[Turnbull Zemi] Schedule for this week
Stephen J. Turnbull
stephenjturnbull at gmail.com
Mon Aug 8 18:34:34 JST 2022
Hi, all
This week we're back to a more normal schedule, with more flexibility.
As usual, if the schedule is inconvenient for you, I would prefer that
you find somebody to trade with. Just let me know who is trading with
who. If you talk to all the people with times you can make it, and
nobody is able to trade, I can manage other times.
This week's Zemi is on Wednesday at the usual time (13:30). First, I
will talk 15-20 minutes about use of git to communicate your documents
to me, and we will spend time in individual consultation getting set
up if necessary. Then we will have individual presentations of data
analysis and previous research. (Details of presentation structure
follow the schedule.)
1. Mr. Hua will present a previously published paper on his research topic.
2. Mr. Nishi will present a statistical analysis of data about used cars.
3. Ms. Xu will present a statistical analysis of data about
public-private research partnerships.
4. Mr. Ma will present a statistical analysis of data about real
estate transactions in Beijing.
Schedule
========
日付(曜日)時間 時間 学生
08/10 (W) 13:30-16:00 150分 ゼミ
08/11 (R) 10:30-10:50 20分 周 雅佩
08/11 (R) 11:00-11:20 20分 王 天舒
08/11 (R) 13:30-13:50 20分 花 季
08/11 (R) 14:00-14:20 20分 王 巧
08/11 (R) 15:00-15:20 20分 西 康成
08/12 (F) 10:30-10:50 20分 李 悦祺
08/12 (F) 11:00-11:20 20分 徐 慧
08/12 (F) 13:30-13:50 20分 金 明旸
08/12 (F) 14:00-14:20 20分 马 弢
08/12 (F) 15:00-15:20 20分 刘 梁
Requirements for presentations of published papers
==================================================
Your presentation time should be 5-10 minutes, try to keep it to
around 5 minutes. This is more than you would spend on previous
literature midterm or final presentations, but less than you would
spend on your own research in those presentations. You may prepare
more if you can, but at minimum you should have one slide each for
1. Title: author(s), title, publication data (journal etc), and
"presented by $YOUR_NAME on July 13, 2022".
2. Background: why did the author(s) write this paper, how is it
related to your research theme.
3. Previous research (1 (or 2) paper(s), main result, how does the
paper you present improve on this (these) paper(s))
4. Methodology of the paper: main dependent variable(s), main
explanatory variable(s), data source (if relevant), main model
relationship(s)
5. Results.
Requirements for presentation of statistical analysis
=====================================================
Presentations should take about 5-10 minutes.
The content should be a statistical analysis based on a model. That
is, a cause and effect relationship. The dependent variable (従属変数)
should be the principal dependent variable in your thesis. The
independent variables (独立変数) can be whatever you have data for.
(Of course if you can do a regression that is a statistical model
relevant to your thesis, that's great. But that is not the main point
of this assignment. It's a preparatory exercise to use statistical
software and do basic interpretation.) The statistical technique must
be based on a scientific model (cause and effect). You may use
techniques like regression, factor analysis, and SEM. I'm going to
assume regression (eg OLS) since all of this session's presenters have
data suitable for regression.
I haven't done this assignment this way before, so I'm not sure about
how you should organize your slides. You will likely need more than
one slide for one or two of the following topics:
1. Title slide including your self-introduction, the date, and a
title describing the analysis you do.
2. Background, including your motivation (why you think this research
is important), description of statistical analysis used in
previous research (this can be very brief: "most previous research
has used time series regression that accounts for serial
correlation"), and a description of the analysis you will do.
(You do not need to explain why your analysis is better or how it
is related to prior research. We will work on that later.)
3. The theoretical model, which explains (1) the dependent variable,
its units and how it is measured (data source), (2) the principal
*explanatory* variables (the independent variables that motivate
your research), and (3) the control variables (independent variables
that are used to make your estimates more accurate, but are not
central to your research theme). If you do not yet have good data
for "your" explanatory variables, pick a couple of "interesting"
control variables and use them as "explanatory" variables. The
important content of this task is to explain why each variable
falls into group (1), (2), or (3).
4. The statistical model, which explains (1) the basic statistical
analysis (for example, regression), (2) the algorithm (for example
OLS = ordinary least squares), and (3) the data generating process
(why isn't the model exact, what causes the disturbance(s) or
error(s) in the data so that R^2 < 1, or other goodness of fit
measure indicating imperfect fit).
5. Report of parameter estimates and statistical diagnostics
(goodness of fit, parameter significance, normality,
independence). For regressions you should provide a plot of
residuals.
6. Interpretation of the results in terms of the relationship between
the dependent variable and the explanatory variable(s).
More information about the G-Zemi
mailing list