1/90
Looks like no tags are added yet.
If you get interactions and main effects in the same study
always interpet intercations first
Factorial Matrix
table that represents the combinations of different independent variables (factors) and their levels in a factorial design experiment. This matrix helps researchers visualize and analyze how these factors interact and influence the dependent variable being studied.
generate a data set that matches 2x2 a main effect of factor 1 only
temp
test 8 20
8 20
Gladue and Delaney an example of a Factorial Design
study at a bar at diffrent times they asked people to rate attractiveness

Gladue and Delaney interaction follow up
they pulled out a book and specifcally asked to rate people

How many conditions

intercetion graph
they often show a crossing path
Results of a factorial design
Main effects (the overall effect of a single IV) EX is there a diffrence between 50 and 90 degree group) and Interactions (when the effect of one varible depends on the level of another independent varible) EX

What is a factorial design
It has both between and with inn subjects. Has to have more than 1 IV. the IV and condtion are no longer synonymous. 2x3 2 is # of IV and 3 is # of conditions
what is the effect of temp

jenkins and Dallenbach (1924)-sleep and memory
2 × 4 factorial design- 2: activities between study and test (classes versus sleep) 4 retention intervals (1,2,4,8) 2 partipants live in the lab for the month RESULTS main effect of activities. Retroactive interfrence.
Types of factorial designs
All IVs between subjects
all with in subject varibles
at least one of each
Px E person by enviroment factorial (factorial design)
some manipulated, some subject BETWEEN SUBJECTS EX The bar study
Ex-post facto factorial (factorial design)
Subject varible BETWEEN SUBJECTS
Matched group factoiral (factorial design)
manipulated and matching BTEWEEN SUBJECTS
Independent groups factorial (factorial design)
Random assigment, manipulated, Between Subjects
Reverse/Block CB (counterblancing)
Test more than once per condition (WITHIN subject)
Full/Partial CB (counterbalancing)
test once per condition Within subjects
mixed PxE factorial (factorial design)
subject, at least one of each type (between and within)
Mixed factorial (factorial design)
manipulated, at least one of each type of IV (between and within) Ex spider looming study
Riskind and Maddox, 1993 spider (mixed factorial)
How do we manage our emotions under scary circumstances
self efficacy- a persons sense of competence in dealing with lifes problems. (between subjects/manipulated)
C1 looming spider C2 backward (within subjects)
REULSTS for high efficacy part. fear was rlatively low regardless of motion condition

Mixed factorial w/out CB Crum and Langer (2007)
2×2 mixd factorial info: 2 levels (informed, not informed BETWEEN) Time: 2 levels (start, 4 wks later WITHIN) DV: health (weight, BP, etc) RESULTS how you think about your daily activities matters- inncrease in health when informed
PxE designs EX anxiety
people are first sorated into high and low anxierty groups- they then meet people wwho will either talk close of far from them- A man effect of env regardless of anx level
women scientist Inzlicht and Ben-Zeev (2000)
PXE design- Lv1: gender, Lv studying sterotypes and math preformance in women IV1: gender, IV2: group composition. STUDY take math test in a room with teo men or two women. RESULTS: when females are in rooms with the same sex they have higher accuracy whereas when they are in the room withe men they preform poorlly
EXPOSURE effect example bad press
they show each persson the product ut each diffrent amounts of time (how to calculate) make the scatterplot first.)

how to calcuate regression
step 1 chnage the scores on each varible to devation scores (X-M) (( to calcuate m you add them all then devide by the sum)) (exposire and liking mean is diffrent). Step 2 then you pultiply across ex. -1.50 x -.55 =.83 (THEN you add them ann up and divide them by 4). STEP 4 saure each deviation scores from before EX -1.50 swaured is 2.25 THEN you add them again and devide by sum. STEP 5 divide the sum of the products of deviation scors from step 3 by the sum of the squares dev of x i step 5 (b=2.5/5=.5) then you have b so now for a a=My-(.5) (2.5) = -1.25 then a 2.05-1.25 =.8
a note of ccaution
th4e linear prediction rule we have learned is scale sensitive- b in particular. Another rearcher does the same but gets another score- they used a diffrent mood scale. a closer look revals they used a diffreent scalewe can resolve this standarized reg coefficent
standardied regression coefficent
if you use the formula you can see that both study when calcuated come out to the same thing. if the scores had been diffrent then there is no consistency. you would have to go bacck and maybe there was a population diffrence.

