Summary Exam 3 Scientific and Statistical Reasoning UvA Year 2 PDF

**Summary Exam 3 Scientific and Statistical Reasoning UvA Year 2** geschreven door lottepeerdeman www.stuvia.com Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912...

**Summary Exam 3 Scientific and Statistical Reasoning UvA Year 2** geschreven door lottepeerdeman www.stuvia.com Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen **W1.1** **ARTICLE BY FOSTER (2010) -- CAUSAL INFERENCE AND DEVELOPMENTAL PSYCHOLOGY** Fundamental problem of causal inference = Moving from association to causation ↓ It is not that causal relationships can never be established outside of random assignment, but that they cannot be inferred from associations alone -- Additional assumptions are required Current practices: 1\. Authors hold causal inference as unattainable -- One is left to wonder about the usefulness of such information 2\. Authors embrace causality, but apply tools that have limitations for performing causal inference 3\. Authors stray into causal interpretations of what are associations -- Usage of other terms to describe the relationships that are identified, such as 'predictive' Causal inference is essential to accomplishing the goals of developmental psychology: 1\. A major goal of psychology is to improve the lives of humanity 2\. Causal thinking is unavoidable -- **Sloman**: Causality is one of the fundamentally invariant relationships that humans use to make sense of the world 3\. Even if researchers can distinguish associations from causal relationships, lay readers, journalists, policymakers and other researchers generally cannot -- Bad causal inference can do real harm Two conceptual tools moving from associations to causal relationships: 1\. Directed acyclic graph (DAG): A graphical representation used in statistics and causal inference to model the relationships between variables and to illustrate the possible causal pathways among them \- *Directed edges*: Arrows in a DAG represent the direction of causal influence - *Acyclic*: There are no loops or cycles in the graph -- One cannot start at a node and follow the arrows in a closed path to return to the same node \- The DAG cannot contain bidirectional arrows implying simultaneity - Fewer linkages are preferred to the more complex 2\. Potential Outcomes Framework: Defines the causal effect as the difference between the outcomes that would be observed versus without the intervention under consideration \- Counterfactual: Something that did not actually happen but is used to assess the causal impact of an intervention Three different variables: 1\. Confounder: A variable that is associated with both the independent variable and the dependent variable in a study -- Not controlling for the confounder can create a spurious or misleading association between the independent and dependent variables and distort the true causal relationship Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen 2\. Collider: A variable that is influenced by two or more variables -- Controlling for a collider can create a spurious or misleading association between the independent and dependent variables and distort the true causal relationship 3\. Mediator: A variable that lies on the causal pathway between the independent variable and the dependent variable -- Provides insight into the underlying causal mechanisms -- Controlling for a mediator prevents information to flow from X to Y **CHAPTER BY PEARL (2018) -- CONFOUNDING AND DECONFOUNDING: OR, SLAYING THE LURKING VARIABLE** Confounding bias: Occurs when a variable influences both who is selected for the treatment and the outcome of the experiment 'Adjusting for *x* '/ 'controlling for *x* ': Taking *x* into account or adjusting for its influence when examining the relationship between other variables Two issues: 1\. Overrating the importance of adjusting for possible confounders = Controlling for many more variables than needed; even for variables that should not be controlled for 2. Underrating the importance of adjusting for possible confounders Deconfounders: The methods that are employed to address confounding and isolate the true causal relationships between variables Latin square design: A type of experimental design used in statistical research and experimental studies that is particularly useful when there are two sources of variation, and researchers want to control for both of them efficiently -- The sources of variation are often referred to as 'rows' and 'columns' Randomisation brings two benefits: 1\. Eliminates confounder bias -- Randomisation is a way of simulating the world we would like to know -- Severs every incoming link to the randomised variable, including the ones we do not know about cannot measure 2\. Enables the researcher to quantify his uncertainty -- The uncertainty comes from the randomisation procedure, which is known Do-operator: A symbolic notation used in the field of causal inference and the study of causality -- Represents an intervention or an action that sets a particular variable to a specified value in a causal model P(Y \| *do*(X = *x*)) -- Used to express, in precise mathematical language, what counterfactual interventions would look like -- Provides scientifically sound ways of determining causal effects from non-experimental studies Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen Surrogate definitions of confounding → Both wrong! 