Image Attribution: UnSplash (public use)

research overview

General Research Highlights

General Research Design Steps (not necessarily linear but can be viewed as circular):

• Select a research problem.
• Conduct a literature review.
• Develop research questions or hypotheses.
• Examine the data needed and how to collect the data.
• Determine data analysis (i.e., how you will analyze the data).

Types of Research:

• Descriptive: Describes or defines a particular phenomenon.
  • “What percentage of clients drop out of services within 3 sessions?”
• Explanatory: Explains why particular phenomena work in the way they do; answers ‘why’ questions.
  • “Does trauma history predict relapse in substance use recovery?”
• Exploratory: Conducted during the early stages of a project, usually when a researcher wants to test the feasibility of conducting a more extensive study.
  • “How do youth in foster care describe their transition to adulthood?”

Types of Experimental Research:

• True Experimental- Some elements of a true experimental research study include randomization, control vs. experimental group, hypothesis, independent variable, and dependent variable.
• Quasi-experimental- Similar to experimental research, but there is no random assignment. Example: Nonequivalent groups and pretest-posttest design.
• Pre-experimental- Typically, this type of experimental research is utilized before a researcher decides to conduct a true experiment. An example of a pre-experimental design is a one-group pretest post-test.

single-subject design

Single-subject research can be conducted with one client, an agency, a group, or a community. Single-subject research is considered a type of quantitative research. The goal is to examine whether an intervention is helping clients. The client system in single-subject research is their own control; there is no comparison group like in true experimental research. Examples of single-subject research include a single-case study or a pre- and post-test. Single-subject research design includes two primary phases: Baseline (A) and Treatment (B). There are different types: AB, ABA, and ABAB.

ABA and ABAB are sometimes referred to as a reversal or withdrawal design.  An ethical issue with this type of research is that it can be considered unethical to remove an intervention helping with symptom management. Another ethical issue is that if a client is experiencing a crisis, it would not be helpful to ask the client to track behaviors/outcomes before engaging in an intervention.

Example of ABAB:

Structure: Single-subject research typically unfolds in two main phases:

1. Baseline Phase

   • This phase occurs before any intervention begins.

   • The goal is to observe and identify patterns in the client’s behavior (e.g., drinking more on weekends).

   • Ideally, data is collected in real time, but when that is not possible, retrospective self-reports may be used, though they are less reliable.

   • Since there’s no control group, the client’s behavior during this phase serves as the comparison point for measuring treatment effects.

2. Treatment Phase

   • Begins when the intervention is introduced.

   • Ongoing data collection helps determine whether the treatment is making a difference.

   • For instance, a social worker might track the number of alcoholic drinks consumed daily to see if there is a decrease in binge drinking.

Reference: DeCarlo, M. (2018). Scientific inquiry in social work (Adapted version). Open Social Work Education. https://scientificinquiryinsocialwork.com/

Unlike group research, which aggregates data from multiple participants and uses statistical analyses like means, standard deviations, and correlations to identify general trends, single-subject research emphasizes analyzing data from individuals. Rather than relying on inferential statistics, single-subject researchers primarily use visual inspection. This involves graphing a participant’s data and evaluating visual patterns to determine if the independent variable influenced the dependent variable.

When visually analyzing data, researchers consider several key elements:

• Level: This refers to changes in the overall magnitude of the behavior or outcome when moving from one condition to another. A noticeable increase or decrease suggests an effect of the intervention.

• Trend: This is the direction of change (e.g., upward or downward) in the dependent variable across time. A shift in trend, especially one that reverses direction after treatment, is a strong indicator that the intervention may be effective.

• Latency: This is the time it takes for the behavior to begin changing after the intervention is introduced. A quick response suggests a likely link between the treatment and the observed change.

Overall, single-subject research relies on direct observation of patterns within an individual’s behavior to assess treatment impact, without using statistical generalization.

Reference: Price, P., Jhangiani, R., & Chiang, I. (2015). Research Methods of Psychology (2nd Canadian Edition). Retrieved from https://opentextbc.ca/researchmethods/

Data collection and analysis methods

Quantitative Methods:

• Structured surveys and questionnaires.

