Flashcards

Risk, safety, and security research can be categorized into applied vs. fundamental, descriptive vs. analytical, quantitative vs. qualitative, and conceptual vs. empirical. These categories are not mutually exclusive, and research often combines elements from different types.

Social Identity:

• Definition: Social identity refers to the aspects of an individual’s self-concept that are derived from their membership in social groups. It involves how one perceives themselves in relation to broader social categories, such as nationality, religion, or organizational affiliations.

Personal Identity:

• Definition: Personal identity pertains to the individual characteristics that distinguish a person from others. It encompasses unique qualities, experiences, beliefs, and traits that define who a person is as an individual.

Relation to Agent and Actor:

• Agent:
  • Social Identity: Acts based on group affiliation.
  • Personal Identity: Acts based on unique personal characteristics.
• Actor:
  • Social Identity: Performs roles aligned with group norms.
  • Personal Identity: Expresses behaviors based on individual traits and experiences.

Lecture 4 Interviews, Focusgroups and additional methods of data collection.

Fully Structured Interview:

• Advantages:
  • Consistency: Ensures consistency in data collection as all participants are asked the same set of predetermined questions.
  • Quantitative Analysis: Facilitates quantitative analysis due to standardized responses.
• Disadvantages:
  • Limited Flexibility: Offers limited flexibility for probing or exploring unexpected insights.
  • Lack of Depth: May lack depth in understanding complex responses.

Semi-Structured Interview:

• Advantages:
  • Flexibility: Allows the interviewer to follow up on interesting responses, providing flexibility in the conversation.
  • In-Depth Exploration: Permits a more in-depth exploration of responses compared to fully structured interviews.
• Disadvantages:
  • Potential Variability: Responses may vary between participants, making it challenging to standardize data.

Unstructured Interview:

• Advantages:
  • Rich Data: Generates rich and detailed data, allowing for a deep understanding of the interviewee’s perspective.
  • Exploratory Nature: Encourages exploration of diverse topics and unexpected insights.
• Disadvantages:
  • Subjectivity: Results in subjective data collection, making analysis and comparison more challenging.
  • Time-Consuming: Can be time-consuming due to the open-ended nature of the conversation.

Advantages
•A highly efficient technique (data collected from several participants at the same time)
•Participants enjoy the interaction
•Participants are stimulated by each other to share opinions
•Relatively inexpensive and less time-consuming

Disadvantages
• The number of questions covered is limited
• The role of a moderator may be challenging
•Conflicts may arise
•Confidentiality can be a problem
•The results are difficult to generalize

Lecture 5 Fieldwork and Ethnography

Lecture 6 The analysis and interpretation of qualitative data

Quasi-Statistical Approaches:
Analyzes qualitative data using word or phrase frequencies and inter-correlations to assess the importance of terms and concepts.
Example: Content analysis, where researchers quantify and analyze the occurrence of specific words or phrases to draw insights from the data.

Thematic Coding Approach:
Involves coding parts of qualitative data, grouping codes with the same label into themes.
Example: Researchers identify and label portions of data, grouping similar codes under broader themes. These themes then guide further analysis.

Grounded Theory Approach:
Generates codes directly from qualitative data, emphasizing the researcher’s interpretation of meanings or patterns.
Example: Researchers develop a theory based on emergent patterns and meanings observed in the data, creating codes grounded in their interaction with the information.

Familiarizing Yourself with Your Data:

• Process: Transcribe data (if necessary), read and re-read the data, and note down initial ideas.
• Importance: Immerse yourself in the data to gain a deep understanding, particularly after initial data collection.

Generating Initial Codes:

• Methods: Devise a framework or template, or generate codes inductively through interaction with the data.
• Coding: Systematically assign codes to extracts across the entire data set, grouping similar extracts under the same code.

Identifying Themes:

• Process: Collate codes into potential themes, gathering all relevant data for each theme.
• Validation: Check if themes align with coded extracts and the entire data set; revise codes or themes if necessary.

Constructing Thematic Networks:

• Visualization: Develop a thematic ‘map’ or network of the analysis.
• Representation: Create a visual representation of how themes connect and interact within the data.

Integration and Interpretation:

• Comparison: Make comparisons between different aspects of the data using techniques like tables and networks.
• Exploration: Explore, describe, summarize, and interpret patterns within the data.
• Quality Assurance: Demonstrate the quality of the analysis through thorough exploration and interpretation.