1\. Declarative → Classical epidemiological definition of confounding: 'A confounder is any variable that is correlated with both X and Y' \- *Association with X*: A confounder is a variable that is associated with the exposure or risk factor under investigation \- *Association with Y*: A confounder is also associated with the outcome or health effect being studied \- *Not on the causal pathway*: A confounder is not an intermediate variable or on the causal pathway between the exposure and outcome -- It is not a variable that is affected by the exposure and, in turn, influences the outcome, but operates independently of the exposure-outcome relationship 2\. Procedural → Non-collapsibility: 'If you suspect a confounder, try adjusting for it and try not adjusting for it; if there is a difference, it is a confounder, and you should trust the adjusted value; if there is no difference, you are off the hook' (**Hernberg**) Proxy variable: A variable that is used as a substitute when it is difficult or impossible to directly measure a certain concept of interest **Greenland & Robins** → Exchangeability: No confounding exists in a study when the treatment group is considered, one imagines what would have happened to its constituents if they had not gotten treatment, and then judges whether the outcome would be the same as for those who actually did not receive treatment -- Counterfactual framework Backdoor path: A path between an exposure variable and an outcome variable that is not a direct path M-bias: A type of bias that can occur in observational studies when there is an uncontrolled or inadequately controlled mediator on the causal pathway between the exposure and the outcome **CHAPTER BY SHADISH (2008) -- CRITICAL THINKING IN QUASI EXPERIMENTATION** **Locke**: 'A cause is that which makes any other thing begin to be, and an effect is that which had its beginning from some other thing' INUS-condition: An *insufficient* but *non-redundant* part of an *unnecessary* but *sufficient* condition (**Mackie**) (E.g.: a fire started by lighting a match) 1\. *Insufficient*: By itself not enough (in the example, other factors such as oxygen and inflammable materials are needed) 2\. *Non-redundant*: It is an important factor (in the example, lighting the match is important because without it, without it, the rest of the conditions are not sufficient for the fire 3\. *Unnecessary*: There are other causes possible (in the example, there are other factors that could cause the fire) 4\. *Sufficient*: Part of a set of required factors (in the example, lighting the match could start a fire with the other conditions present) **Hume**: 'An effect is the difference between what happened and what would have happened' → The discrepancy between reality and the counterfactual → Random assignment is the best approximation to the counterfactual that we can obtain Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen Quasi-experiment: A research design that shares similarities with both experimental and non experimental designs, which incorporates some features of experimental research, but lacks the full random assignment of participants to conditions -- Often conducted in real-world settings where random assignment may be impractical or ethically questionable 1\. In quasi-experiments the differences between treatment and controls are usually systematic, not random! 2\. Quasi-experiments improve over correlational studies by: 1\. Forcing cause to precede effect by first manipulating the presumed cause and then observing an outcome 2\. Allowing the researcher to control some of the third-variable alternative explanations Causal relationship → **Mill**: A causal relationship exists if: a\) Priority: The cause preceded the effect -- Hardest to prove, since it is often near impossible to find out which event happened first, especially in correlational studies b) Consistency: The cause is related to the effect c\) Exclusivity: There is no plausible alternative explanation for the effect other than the cause **Campbell** → Threats of (quasi-)experimental research: 1\. *History*: The influence of external events or changes over time that could affect the dependent variable (E.g.: in a study that assesses the impact of a new workplace training program on employee job satisfaction, a company merges with another company, which could influence job satisfaction too) 2\. *Maturation*: Natural changes or developments that occur within participants over time (E.g.: in a study investigating the