Theoretical Guidance

Question 1

Deficient quantitative research

Variable reserach and what its missing

Answer 1

• download an available dataset (i.e. Allbus) and apply fancy statistical models  u think of you’re outcome variable and what could be plausible predictors, put all these in a regression model and interpret the results
• What is missing?
  1) Correlation not enough  Theory to interpret and understand relationships between variables
  2) Bias: Personal bias - want to see a certain outcome
  3) Only based on observational data - not experimental, u can make only causal statements if u subscribe to very strong assumptions

Question 2

Causal/credibility revolution

How can we draw causal inferences?

Answer 2

• If feasible, go for an experimental or quasi-experimental design
• explain for each control variable why you include it
• find every confounder possible
• make sure to not include a mediator

Question 3

Example: Reducing Crime Through Environmental Design. Street Lighting in New York City

What was good, what not so much?

Answer 3

• Problem: Spatial Spillover - offenders realize there is greater risk so they move to a different area  violates statistical principles bc you also mess with control group
• Results confirm: very local effect (36% reduction of nighttime outdoor crime), can be seen with on-campus crimes but the further away, the less strong the effect
• Really impressive and hands-on research - but if everybody would do this - what would be missing? Innovative theory, we learn very little from this, every crime theory - to a certain degree - takes already into account the local opportunity structure

“The present methodological and statistical solutions to the replication crisis will only help ensure solid stones; they don’t help us build the house.” (Muthukrishna & Henrich, 2019) -> helps to hands-on reduce crime, but does not help in understanding crime more profoundly

Question 4

What do we need for really good research?

Answer 4

• Estimation, Data (that realls represents the phenomena we are interested in!), Identification AND theory (at least some scholars that specialize in theory and organize the empirical results)
• Questions we should ask
  1. Do the data capture the phenomena of interest?
  2. What about external validity (generalizing towards the population and other populations)
  3. What is the theoretical significance of the results?

Question 5

Kroneberg model of crime

implications + problematics

Answer 5

• Person AND environment must converge - both inducive to crime - then crime happens  research that does not include both elements = incomplete
• What are the implications of this theoretical view of crime causation? Data on both - person and environment
  1. Person: questionnaire or self-control tests
  2. Environment: another questionnaire - where do u spend your time
  3. Crime: then ask about criminal offenses
• What is still problematic? No data of the coming together of all these parts, esp. the triggering part  ideally, we would analyze situational data to capture the person-environment interaction (i.e. cheating experiments, analyze video-research)

Question 6

Map of research designs: analytical criminology

Answer 6

• Unit of analysis: Person vs place oriented  ideally both (space-time budget: where have been, with whom, what kind of behavior - relate personality of person with places  hard to measure causal effect bs observational)
• Internal vs external validity (causal effect with lab experimental designs)  hard to balance those, if artificial student brain experiment in lab, this is not related to real world aka external validity
  –> No goal standard  hopefully, bigger picture emerging from all the different kinds of research

Question 7

Example: Confiding in others
Small (2017)

What is the problem?

Answer 7

• The general social survey measure of close personal ties (aka the “core discussion network”): “From time to time, most people discuss important matters with other people. Looking back over the last six months, who are the people with whom you discussed matters important to you?”
• Small (2017) critique  Question 2: “From time to time, most people discuss important matters with other people. Looking back over the last six months, think about the last time you discussed something that was important to you. That time, what did you talk about? Two or three words are sufficient.” If yes: “Who did you talk to?”
  –> Incidental activation of ties get included, would never be mentioned in the question above + less normative pressure
• idea was just to measure strong ties of a person and therefore you used the item (discuss important matters) and it gives you those strong ties but just gives you half the story, also just random people -> theory of strong ties proved wrong, we also confide in weak ties

-> moral of the story: BE critical of the measures we use!

Question 8

How can we further develop our theories?

How to relate theories and empirical research

Answer 8

Implication analysis (Lieberson and Horvich)

• Evaluation in the social sciences commonly entails deriving implications from a theory (usually in the form of (risky) hypothesis that fail or stand) and then in turn ascertaining how closely the empirical evidence meets these implications
• implications = as testable consequences of the theory = bridge between the theory and the data (appropriateness is key factor)
• Questions for us: Are there any implications of a theory that haven’t been tested? Or maybe the hypothesis that have been tested have only been interpreted from the lens of a specific theory but also could be aligned with other competing theories - equally plausible

Question 9

Theory and implications - wrong false

Answer 9

• If the evidence fails to support a theory, it does not follow that the theory is false: implication may have been incorrect, or the data of poor quality or insufficiently big, or the bounds of the theory inadequately drawn
• On the other hand, if the evidence supports the implications of the theory: we have to consider whether another theory would have generated the same outcome
• Finally, a theory may be false, even if the results are congruent with the hypothesis (which is an appropriate implication): This is because the theory may be in error— even if in the specific case the outcome is correctly implied. These problems exist, but they should not be severe if the theory’s implication is examined under a variety of contexts. In that circumstance, it becomes highly unlikely that a “true” theory will be persistently rejected or a “false” one persistently accepted

-> It’s a multi-study process! Jury trial model of scientific research

Question 10

Siblings of theoretical building block

Answer 10

Causal mechanism - means that there could be mediators (not itself causal, just a variable we think about as in between cause and effect)

• Example: Theory of intergroup contact reducing prejudice through reduction of anxiety, does not hold under any scope conditions, one has to associate the other one as belonging to a certain group or “typical” for the group in order to reduce anxiety concerning the whole group

Same goes for scope conditions and moderators

• Example: labor shortage as a moderator that conditions whether preferences lead to discriminating practices of labor market

Question 11

Three “generations” of research (distinction in social psychology):

Answer 11

1. Establish causal effect
2. Investigate the mediating mechanisms
3. Address their scope conditions, i.e., identify moderators