Psychological Methods: Wiki: Difference between revisions
| (4 intermediate revisions by the same user not shown) | |||
| Line 53: | Line 53: | ||
==Learning and exploring statistics== | ==Learning and exploring statistics== | ||
A lot of statistics and what makes learning it and dealing with it so difficult is that it is based on quite abstract assumptions. For example, it is often difficult to grasp that distributions that statistical tests are based upon (e.g., t, F or χ² (chi-square) distributions) are a mathematical description (i.e., a formula) that is based upon empirical distributions (e.g., having an incredibly large number of trials with tossing a coin or throwing a dice). [https://garthtarr.com/about/ Dr. Garth Tarr] developed | A lot of statistics and what makes learning it and dealing with it so difficult is that it is based on quite abstract assumptions. For example, it is often difficult to grasp that distributions that statistical tests are based upon (e.g., t, F or χ² (chi-square) distributions) are a mathematical description (i.e., a formula) that is based upon empirical distributions (e.g., having an incredibly large number of trials with tossing a coin or throwing a dice). [https://garthtarr.com/about/ Dr. Garth Tarr] developed a [http://statstar.io/ helpful web page] where students can play around with, e.g., how the mean and the standard deviation a sample of N=20 could look like if an experiment in this group is repeated again and again or how likely outliers are and how this can influence whether a statistical test becomes significant or not. | ||
==Organizing and storing your data== | ==Organizing and storing your data== | ||
| Line 64: | Line 64: | ||
===Quantitative data analyses=== | ===Quantitative data analyses=== | ||
Statistical analyses are based upon a number of concepts and underlying assumptions. The [[Media:CrashCourseRefresher_Concepts.pdf|following lecture]] aims to give an introduction into such general statistical concepts, discussing, e.g., population and sample, levels of measurement, how to organize and first explore your data (using descriptive statistics), and prerequisites for conducting inference statistics such as a description of what research vs. statistical hypotheses are, several conceptualizations of the p-value and a discussion of statistical significance vs. effect size. | |||
[[Media:CrashCourseRefresher_Concepts.pdf]] | |||
When choosing your evaluation method, a key criterion is whether your variables (predictor/independent and outcome/dependent) are categorical or continuous. Most analysis methods are parametric statistics (i.e., they rely on the assumption that the data are drawn from a distribution, e.g., a standard normal distribution) and based upon the [https://en.wikipedia.org/wiki/General_linear_model General linear model]. | When choosing your evaluation method, a key criterion is whether your variables (predictor/independent and outcome/dependent) are categorical or continuous. Most analysis methods are parametric statistics (i.e., they rely on the assumption that the data are drawn from a distribution, e.g., a standard normal distribution) and based upon the [https://en.wikipedia.org/wiki/General_linear_model General linear model]. | ||
| Line 71: | Line 70: | ||
[[Correlation and regression analysis]] can be used to explore the relationship between continuous predictor and continuous outcome variables. Logistic regression, where the relationship between continuous predictor and categorical outcome variables is explored, is a special case of the regression.<br> | [[Correlation and regression analysis]] can be used to explore the relationship between continuous predictor and continuous outcome variables. Logistic regression, where the relationship between continuous predictor and categorical outcome variables is explored, is a special case of the regression.<br> | ||
To explore the relationship between categorical predictor and continuous outcome variables, we use [[t-test and Analysis of Variance (ANOVA)]]. It is (in an ANCOVA) also possible to include continuous predictor variables, however the main focus in those analyses is typically on the categorical predictors as those represent the experimentally manipulated variables (e.g., treatment vs. control group).<br> | To explore the relationship between categorical predictor and continuous outcome variables, we use [[t-test and Analysis of Variance (ANOVA)]]. It is (in an ANCOVA) also possible to include continuous predictor variables, however the main focus in those analyses is typically on the categorical predictors as those represent the experimentally manipulated variables (e.g., treatment vs. control group).<br> | ||
To explore the relationship between categorical predictor and categorical outcome variables, we use frequency analyses. Those are covered as the | To explore the relationship between categorical predictor and categorical outcome variables, we use frequency analyses. Those are covered as the [[Media:CrashCourseRefresher_jamovi_Analyses.pdf|first part of the following lecture]].<br> | ||