when should you report b when should you report beta
if you are talking about a applied context- you developed this reg line for willamette usig willamette student you woould use b you have speciic pople for speecifc stuents. if you want to see how high school gpa determines college gpa then you would use beta. if you are not sure then you can use both
how accurate are the prediction yu are making with your reg line
pre-portionate reduction error- compare the total error of your regression line with wrorr without your reg line. The best pred of a scire you can make without regression line is the mean.
porportion ate reduction error
tells us how much better our regression does than simply predicting the mean. SS total- SS error/ SS total. steps figure the sum of square eroor usig mean to for 2 each score - predicted score sqaure that then add that step 3 then plug that that step 1 - sep 2) step 3 figutr
pre-using our beloved mood and sleep data ex..
if reg model has 0 error
means perfect ss tot - sstot =1, or 100% reduction of error.
PRE and r2
if you know the r value then if you sqaure it then it = to PRE (r squareed is called the coffiennt of determinatio)(pre tells you how much better your regresssion model is to the normal one)
follow up to women and math study Johms, Schmader, and Martens (2005)
the study is the same except for an addiitional condition where they added a part where they teach each gender about the math sterotype.
analyzing factorial designs
N (any number) way anova. ANOVA is a hypotheses about the effects of factors A, B, C, and their interactions on the response variable y. A larger F value indicates a greater likelihood that the observed differences are not due to random chanc (f stands for fisher)

correlations
a descrptive stat to characterie the relationship between two varables. negatve corecc;ation between compassion and anxiety. postive between compassion and determination/self confidence
Galton
inveneted fingerprint idetification, invented correlation. rember his prayer study. corelaations or co-relations. finds a new way to organize the data- childrens height.

Steps for making a scatterplot
draw the axes and decide whih varible goes where. determine the rnage of values and mark them on your axes 3. mark a dot for pair of scores
a postive corelation
high scores on varible x correspond with high scores on varible y. as score on x go up, scores on y also tend to go up Ex study hous and grade on test sat and IQ. 1 is the highest.

a correlation of 0
no relationship between the variables. EX shoes size and score on methods tests
a negative relationship
high scores on variable x correspond with low scores on varible y EX hours partying and test scores

assuming linearity
pearsons r is not meaningful for data sets that havee a curvilinear relationship. Also should not be calculated for ratio or interval sales. EX arousal and preformance- the inverted U hypo. r close to 0 but this relationship is not chaos. pearsons r is not appropriate.
mis interperting r
coefficient of determination (r2)
the proportion of variability in one of the variables in a correlatio that can be accounted for by variability in the second variable. r=.50 r2= .25 25% of the variability in star trek coventio attendance is associated with nerdinness.
correlations can lead to predictons
if we know there is a correlation between sat score and collge gpa, we can make predictions about an incoming class of freshamn future ( regression analysis)
regression
regression line-the best fitting line for a scatterplot.
regressio line the best fitting line
y^ = a +(b)(X) - a is the y intercpt b is the slope x= kknown value (predictor value) Y= value you are trying to predict
2 types of problems interpting correlation
directionality and third varible probelm-
the 3rd variable problem
when teo varibles appear to be relatted to each other but there id an unknown other varible that is the real source of the link EX hair length and weight and crime and # of churches in a community
what to do when you suspect the influence of a 3rd varible
partial correlation it is a stat technique for measuring infleuce of 3rd varibale
directionaity
hallenge of determining which variable influences the other in a relationship, particularly in correlational research EX beer consumption and publication
what to do about directionality problem
cross lagged panel correelation. looks at correlations at various points in time (longitudnal design) how two variables measured across the same people, groups, or units are related over time. In a cross-lagged panel model, it often refers to whether one variable at an earlier time predicts the other at a later time, while accounting for their earlier levels 3rd VARAIBLE STILL HURTS INTERPATION
conditions for making conclusions about causality
-A and B occur together with regularity -A precedes B in time -A causing B makes sense with regard to a theory -Other explaations for their co-occurnce can be ruled out
multiple regression bivariate
relationship between 2 varibles o analyses that involve more than one random variable, allowing researchers to examine how multiple variables interact and relate to each other simultaneously. (one critetion varible and at least 2 predictor varaibles- genearlly give us more accuracy))
R-
correlation between the combined predictors and the criterion application a company might predict sales from a # of known factors
factor analysis
a large # of varibles are measured and correlated with each other, it is then determined whether groups of these varbls cluster togther to form “factors” "big 5 personality” research. Factor analysis is a statistical method for finding hidden underlying dimensions, called factors, that explain patterns of correlation among observed variables

