Chapter 4 Flashcards
Advantages field experiment over lab experiment:
- Ability to observe actual behaviour.
- Ability to investigate more interesting outcomes (click-through rate, purchase likelihood, etc.).
- Participants do not know they are taking part in an experiment, limiting changes of hypothesis guessing.
Advantages lab experiment over field experiment
- Full/easier control over all factors.
- Easier to replicate (copy).
- You can use filters, so that they are representative of a given market.
Reasons to bid for branded keywords:
- More control over the content of the ad.
- Having both sponsored and organic search results will lower the position of competing search results.
- The cost per click (CPC) for branded keywords is relatively low.
Difference in difference analysis
Approach that compares the changes in outcomes over time between a population enrolled in a program (treatment group) and a population that is not (control group).
This analysis is especially useful to capture the causal effect when both groups are very similar to one another.
Search Engine Advertising (SEA)
Advertisers who users the show their ads to users who are actively searching for specific keywords, and they pay a fee every time someone clicks on the ad.
Search Engine Optimisation (SEO)
The process of improving your site to increase its visibility when people search for products or services related to your business in Google, Bing, or other search engines.
Disadvantage before (during) after analysis
Over time, there might also be other changes which are not due to the experiment. E.g., the weather that influences the result instead of the expected factor.
Moreover, it can be hard to exclude everything else.
Examples dependent variables
Sales
Revenue
Traffic to website
Amount of users
Quasi experiment
The control and treatment group differ not only in terms of the experimental treatment they receive (as with the true experiment), but also in other, often unknown, ways.
The researcher must try to statistically control for as many of these differences as possible.
Natural experiment
An experiment conducted in the everyday environment of the participants, but here, the experimenter has no control over the independent variables as it occurs naturally in real life.
Lab experiment
An experiment executed under highly controlled conditions, where accurate measurements are possible.
The researcher decides where the experiment will take place, at what time, with which participants, under which circumstances, etc.
Field experiment
Experiments done in the everyday environment of the participants. The experiments still manipulate the independent variable, but in a real-life setting.
Vector Auto Regressive (VAR) model
A forecasting algorithm that can be used when two or more time series influence each other. That is, the relationship between the time series involved is bi-directional. It takes into account the patterns of time.
Why the VAR model is better than the standard regression model:
It is more ideal to see if factors are causal, because it takes into account the PATTERNS OF TIME.
Three conditions to proof causality:
- Order in time: shock in independent variable must occur before shock in dependent variable.
- Covariation: change in independent variable must lead to change in dependent variable.
- No third variable: other variables found have to be excluded.
E.g., advertising and sales both go up, both are influenced by seasonality. There can thus not be a guaranteed causality between sales and advertising.
