Samenvatting
Summary ARMS - overview of all statistical analyses including steps and key points.
An overview of all analyses covered in the ARMS course at the UU, the 3/3 statistics course in the bachelor psychologie.
[Meer zien]Voorbeeld 2 van de 5 pagina's
Voorbeeld 2 van de 5 pagina's
In winkelwagengesponsord bericht van onze partner
Voorbeeld van de inhoud
Initial assumptions1. DV is continuous: interval or ratio
2. IV’s are continuous or
dichotomous (nominal with two
Multiple Linear Regression (MLR)
categories) Book summary
3. There is a linear relationship
SSR = Goodness of fit (small is good)
between all IV’s and the DV (an
oval shaped scatterplot) OLS = ordinary least squares = finds
4. There are no outliers the b-value with the least residuals. Interpreting
SST= how good is the mean as a model R = multiple correlation coefficient = Basic equation
Statistical assumptions for observed scores = difference correlation of Y and ŷ = how good is
between observed values and model = the prediction of Y. Yi = B0 + B1X1i + B2X2i + B3X3i + ei
1. Absence of outliers: Σ (y - 𝑦̅)2
- On X: Mahalanobis distance is
R2 x 100 = variance explained
< 10 + 2*amount of IV’s SSM = differences between the mean Comparing models by pressing next Dummy coding
- On Y: standardized residuals and model = Σ (ŷ - 𝑦̅)2 after adding each variables, will Dummy variables need to be
are not < - 3.3 or > 3.3 give you change statistics. dichotomous (2 categories).
SSR = differences between observed
- On XY: Cook’s distance < 1 R2 change = amount of variance
values and the model = Σ (y - ŷ)2 Steps:
- If theoretically possible explained extra compared to
compare analysis with and 1. Number of groups – 1
R2 = SSM/SST previous model (check sig. F
without the subject. 2. Create this number of variables
𝑒𝑥𝑝𝑙𝑎𝑖𝑛𝑒𝑑 𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 𝑀𝑆𝑀 change). 3. Choose your reference group
2. Absence of multicollinearity: F = 𝑢𝑛𝑒𝑥𝑝𝑙𝑎𝑖𝑛𝑒𝑑 𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 = 𝑀𝑆𝑅 ANOVA F-test is different and (which is most relevant?)
- Coefficients table: tolerance
values < .2 are potential Good = MSM is large and MSR is small calculated per model. 4. Give reference group a value
problems. <.1 is a problem. of 0 on all variables
SS .. → MS .. = SS/df 5. Give the other groups a value
- VIF > 10 = problem Extra
- VIF = 1/tolerance MSM = SSM/k & MSR = SSR/N-k-1 of 1 for their variable
Generalisability: adjusted R2 or use 6. Change dummy names to
- Relationship > .8 is too strong data splitting (are halves very different)
- In case of multicollinearity: N = number of observations which comparison it is
unreliable B’s, limitations of R K = number of predictors/IV’s Flat model = regression coefficient Procedure:
and no importance of B1 = 0
- Transform → recode into
individual variables is known F is also used to test the null hypothesis
Assumptions not met? Use bootstrap or different variables.
3. Homoscedasticity: see plot of R2 = 0. robust regression. - Old → new name and change
4. Normally distributed residuals: see t-test = null hypothesis that B1 = 0 → - Group ,, → 1
plot Eigenvalue > 1 will be included.
- System/user missing → system
- Plot: ZRESID = Y & ZPRED = 𝑏𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 − 𝑏𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 Casewise diagnostics: cases that need missing
X 𝑆𝐸𝑏 extra attention due to values. - All other values → 0
, -----------------------------------------------------------------------------------------------------
In SPSS
Moderation
PROCESS tool is needed
Simple slope analysis: looking at the
Moderation
effect of the moderator at different
Moderation = a moderator affects the levels (low, mean, high).
direct effect of X on Y, in strength or
Zone of significance: when looking at a
direction.
lot of values for the moderator, two Mediation
values will be most significant. The in
between bit isn’t as much influenced by Mediator: the relationship between x
the moderator. and y can be explained by this variable.