• Standardized assessment tools.

• Numerical, measurable outcomes.

Qualitative Methods:

• Interviews (open-ended).

• Focus groups, case notes.

• Rich, descriptive client narratives.

Mixed Methods:

• Combines qualitative + quantitative approaches.

• Offers both depth and measurable insight.

Exploring Real-World Application

The following example is fictitious and not based on a real social worker.

Meet Jasmine. She is a licensed social worker at a community mental health clinic that serves young adults dealing with anxiety and depression. Over the past year, Jasmine noticed something: clients who attended the new mindfulness group seemed to stick with therapy longer than those who did not. Curious and committed to ethical practice, Jasmine decides to explore this pattern further, not just to satisfy her own curiosity, but to support her clients and her team.

Step 1: Identifying the Research Question: Jasmine frames a simple question: Does participating in the mindfulness group improve client retention in therapy?

Step 2: Choosing a Research Design: Since she cannot randomly assign clients to groups (that would not be ethical), she chooses a quasi-experimental design. She compares outcomes between those who chose to attend the mindfulness group and those who did not.

Step 3: Collecting Data: She pulls attendance records and client progress notes. She also sends out a short survey to measure clients’ self-reported stress levels. She codes the survey responses (e.g., “Very Stressed” = 5, “Not Stressed” = 1) and enters all the data into Excel.

Step 4: Analyzing the Data: Using descriptive statistics, Jasmine finds that mindfulness group members attended more sessions on average. With help from a colleague familiar with SPSS (statistical software used to edit and analyze data), she runs a t-test to compare session attendance between the two groups. Her p-value is 0.03, suggesting the difference is statistically significant.

Step 5: Adding Qualitative Insight: Wanting to go deeper, Jasmine also holds informal interviews with a few participants. She codes the transcripts with two of her co-workers and finds themes like “sense of community” and “feeling more in control.” These insights help explain why the group might be working.

Step 6: Sharing and Applying Findings: Jasmine shares her results in a team meeting. The findings support expanding the mindfulness group and tailoring intake to highlight it as an option. She even writes a short article for her agency’s newsletter.

qualitative data analysis

Qualitative analysis is all about making sense of words, patterns, and meaning. Whether reviewing interview transcripts, client journals, or case notes, the goal is to identify themes. For qualitative data, more than one person should always work together on coding to reduce the potential for bias. For instance, let’s say two social workers are reviewing interview transcripts from clients who participated in a community food program. One social worker highlights a quote that says, “I didn’t have to worry about groceries for once,” and labels it as relief from food insecurity. The other social worker reads the same quote and codes it as financial stress. They need to talk through their different interpretations to determine whether they need a broader theme. By working together, they help ensure the analysis reflects what clients actually meant, not just one person’s opinion.

Examples of Qualitative Analysis:

Qualitative researchers focus on trustworthiness, not just validity.

• Credibility: Are the findings believable? Did they check with participants or use peer review?

• Dependability: Would others see what they see if they looked at the data?

• Confirmability: Can someone trace their conclusions back to the original data?

Quantitative data analysis

Descriptive and inferential statistics are both types of quantitative analysis. Descriptive stats summarize the data, while inferential stats help test hypotheses and make predictions. In social work research and practice, statistics help professionals make sense of data.

Descriptive statistics are used to summarize and organize data so that patterns and trends can be easily understood. This type of analysis helps social workers describe what is happening within a client group or program.

Inferential statistics allow researchers to use data from a sample to make broader conclusions or predictions about a larger population. Inferential statistics help answer questions such as whether a treatment made a significant difference or if outcomes differ between groups. Both types of statistics are important in evidence-informed practice and are commonly seen in social work evaluations and assessments.

Common Types of Inferential Statistics

Univariate vs. Bivariate Analysis

Before diving into complex data, social work researchers can begin with univariate analysis to examine one variable at a time. This helps identify basic trends like averages or frequency. When the goal is to explore relationships between two variables, bivariate analysis is used. For example, checking whether stress levels are related to how often clients attend sessions. Both methods are used in practice and program evaluation.

