PSYCH 2999 Lecture 2: PSYC 2890 - Fundamental of Psychology - Test 2
29 Apr 2018
School
Department
Course
Professor
Fundamental of Psychology Test 2
TEST QUESTIONS
1/31/18
Extreme groups
Regression to the mean!
Two types of statistics
• Descriptive statistics- things that tell you what the distribution looks like; where the
distribution falls; mean, median, mode, and how spread out the numbers are
• Inferential statistics- tells you if there is a statistical difference between two things on
hateer easure ou’re looking at; hard to calculate
Which one is most important?
• Descriptive statistics because what you want to know is not necessarily if things are
different but tells you more important things
ALWAYS include direction of effect
• Should it be positively correlated or negatively correlated?
• Even if it seems like it should be obvious, it is important
o E: iagie ou’re sik ad ou go to the dotor, dotor sas the a desrie
tablets or capsules and they are extremely different- that does’t help, ou eed
to know the direction
• Do’t just sa there is a relatio, tell hat the relatio is
Tolerance for ambiguity
• The further out you go, the harder it is to make a prediction
• The further out you go, the more error you will have
o Ex: if we want to predict the eather o Ma 14, e proal are’t goig to e
that accurate
The est researh orks for soe of the people soe of the tie- Acacia Parks
• There are no introverts and extroverts- there are people who are more introverted-
most people are in the middle
• No such thing as types
• How do I feel on a scale of 1-10 RIGHT NOW= 5
• How will I feel on February 12th= 4
2/2/18
Research methods
• Psholog is ath applied to people
• Lots of other people are jupig o the psholog researh adago…
• Economics Behaioral eoois, The Big “hort
• Engineering (re: NSF, car following, etc.)
• Two very good reasons to learn research methods
• You want a good job with your psych degree.*
o But you have to have some marketable skills that distinguish you from Liberal
Arts majors. Those skills are:
▪ Critical Thinking
▪ Ability to read and interpret original Psychology research
find more resources at oneclass.com
find more resources at oneclass.com
• You might want to go to graduate school in Psych
o EVERY respectable graduate program requires graduate level research methods
• LOTS of misunderstanding of how science actually works.
• Eaple: proof
• Nood sas proof i siee. We sa eidee.
• There’s a differee. Thik aout a footall gae at half-tie…
Lots of different research methods in psychology
• Hence Psych 3010
• Everything from case studies to sures to eperiets to ig data…
• We’re goig to talk aout the gold stadard- EXPERIMENTS
Why are experiments so great?
• Because THEY tell you DIRECTLY about cause and effect.
• (the others give you important clues, but a good experiment is usually a direct test of
cause and effect)
• But i geeral, here’s ho eperiets ad Psh siee i geeral ork…
Doing the science thing...
• Theory
o An explanation using an integrated set of principles that organizes and predicts
observations
o AGAIN, theories are’t right or rog or proe or disproe …
• Hypothesis
o A testable prediction
Your research question
• Note the ord preise i the preious defiitio
• Poor hpothesis: I thik paper olor has a effet o readig speed.
• Better hpothesis: I thik people ill read faster o gree paper tha o hite paper.
• Even better: write a whole Methods section detailing EXACTLY what kind of paper, the
size, the lighting, the reading speed test, the kinds of participants etc.
Experiments
o Chage soethig the independent variable
o Cotrol other fators usig controls
o Osere the effet o soethig the dependent variable- always a measurement
o *can tell you cause & effect!
Experiment results
• 3 things can happen in an experiment:
• It turs out eatl as ou predited. our hpothesis as supported- learn
something
• NO effect was observed- B did’t hage or did’t hage i the a ou predited. the
null hypothesis as supported- learn something
• B (the effect) DID happen as you predicted, but it changed for other reasons besides A.
a alternate hypothesis as operatig- DON’T lear athig
• You haged /ipleeted aipulated ariale A, ut othig happeed to B…
• Does absolutely nothing usually happen?
find more resources at oneclass.com
find more resources at oneclass.com
• So, the unusual question is: Did B change ENOUGH to rule out random chance?
• Or to put it aother a…
• Ca e rejet the ull hpothesis?
• In other words, did something really happen (to variable B) in our experiment or was it
just luck, lack of reliability in the measurement, or other sources of random error?
• Note that this random chance/random error stuff is the #1 threat to your experiment.
• The fix for the #1 threat:
o Inferential statistics (also known as Psych 3090)
• More o that later, ut first let’s talk aout the other threats to your experiment.
• Remember that possibility #3 was: B actually does change but for some reason BESIDES
A.
• BTW, this possiilit is also ko as: false positive or type I error
• BTW, there is something called Type II error
• This is a special circumstance in which your hypothesis was right, but for some reason*
our eperiet did’t sho it.
• *the usual reason is your sample size was too small. Sometimes its other things (like
reliability and validity in your measurement of B).
• Okay, back to Type I error:
o Your experiment LOOKED like it worked but actually some OTHER variable
(besides A) caused the effect on variable B.
o That sterious other ariale is alled a confounding variable
o A- bad thing
o Cofoudig ariales are threats to internal validity.
o Internal validity = can I believe the results of my experiment?
o BTW, are there threats to external alidit? Yes, ut the do’t hae a speial
name.
• Controls = methods to protect against:
• ofoudig ariales
• alteratie hpotheses
• opetig hpotheses (all three mean the same thing^^^)
• Such as:
o Demand characteristics
o The placebo effect
o And the big daddy of confounds: Chance
Controls
• BTW, how do you control for the big daddy of confounds, Chance?
• Answer:
o Random assignment to conditions PLUS
o Inferential Statistics
2/5/2018
Correlations
• Another MAJOR technique in our research toolbox
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
Two types of statistics: descriptive statistics- things that tell you what the distribution looks like; where the distribution falls; mean, median, mode, and how spread out the numbers are. Inferential statistics- tells you if there is a statistical difference between two things on (cid:449)hate(cid:448)er (cid:373)easure (cid:455)ou"re looking at; hard to calculate. Which one is most important: descriptive statistics because what you want to know is not necessarily if things are different but tells you more important things. Lots of different research methods in psychology: hence psych 3010, everything from case studies to sur(cid:448)e(cid:455)s to e(cid:454)peri(cid:373)e(cid:374)ts to (cid:271)ig data , we"re goi(cid:374)g to talk a(cid:271)out the gold sta(cid:374)dard- experiments. Doing the science thing: theory, an explanation using an integrated set of principles that organizes and predicts observations, again, theories are(cid:374)"t (cid:862)right(cid:863) or (cid:862)(cid:449)ro(cid:374)g(cid:863) or (cid:862)pro(cid:448)e(cid:374)(cid:863) or (cid:862)dispro(cid:448)e(cid:374)(cid:863) , hypothesis, a testable prediction.