When interaction effect is significant
make a grouped scatter plot where Xi
on the x-axis and Yi on the y-axis. Put
Effects of a moderator can depend on
the moderator on set colour.
the value of that variable or be
categorial. → Check if the effect between groups Mediation
is different.
Equation: (M being the moderator) c = normal effect of X on Y.
PROCESS (model 1): In SPSS
ŷ = b0 + b1X1i + b2Mi + b3X1iMi With analyses you want to get the b-
- Options: choose generate code, values for a, b, c and c’. Use PROCESS + choose model 4
You will be testing three effects: main
mean center, p < .05 & -/+ 1 SD.
effect of X, main effect of M and Via linear models you test the Choose effect size and total effect
- What does the relationship look
interaction of X*M. significance of these effects. boxes.
like? Write out regression
B1 = effect on Y when other b’s are 0. equations, check conditional Complete mediation when c’ = 0. Output per table:
effects or visualise (put the part of
- Theoretically can be impossible. ‘data list free/’ in new syntax) Effect sizes of mediation: 1. Effect a
- Interaction variable makes it hard - Johnson-Neyman method when the 2. Effect c’ and b
to look at the other effects - Indirect effect = a*b
moderator is continuous (zone of 3. Path c (total effect model)
separately. So you centre your - Partially standardized indirect
significance). Check this box! 𝑎∗𝑏 4. All effects
variables to z-scores (= Grand effect = 𝑆𝐷𝑜𝑢𝑡𝑐𝑜𝑚𝑒
mean centering). If you have 1 categorical variable, a - Fully standardized = index of Completely standardized indirect effect
- B2 = what is the effect of b1 on Y value of 0 on that will make the mediation = when X is dichotomous.
for an average score on b2. interaction 0. If you have 2 continuous 𝑎∗𝑏
𝑥 𝑆𝐷𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑜𝑟 Partially standardized indirect effect
→ This makes low-order b’s variables the interaction scores will 𝑆𝐷𝑜𝑢𝑡𝑐𝑜𝑚𝑒
𝑎∗𝑏 𝑎∗𝑏 when you want to compare effects
interpretable. differ more. - Can be estimated by 𝑜𝑟
𝑐 𝑐′ across studies.
Use a difference variable (follow-pre) - Standard error = the SD for sample
Order = higher the more variables you
when looking at improvement. means in a distribution. Use bootstrap → look at CI’s for sig.
have, less influence on centering.
Voordelen van het kopen van samenvattingen bij Stuvia op een rij:
Verzekerd van kwaliteit door reviews
Stuvia-klanten hebben meer dan 700.000 samenvattingen beoordeeld. Zo weet je zeker dat je de beste documenten koopt!
Snel en makkelijk kopen
Je betaalt supersnel en eenmalig met iDeal, creditcard of Stuvia-tegoed voor de samenvatting. Zonder lidmaatschap.
Focus op de essentie
Samenvattingen worden geschreven voor en door anderen. Daarom zijn de samenvattingen altijd betrouwbaar en actueel. Zo kom je snel tot de kern!
Veelgestelde vragen
Wat krijg ik als ik dit document koop?
Je krijgt een PDF, die direct beschikbaar is na je aankoop. Het gekochte document is altijd, overal en oneindig toegankelijk via je profiel.
Tevredenheidsgarantie: hoe werkt dat?
Onze tevredenheidsgarantie zorgt ervoor dat je altijd een studiedocument vindt dat goed bij je past. Je vult een formulier in en onze klantenservice regelt de rest.
Van wie koop ik deze samenvatting?
Stuvia is een marktplaats, je koop dit document dus niet van ons, maar van verkoper hmldekruijff. Stuvia faciliteert de betaling aan de verkoper.
Zit ik meteen vast aan een abonnement?
Nee, je koopt alleen deze samenvatting voor €6,49. Je zit daarna nergens aan vast.
Is Stuvia te vertrouwen?
4,6 sterren op Google & Trustpilot (+1000 reviews)
Afgelopen 30 dagen zijn er 53340 samenvattingen verkocht
Opgericht in 2010, al 14 jaar dé plek om samenvattingen te kopen