Educational Research Methods Dr K A Korb University

Educational Research Methods Dr. K. A. Korb University of Jos

Introduction • • There is no point to analyzing data from a study that was not properly designed to answer the research question under investigation. In fact, there’s a real point in refusing to analyze such data less faulty results be responsible for implementing a program or policy contrary to what’s really needed. Two of the valuable things a researcher can possess are: 1. Knowledge of the principle of good study design 2. The courage to refuse to cut corners (Dallal, 1998) Dr. K. A. Korb University of Jos

Scientific Method 1: Ask Question 6: Share Findings 2: Design Study 5: Reach Conclusions 3: Collect Data 4: Analyze Results Dr. K. A. Korb University of Jos

1. Ask a Question • What is your general research question? • What is the independent variable in the research question? • What is the dependent variable in the research question? Dr. K. A. Korb University of Jos

2. Design the Study • Research Design: Structure the research to show the major parts of the research project work together to address the central research questions – Participants • What is the population of people I am interested in? – Instruments • How will I accurately measure my independent and dependent variables? – Procedure • How will I collect data in an appropriate manner? Dr. K. A. Korb University of Jos

Outline • Types of Research Methodologies – Descriptive – Causal Comparative – Correlation – Experimental – Quasi-Experiment • Threats to valid Research Design Dr. K. A. Korb University of Jos

Descriptive • Purpose: Carefully describe a naturally occurring educational phenomenon through systematic observation • Key characteristic: Describes one (sometimes more) variable within a particular population • Descriptive research designs are sometimes also called Survey Research Designs. Dr. K. A. Korb University of Jos

Descriptive • Research hypotheses are impossible for descriptive designs. Only research questions should be posed. • There are no independent or dependent variables, simply the key psychological variables that the researchers is interested in. – Example questions: • What are teachers’ attitude towards a new moral educational program? • What would counselors recommend be included in an HIV/AIDS curriculum? • How many students engage in exam malpractice? Dr. K. A. Korb University of Jos

Descriptive 1. Select participants: Define the group that possess the variables you want to study 2. Data collection: Administer valid measures of the variables of interest 3. Data analysis: Compute descriptive statistics Dr. K. A. Korb University of Jos

Descriptive Dr. K. A. Korb University of Jos

Causal Comparative • Purpose: Determine a cause/effect relationship where the independent variable cannot be manipulated • Key characteristic: Compare two or more naturally-occurring groups on the dependent variable • Causal Comparative designs can also be called ex post facto designs Dr. K. A. Korb University of Jos

Causal Comparative • Null hypothesis: There is no significance difference between GROUPS on DEPENDENT VARIABLE. • Independent variable: Group category • Dependent variable: What the groups are hypothesized to differ on. – Examples: • There is no significant differences between boys and girls on interest in mathematics. • There is no significant differences between children from low and high socioeconomic status in the number of books read in a month. • There is no significance differences between Nigeria and British children in academic self efficacy. Dr. K. A. Korb University of Jos

Causal Comparative 1. Select participants: Obtain a critical number of participants in each group 2. Data collection: Administered valid measures of your variables to all groups of participants 3. Data analysis: Calculate the mean score for each group on the dependent variable. – – However, any difference in the mean score between groups might be due to error. Therefore, inferential statistics are necessary to determine if the difference are significant. Dr. K. A. Korb University of Jos

Inferential Statistics – The inferential statistics to be depends on how many groups are being compared • Two groups: t-test • Three or more groups: ANOVA – If the value is greater than. 05, retain the null hypothesis. • There is no significant difference between boys and girls on interest in math. – If the p-value is less than. 05, reject the null hypothesis. • There is a significant difference between boys and girls on interest in math. – Look at the mean score to determine which group has the stronger interest. Dr. K. A. Korb University of Jos

Causal Comparative Dr. K. A. Korb University of Jos

Correlational • Purpose: Quantity the extent to which two variables are associated • Key characteristic: The same group of participants are given measures of both key variables in order to calculate the correlation coefficient • Because of the directionality and third variable problems, there are no independent or dependent variables in correlational designs Dr. K. A. Korb University of Jos

Correlational • Null hypothesis: There is no significant relationship between VARIABLE 1 and VARIABLE 2. • Examples – There is no significant relationship between number of books a child reads at home and reading ability. – There is no significant relationship between WAEC scores and university GPA. – There is no significant relationship between the amount of time spent in lecture and frequency of exam malpractice. Dr. K. A. Korb University of Jos

Correlational 1. Select participants: A homogeneous sample 2. Data collection: Administer valid measures of the variables 3. Data Analysis: Calculate the correlation coefficient and the significance test for correlation Dr. K. A. Korb University of Jos

Dr. K. A. Korb University of Jos Interpreting Correlations Nature: Negative Strength: -1 Positive 0 +1 • Nature – Positive: Two variables increase or decrease together – Negative: As one variable increases, the other decreases • Strength – Closer to -1 or +1 is stronger relationship – 0 is no relationship Dr. K. A. Korb University of Jos

Correlation =. 78 Dr. K. A. Korb University of Jos

Correlation = -. 86 Dr. K. A. Korb University of Jos

Correlation: . 04 Dr. K. A. Korb University of Jos

Research Example • The effect of regular leisure reading on reading achievement in primary school. • Students in a primary school will be randomly assigned to either a treatment or a control group. – The treatment group will spend 10 minutes everyday reading a book with an adult. – In an attempt to keep the treatment and control groups as similar as possible, the control group will also spend 10 minutes with an adult everyday, but will instead do math problems. Dr. K. A. Korb University of Jos

Correlational Correlation does not prove causation. Interest Amount of time spent studying Academic Achievement Interest Academic Achievement Directionality Problem Third Variable Problem Dr. K. A. Korb University of Jos

Experimental • Purpose: Establish cause and effect • Key characteristic: Statement about how one variables affects (causes) another • Three essential components of an experiment – Treatment – Control – Random assignment to treatment and control groups Dr. K. A. Korb University of Jos

Experimental • Null hypothesis: There is no significant effect of TREATMENT on DEPENDENT VARIABLE. – Independent variable: Treatment – Dependent variable: What the treatment should change • Examples – There is no significant effect of a counseling intervention on homework completion. • Treatment: Counseling – There is no significance effect of reading a book a day on reading achievement. • Treatment: Reading a book a day – There is no significant effective of an anti-exam malpractice campaign on the frequency of exam malpractice. • Treatment: Anti-exam malpractice campaign Dr. K. A. Korb University of Jos

Experimental • Goal: Keep the experiences of the treatment and control groups as identical as possible except for the treatment in order to conclude that the change in the dependent variable is the result of the treatment. Experimental Group Read a book every day Reading Achievement Reading Class Control Group Reading Class Reading Achievement Dr. K. A. Korb University of Jos

Experimental • Random Assignment: Each subject has an equal chance of being assigned to either the treatment or control group – Minimize individual differences in extra variables that might influence the dependent variable Interest Experimental Group Read a book every day Reading Achievement Reading Class Control Group Reading Class Reading Achievement Dr. K. A. Korb University of Jos

Random Assignment • Draw names out of a hat • Random number table S/No Random Number Group 1 1 Treatment 2 1 Treatment 3 0 Control 4 1 Treatment 5 1 Treatment 6 0 Control 7 1 Treatment 8 0 Control 9 0 Control 10 0 Control Dr. K. A. Korb University of Jos

Experiment • Pretest, Post-test Procedure 1. Random assignment of participants to experimental and control groups 2. Administered valid measure of DV as pretest to all groups 3. Administer treatment to experimental groups(s) 4. Administer identical (or conceptually identical) measure as posttest 5. Statistical Analysis – – Descriptive statistics of pretest and post-test score for experiment and control groups separately Significance tests to determine significance between differences (Analysis of Covariance: ANCOVA) Dr. K. A. Korb University of Jos

Experimental Dr. K. A. Korb University of Jos

Experimental • Post-test Only Procedure 1. 2. 3. 4. Random assignment of participants to experimental and control groups Administered treatment to experimental group(s) Administer valid measure of DV post-test Statistical Analysis – – – Descriptive statistics of post-test score for experimental and control groups separately Significance tests to determine significance between differences (Analysis of Variance – ANOVA or t-test) The post-test only procedure is typically recommended because of the possibility that the pretest may have an effect on the experiment treatment Dr. K. A. Korb University of Jos

Experimental Dr. K. A. Korb University of Jos

Quasi-Experiment • Purpose: Conduct an experiment when random assignment is not possible • Key characteristic: Treatment and control groups, but no random assignment • The null hypotheses are exactly the same as the Experimental method Dr. K. A. Korb University of Jos

Quasi-Experiment • Pretest-Posttest Procedure 1. Administer valid measure of DV as pretest 2. Administer treatment to experimental group(s) 3. Administer identical (or conceptually identical) measure as posttest 4. Statistical Analysis • • Descriptive statistical of pretest and posttest score for experiment and control groups separately Significance tests to determine significance between difference (ANCOVA) Dr. K. A. Korb University of Jos

