SOCI209 - DESCRIPTION OF MIDTERM
The midterm for SOCI 209 has 42 multiple-choice
questions.
The main difficulty in anticipating, and
studying for, such an exam is psychological: human beings do not have built-in
internal gauges of the extent of their knowledge on any topic. So
when one introspects and tries to assess how much one knows, the internal
image is often a blank: it feels as if one knows nothing. This, however,
is an entirely spurious impression. Anyone who has been taking this
course so far knows a lot about regression models. The following
is a sample of the kind of knowledge on which the multiple-choice questions
in the midterm will bear.
-
meaning of the terms dependent and independent
variable
-
meaning of the error term epsilon
-
what are the parameters of a regression
model? (This is a bit tricky because the variance of e,
s2,
is also considered a parameter, in addition to the bk.)
-
what does the notation X ~N(m,
s2)
mean?
-
in what sense is the OLS estimator "least squares"?
-
what does BLUE mean?
-
knowing the estimated regression function and
the values of X and Y, calculate the residual
-
what does SSE measure?
-
what does MSE estimate?
-
know how to recognize a sum of squares when you
see its formula
-
what are the ranges of possible values for the
coefficient of correlation, for the r-square?
-
what patterns of the residual plot suggest non-independence
of the errors, heteroskedasticity, etc.?
-
what patterns of the residual plot suggest that
errors conform to the assumptions of the regression model?
-
know what tests or procedures can be used to
identify outliers or diagnose a pathological condition such as heteroskedasticity
or non-normality of residuals
-
given a matrix expression, determine the dimension
of the result; there are 5 questions of this type, involving matrix expressions
used in the regression model
-
what is the substantive interpretation of a regression
coefficient?
-
in a multiple regression model, what df are associated
with regression, error, and total?
-
calculate the R-square in terms of sums of squares
-
know the general properties of R2
and of Ra2
-
in a multiple regression model, knowing regression
coefficient and s.e., test the coefficient for significance at the .01,
.05 or .10; the necessary statistical calculations are provided in an Appendix
to the exam, in SYSTAT and STATA language
-
know how to conduct an F test of the existence
of a regression relation between Y and the independent variables
-
interpret the coefficient of an indicator variable
-
now what the coefficient of multiple correlation
is, and what values it can assume
-
how many indicators are needed to represent a
nominal (categorical) variable with k categories?
-
knowing the regression estimates, calculate the
response function for a specific category of a nominal variable
-
understand the meanings and properties of a polynomial
regression, and of a regression model with a interaction term
-
interpret the coefficient of an interaction term
involving an indicator variable
Last modified 3 March 2003