putting correlational and experimental methods together Epley et al 2008
anthropomorphism- are people whp are more prone to anthropormorphz more lonley. Yes people generally are with a r value of .53
epley et al 20008 follow up supernatural
tak a flase persnality test and get fake feedback- told withrer will spend life alone or told will have many future satisfying social relationships- then asked to rate how much thye believe in the supernatural. Those in disconnected group reported higher levels of belief in supernatural beings. and watched movie about lonliness or togtherness and then rated pets on anthropomorized traits like thoughtliness RESULTS thosee who watch isolating movie clip rate pets higher on anthro traits
steps in calcuating r
draw scatterplot -chnage all scores to z scores- figure the cross priduct of the z score- for each person add thr products of the scores -then dividde by the # of people in the stuy

have the same sign +/+
therfore the cross product wil be postive
if one z is 0
then the cross products wioll be 0

regression making predictions

the regression constant (a)
predicted raw score on criterion varible when raw score on preictor varible is 0
raw scor regression coefficent (b)
how much the predicted criterion varible increases for every increase of 1 on the predictor varible
draw the regression line
draw and label the axes for a scatter plot 2 figure predicted value on criterion varble for a low value on predictor var.-mark point on grapj 3 repest step 2 with a high value on predictor varible 4 draw a line passing through two marks

error
diffrence between predi cted score and acutal score
the least squared error princple
The least-squares error principle chooses the model whose predictions minimize the sum of the squared differences between observed and predicted values. Squaring prevents positive and negative errors from cancelling and gives larger errors greater influence.
steps calc b
change the scores for each variable to deviation scores
▪ 2. figure the product of the deviation scores for each pair of scores
▪ 3. Add up all of the products of the deviation scores
▪ 4. Square the deviation scores of the predictor variable
▪ 5. Add the squared deviation scores for X
▪ 6. Divide the sum of the products of the deviation scores by the
sum of the squared deviations for X
steps calc a
Steps for Calculating a
▪ Multiply the regression coefficient , b, by the mean of the X
variable.
▪ Subtract the result of step 1 from the mean of variabl
Quasi experimental study
Exisit when causal conconclusion cannot be drawwn because we dont have complete control over thee variables in the stuy. we already know of some varibles and designs that will result in a quasi experimental study. There are more common quasi exp designs. EX for gov study
high internal validity - high external validity
true expeiment, quasi experirmments
correlational study
naturalitic observaitons

