module 4 Flashcards
choosing a research strategy
By choosing a research strategy, you adress the question of how you are going to reach the conclusion of your study through the use of data
What is survey research
A survey is a way to collect data from respondents by asking them questions. Respondents typically read the questions and record their answers themselves. Alternatively, the questions can be asked by the interviewer, with the interviewer also recording the answers
Cross-sectional surveys
Conducted at a single point in time
Longitudinal information
When researches collect data from the same respondents at regular time intervals
example of survey research link
https://www.pewresearch.org/global/2022/08/31/climate-change-remains-top-global-threat-across-19-country-survey/
Experimental research
Experimental reasearch is a research strategy in which one or more independent variables are manipulated
Manipulation
Means that the researcher creates different levels/categories of the independent variable. After manipulating the independent variable(s), the dependent variable is measured
A variable is measured when
its levels are recorded as they occur naturally. For example, variables such as EBITDA and stock price a re measured. Varaibles such as organizational trust and job stress can also be measured – researchers can devise questions to represent the various levels
A variable is manipulated when
The researchers assign study participants to the different levels of the variable. for example, the researchers may assign some participants to watch one coca cola ad and assign others to watch another coca cola ad. The participants could end up in any of the two levels because the researchers do the manipulation. That is, they assign participants to be at one level of a variable or another
Archival research
Archival research is research that capitalizes on archival data.
Archival data are data that already exist. These data were initially collected and stored for purposes other than addressing the business problem. Archival data are also referred to as secondary data.
Some research problems can be approached through more than one research strategy: Hypothetical example
You would like to test the relationship between the number of beers people drik in the evening on their hangover feeling the following day. You could send out a survey and ask people to recall their last evening in a bar. You could ask them to fill out questions on (i) the number of beers they drank that evening and (ii) their hangover feeling the following day.
Alternatively, you could address the same research problem through an experiment by inviting people to a mock party and having them drink different numbers of beer (e.g. 1 vs. 5 vs. 10) The next day, you would ask them to rate their hangover feeling.
Conditions for causality: X and Y co-occur
A change in the DV should be associated with a change in the IV
Reflected in a significant correlation
E.g. X increases and Y decreases
Conditions for causality: A logical explanation for the effect of X on Y
A theory / logical argument explains why the IV affects the DV
Pittfall 1: Using “authorities”
The CEO says so it must be true
Pittfal 2: Being selectively blind for counterexamples / results
Reporting walters (2007) but not rogers (2010)
Conditions for causality: X precedes Y in time
The cause must come before the effect
E.g. if you drink beer you have a hangover
you cant have a hangover before drinking a beer
Conditions for causality: There are no other causes (Z) that explain the co-occurrence of X and Y
No other variable should be a possible cause of the change in the DV
The four conditions for causality
X and Y co occur (correlate)
A logical explanation for the effect of X on Y is needed
X proceeds Y in time
No other cause (Z) explains the co-occurrence of X and Y
Causal research strategies
Test whether changes in one variable actually result in (cause) changes in another variable
Experimental research is a causal research strategy. In experiments, researchers actively change one or more X variables (manipulate one or more vairables) Thus, it is the researchers who assign values of the X variable(s) to the participants in the experiment.
The key to ensuring that participants differ only in terms of X (and to rule out that third factors – Z – account for the relationship between X and Y) lies in how the experiment is carried out: randomly.
Randomized experiments to the levels of the manipulated variables (in the previous example, to all four types of ads) automatically controls for all third factor Z.
Why not always run experiments
To establish causality, experiments are the golden standard because they randomly assign subjects to the levels of the manipulated variable(s), thereby ruling out the influence of third factors.
However, random assignment is often impossible in business settings. Suppose researchers would like to find out whether large companies are more profitable than small companies. Try manipulating a company into being small or large - that is impossible! Company size is a variable that cannot be manipulated by researchers, so an experiment is impossible
Correlation studies cannot state:
That the association identified is a cause-and-effect relationship, but often they are the only recourse. In such settings, you must do the best you can by controlling for those third factors that you can identify.
For exammple, in the study on the effect of ice cream sales on the number of drownings, the intern could have controlled for the third factor “temperature”. in particular, he could have estimated a linear regressions with number of drowinings as the dependent variable ice cream sales as an independent variable and temperature as a control variable.