Lecture 7 The analysis and interpretation of qualitative data part II

Characteristics of a Good Flexible Design:

1. Multiple Data Collection Techniques: Utilizes various qualitative data collection techniques, and can include quantitative data. The data is summarized and information is given on how the data is collected.
2. Framed within Flexible Design Assumptions: Inclusion: Aligns with the assumptions and characteristics of flexible design, including an evolving nature, presentation of multiple realities, researcher as a data collection instrument, and a focus on participants’ views.
3. Informed by Existing Research Traditions: Acknowledges and incorporates one or more traditions of research.
4. Starts with a Single Idea or Problem: Begins with a singular idea or problem that the researcher aims to understand.
5. Rigorous Approach to Data Collection and Analysis: The study shows a rigorous approach to data collection, data analysis, and report writing.
6. Analyses Using Multiple Levels of Abstraction: Analyzes data using multiple levels of abstraction, often presenting studies in stages or layering analyses from specific themes to larger perspectives.
7. Clear and Engaging Writing: Ensures clear, engaging writing that immerses the reader in the research context.

To establish trustworthiness in fixed research designs, the process begins with common-sense actions such as conducting a thorough and honest job, and maintaining an unbiased and objective approach throughout the study. Here’s a detailed breakdown of how trustworthiness is supported by two key concepts: validity and reliability.

Validity involves several components:

1. Justification of Conclusions: Ensuring the conclusions drawn from the research are well-supported by evidence.
2. Measurement Alignment: Confirming that the metrics used in the study accurately reflect what they are supposed to measure, known as construct validity.
3. Study Design and Findings: Checking that the research design and its findings reliably support the conclusions, referred to as internal validity.
4. Extent of Applicability: Assessing how well the findings can be generalized to other settings or groups, which relates to external validity.

Reliability focuses on consistency:

• This includes maintaining dependability in the data collection and analysis processes to ensure that results are consistent over time and replicable under similar conditions.

Experimental Designs:
Experimental designs involve the assignment of participants to different conditions, manipulation of independent variables by the experimenter, and the measurement of the effects on dependent variables, while controlling for all other variables.

True Experiments and Randomized Controlled Trials (RCT):
True experiments, exemplified by Randomized Controlled Trials (RCT), employ random allocation of participants to intervention/treatment and control groups. Examples of true experiments include post-test-only RCTs, pre-test post-test RCTs, and matched pairs experiments, among others.

Quasi-Experiments:
In the context of studying the impact of a new job safety analysis (JSA) template, the following designs should be avoided:

• A single group of workers asked about the usefulness of using the new template.
• Two groups representing different employers asked about usefulness after one group used the new template.
• A single group asked about usefulness before and after template implementation.

Experimental Design without Random Allocation:
While experimental designs without random allocation can be useful, they must be approached with caution. Single case experiments, a class of experimental designs studying a single group, use baseline conditions (A) and intervention phases (B) to measure changes in a dependent variable under different conditions.

Non-Experimental Fixed Designs:
Non-experimental fixed designs lack manipulation and are suitable in situations where manipulation is not possible, feasible, or ethical. Surveys are examples of non-experimental fixed designs.

lecture 9 Collection of quantitative data

Elaborate Answer:
Surveys are a common methodology in applied risk and safety research due to their versatility and effectiveness in gathering data from diverse populations. The four typical features of surveys include:

1. Fixed Research Design: Surveys usually follow a structured format where the design does not change once established. This allows for consistent data collection across various subjects.
2. Non-experimental Design: Most surveys are observational and do not involve the manipulation of variables, which is characteristic of non-experimental designs. However, they can also be adapted to experimental setups where certain conditions are manipulated.
3. Standardized Data Collection: Surveys collect data using standardized methods such as questionnaires or interviews that are administered to a large number of participants. This standardization ensures that the data are comparable and measurable across all respondents.
4. Representative Samples: Surveys often use representative samples that reflect the characteristics of a larger population.