All analyses described above are used to explore manifest variables, i.e. variables that we can directly observe (and analyse). However, there are also [https://en.wikipedia.org/wiki/Latent_and_observable_variables latent variables] that are inferred indirectly through a mathematical model from observable variables. A typical example for such latent variables are personality dimensions, where we measure characteristics of that personality dimension with several items and afterwards combine them (e.g., by summing them up). A method to find or extract such dimensions based upon the pattern of correlations among questionnaire items is called (exploratory) factor analysis. The [[Media:FactorAnalysis_jamovi_Lecture.pdf|following lecture]] first describes the principles and inner workings of factor analysis (FA) before turning to how different factor-analytical methods (exploratory FA, principal component analysis, confirmatory FA and reliability analysis) are carried out in jamovi.<br> | All analyses described above are used to explore manifest variables, i.e. variables that we can directly observe (and analyse). However, there are also [https://en.wikipedia.org/wiki/Latent_and_observable_variables latent variables] that are inferred indirectly through a mathematical model from observable variables. A typical example for such latent variables are personality dimensions, where we measure characteristics of that personality dimension with several items and afterwards combine them (e.g., by summing them up). A method to find or extract such dimensions based upon the pattern of correlations among questionnaire items is called (exploratory) factor analysis. The [[Media:FactorAnalysis_jamovi_Lecture.pdf|following lecture]] first describes the principles and inner workings of factor analysis (FA) before turning to how different factor-analytical methods (exploratory FA, principal component analysis, confirmatory FA and reliability analysis) are carried out in jamovi.<br> | ||
For a couple of years ago, we have changed from SPSS to jamovi for teaching. [https://www.jamovi.org/ jamovi] is a statistics package that is based upon R and quite similar in functionality to SPSS. For further information, please consult the [https://jamovi.readthedocs.io jamovi-documentation] which is available in several languages such as [https://jamovi.readthedocs.io/en/latest/ English], [https://jamovi.readthedocs.io/ | For a couple of years ago, we have changed from SPSS to jamovi for teaching. [https://www.jamovi.org/ jamovi] is a statistics package that is based upon R and quite similar in functionality to SPSS. For further information, please consult the [https://jamovi.readthedocs.io jamovi-documentation] which is available in several languages such as [https://jamovi.readthedocs.io/en/latest/ English], [https://jamovi.readthedocs.io/no/latest/ Norwegian], and [https://jamovi.readthedocs.io/de/latest/ German].<br> | ||
It is not only that jamovi implements its analyses as R-code under the hood, it also offers plenty of opportunity to combine the “safer” option of assembling analyses using a graphical user interface (for standard analyses) with the much broader range of analyses available in R. The [[Media:JamoviToR.pdf| following lecture]] first gives a brief introduction into R (what is it about, variable types, etc.) before demonstrating how to use R from within jamovi (using the [Rj editor]) and how to use jamovi within R (jamovi can output the syntax for the analyses conducted and all analyses are available as [https://cran.r-project.org/package=jmv jmv] R-package, and the R-packages [https://cran.r-project.org/package=jmvconnect jmvconnect] and [https://cran.r-project.org/package=jmvReadWrite jmvReadWrite] help with handling datasets opened in jamovi and jamovi-files). The lecture ends with presenting two typical use cases: using analysis results (e.g., regression coefficients) and importing text data. | It is not only that jamovi implements its analyses as R-code under the hood, it also offers plenty of opportunity to combine the “safer” option of assembling analyses using a graphical user interface (for standard analyses) with the much broader range of analyses available in R. The [[Media:JamoviToR.pdf| following lecture]] first gives a brief introduction into R (what is it about, variable types, etc.) before demonstrating how to use R from within jamovi (using the [https://jamovi.readthedocs.io/en/latest/usermanual/um_6_jamovi_and_R.html#rj-editor Rj editor]) and how to use jamovi within R (jamovi can output the syntax for the analyses conducted and all analyses are available as [https://cran.r-project.org/package=jmv jmv] R-package, and the R-packages [https://cran.r-project.org/package=jmvconnect jmvconnect] and [https://cran.r-project.org/package=jmvReadWrite jmvReadWrite] help with handling datasets opened in jamovi and jamovi-files). The lecture ends with presenting two typical use cases: using analysis results (e.g., regression coefficients) and importing text data. | ||
===Qualitative or mixed-method analyses=== | ===Qualitative or mixed-method analyses=== | ||
Latest revision as of 13:45, 3 September 2024
Planning your study
Literature search
The following lecture gives an overview: (1) on the differences between search engines (e.g., Google Scholar) vs. databases (e.g., PsychINFO, PubMed); (2) on the choice of search terms: their selection, combination (boolean), and further operators (e.g., wildcards) to help with the search; (3) a comparison of systematic reviews vs. meta-analyses (with a focus on aims and procedure; (4) on the use of Google Scholar, Oria, Web of Science, and PubMed (incl. some practical hints); and (5) on different reference management software packages: Zotero, Mendeley, and EndNote (see here for a more detailed overview).
There is also a fantastic introduction on cochrane.org. Cochrane is an organization that summarizes scientific evidence and publishes them as literature reviews / meta-analyses aiming to enhance healthcare knowledge and clinical decision making.
On the PRISMA-web page will you find a checklist, a template for a flow diagram, and their guidelines. These materials should help and guide you when creating systematic reviews or a meta-analyses with high quality.
Finally, the APA web page, especially the Journal Article Reporting Standards for manuscripts using quantative methods, also give helpful advice and a checklist what should be included in a meta-analysis.
Experimental design
The following lecture gives an introduction into experiments as a method to explore cause-effect-relationships, different types of validity related to the experiments and what might be threats to these types of validity. The first part explores the concept of causation, how cause-effect-relationships can be explored using experimental methods, and what the conditions for generalizing the cause-effect-relationship (explored in the experiment). The second part concentrates on the validity types related to the experiment: internal and statistical conclusion validity. The third part focusses on validity types related to the generalizability of the findings from an experiment: external and construct validity.
Preparing physiological and neurophysiological measurements
Physiological measurements
TBA
Neurophysiological measurements: EEG
TBA
Neurophysiological measurements: MRI
TBA
Preparing and conducting your study
Experiments
e-prime
PsychoPy
Web experiments
Questionnaires
SurveyXact (web questionnaire; licensed for UiB)
Pavlovia (a web interface to run experiments created in PsychoPy)
Helsebibliotek (health-related questionnaires)
Acquisition of qualitative / mixed-method data
TBA
Collecting physiological and neurophysiological data
Physiological measurements
TBA
Neurophysiological measurements: EEG
TBA
Neurophysiological measurements: MRI
TBA
Communicating with your participants
TBA
Conducting your research in accordance with legal requirements
The following lecture describes some practical considerations regarding research ethics and data protection. It includes some overview about ethical principles, how and from whom to obtain informed consent, an overview about data protection regulations for research data, a discussion which studies have to be approved by the regional research ethics committee (REK) or the Norwegian Data Protection Agency for Research Data (NSD) and what documents to include in such applications.
Analyzing your data
Learning and exploring statistics
A lot of statistics and what makes learning it and dealing with it so difficult is that it is based on quite abstract assumptions. For example, it is often difficult to grasp that distributions that statistical tests are based upon (e.g., t, F or χ² (chi-square) distributions) are a mathematical description (i.e., a formula) that is based upon empirical distributions (e.g., having an incredibly large number of trials with tossing a coin or throwing a dice). Dr. Garth Tarr developed a helpful web page where students can play around with, e.g., how the mean and the standard deviation a sample of N=20 could look like if an experiment in this group is repeated again and again or how likely outliers are and how this can influence whether a statistical test becomes significant or not.
Organizing and storing your data
TBA
Evaluation methods
Extracting data
This presentation provides an overview how shell scripts can be used to extract data from log files. The commands in the presentation can be tested using these example data.
Quantitative data analyses
Statistical analyses are based upon a number of concepts and underlying assumptions. The following lecture aims to give an introduction into such general statistical concepts, discussing, e.g., population and sample, levels of measurement, how to organize and first explore your data (using descriptive statistics), and prerequisites for conducting inference statistics such as a description of what research vs. statistical hypotheses are, several conceptualizations of the p-value and a discussion of statistical significance vs. effect size.
