developing symptoms of social anxiety
nations). among first-year university
Contradictory Findings: students?".
Identifying areas where 30. Formulating Null and
previous studies show Alternative Hypotheses
conflicting results, justifying
a new study to clarify these In quantitative research,
inconsistencies. hypotheses provide a formal
Temporal Gaps: Arguing bridge between the research
that older research may no question and the statistical
longer be relevant due to analysis.
rapid changes in technology Null Hypothesis ($H_0$):
or social norms, This predicts that there is no
necessitating an updated relationship or no effect.
study. Example: "Increased social
29. Characteristics of a media use has no significant
Strong Research Question effect on the stress levels of
university students".
A strong research question Alternative Hypothesis
serves as the foundation for ($H_a$): This predicts that
the entire project. For a there is a relationship or a
study on social media and specific effect.
mental health, the question Example: "University
must be: students with higher social
Clear: It must avoid vague media usage will report
language and clearly define significantly higher stress
the variables, such as levels compared to those
"anxiety levels" instead of with lower usage".
just "mental health". Guiding the Design: These
Focused: It should narrow hypotheses determine which
the scope to a specific statistical tests are used
group (e.g., "first-year (e.g., a Correlation test or a
university students") rather T-test) and help the
than "all people". researcher decide whether
Researchable: The to "reject" or "fail to reject"
researcher must be able to the null hypothesis based on
collect data to answer it the collected data.
within the given time and
resource constraints.
Significant: The question
must address a problem that
matters to the field of
psychology, education, or
social policy.
Example: "To what extent
does daily Instagram usage
(exceeding 3 hours)