impact of an exercise program on weight loss, changes in participants' weight may be confounded by natural processes, such as aging or hormonal changes) 3\. *Selection*: Occurs when the groups being compared in a quasi-experiment are not equivalent at the outset (E.g.: if a researcher is studying the impact of a tutoring program on academic achievement and selects participants based on their willingness to participate, there may be a self-selection bias) 4\. *Attrition*: The loss of participants over time (E.g.: in a longitudinal study participants may pass away or drop-out over time) 5\. *Instrumentation*: Changes in the measurement instruments or procedures used to assess the dependent variable (E.g.: in a study assessing employee productivity before and after the implementation of a new management strategy, there may a change in the way productivity is measured, such as a shift from subjective evaluations to objective metrics, which introduces variability that is not related to the management strategy) 6\. *Testing*: Occurs when repeated measurement of the same participants on the dependent variable influences their responses, as participants may become more competent at taking a test or experience test-fatigue 7\. *Regression to the mean*: Occurs when extreme scores on a variable tend to move closer to the mean upon retesting (E.g.: in a study evaluating the effectiveness of a counselling program for stress reduction, participants with extreme stress levels at the outset may naturally regress toward the average in subsequent measurements, giving the appearance that the counselling program had a therapeutic effect) Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen It is neither feasible nor desirable to rule out all *possible* alternative interpretations of a causal relationship; instead, only *plausible* alternatives are of concern Falsification: Prior to **Popper**, many philosophers emphasised the importance of using data to confirm hypotheses, instead of falsifying the conclusions ↓ Assumptions: 1\. The causal claim must be clear, complete and agreed upon in all its details -- When data suggest the hypothesis is wrong, theorists respond by making a small adjustment to their causal theory while maintaining that the overall theory is still correct 2\. Requires observational procedures that perfectly reflect the theory that is being tested, but observations are never that perfect -- Tests can never provide fully definitive results **LECTURE W1.1** \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **W1.2** **CHAPTER 8 (FIELD) -- CORRELATION** If two variables are related, then changes in one variable should be met with similar changes in the other variable -- When one variable deviates from its mean we would expect the other variable to deviate from its mean in a similar way Covariance: A statistical measure that quantifies the degree to which two variables change together -- The deviations of one variable is multiplied by the corresponding deviations of a second variable, and these cross-product deviations are divided by N - 1 (covariance = *^∑^*~(~ *x~i~*−*x* ~)(~ *y~i~*−*y* ~)~ ~*N*−1~) \- A positive covariance indicates that as one variable deviates from the mean, the other variable deviates in the same direction \- A negative covariance indicates that as one variable deviates from the mean, the other deviates from the mean in the opposite direction Drawback = Covariance depends upon the scales of measurement used; it is not a standardised measure ↓ Correlation coefficient/ Pearson's *r*: Standardised covariance (*r* = *^covariance^* *~multiplied\ standard\ deviations~*) -- A value that has to lie between -1 and +1 -- Requires interval or ratio data! Testing the hypothesis that the correlation is different from zero: 1\. Using *z*-scores -- Useful because the probability of a given value of *z* occurring is known if the distribution from which it comes is normal -- However, Pearson's *r* is not normally distributed, so we adjust for it 1\. Adjust *r* →*z~r~*~=~^1^~2~log*^e^*(^1+*r*^ ~1−*r*~) 2\. Adjust the SE → *^S\ E^~zr~*~=~1 √*N*−3 3\. Compute the *z*-score → *z* = *^z^r* *S E~zr~* 4\. Look up the *z*-score in the table for the normal distribution get the one-tailed probability → Multiply this value by 2 to get the two-tailed probability 5. If the value is \> 0,05 then the correlation is not significant 2\. Using *t*-statistics → *^t^~r~*~=~*r* √*N*−2 √1−*r*^2^ Confidence intervals for *r*: \- Lower bound: *z~r~*−~(~1,95*⋅ S E~zr~*) - Upper bound: *z~r~*+~(~1,95 *⋅ S E~zr~*) Need to be converted back into the correlation coefficient through ~*r*=~ⅇ^2\ *zr*^−1 ⅇ^2*zr*^+1 Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen When a confidence interval crosses zero it means that: 1\. The population value could be zero 2\. It is not certain if the true relationship is positive or negative because the population value could plausibly be a negative or positive value Coefficient of