When choosing your evaluation method, a key criterion is whether your variables (predictor/independent and outcome/dependent) are categorical or continuous. Most analysis methods are parametric statistics (i.e., they rely on the assumption that the data are drawn from a distribution, e.g., a standard normal distribution) and based upon the General linear model.
Correlation and regression analysis can be used to explore the relationship between continuous predictor and continuous outcome variables. Logistic regression, where the relationship between continuous predictor and categorical outcome variables is explored, is a special case of the regression.
To explore the relationship between categorical predictor and continuous outcome variables, we use t-test and Analysis of Variance (ANOVA). It is (in an ANCOVA) also possible to include continuous predictor variables, however the main focus in those analyses is typically on the categorical predictors as those represent the experimentally manipulated variables (e.g., treatment vs. control group).
To explore the relationship between categorical predictor and categorical outcome variables, we use frequency analyses. Those are covered as the first part of the following lecture.
All analyses described above are used to explore manifest variables, i.e. variables that we can directly observe (and analyse). However, there are also latent variables that are inferred indirectly through a mathematical model from observable variables. A typical example for such latent variables are personality dimensions, where we measure characteristics of that personality dimension with several items and afterwards combine them (e.g., by summing them up). A method to find or extract such dimensions based upon the pattern of correlations among questionnaire items is called (exploratory) factor analysis. The following lecture first describes the principles and inner workings of factor analysis (FA) before turning to how different factor-analytical methods (exploratory FA, principal component analysis, confirmatory FA and reliability analysis) are carried out in jamovi.
For a couple of years ago, we have changed from SPSS to jamovi for teaching. jamovi is a statistics package that is based upon R and quite similar in functionality to SPSS. For further information, please consult the jamovi-documentation which is available in several languages such as English, Norwegian, and German.
It is not only that jamovi implements its analyses as R-code under the hood, it also offers plenty of opportunity to combine the “safer” option of assembling analyses using a graphical user interface (for standard analyses) with the much broader range of analyses available in R. The following lecture first gives a brief introduction into R (what is it about, variable types, etc.) before demonstrating how to use R from within jamovi (using the Rj editor) and how to use jamovi within R (jamovi can output the syntax for the analyses conducted and all analyses are available as jmv R-package, and the R-packages jmvconnect and jmvReadWrite help with handling datasets opened in jamovi and jamovi-files). The lecture ends with presenting two typical use cases: using analysis results (e.g., regression coefficients) and importing text data.
Qualitative or mixed-method analyses
TBA
Literature review and meta-analysis
An overview on literature search is given at the top of this page.
Types of literature reviews
TBA
Meta-analysis
TBA
Summarizing and publishing your study
Obeying the standards of the APA publication manual
A series of lectures dealt with how to obey the standards of the APA publication manual:
The first lecture explores the questions: Why publishing? Why a rule system? before turning to the structure of a manuscript, proper language use and some mechanics of style (i.e., the use of period (.), comma, abreviations, parentheses, etc.).
The second lecture shows how to display results in figures and tables and provides some practical hints to help with writing manuscripts.
The third lecture demonstrates why, when and how to use references.
The fourth lecture gives practical hints for writing manuscripts and term papers, gives and overview how the publication process works and discusses ethical issues with publication (authorship, consent, plagiarism).
Recently, the 6th edition of the APA-manual has been replaced by the 7th edition. The amount of changes was rather moderate (summarized in a list with the most notable changes).
APA also provides the APA-style web page with plenty of very helpful information, incl. a steadily increasing numbers of examples, e.g., for papers, figures and tables, and for references. The web page also provides what should be included in a manuscript, denoted as Journal Article Reporting Standards (JARS). JARS includes three checklists for manuscripts using quantitative, qualitative and mixed-method approaches (in addition there are specific checklists,. e.g., for quantitative and qualitative meta-analyses).
Please note that there is also an official recommendation / standard on how to use the APA-style in Norwegian.
Presenting your results
This lecture deals with how to present, covering topics such as the structure of a presentation and the use of graphics, as well as personal factors such as dealing with nervousness.
Software
What is open source software, why should you use it and what software packages are available for standard tasks (office suites, working with graphics, doing statistics)
Reference management
LaTeX
Tips and tricks for standard programmes
