302

What is perhaps the single most important element in the scientific method?

Control

What is control?

- any technique/method/procedure put in place to minimize threats to the research design

- APPLIES TO ALL RESEARCH DESIGNS (exper and non- exper)

What are threats of control?

- extraneous variables and confounding variables

What is the difference between a control experiment and a experimental control?

- control experiment ( type of research design, trying to describe a specific type of experimental design/method)

- experimental control ( what other control techniques can be used to minimize threats, essentially the same as normal control def.)

what defines an experiement?

- manipulation

- random assignment

What strategies help achieve/enhance control?

- random assignment to groups

- matching

- subject as own control (w/n subject designs - non-experimental)

- manipulation checks

- instrumentation of response

- building nuisance variables into the experiment

- statistical control

What is random assignment?

- the most powerful control technique for experimental studies

- theoretically ensures equivalency between groups and conditions

- controls for BOTH KNOWN AND UNKNOWN factors

- allows us to make causal inferences that x-\>y

what are some limitations of random assignment?

- requires a large \# of participants

- can be invasive at times

- not used in every design because some do not allow it (spanking of kids leads to aggression being too invasive)

What is matching?

- procedure where participants are matched on some variables or characteristics of interest

-\> trying to match pre-existing groups on other factors (KNOWN FACTORS ONLY CAN BE USED)

- non-experimental designs

- tries to approximate a relation of pre-existing groups

What are some limitations of matching?

- not as powerful of a control as rand. assign.

- also associated with quasi-experimental designs which are weaker than non-experimental

- can only control for known factors

Random assignment and matching on the nature of equivalency

- both are trying to enhance the equivalency of groups

- trying to make experimental and control groups as close together that any observed differences can be attributed to the independent variable

(leads to within and between subject designs)

Order of most to least powerful in terms of equivalency

- random assignment

- within-subjects design

- matching

What is subject as own control?

- within-subjects design and single-subject designs

- achieves equivalency by using the same participants (100% equivalency)

- dependent/repeated measures

What are some limitations of within-subject control?

- practice effects

- irreversibly of treatment effects

- dependability of treatment effects

*order effects

*carry over or sequence effects

Practice effects

- change is not due to the treatment, its due to the exposure to the same treatment multiple times

- longitudinal studies - over time one will take the test again

- covaries with time (the shorter the time interval between tests, the greater the practice effects)

- counterbalance do condition B-\>A instead of A-\>B again

Irreversibility of treatment effects

- cant reverse the effects of time 1 from the outcomes of time 2

- cant use the within subjects design if there are irreversibility effects

- example: use paper and pencil to take test

- come back 2 weeks later and take same test on computer

- bad results because participants already learned the material \= gives inaccurate results

Order effects (dependability of treatment effects)

- when you have sequencing of treatments that affect the ratings

- the position of the condition affects the dependent variable

- 3 movies, each 2 hours being rated example

Carry over or sequence effects (dependability of treatment effects)

- when the content of the treatment carries over to influence the subsequent treatment

- (toothpaste influences orange juice taste example))

Controls for some within subjects design threats

- increase the time intervals between test/treatment for practice effects

- if can't use within-subjects switch to between-subjects for irreversibility of treatment effects

- dependability of treatment effects (reverse order of treatments, intrasubject control, intersubject control, complete (use of all possible combinations of a sequence ABC), incomplete (use of a subset of the ABC sequence))

What are manipulation checks?

- attempt to assess the efficacy of the independent variable or manipulation variable

- did the manipulation work as intended? (white noise example, did Iv result in different levels)

Why are manipulation checks important?

- based in argument that physiological processes affect the DV (white noise example on levels of stress)

What are manipulation techniques of assessment and when are they used?

- self-reports of participants after intervention has been implemented

What does one do if manipulation did not work?

- drop the data of anyone failing the manipulation check

- pilot phase data collection -\>pilot study done to see how well the manipulation check worked before actual experiment (make stronger if not already)

What is instrumentation of response? *look at slides for examples*

- using a measurement/instrumentation process that is objective, standardized, sensitive, and both reliable and valid

- self-report measures

(assumes you are aware of your state of mind & are willing to share information - if unwilling you will likely lie in the study)

- focus on the measurement process - how do we ensure these processes are the very best

Which is stronger, self-reports or fMRI brain scans to get activity reports?

- fMRI scans are stronger to measure personality, working memory, and brain nerve potentials

- are they reliable? (can be unreliable like a polygraph)

What are the strengths of self-reports?

- cheap and easy to use

Threats of self-reports?

- might not be as reliable and valid

- likert scale based

- subjective so its inferior -\> many things can go into scores that are threats

- moving to brain scans lowers the error

- has some utility, we try to improve them, not as bad as one may think

What is building a nuisance variable into the study?

- MODERATOR VARIABLES! ( a nuisance variable (another IV) that moderates or influences the relationship between the IV and DV being studied, gets built into the study as a control)

- initially a potential extraneous variable being built into the study as a 3rd variable by collecting empirical data on the variable

- example: age -\> income

- nuisance: education

- solution -\> collect data and implement as 3rd variable

empirical assumption

- determined if relationship of interest is influenced by E.V.

Limit of moderator variables?

- only controls for known factors

- cant do anything for a confounding variable (throw away study if theres one)

Differences between moderators and mediators?

- moderators (relationship between x&y is moderated by z-\> a (IV2), function of what z is doing) (cant control moderator only can test for it)

- mediators ( sits between IV and DV and IV passes through it to influence the DV - if take mediator away, the relationship of IV and DV goes away)

What is statistical control

- statistical method or any analytical procedure

- the statistical power of a study (power covaries with sample size)

- use of specialized statistical techniques to enhance control (know (conceptual) difference between correlation, regression, t-test, ANOVA, chi-square)

- increase number of items to account for measurement error - reliability of scores of measurement tools

What do all statistical controls do?

- removed the effects of 3rd variables on the relationship between X-\>Y

What is ANCOVA?

- analysis of covariance

- statistical control of a covariate by holding it constant

What is chi-square?

- has no real analog nut needs to make sure tools are valid, rely on analogs of statistical research

Partial correlation

- partials out the effects of the 3rd variables - (partial out edu. in age-\> income relationship)

- correlation should drop and the magnitude if the drop shows effect of the 3rd variable

Greater degrees of control result in?

- higher levels of internal validity (lower levels of external validity of experimental lab)

What are experimenter effects?

- includes both experimenter characteristics and expectancy