Key Takeaway: Univariate analysis helps you understand what is happening with one variable. Bivariate analysis helps you understand how two variables relate to or influence each other.

Self-Check

methods to assess reliability and validity in social work research

Reliability refers to the consistency of a measurement tool. For example, if we measure alcoholism by asking individuals, “Have you ever had a problem with alcohol?” we might assume this would effectively identify those with alcohol use issues. However, responses may vary depending on context. An individual might answer “yes” after a night of heavy drinking, but “no” on a typical day (DeCarlo, 2018). These inconsistencies suggest that the measurement tool may not be reliable.

Memory is another common challenge in obtaining reliable data. When participants are asked to recall their past behaviors, the accuracy of their responses can vary. Asking someone to remember how much alcohol they consumed each day over the past three months may lead to errors unless they kept a detailed record. A more reliable method may be to ask how many drinks they had over the past week, as the shorter recall period is more manageable (DeCarloo, 2018). Even when observations are used rather than self-reporting, reliability can still be a concern. For instance, a researcher observing drinking behaviors in a bar may miss certain events, such as someone consuming multiple drinks while the researcher is temporarily away. The reliability of the observation depends on the researcher’s ability to consistently capture all relevant behaviors.

One way to assess reliability is through test-retest reliability, which examines whether a measure yields similar results over time. For example, if a student takes the SAT multiple times, the scores should be relatively stable. Likewise, a person with clinical depression should score similarly on a depression scale over a few days (DeCarlo, 2018).

Inter-rater reliability applies when multiple observers are involved. It reflects the degree to which different raters agree on their observations. For instance, when observing a parent interacting with a child, two raters should have similar evaluations of the interaction (DeCarlo, 2018). Discrepancies suggest lower inter-rater reliability.

Internal consistency reliability is relevant when using scales. The items should measure aspects of the same underlying concept. For example, a depression inventory like the Beck Depression Inventory (BDI) includes items on mood and pleasure. While these are not identical constructs, they are related, and their scores should correlate (DeCarlop, 2018). Cronbach’s Alpha is a statistical method used to evaluate internal consistency.

While reliability is about consistency in measurement, validity focuses on whether a tool or question truly measures what it is intended to measure. A measure can be consistent without being accurate; validity ensures that the results actually reflect the concept being studied. Validity is about accuracy.

Source of the following image (DeCarlo, Chapter 9, 2018):

This image consists of four target diagrams arranged in a 2×2 grid. Each target uses black dots to represent individual data points (or measurements), and each is labeled to illustrate different combinations of reliability and validity:

1. Top Left: “Unreliable & Invalid”
   The black dots are widely scattered across the target with no clear pattern and are not centered around the bullseye. This represents a measure that is neither consistent (unreliable) nor accurate (invalid).

2. Top Right: “Unreliable, But Valid”
   The black dots are also widely scattered, but they generally cluster around the bullseye in the center of the target. This represents a measure that is somewhat accurate (valid) but lacks consistency (unreliable).

3. Bottom Left: “Reliable, Not Valid”
   The black dots are tightly clustered together, but the cluster is off to one side of the target, away from the bullseye. This represents a measure that is consistent (reliable) but not accurate (invalid).

4. Bottom Right: “Both Reliable & Valid”
   The black dots are tightly clustered directly at the bullseye of the target. This represents the ideal measurement—both consistent (reliable) and accurate (valid).

evidence-based practice (EBP)

Definitions:

• EBP (noun): Prevention/treatment approaches that have been supported by multiple randomized clinical trials.  For instance, these are interventions that have empirical evidence. Examples: Motivational interviewing for individuals with alcohol dependence; cognitive-behavioral therapy for individuals with depression; assertive community treatment for individuals with serious mental illnesses.
• EBP (verb): A process whereby clinicians consider the values and preferences of the client system, the best research available, and clinical practice expertise.
• Evidence-informed: A term used to denote the ever-evolving nature of evidence-based practice in social work.  There are many research-supported interventions with varying client populations.  Thus, the intervention we choose may not be the “best” or “most important” component. The term EI practice captures our evolving understanding of the science of social work practice.

