Advanced ANOVA/Feedback

James Neill's suggestions

edit
  1. Electronic materials: Attempt was made to convert some of the materials into electronic format and to present 2008 materials electronically:
    1. Scanned notes: The previous hard-copy notes (used as overheads) were scanned and are available on the shared psychology drive, along with all other related materials e.g., handouts - under Teaching/Units/4th year/7410/2008.
    2. ucspace and Wikiversity: These wiki-sites were used to present an organising structure, with additional notes and exercises in 2008 (e.g., downloadable data sets). Future units could build on these materials and/or use Moodle.
  2. Timetabling:Thu 9.30-12.30 and 12.30-3.30 for the first 7 weeks of Semester 2. Median attendance was approx. 18 for the first tutorial and 7 for the second tutorial, with approx. 3 to 5 students missing on average (attendance wasn't kept).
    1. Could it have been collapsed into one class? Borderline and the decision was to err on the side of safety since it was the first time James taught the class. With some more experience and planning, a similar-sized cohort could have been taught as one class in 7D4a, and this is worth planning towards, as an option.
    2. 6 or 7 tutorials?: 6 tutorials + 1 one review tutorial were conducted. This could have been taught as 6 tutorials. A review/recap was helpful, but did feel a bit like one class too many and attendance was lowest for this class. Alternatively, the 7th class (or the first class) could be dedicated somewhat to a specific skill e.g. lab report/thesis preparation.
  3. Consultation: The instructor should clarify his/her preferred consultation channels and out-of-class consultation availability to students at the beginning, and particularly towards the due date for the assessment. Normally this goes in the Unit Outline, except where a temporary instructor is used. Awareness of this need arose because at least one student indicated that he didn't receive a requested level of face-to-face support during the three days before the assignment was due.
  4. Write-ups: Example write-ups could continue to worked on and developed. Students really benefited from these and wanted more. Providing multiple sample write-ups of different types of ANOVAs would be ideal, and would also help to discourage over-reliance on a single approach. In-depth write-ups in downloadable word processing format were developed for t-test (partial) and one-way ANOVA (complete)); the aim was to use APA style headings for a lab report, with tables and figures etc.. In addition, checklists (e.g., Advanced ANOVA/Factorial ANOVA#Write-up could be further developed). The brief write-ups from previous years' handouts are a good starting point - note that there are some variations between the electronic versions of these (provided as part of the tutorial .doc handouts) and the tutorial overheads - these could be checked and made consistent.
  5. Advanced ANOVA/Tutorials
    1. Handouts: Students liked having printable handouts available well ahead of each tutorial (the handouts from previous years were used and made available via ucspace).
    2. Break: Students liked to have a 10-min break during the 3-hour tutorials.
    3. Summary: Students generally seemed to like/benefit from a brief summary/wrap-up at the end.
  6. Textbook:
    1. Whitley - didn't use/refer to this very much, but it is important that students have and read such a text.
    2. E-reserve - In addition or alternatively, e-Reserve resources for the unit could be enhanced. Wilkinson's (1999). Statistical methods in psychology journals: Guidelines and explanations was added to e-Reserve.
  7. SPSS
    1. Lab manual: It might have been somewhat confusing for students to have been prescribed Allen and Bennett as their SPSS lab manual, but for the temporary instructor to have preferred use of (and taught the classes from) an alternative text, Francis. On the other hand, at fourth year level, it is probably advisable that students should not become overly reliant on a single SPSS manual. In fact, my preferred approach would be that students also use a triumberant which also included Pallant's SPSS Survival Manual. However, note that the 3rd year Survey research and design in psychology will be using Allen and Bennett from 2009 onwards, so this will bring greater consistency/overlap from 2010 for this unit.
    2. Syntax: Students were largely unfamiliar with, and lacking in confidence use of, SPSS syntax. The efficiency and utility of syntax (particularly for large projects such as theses) was emphasised and most students subsequently became adept at using syntax - mainly because it was demo-ed consistently and encouraged. Development of this skill could be improved by teaching use of syntax during the preceding multiple regression component. Also note that rather than requesting SPSS output for lab report appendices, the accompanying data and syntax files were requested.
