[Grad2022] [Turnbull Zemi] Notes from 12/29 and 12/22 Zemi

[Stephen J. Turnbull]

FYI

Zemi notes 2021/12/29

Schedule for individual consultation:

2021/12/30（木）  15:00 櫻井 15:30 陶  16:15 陸
2022/01/03（月）  13:00 陶  16:00 陸
2022/01/04（火）  １年生／研究生
2022/01/05（水）  １年生／研究生
2022/01/07（金）  櫻井

M2
- Goal: Submit thesis by 01/06.
- Task: Send me your Response to Comments on the "karitoji" draft.
- This is different from the Response to Comments you will attach to
  the submitted paper.  That version should provide concise
  descriptions of the changes you made in the paper in response to
  each comment, and an accurate reference to the place where each
  change was made.
- In this version, I want to know your current status on each comment,
  so that we can efficiently discuss how to improve your thesis.  You
  do not need to do more work on the paper for this task.  I just want
  the list of comments in a document, and for each comment, the
  current status.  Typical responses are:
  1.  I don't understand the question/comment.
  2.  I understand the comment, but I don't know what to do about it
      yet.
  3.  I understand the comment, and I think I need to do ... about
      it.  Or I have an idea but I don't know how to explain what I
      want to do.
  4.  I know what I should do, and it is ... .
  5.  I have already addressed this by doing ... .
- For each kind of response, reference the place the paper will change
  with page # and a brief description of existing content.  For
  example:
  COMMENT: Sample size is insufficient.
  RESPONSE: I have added 200 more observations.  See p. 12, the
  updated data description.
  COMMENT: Variable z is not defined.
  RESPONSE: z was defined on p. 17, so I moved the definition to p. 11
  after the definition of y.

M1
- Goal: a specific research topic, to begin serious library research
  and study of theory and methods
  Task: write a half-page description: background (why you are
  interested, and why readers would be interested = contribution),
  high-level method (theory, statistical analysis, simulation) and
  data source for empirical studies, theme (research question: is
  there a cause and effect relation between two variables, are there
  conditions where an relationship in the literature is particularly
  strong or weak = mediation)
- If you have a topic, what is your research plan?
  1.  Theme (as above)
  2.  What is your method?
  3.  What is your model (list of cause and effect relationships)?
  4.  As you build the model, you will build a list of variables.
  5.  What resources do you need?
  6.  Construct specific research questions (eg, statistical
      hypotheses about sign or magnitude of coefficients in
      equations).
  7.  How do you plan to use the resources to answer the research
      questions?  For example, ordinary least squares regression.
  Task: write up as much of 1-7 as you know now.

研究生

Like M1, but more time.


Zemi notes 2021/12/22

Abstract contents

普通に
- 短く背景
- 新規性の根拠（先行研究を越えたなど）
- 方法（名前ぐらいで充分）
- 主な結果（２、３件ぐらいまで）
- 意義＝誰に役立つか
を述べて読者の興味を湧かすためにアピールする意図です。
2. Abstract
Overall this would be a good introduction, or part of a good
introduction, but it is not appropriately constructed for the
abstract.  The purpose of the abstract is to excite the reader's
desire to read the rest of the paper.  The usual contents are
- Background, explaining your motivation for the research.  This may
  overlap the policy (or strategic) implications, or be derived from a
  desire to improve on previous research.
- Explain the original contribution (such as improving on the previous
  research).
- Mention the method by name.  Details of how the results were
  achieved is relevant only if the paper is developing the
  methodology.
- The principal results (about 2 or 3).  The idea is to emphasize the
  importance of these results, not to claim that you did many things.
- Implications for private strategy or public policy.  Think about
  this as explain who would benefit from reading your thesis.

Appendices

- Data
- Data collection instruments
  - Simulation: program, initialization parameter
  - Questionnaire surveys and (semi-)structured interviews
    - Original language of questions should be in the Appendix
      (free form) for multilingual researchers to verify translation
    - The full survey in all languages presented to subjects should be
      reproduced in the Appendix (preferably WYSIWYG text)
    - Full citation to original sources (in the same form as text, but
      specify question numbers) etc.

Bibliographies

- Spell author names correctly!
- Use a consistent style.
- Provide translations of author names, titles of papers, and journal
  names for languages not commonly used in the community.
- Need to be complete = identify the resource so readers can acquire
  it efficiently, and recognize it.
  - Exceptions: doi/website URLs/ISBN not enough, for books the city:
    publisher part is traditional, not useful, but do it anyway

In hypothesis statements
- Especially useful in degree thesis to communicate what was done, and
  help verify correct execution
- Precise statements
  - Poor example: "social identity tends to have a plus effect on
    purchase intention."
  - Everybody has a social identity -- what variable is meant here?
  - How is it measured (if there are multiple ways and the text does
    not carefully specify)?
  - "Plus effect on purchase" -- total expenditure, number of items,
    substitution from other platforms to this one (expenditure, items),
    satisfaction/utility, time spent on platform

In stating results:
- long lists of signs or coefficients -> better summarized in a table?
- words can express theories/models
- words can express strategic or policy implications

In statistics:
- Don't p-hack!
- Rejecting the null (usually boring) and accepting the alternative
  (more interesting) tends to be valued.  But accepting the null is
  *useful information* (especially if frequently repeated in the
  literature).
- Significance isn't everything.
  - Men v. women: difference of average is significant at p=.002 for
    average man > average woman, but s.d. is large enough that 30% of
    women have value > average man.
  - Is this important?  Depends on the case, and it has to be argued
    in words.  Sometimes "the difference is statistically significant,
    but not practically significant".

In reporting survey results:
- How are subjects recruited?  How do they apply?  (even if using a
  outside service)

Response to comments
- The comments are not questions to be answered directly, but rather
  are intended to improve the thesis itself.
- It follows that the full response to comments or questions should be
  in the thesis itself, *not* in the response document.
- The response document most importantly documents *where* you changed
  the thesis document so that the examiners can determine whether your
  changes are adequate.
- A very brief description of how you addressed the criticism is
  appropriate, but only rarely is a long answer in the response
  document useful.  Most commonly if the response is that the thesis
  needs no changes (but be very careful about making that claim!)