The ASWB exam may include traditional questions about evidence-based practice (EBP), such as how social workers integrate research evidence with clinical expertise and client preferences. However, it is also important for social workers to critically examine the research itself, especially the sample used. For example, much of the research on bereavement interventions is based on non-diverse samples, often composed primarily of White, female participants. When applying EBP in practice, social workers must consider whether the evidence reflects the populations they serve and whether the intervention is culturally responsive and appropriate.

Three Pillars of EBP:

• Best available research evidence

• Clinical expertise

• Client values and preferences

Steps of EBP Process:

1. Formulate a clear practice question.

2. Search for relevant research.

3. Critically appraise the evidence.

4. Integrate with client needs and clinical context.

5. Evaluate outcomes and adjust as needed.

Ethical Considerations:

• EBP is rooted in the NASW Code of Ethics, especially the principles of competence and client self-determination

• Social workers must critically assess whether an intervention aligns with a client’s identity, values, and lived experience.

Real World Social Work Application

The following scenario is not based on a real social worker.

Tanya is a licensed social worker at a grief support center serving a diverse urban population. A new client, Amina, a 32-year-old Muslim woman whose father died three months ago, comes in for counseling. Tanya is trained in a particular grief therapy modality that has supportive evidence. As Tanya prepares to use the model, she reviews the research behind it. She notices that the original studies primarily involved older White women from suburban areas and lacked cultural or religious diversity in the participant samples. Tanya recognizes that some of these methods may not align with Amina’s spiritual or cultural beliefs around grief expression and mourning rituals. Tanya still draws from the core principles of the grief modality, like helping clients process the pain of loss and reconnect with life, but she integrates it flexibly, guided by Amina’s values and voice.

This scenario shows how EBP is not a “one-size-fits-all” tool. Social workers must balance research evidence with cultural responsiveness and clinical judgment.

Program evaluation

Evaluation is an essential process in the field of social work practice across all levels of practice (micro, mezzo, and macro).  Program evaluation helps social workers and organizations assess whether their efforts are actually making a difference. Evaluation is not just about tracking activities, but it is also about measuring outcomes in a structured and intentional way. Before beginning any evaluation, it is important to clearly define what success looks like and how it will be measured.

Program evaluation helps social workers:

• Demonstrate accountability to funders and communities.

• Identify strengths and gaps in services.

• Make data-informed decisions to improve outcomes.

• Ensure that time and resources are used effectively.

It also aligns with the NASW Code of Ethics, which emphasizes competence, integrity, and evaluation of one’s own practice.

Steps in Evaluation:

1. Engage Stakeholders
2. Describe the Program
3. Focus the Evaluation Design
4. Gather Credible Evidence
5. Justify Conclusions
6. Ensure Use and Share Lessons Learned

Program Evaluation Examples:

• Needs assessment
• Outcome assessment
• Cost-benefit analysis
• Cost-effectiveness analysis
• Formative assessment
• Summative assessment

Information Needed for Different Types of Program Evaluations:

Note: Click the dropdown menus to read the content.

Self-Check

research ethics

Social workers must consider the potential consequences of research and follow professional ethical guidelines. When conducting research, they should consult institutional review boards (IRBs) to ensure ethical compliance. Social workers should ensure access to supportive services and avoid causing any harm, distress, or deprivation.

An essential issue to remember about research ethics is the importance of informed consent. Informed consent should include the nature, purpose, duration, risks, and benefits of participation. The standard is to obtain written consent from research participants and to explicitly state that their participation is voluntary and that they can withdraw at any time. Consider the importance of assent, even if the client’s legal guardian/proxy has signed written consent. Social workers should still seek assent from the client. Confidentiality and anonymity should be upheld, and any limits must be clearly explained. Identifying information must not be shared without explicit permission.

When using technology for research or evaluation, social workers must ensure participants understand and consent to its use, offering alternatives if needed.

self-check