determination (*R*^2^): Squared Pearson's correlation coefficient -- A measure of the amount of variability in one variable that is shared by the other -- A value between 0 and 1 which indicates the proportion of the total variability in one variable that is shared by the other variable Spearman's correlation coefficient (*ρ*): A non-parametric measure of statistical dependence between two variables which assesses the strength and direction of the monotonic relationship between the ranks of the paired data points -- Used when data is interval or ratio -- Ranges from -1 to +1 → For ordinal data, a negative coefficient might seem contrary to what is predicted, but a low number might mean something positive Kendall's tau (*⊤*): A non-parametric measure of statistical dependence between two variables which assesses the strength and direction of the monotonic relationship between the ranks of the paired data point -- Should be used when you have a small data set with a large number of tied ranks (i.e. many scores with the same rank) → For ordinal data, a negative coefficient might seem contrary to what is predicted, but a low number might mean something positive Computing a bivariate correlation coefficient in SPSS: 1\. Check for sources of bias: \- *Linearity*: The data points should roughly follow a straight line pattern on a scatterplot \- *Normality*: The distributions of the variables being correlated should be approximately normal → P-P plot: A graphical method to assess whether a sample follows a particular distribution (in SPSS: 'Analyze' → 'Descriptive statistics' → 'P-P plot') -- If the points on the plot closely follow the diagnonal line, it suggests that the sample distribution is in agreement with the chosen distribution 2\. Conduct bivariate correlation using the dialog box accessed through 'Analyze' → 'Correlate' → 'Bivariate' 3\. Drag the variables of interest to the dialog box 4\. Choose between the three correlation coefficients (i.e. Pearson's *r*, Spearman's rho and Kendall's tau) 5\. Under 'Style', click on 'Correlation' under 'Value' 6\. Under 'Options', click on both the 'Statistics' options when Pearson's correlation is selected → The cross-product deviations are the top-half of the equation for the covariance mentioned above 7\. Under 'Bootstrapping', click on 'Perform bootstrapping' and on 'Bias corrected accelerated' → BCA: An approach to constructing bootstrap confidence intervals -- Particularly useful when dealing with skewed or non-normally distributed data Point-biserial correlation: A measure of association used when one variable is dichotomous and the other variable is continuous -- Used when one variable is a *discrete* dichotomy (such as being pregnant or not being pregnant) \- Point-biserial correlation in SPSS: Compute a Pearson's correlation (as stated above) Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen \- The sign of the coefficient is dependent on which category is assigned to which code, so all information about the direction of the relationship can be ignored Biserial correlation: A measure of association used when one variable is dichotomous and the other variable is continuous -- Used when one variable is a *continuous* dichotomy (such as passing or failing a test; some people will only just fail while others will fail by a large margin; likewise some people will scrape a pass while others will excel) -- Cannot be calculated directly in SPSS Semi-partial correlation: A statistical technique that measures the strength and direction of the linear relationship between two variables (X and Y) while controlling for the influence of a specific variable on either X or Y, but not both -- The uniquely shared variance Partial correlation: Quantifies the relationship between two variables while accounting for the effects of a third variable on both variables in the original correlation -- The unique relationship between X and Y as a function of the variance in Y left over when other variables have been considered -- Controls for the influence of one or more variables on both X and Y simultaneously \- Partial correlation in SPSS: 'Analyze' → 'Correlate' → 'Partial' \- Zero-order correlation: The Pearson correlation coefficients without adjusting for other variables \- First-order correlation: Adjusting for one variable \- Second-order correlation: Adjusting for two variables Comparing independent correlations → Two samples with different entities 1. Compute *z~r~*-scores for both correlations 2\. Compute the *z*-score of the difference (*^z^~difference~*~=~*z~r~* ~1~−*z~r~*~2~ ) √^1^ *N*~1−3+~1 *N*~2~−3 3\. Look up this value of *z* (ignoring the minus sign) in the table for normal distribution and get the one-tailed probability 4\. Double this value to get the two-tailed probability Comparing dependent correlations → The same entities 1\. Compute the *t*-statistic (*t~d~*=~(~*r~xy~*−*r~zy~* ~)~*^⋅^*√^(\ *n*−3)^ (1+*r ~xz~*) 2 (1−*r~xy~*^2^ −*r ~xz~*^2^ −*r*~2\ *y*~ ^2^+2*r~zy~ r ~xz~ r~zy~* ) 2\. Look up this value against the appropriate critical value for *t* with N - 3 degrees of freedom 3\. Double this value to get the two-tailed probability Effect sizes: \- For Pearson's *r* → Square *r* to get *R*^2^ **-** For Spearman's *ρ* → Square *ρ* to get *R~s~*^2^ → The proportion of variance in the ranks that two variables share Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **CHAPTER 9.1-9.8 (FIELD) -- THE LINEAR MODEL (REGRESSION)** Simple linear regression: A linear approach for modelling the relationship between a response and *one* predictor variables ↓ *outcome* ( *y* )=*ax*+*b*+*ε~i~* **OR** *outcome* ( *y* )=*β*~0~+*β*~1~ *x*~1\ *i*~+*ε~i~* \- Regression coefficients: \- *a* **OR** *β*~1~= S lope (i.e. the rate at which *y* changes with respect to changes in *x*) -- Rate at which the dependent variable (*y*) changes for a one-unit change in the independent variable (*x*) -- If *a* is positive, it indicates a positive relationship, and if it is negative, it indicates a negative relationship \- *b* **OR** *β*~0~ = *Y* -intercept (i.e. the value of *y* when the predictor variable is 0) -- Starting point of the line on the *y*-axis \- *ε~i~* = Error term (i.e. the discrepancy between the observed *y* and the value predicted by the linear model \- *x*~1\ *i*~ = Observed value of the independent variable Multiple regression: A statistical technique used to analyse the relationship between a response variable and *two or more* predictor variables ↓ *outcome* ( *y* )=*β*~0~+*β*~1~*x*~1*i*~+*β*~2~*x*~2\ *i*~+*ε~i~* Regression plane: A three-dimensional plane that represents the relationship between the dependent variable and two predictors Estimating the model: \- Method of least squares: The fit of a model can be assessed by looking at the deviations (= residuals) between the model and the data collected ↓ Sum of squared residuals (*S S~R~*): A gauge of how well a linear model fits the data -- If the squared differences are large, the model is not representative of the data; if the squared differences are small, the line is representative ↓ Ordinary least squares (OLS) regression: A mathematical technique for finding maxima and minima to find the slope-values that describe the model that minimises the *S S~R~* Assessing the goodness of fit: Must be assessed even though the model is the best one available, as it can well be the best of a bad bunch → The model has to be compared to a baseline to see whether it improves how well we can predict the outcome → The mean can be used as a baseline 1\. Using *R*^2^ and *r*: \- Compute *S S~T~*: The difference between the observed values and the values predicted by the mean \- Compute *S S~R~*: The differences between the observed values and what the model predicts Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen \- Compute *S S~M~* : The improvement in prediction resulting from using the linear model rather than the mean is calculated as the difference between *SS~T~* and *SS~R~* (*S S~M~* = *S S~T~* - *S S~R~*) \- If *S S~M~* is large, the linear model is very different from using the mean to predict the outcome variable \- If *S S~M~* is small, then using the linear model is little bit better than using the mean \- Compute *R*^2^: The proportion of improvement due to the model (*R*^2^ = *^S\ S^M* *S S~T~*) -- The amount of variance in the outcome explained by the model relative to how much variation there was to explain in the first place ^-\ Compute\ *r*:\ Pearson's\ correlation\ coefficient\ (*r*\ =^ √ *R*^2^) 2\. Using the *F*-statistic: \- *F* = *^M\ S^M* *M S~R~* \- *M S~M~*~=~*S S~M~* *~k~*, where *k* is the number of predictors \- *M S~R~*~=~*SS~R~* ~*N*−*k*−1~, where *N* is the number of observations and *k* is the number of predictors For a good model the numerator will be bigger than the denominator, resulting in a large *F*-statistic \- *F* = ^(\ *N*−*k*−1)\ *R*2^ *k* (1−*R*^2^) Assessing individual predictors: \- If a variable significantly predicts an outcome, it should have a regression coefficient that is different from zero → Tested using a *t*-statistic that tests the null hypothesis that the value of the slope is 0 \- *t* = *^b^*~0\ *bserved*~−*b~expected~* *S E~b~*, where *^b^~expected~* is the value that would be expected if the null hypothesis were true (i.e. it is 0) Biases in linear models: \- Outliers: Cases that differ substantially from the main trend in the data - Convert the cases that differ substantially from the main trend in the data to standardised residuals (= the residuals converted to *z*-scores) 1\. Standardised residuals \> 3,29 are cause for concern because in an average sample a value this high is unlikely to occur; (2) if more than 1% of our sample cases have 2\. Standardised residuals \> 2,58 are evidence that the level of error within our model may be unacceptable 3\. If more than 5% of cases have standardised residuals \> 1,96 then the model may be a