From the lecture slides:
• Ambiguous or incomprehensible questions
• Low response rates (Note from me: This is a reliability issue, not a validity issue. Reliability is about the replicability of the study, validity is about measuring the right thing. The response rate does not tell if the right thing was measured or not).
• Bias in the sample
• Differences in what participants answer and what they do/how they feel/what they believe

1. Variables Correlate: Suppose a survey investigates the relationship between the use of personal protective equipment (PPE) and the incidence of workplace injuries. To establish validity, there must be a demonstrable correlation where increased use of PPE correlates with decreased injuries.
2. Cause Precedes Effect: For the survey findings to be valid, it must be shown that the use of PPE (cause) occurs before the reduction in injury rates (effect).
3. Other Explanations Ruled Out: It must also be demonstrated that other potential causes of the reduction in injury rates, such as improved training or safer workplace practices, are not the actual reasons behind the observed correlation. This involves controlling for these variables or showing through statistical analysis that their impact is negligible compared to that of PPE.

1. Online Surveys

Advantages:

• Cost-effective: Low cost as there are no physical materials or travel expenses.
• Broad Reach: Can reach a large and geographically dispersed audience quickly.
• Speed: Rapid deployment and real-time data collection.

Disadvantages:

• Sampling Bias: May not represent populations without internet access.
• Low Response Rates: Often lower than other methods due to survey fatigue or lack of personal engagement.
• Quality of Data: More prone to careless responses as participants may rush through the survey.

2. Telephone Surveys

Advantages:

• Personal Touch: Can increase the likelihood of participation through direct interaction.
• Higher Response Rates: Typically higher than online surveys.
• Controlled Sampling: Easier to target specific demographic groups.

Disadvantages:

• Cost: More expensive due to the need for trained staff and telecommunication costs.
• Time-Consuming: Takes longer to collect data.
• Potential Bias: Call screening or refusal to participate can skew results.

3. Mail Surveys

Advantages:

• Wide Coverage: Can reach individuals without phone or internet access.
• Visual Design Flexibility: Allows for complex question formats and branding.
• Response Consideration: Respondents can take their time to consider answers thoroughly.

Disadvantages:

• Low Response Rates: Typically lower than telephone surveys.
• Slow Data Collection: Takes longer to send, receive, and process.
• Higher Costs: Printing, mailing, and data entry costs are significant.

4. Face-to-Face Surveys

Advantages:

• High Quality Data: Interviewers can ensure comprehension and thoroughness.
• Flexibility: Questions can be adapted or clarified on the spot.
• High Response Rates: Personal interaction boosts participation.

Disadvantages:

• High Cost: Travel, training, and time make it the most expensive method.
• Time-Intensive: Requires significant time for both preparation and execution.
• Potential Interviewer Bias: The presence or demeanor of the interviewer might influence responses.

• Simplicity in Language: Questions should be straightforward and free from jargon to be universally understandable.
• Relevance to Research Questions: Each question should directly contribute to answering the main research questions, ensuring all data collected is relevant.
• Avoidance of Biased Questions: Questions should be neutral, avoiding any wording that might lead or influence respondents’ answers.
• Clarity and Specificity: Questions should be specific and clear, avoiding any ambiguity that could confuse respondents or lead to varied interpretations.

To develop a Likert scale in survey research, researchers typically follow these steps:

1. Item Creation: Develop a series of statements that respondents can agree or disagree with, relevant to the concept being measured.
2. Pilot Testing: Administer these items to a representative sample to gather initial data.
3. Scoring Responses: Assign numerical values to the degrees of agreement or disagreement (e.g., Strongly Agree = 5, Strongly Disagree = 1).
4. Analyzing Item Responses: Use statistical methods to analyze the responses, identifying items that effectively discriminate between different levels of the attribute or construct being measured.
5. Scale Refinement: Remove or revise items that do not contribute effectively to the scale’s overall reliability and validity.

Lecture 10 Guest lecture Safetec on quantitative data 1

Creating a robust dataset involves several key activities:

1. Data Export: Efficiently transfer data from collection platforms (like SurveyXact) to analysis software (e.g., SPSS).
2. Data Cleaning: Address incomplete entries, remove or recode irrelevant responses like ‘Don’t know’, and ensure that all variables are correctly named and formatted.
3. Variable Adjustment: Recode data to create groups or categories that are of interest for the specific analysis.
4. Quality Control: Use statistical methods like frequencies and cross-tabs to verify the data’s integrity and consistency.