Review: Research Designs • • • Descriptive: Carefully describe a naturally occurring educational phenomenon through systematic observation Causal Comparative: Compare two or more groups on a dependent variable Correlational: Quantify the extent to which two variables are related Experiment: Establish cause and effect – Three requirement: 1. Treatment Groups(s) 2. Control Group 3. Random Assignment Quasi-Experiment: Experiment where random assignment is not possible Dr. K. A. Korb University of Jos

Research Designs Conclusion • Testing the effectiveness of a counseling or teaching intervention MUST either use an experimental or quasi-experimental design • Simply administering surveys CANNOT establish the quality of a counseling or teaching intervention Dr. K. A. Korb University of Jos

Research Example • The effect of regular leisure reading on reading achievement in primary school • Students in a primary school will be randomly assigned to either a treatment or a control group. – The treatment group will spend 10 minutes everyday reading a book with an adult. – In an attempt to keep the treatment and control groups as similar as possible, the control group will also spend 10 minutes with an adult everyday, Dr. K. A. Korb University of Jos but will instead do math problems.

Research Example IV: Regular Leisure Reading DV: Reading Achievement Experimental Group Reading Achievement Read Book Every Day with Adult Reading Achievement Control Group Reading Achievement Do Maths Every Day with Adult Reading Achievement Dr. K. A. Korb University of Jos

Construct Validity • Construct Validity: Establishing valid operational measures for the concepts being studied – This relates to: • Independent Variable • Dependent Variable • Definitions of Variable: – Construct Definition: General explanation of the construct – Operational Definition: Statement of specifically how the construct will be measured or implemented in the study Dr. K. A. Korb University of Jos

Construct vs. Operational Definitions Construct Operational Definition WAEC Scores Academic Achievement Final Course Exam NECO Scores State Exams Economic Status Self-Report of Parents’ Yearly Income Public/Private School Dr. K. A. Korb University of Jos

Research Example Construct validity focuses on: IV: Regular Reading DV: Reading Achievement Experimental Group Reading Achievement Read a Book Every Day Reading Achievement Control Group Reading Achievement Dr. K. A. Korb University of Jos

Construct Validity • Research studies begin with hypotheses about psychological construct. – There is no significant effect of regular leisure reading on reading achievement scores. • Independent variable: Regular leisure reading • Dependent variable: Reading achievement score – The psychological constructs are then operationalized into manipulatable, measurable terms. • Regular leisure reading: Reading a book with an adult everyday for 10 minutes • Reading achievement score: Classroom exam scores Dr. K. A. Korb University of Jos

Construct Validity • When finished analyzing the data, the result will be translated back into psychological constructs. – The conclusion should not be limited to Reading a book for 10 minutes everyday with an adult leads to higher score on a classroom reading exams. – The conclusion should be Regular leisure reading will lead to higher reading achievement scores in general. • However, in order to translate the results back to psychological construct, the study must have construct valid measures of the independent and dependent variables. Dr. K. A. Korb University of Jos

Construct Validity • When you develop your research project, you start with a theory. – For example, regular leisure reading will lead to increased reading achievement scores. • Then you operationalize your theory into your research study. – You translate “regular leisure reading” into how it will be implemented in your study – reading a book with an adult everyday for 10 minutes. – You translate “reading achievement scores” into how it will be measured in your study – classroom exam scores. • When you finish collecting data, you hope to be able to move back to theory. – You want to be able to say that not only did reading a book for 10 minutes everyday with an adult lead to higher classroom exam scores on reading achievement, but also that regular leisure reading will lead to higher reading achievement scores in general. – However, in order to do that, you must have construct valid measures of your independent and dependent variables. Dr. K. A. Korb University of Jos

Research Example Constructs: Operationalized: Constructs: IV: Regular Reading DV: Reading Achievement Read book every day with an adult for 10 min. Scores on reading section of classroom exams IV: Regular Reading DV: Reading Achievement Dr. K. A. Korb University of Jos

Construct Validity • • Construct Validity of Research Designs: Allows generalization from the operationalized treatment and measures to the general psychological constructs Threats to Construct Validity – – Poor construct definitions in the paper Flawed matching of operationalization to constructs • The measures do not align with the constructs they were designed to measure. Dr. K. A. Korb University of Jos

Construct Validity • Research studies begin with hypotheses about psychological construct. – There is no significant effect of regular leisure reading on reading achievement scores. • Independent variable: Regular leisure reading • Dependent variable: Reading achievement score – The psychological constructs are then operationalized into manipulatable, measurable terms. • Regular leisure reading: Reading a book with an adult everyday for 10 minutes • Reading achievement score: Classroom exam scores Dr. K. A. Korb University of Jos

