Introduces methods to empirical research in all fields to counteract errors being a cause of bias in study results

Covers of all aspects of behavioral studies such as sampling, designs, operationalization, data analysis, and reporting

Delivers practical guidance to subfields of the behavioral sciences and related fields

Autorentext

Gideon J. Mellenbergh is emeritus professor of Psychological Methods at the University of Amsterdam, former director of the Interuniversity Graduate School of Psychometrics and Sociometrics (IOPS), and emeritus member of the Royal Netherlands Academy of Arts and Sciences (KNAW). His research interests are in the construction of psychological and educational tests, psychometric decision making, measurement invariance, and the analysis of psychometrical concepts. His teaching was on a large number of methodological topics (design, measurement, and data analysis) for audiences that vary from freshmen to dissertation students. He (co-) supervised 89 PhD students who successfully defended their thesis. Recently, he taught courses on methodological consultancy for research master and dissertation students. He published in international methodological journals (e.g., Applied Psychological Measurement, Journal of Educational Measurement, Multivariate Behavioral Research, Psychological Bulletin, Psychological Methods, and Psychometrika), contributed to methodological books, and published the introductory textbook A Conceptual Introduction to Psychometrics.

Klappentext

This book describes methods to prevent avoidable errors and to correct unavoidable ones within the behavioral sciences. A distinguishing feature of this work is that it is accessible to students and researchers of substantive fields of the behavioral sciences and related fields (e.g., health sciences and social sciences). Discussed are methods for errors that come from human and other factors, and methods for errors within each of the aspects of empirical studies. This book focuses on how empirical research is threatened by different types of error, and how the behavioral sciences in particular are vulnerable due to the study of human behavior and human participation in studies. Methods to counteract errors are discussed in depth including how they can be applied in all aspects of empirical studies: sampling of participants, design and implementation of the study, instrumentation and operationalization of theoretical variables, analysis of the data, and reporting of the study results. Students and researchers of methodology, psychology, education, and statistics will find this book to be particularly valuable. Methodologists can use the book to advice clients on methodological issues of substantive research.

Inhalt

Preface

1 Random and systematic errors in context

1.1 Research objectives

1.2 Random and systematic errors

1.3 Errors in context

1.3.1 Research questions

1.3.2 Literature review

1.3.3 Sampling

1.3.4 Operationalizations

1.3.5 Designs

1.3.6 Implementation

1.3.7 Data analysis

1.3.8 Reporting

1.4 Recommendations

References

2 Probability sampling

2.1 The elements of probability sampling

2.2 Defining the target population

2.3 Constructing the sampling frame

2.4 Probability sampling

2.4.1 Simple random sampling

2.4.2 Sample size

2.4.3 Stratification

2.4.4 Cluster sampling

2.5 Obtaining participation of sampled persons

2.6 Recommendations References

3 Nonprobability sampling

3.1 The main elements of nonprobability sampling

3.2 Strategies to control for bias

3.2.1 Representative sampling

3.2.2 Bias reduction by weighting

3.2.3 Generalization across participant characteristics

3.2.4 Comments

3.3 Recommendations

References

4 Random assignment

4.1 Independent and dependent variables

4.2 Association does not mean causation

4.3 Other variable types

4.4 Random assignment to control for selection bias

4.5 Reducing random error variance

4.5.1 Blocking

4.5.2 Covariates

4.6 Cluster randomization

4.7 Missing participants (clusters)

4.8 Random assignment and random selection

4.9 Recommendations

References

5 Propensity scores

5.1 The propensity score

5.2 Estimating the propensity score

5.3 Applying the propensity score

5.4 An example

5.5 Comments

5.6 Recommendations

References

6 Situational bias

6.1 Standardization 6.2 Calibration

6.3 Blinding

6.4 Random assignment

6.5 Manipulation checks and treatment separation

6.6 Pilot studies

6.7 Replications

6.8 Randomization bias

6.9 Pretest effects

6.10 Response shifts

6.11 Recommendations

References

7 Random measurement error

7.1 Tests and test scores

7.2 Measurement precision

7.2.1 Within-person precision

7.2.2 Reliability

7.3 Increasing measurement precision

7.3.1 Item writing

7.3.2 Compiling the test

7.3.3 Classical analysis of test scores

7.3.4 Classical item analysis

7.3.5 Modern item analysis

7.3.6 Test administration

7.3.7 Data processing

7.4 Recommendations

References

8 Systematic measurement error

8.1 Cheating

8.2 Person fit

8.3 Satisficing

8.4 Impression management

8.5 Response styles

8.5.1 'Plodding' and 'fumbling'

8.5.2 The extremity and midpoint style

8.5.3 Acquiescence and dissentience

8.6 Item nonresponse

8.7 Coping with systematic errors

8.8 Recommendations

References

9 Unobtrusive measurements

9.1 Measurement modes

9.2 Examples of unobtrusive measurements

9.3 Random error of unobtrusive measurements

9.4 Systematic errors of unobtrusive measurements

9.5 Comments

9.6 Recommendations

References

10 Test dimensionality

10.1 Types of multidimensionality

10.2 Reliability and test dimensionality

10.3 Detecting test dimensionality

10.3.1 Factor analysis of inter-item product moment correlations

10.3.2 Factor analysis of inter-item tetrachoric and polychoric correlations

10.3.3 Mokken scale analysis

10.3.4 Full-information factor analysis

10.3.5 Comments

10.4 Measurement invariance

10.4.1 Measurement bias with respect to group membership

10.4.2 Measurement invariance and behavioral research

10.5 Recommendations

References

11 Coefficients for bivariate relations

11.1 Bivariate relation types

11.2 Variable types

11.3 Classification of coefficients for bivariate relations

11.4 Examples of coefficients

11.4.1 Dichotomous variables and a symmetrical relation

11.4.2 Dichotomous variables and equality of X- and Y-categories

11.4.3 Dichotomous variables and an asymmetrical relation

11.4.4 Nominal-categorical variables and a symmetrical relation

11.4.5 Nominal-categorical variables and equality of X- and Y-categories

11.4.6 Nominal-categorical variables and an asymmetrical relation

11.4.7 Ordinal-categorical variables and a symmetrical relation

11.4.8 Ordinal-categorical variables and equality of X- and Y-categories 11.4.9 Ordinal-categorical variables and an asymmetrical relation

11.4.10 Ranked variables and a symmetrical relation

11.4.11 Continuous variables and a symmetrical relation

11.4.12 Continuous variables and equality of X- and Y-values

11.4.13 Continuous variables and an asymmetrical relation

11.5 Comments

11.6 Recommendations

References

12 Null hypothesis testing

12.1 The confidence interval approach to null hypothesis testing

12.1.1 Classical confidence intervals of the means of paired scores

12.1.2 Classical confidence intervals of independent DV score means

12.2 Overlapping CIs

12.3 Conditional null hypothesis testing

12.4 Bootstrap methods

12.4.1 The bootstrap t method for paired DV score means

12.4.2 The bootstrap t method for independent DV score means

12.4.3 The modified percentile bootstrap method for the product moment correlati…