    3. Data manipulation: Students needed and benefited from assistance with data manipulation, such as selecting cases, computing new variables etc..
  8. Lab report
    1. Assessment: In the first class, the type of assessment class was discussed for the Advanced ANOVA/Lab report. It was proposed by James Neill that students could use any dataset (including their thesis data) and to present an APA style lab report which presented at least one of each of the main types of ANOVAs taught in the unit. Students seemed to accept the proposal, although this seemed to be largely out of not really knowing much about what was involved! In the end, the assessment exercise worked well and can be recommended again.
    2. Data: In general, students seemed to value and appreciate being able to work on their own assignment data. Note that it was necessary to allow a couple of students to fabricate some data in order to be able to do all six type types of ANOVA, plus a back-up dataset was provided (from James Neill's PhD) for students who didn't have access to suitable data (a couple of students took this option). Interestingly, 4th year thesis marks seemed to be very solid this year, perhaps in part due to allowing students to use their thesis for this lab report.
    3. Submission: There were three noted difficulties with electronic submission of assignments:
      1. One student had a very large file size (> 20MB) due mainly to her putting a lot of SPSS output in the appendix. Students should be warned about and further discouraged from doing this.
      2. Another student had a very large size due to it being in .rtf format. Encourage use of other word processing formats (.doc and/or .odt).
      3. Another student sent her assignment via email (as requested) but the university's student server caused significant delays and multiple emails were needed over several days in order to get the email through.
    4. Marking: Marking took 3 to 4 weeks (involved approx. 2 weeks work). Ideally, this could be a little shorter, perhaps by setting a word limit (students seemed to want an upper limit - perhaps 4,000 words?).
  9. Hypothesis formulation: Several students needed assistance with hypothesis formulation. In retrospect, this could have been emphasised more in the demonstrations, exercises, and write-ups. Some confusion was evident around conflicting advice from thesis supervisors about whether hypotheses were to be stated prosaically or using numbered statements. Consistency between supervisors would be helpful, but also recognition that individual supervisors also have personal preferences and that there is correct way according to APA style. It should be made clear what the options are, what APA style suggests, the pros and cons of each way, with examples, etc..
  10. Descriptive statistics: Ways to calculate and report descriptive statistics were emphasised - i.e., the four moments of normal distributions (mean, standard deviation, skewness, and kurtosis) for the cells, marginal totals, and grand total. Methods for obtaining such results in SPSS and interpretation of skewness and kurtosis were emphasised, along with use of histograms, in testing normality. This was emphasised in an effort to readdress the bad habit many students seem to have acquired of relying on inferential tests of normality rather than more intimate examinations of the distributions. Some demonstration of how to construct effective descriptive statistics table in APA style in MS Word was provided, but even more could be done in this respect (e.g., instructor feedback about how to improve such tables was given on a number of lab reports).
  11. Effect sizes: Some prominence was given to calculation, reporting, and interpretation of effect sizes in 2008. This was emphasised more in the latter tutorials and included explanation of uses of eta-squared, partial eta-squared, and standardised mean effect sizes (Cohen's d). It was suggested that multivariate and bivariate effect sizes be potentially considered for each analysis, depending on hypotheses. It was suggested that ESs should be reported for hypotheses regardless of the statistical significance of inferential tests due to the typical problem of low power and consquently high Type II error rates.
  12. Word processing: Students appreciated and benefited from being taught how to use MS Word to create APA style tables, figures, and automatically generated tables of contents, tables, figures, and appendices. However, this was somewhat difficult to fit in and could perhaps have been done more systematically across the unit and perhaps in a dedicated tutorial.

Student suggestions

edit
  1. The class handouts could be tailored to suit those who perceive themselves as strong in statistical analyses and those who don't.

See also

edit