Construct Validity • When finished analyzing the data, the result will be translated back into psychological constructs. – The conclusion should not be limited to Reading a book for 10 minutes everyday with an adult leads to higher score on a classroom reading exams. – The conclusion should be Regular leisure reading will lead to higher reading achievement scores in general. • However, in order to translate the results back to psychological construct, the study must have construct valid measures of the independent and dependent variables. Dr. K. A. Korb University of Jos

Construct Validity • When you develop your research project, you start with a theory. – For example, regular leisure reading will lead to increased reading achievement scores. • Then you operationalize your theory into your research study. – You translate “regular leisure reading” into how it will be implemented in your study – reading a book with an adult everyday for 10 minutes. – You translate “reading achievement scores” into how it will be measured in your study – classroom exam scores. • When you finish collecting data, you hope to be able to move back to theory. – You want to be able to say that not only did reading a book for 10 minutes everyday with an adult lead to higher classroom exam scores on reading achievement, but also that regular leisure reading will lead to higher reading achievement scores in general. – However, in order to do that, you must have construct valid measures of your independent and dependent variables. Dr. K. A. Korb University of Jos

Internal Validity • Internal validity: Extent to which variables other than the treatment provide plausible explanations to the experimental results – Only relevant for experimental designs • Primary consideration for interventions – Extraneous variable: Any variable other than treatment variable that, if not controlled, can affect the experimental outcome – Hold constant or eliminate all extraneous variables that might affect the posttest – Goal of Research Design: Create set of conditions so any observed changes in your dependent variable can be attributed to experimental treatment instead of extraneous variables • Do variables other than the treatment provide plausible explanations to the experimental results? Dr. K. A. Korb University of Jos

Research Example IV: Regular Reading DV: Reading Interest Achievement Maturation Experimental Group Reading Achievement Read a Book Every Day Reading Achievement Control Group Reading Achievement Dr. K. A. Korb University of Jos Reading Adult Achievement Ability Attention

Threats to Internal Validity • History: If treatment lasts over time, other events may influence dependent variable – To prevent: Keep the control group equivalent in all aspects but treatment • Maturation: Physical or psychological changes may influence the dependent variable – To prevent: Use a control group • Selection Bias: Treatment and control groups are different on an important extraneous variable – To prevent: Random assignment OR Use a pre-test Dr. K. A. Korb University of Jos

Threats to Internal Validity • Testing: Giving a pretest may increase performance on the post-test – To prevent: Don’t use a pre-test OR Use different pre- and post-tests • Instrumentation: Nature of measuring instrument has changed between pre- and post-test – To prevent: Do not use a pre-test OR Use the same for preand post-test • Many of these solutions to prevent threats to internal validity contradict. – Consider the threats that are most relevant to a particular study and then develop a strategy for overcoming the most important threats. Dr. K. A. Korb University of Jos

External Validity • External Validity: Establishing the group of people to which the research findings can be generalized Dr. K. A. Korb University of Jos

Research Example Population Experimental Group Sample Reading Achievement Read a Book Every Day Control Group Reading Achievement Dr. K. A. Korb University of Jos

Types of External Validity • Population Validity: Generalizing the results from an experimental sample to a defined population – To establish: Give a thorough explanation of the constitution of the sample and how the sample was selected. • Ecological Validity: Generalizing the results of a study from the artificial conditions created by researcher to real-life conditions – To establish: • Give an adequate description of treatment • Give an adequate description of the measure for the dependent variable – Threats to Ecological Validity: • Novelty: The treatment was different than typical treatment • Experimenter Effect: The particular person delivering the treatment affects results Dr. K. A. Korb University of Jos

Reliability • Reliability: Demonstrating that the operations of the study can be repeated with the same results – Conduct research as if someone were always looking over your shoulder – Report the procedure of your study in explicit detail Dr. K. A. Korb University of Jos

Other Issues in Valid Research Design • Experimenter Bias: Researcher’s expectations about the outcome of experiment influences participants’ response – To prevent: Use research assistants who are blind to the study (aka do not know the purpose of the study) • Treatment Fidelity: Extent to which treatment conditions are implemented according to the researcher’s specifications – To support: • Carefully train research assistants • Periodically check up on research assistants unexpectedly to determine if they are following guidelines Dr. K. A. Korb University of Jos

Other Issues in Valid Research Design • Strong Experimental Treatments: Developing a treatment that will make a robust effect on the dependent variable – To support: • Develop a thorough understand of the dependent variable and how it can be influenced. • Spend much thoughtful time developing your treatment. Dr. K. A. Korb University of Jos

Revision • Describe the five types of educational research designs. • What is construct validity as related to research designs? How can it be supported? • What is internal validity? What are threats to internal validity? • What is external validity? What are types of external validity? • What are some other issues to consider when conducting educational research? Dr. K. A. Korb University of Jos