Exploring relationships involves:

1. Correlation Analysis: Determines if changes in one variable predict changes in another, with methods tailored to data types (Pearson, Spearman).
2. Cross Tabulation: Examines relationships between two categorical variables by creating a matrix of frequencies for each category.
3. Statistical Tests: Chi-square tests, for example, help confirm the strength of associations between categorical variables.

Analyzing differences involves:

1. T-Tests: Assess if two groups differ significantly on some continuous variable.
2. ANOVA (Analysis of Variance): Used when comparing more than two groups to determine if there are any statistically significant differences between the means.
3. Regression Analysis: Helps in predicting a dependent variable based on one or more independent variables, identifying the strength and nature of the relationship.

Causal inferences require careful design:

1. Control for Confounders: Using statistical techniques like regression to control for variables that might affect the relationship.
2. Temporal Precedence: Ensuring that the cause precedes the effect in time, which is crucial for establishing a causal link.
3. Ruling Out Alternatives: Using experimental or quasi-experimental designs to rule out other potential causal explanations.

Lecture 11 Guest lecture Safetec on quantitative data 2

Factor analysis groups together interrelated variables into factors based on their correlations, simplifying data analysis by reducing the number of variables:

1. Identifying Clusters: Variables that correlate highly are grouped into fewer factors.
2. Principal Component Analysis (PCA): Used to extract factors, where factors are chosen based on their eigenvalues (usually greater than 1) and factor loadings (typically above 0.4).
3. Interpretation: The factors are interpreted to understand the latent constructs they represent, often using varimax rotation to enhance the interpretability.

EFA and CFA serve different purposes in research:

1. Exploratory Factor Analysis (EFA): Often employed when a researcher wants to discover the underlying structure of a dataset without prior assumptions. It helps in identifying new factors.
2. Confirmatory Factor Analysis (CFA): Used to test whether measures of a construct are consistent with a researcher’s understanding based on theoretical knowledge. CFA assesses the goodness of fit of a hypothesized model using various fit indices.

SEM is used to confirm theory and test complex models involving multiple pathways and interrelationships among variables:

1. Model Specification: Involves setting up a model that specifies the expected relationships among observed and latent variables.
2. Model Estimation: The model’s parameters are estimated, usually through maximum likelihood estimation.
3. Model Evaluation: Fit indices such as RMSEA, CFI, and TLI are used to determine how well the model fits the data.

Understanding these errors is critical for research validity:

1. Type I Errors (False Positives): Erroneously concluding that there is an effect when there is none, often controlled by setting a lower alpha level (e.g., 0.05).
2. Type II Errors (False Negatives): Failing to detect an effect when there actually is one, typically addressed by increasing the sample size to enhance statistical power.

1. Linear Regression: Assesses the linear relationship between dependent and independent variables, quantified by coefficients indicating the magnitude and direction of relationships.
2. Multiple Regression: Involves several independent variables to understand their collective impact on the dependent variable, adjusting for other factors.
3. Model Diagnostics: Regression also involves checking for the assumptions of linearity, normality, homoscedasticity, and absence of multicollinearity.

ANOVA identifies group differences under controlled conditions:

1. One-way ANOVA: Tests differences across multiple groups based on a single independent variable.
2. Factorial ANOVA: Allows researchers to assess the interaction effects between two or more categorical independent variables on a continuous dependent.
3. Post Hoc Tests: If ANOVA results are significant, post hoc tests like Tukey’s HSD are performed to pinpoint exactly which groups differ from each other.

Lecture 12 conceptual studies in risk safety and security research

Unlike empirical research that relies on observable data to validate theories or models, conceptual research in risk, safety, and security studies involves:

1. Theoretical Exploration: Engages with theoretical constructs and the formulation of frameworks without the direct use of empirical data.
2. Methodological Development: Focuses on enhancing conceptual understanding and proposing new approaches to tackle complex problems.
3. Normative Analysis: Often involves evaluating and recommending the best practices or definitions, which are inherently normative and not purely factual.

1. Innovativeness: Developing new insights that significantly advance the field.
2. Rigorous Evaluation: Ensuring conceptual clarity, internal consistency, and empirical applicability of the new concepts.
3. Peer Review and Adoption: Navigating the peer review process and gaining acceptance and implementation in practical settings.

Lecture 13 Exam prep: