Advanced ANOVA/Assessment

From Wikiversity
Jump to navigation Jump to search

This page describes the requirements for the ANOVA assignment, 2008.

Requirements[edit]

The lab report requires students to conduct and present in APA style (at least) one of each of the six types of ANOVA models covered in the tutorials, that is:

  1. t-test: one-sample, independent, or paired
  2. One-way ANOVA
  3. Factorial ANOVA
  4. Repeated-measures ANOVA
  5. ANCOVA
  6. MANOVA

Data[edit]

  1. You may use any dataset (providing that it allows you to conduct each of these tests).
  2. You may use your thesis data (or someone else's thesis data with their permission).

Marking criteria[edit]

The sections for the report, the marking criteria, and their weighting will be as follows:

Section Notes Weighting
Introduction Include title and abstract. Identify research question, explain key terms/constructs, and provide major references. State clear hypotheses. 10%
Method Provide brief participant overview, background to instrumentation, and brief procedure. 10%
Results Structure results by the types of analyses; for each, consider recoding, descriptives, graph, table, and description of inferential and effect size results. 60%
Discussion Interpret the results, their strengths and limitations, and implications. 20%

APA style will represent approximately 10-20% of the marks within each section.

Sections[edit]

This section contains some further advice (prompted by student requests) about what to include in the major report sections. The intention of this section is not to change in any way the previously stated requirements, but rather to consolidate and clarify verbal and email advice and feedback given during the first six weeks of the unit.

Please demonstrate independent writing and thinking, reference your sources, and so on (i.e., follow standard academic and APA style requirements). Avoid simply replicating or copying thesis work or others' assignment work (because unnecessarily verbose dumping of content with debatable relevance will be unlikely to meet the report requirements). Instead, demonstrate your independent capacity to succintly formulate, conduct, and report on ANOVA analyses using a specific data set. Highest marks are likely to be obtained by the most disciplined, targeted, and judiciously-edited reports, not the longest ones.

Title and Abstract[edit]

A ~10-15 word title and ~150-200 word abstract should provide a clear, succint, and accurate orientation to the content and findings of the report.

Introduction[edit]

The number one purpose of the introduction (for this exercise) is to present logically-derived and clearly articulated hypotheses. This is your chance to demonstrate an independent capacity to do this within the context of ANOVA approaches to inferential data analysis. It should be succint, not a thesis intro, or a reworded thesis intro (that's what theses are for!).

Tip: Begin with independent formulation of hypotheses, then work backwards to make sure you can explain in your own words the rationale for the hypotheses, explain the constructs involved, and why these might be an interesting/important set of questions/theory to test.

Method[edit]

The Method will be evaluated in terms of whether it provides a reader with a clear understanding of the study's sample, instrumentation, procedure, design, etc. but it should also be short, focused, targeted, etc. for this specific exercise, with relevant references, etc. Avoid extraneous information; focus on helping the reader to understand how the data you analyse was obtained. A reader should have enough info from the method to replicate the study, and should not be distracted by irrelevant info.

Results[edit]

Present the results in sub-sections by each hypothesis. However, avoid redundancy, e.g., it may make sense to present a single table of descriptive statistics up front, with each section referring back to this table. For each of the ANOVAs, consider whether it is appropriate to address (all of these for each hypothesis would probably be over the top - but think about which might be appropriate in each case):

  1. Purpose and design (e.g., IVs and DVs)
  2. Assumptions
  3. Descriptives
  4. Inferential results, including choice of critical
  5. Graphical results
  6. Followup/Post-hoc comparisons
  7. Direction and strength of effect sizes
  8. Rejection / non-rejection of hypothesis
  9. Meaning of results (probably belongs mostly in Discussion)

Discussion[edit]

Demonstrate your understanding and conclusions about the hypotheses, given the statistical testing results (i.e., what do the results mean?). What can be said with reasonable confidence? What caveats should be applied? What were the strengths and weaknesses of the design and analyses with respect to testing the hypotheses? How could such studies (using similar designs, instrumentations, and analysis techniques) be improved in the future? Strive for a balanced critique (i.e., drawing on the study's strengths whilst also providing tangible, specific suggestions for improvement).

References[edit]

A long list of references is not expected or necessary; however diligent referencing of key theoretical references which help to explain/justify the hypotheses, methodology, specific statistical decisions, and conclusions is vital.

Appendices[edit]

Appendices are largely unecessary (avoid appendices for appendices' sake). However, effective, judicious use of appendices could be appropriate, for example, for providing copies of:

  • The instrumentation and procedures used to the obtain the data analysed in the study
  • Graphical or statistical output referred to in the results but the inclusion of which would break the flow of results. Note however that the results section should be complete and should not require a reader to consult appendices in order to make sense of it (e.g., a journal article generally makes little use of appendices because the results section should stand up its own). An example of appropriate statistical output for the appendices are the histograms used as part of the researcher's examination of univariate normality. But please don't dump appendices of all output; plus make sure that all appendices are clearly labelled and referred to in the main body.
  • When submitting electronically, also attach the SPSS data and SPSS syntax. The marker may refer to these if there seems to be a problem with your results.

Word count[edit]

  1. No lower or upper word limit
  2. Parsimony/succintness is ideal, but not to the exclusion of thoroughness, completeness, and clarity.

Due date[edit]

  • 5pm September 11, 2008
  • 5% penalty per 24 hours late

Submission[edit]

  • Submit the lab report as a single word-processing file (with coversheet and any appendices included) via email attachment to James Neill - no hard copies.
  • Also submit the data file(s) and any relevant syntax files to allow checking of your analyses.
  • All attachments should start with your surname; this will help to keep track of things.
  • You should receive feedback and marks via email within two weeks of the due date.

See also[edit]

External links[edit]