100983 Research Methods Seminar

Quantitative Research Assessment
The data for this assignment are on vUWS: quant assessment.sav.
The hypothetical research scenario:
Dr Lang is a new Lecturer in psychology at a university in the United States. Since she
has started working at the university, she has noticed that some of her colleagues’
salaries seem to be going up quicker than others and she wants to know why. She
wants to know what they are doing so that she can also get increased pay. She has
heard rumours that academics are getting raises sheerly on the basis of the number of
publications they produce. She finds it difficult to believe that academics at her university
are getting an increase in pay based on the number of their publications. She thinks that
there are surely other factors that come into play in determining a researcher’s salary
increment.
For example, Dr Lang is aware that a gender gap in salaries exist across many
vocations. Another factor she thinks should play a role in the salary increment is how
long it has been since the staff member obtained his/her PhD. As a new academic, she
hopes that she won’t get penalised for having fewer publications than someone who has
been promoted to Professor and has worked at the university for over 30 years! Finally,
her colleagues are always talking about the impact that a given publication has, and she
knows that one indicator of impact is how often a particular publication is cited.
Dr Lang wants to know whether she can predict her colleagues’ salaries. She obtained
ethics approval to run a survey in which she asked her colleagues some demographic
questions (i.e., sex, number of years since obtaining their PhD) and she also collected
their total number of publications and citations (from the Scopus database). She also
had permission to obtain the current salaries of the participants in her survey from the
Human Resources database. Once her data collection was complete, she de-identified
the data.
She now has a dataset with the following five variables: Time (number of years since
obtaining the PhD), Pubs (total number of publications), Cites (total number of citations),
Sex (0 = female, 1 = male), and Salary (yearly salary).
2
Her overall research question is: Can psychology academics’ salaries be predicted from
the time since obtaining their PhD, total number of publications, total number of
citations, and sex? Dr Lang believes that her school is fairly egalitarian, and thus she
does not think that there are any sex differences in salary. However, she does think that
time since obtaining the PhD, and the total number of publications and citations should
make different contributions to overall salary.
Your task:
You are employed by Dr Lang to analyse her data. You decide to perform a standard
multiple regression analysis on these data in order to answer Dr Lang’s overall
research question. The dependent variable is Salary and the predictor variables are
Time, Pubs, Cites, and Sex.
Next, you decide to analyse these data in such a way as to let Dr Lang know what the
best predictor(s) of Salary are by conducting a stepwise multiple regression
analysis.
Dr Lang contacts you after you have run the standard and stepwise multiple regression
analyses to say that she received some advice from her colleague that she might want
to order her predictors in a logical way, and examine the contribution that each variable
makes to the prediction equation. Specifically, her colleague thinks that a temporal
ordering makes the most sense: demographic characteristics first, then time since PhD
second, and then number of publications and citations last. You now need to conduct a
hierarchical multiple regression analysis in which you enter Sex in the first block,
then Time in the second block, and finally Pubs and Cites in the third block.
After conducting the hierarchical multiple regression analysis, you are contacted once
again by Dr Lang, who tells you that she has given it some thought and it occurred to her
that citations are likely differ as a function of the number of publications a person has.
You decide to create an interaction term between the Pubs and Cites variables, and
include the interaction term as a 5
th predictor in another standard regression analysis.
You are to check all relevant assumptions, and then conduct the corresponding analysis
for each of the four tasks. You will draw the appropriate conclusions, and write up the
analyses in the form of a Results section of a journal article and a partial Discussion.
(Note: If you were to actually write up these results for a journal article, you would not
report all four analyses. For the purposes of this assessment, however, please write
one Results section incorporating all four analyses, and one Discussion section
interpreting all four analyses.) These Results and Discussion sections must strictly
adhere to APA format, and are expected to be as clear and concise as possible. Include
an appendix of essential SPSS output.
The Discussion section needs to provide a nontechnical description of the results,
including the status of any hypotheses and full conclusions with respect to the research
questions. After reading your report, Dr Lang (who is not statistically inclined) should
have a good understanding of what it all means and the implications for her future
salary.