O x O (ovservation manipluation observation design
observation then introudced a maniplution then observation. Ex people voul if oats eating could reduce cholestra; 100 people (manp was free quacker oats and everyone eats them)
gee whiz graphs
the scale in the graph is misleading
quasi- non equaivent control groups deisgn
you are a psych hireed by a comoppmay to evelant a new work program. program is a felxible work program. company has 2 plans. 1 group (company) and the other (cpmapny). A quasi because there is random assigment. intsal program then do post test.
quasi non equaivlent control groups designs
often have to use matching (tecq to try to control for the prob for equ groups- meausre my high and low readers are spread out) which makes them espicially vulerbale to regression effects regression effects (
westinghouse study-head start
attempt to evalute the efficacy of head start programs (they isolate kids that are at risk such as social econmical) they try to give me schoool skills befire school starts. When it was new they hired consulants- the study fnins that the head start program is not super effective. it helps kids short term but not long term. critqued as being misleading because of the they matched kids. they took the highest with the lowest scores (the regression towards the mean effects made it look like it did not work the program)
d heaese et al 2015 play street
looking at a program called play strreet (akllows for more physical activituy). they shut down a city streeet and make it into a playground. suppose to give kids in the neighborhood a chnace to room free with peers. pla ystreet- exp group lives in neighborhoods next to play street, in control groups kids are kkids wo dont live near play stree. RESULTS the play street kids seems to be enganing in more excerise.
quasi wood and bootin 1990
not all sqi groups have a pre and post test. non-equivalnt control group design. eathquake in san fran- in the aftermath are people dreams influence by theis powerful exp- looking at nightmes- ask peoplee who are near acciedent then group- expeimental groups are standford, san jose, and arizon. we look at previlence on earthquake nightmares. Results- people from cal have more nightmares but we cant rule out alternative edpimations. possible that califronias are more earthquake aware.
bush adminstration quasi
bush admin had a sex education progrma that dilberatley when it came to STD protection it was absence was the only posibilty. People who partipated in the program- you had to be asked? RESULTS the program did nothing-
quasi interupted time seires data
you collect data for a big chunck of time think it as pre pre pre manp post post post. an ex is exp deisgn to test the affectivdness off drop dead days. specifcally of smoking. what % of people then make the represtive amount and lay on the ground for a long time. evaluate the eeffect of a tretament by takinf a large # of observaito nbefore the program is in effect and again many times after. do we see if the drop dead day influence people over time. RESULTS POSSIBLE outocme is first graph not efffect seconf had a little effect.

what is applied research
designed to increase our knowlege of a specific real world problem? connducted outsdir of the lab. hekps us adress real life problems. American psych has always had an appplied aspect. eaiser to get funding.
walter milies sports
researcher that pionered that you could quantify reaction time/ sports players. droped golf ball and they need to grab it. applied research
hugo munsterberg- garvard
he was one of the first promept applied psychologit..developed similation tequinnes for commpanys to hire good stem train simmmilator. fund test to find the best phone operators. He measured memoory attentioin, innteligence, exactitude, and rapidity. He was from germany- argued for german side.
modern applied resarch
major catrgories of applied psycholgy (health psych, inndustrial and orgnizational psych, forensic psych, and human factors)
health psych
how do we know when to stop eating? STUDY what kinds of ques do people decide when to stop eatinnggg. internal and external ques. if you take away visual ques (unlimmited soup) then anotherr rigged up soup when you take a sip a sip gets refledd from under. Then we asked people ho emcuhh they ate etc RESULTS they ate more soup in the refilling soup condition. there estimated consumption was undr estaimtres. PROBLEMS
james theathers (sprite)
he looks at all the means and n's. it can calcuate the mean if it is even possible. he went through the soup study and said it should be retracted.
health psych kiecolt- Glaser et a., 1996
Does psych stress ifluence vaccine effectiveness. theyrecruit older adults who are either caring for a loved one with alzhemiers (level of stress high) or not. We will give them flu shots. Are people who are stres out will they produce anti bodies that will help fight off the flu. RESULTS the caregiivers the flue shots are not as efffective.
IO psych (industrial orginzation)
sompany wide psych quesions like what influences the workers and or company. EX depresssion in the workplace. Estimates in the total cost of depression to the nation in 1990 rannge from 30-40 billion. missed work/ decreased productivity. how will IO psych deal with this. Review coirpate medical program. hwlp engage your mental health. the IO might be in charge of developing the program.
greden (2004) teted a simple education plan in combination with making it easier for atidepresants to be covered unnder hralth plans. Sig decrease in lost time, sick leave, cut diability clais, and decrease prresenteeism.
teted a simple education plan in combination with making it easier for atidepresants to be covered unnder hralth plans. Sig decrease in lost time, sick leave, cut diability clais, and decrease prresenteeism.
wong et al
formed a sries of structured inteerviews with people as they transtion to worrk feom home during the pandemic. challenges (work home interfence, inefefctive commuication, procsation, loneliness). they idenifd factors tht influnece they challgens (social support, job autonomy, monitoring, workload) RESULTS people with