poor representation of the data \- Analysing how different the regression coefficients would be if a certain case were to be deleted \- Deleted residual: The difference between the adjusted predicted value and the original observed value -- The difference between the residuals Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen before and after removing the potential outlier → Studentised deleted residual: Deleted residual divided by the standard error \- Cook's distance: A statistical measure used to identify influential observations in regression analysis -- A larger Cook's distance indicates that removing that observation would have a more substantial impact on the model\'s coefficients -- Values \> 1 may be cause for concern \- Leverage values: Measure how much an individual data point can influence the estimated coefficients of the regression model \- Mahalanobis distance: A measure of how far an observation is from the centre of a multivariate distribution -- Used to identify outliers or observations that deviate from the expected patterns across multiple variables \- How the estimates of *b* change as a result of excluding a case \- Covariance ratio (CVR): Quantifies the degree to which a case influences the variance of the regression parameters \- Generalisability -- For a linear model to generalise the underlying assumptions must be met, and to test whether the model does generalise it must be cross-validated - Assumptions: \- *Additivity and linearity*: The outcome variable should be linearly related to any predictors, and, with several predictors, their combined effect is best described by adding their effects together \- *Independent errors*: For two observations the residual terms should be uncorrelated → Durbin-Watson test \- *Homoscedasticity*: The residuals at each level of the predictor(s) should have the same variance *- Normally distributed errors*: The differences between the predicted and observed data are most frequently (close to) zero \- Cross-validation: A statistical technique used to assess the performance and generalisation ability of a predictive model by diving a dataset into two main subsets: a training set (calibration) used to educate the model and a testing set (validation) used to evaluate the model's performance Regression analysis with one predictor in SPSS: \- 'Analyze' → 'Regression' → 'Linear' \- Define the outcome variable in 'Dependent' \- Define the predictor variable in 'Independent(s)' \- Request bootstrapped confidence intervals for the regression coefficients by clicking 'Bootstrap' and then selecting 'Perform bootstrapping' SPSS output: **-** The table 'Model Summary' provides the value of *R* and *R*^2^ **-** The table 'ANOVA' provides the various sums of squares, the degrees of freedom associated with each and the resulting mean squares and the *F*-statistic → If the value under 'Sig.' \< 0,05, then the model results in a better prediction than the mean value **-** The table 'Coefficients' provides the estimates of the model parameters → 'B (constant)' is the *y*-intercept; the other value is the slope Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen **LECTURE W1.2** \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **W1.3** **CHAPTER 6 (FIELD) -- THE BEAST OF BIAS** Bias: Systematic error or deviation from the true value in the data collection or analysis process ↓ Statistical bias: 1\. Outliers: A score very different from the rest of the data 2\. Violations of assumptions: \- Additivity and linearity: Changes in the independent variable result in a constant and consistent change in the dependent variable -- If there are several predictors then their combined effect is best described by adding their effects together \- Normality: Based on the idea that the data or the distribution of the data in the population follows a normal distribution (= Gaussian distribution/bell curve) -- The data is symmetrically distributed around a central value and extreme values are less likely \- Homogeneity of variance: Suggests that the spread or dispersion of the scores in different groups or conditions being compared is roughly the same -- The variances of the dependent variable are approximately equal across all levels of the independent variable(s) \- Independence: The values of one observation or data point are not influenced by the values of other observations -- The occurrence or measurement of one data point does not affect the occurrence or measurement of another **CHAPTER 9.9-9.11 + 9.14-9.17 (FIELD) -- THE LINEAR MODEL (REGRESSION)** Only select predictors based on a sound theoretical rationale or well-conducted past research that has demonstrated their importance! The order of the predictors → When predictors are completely uncorrelated the order of variable entry has very little effect on the parameters estimated → Predictors are rarely uncorrelated, so the method of variable entry has consequences and is, therefore, important \- Simultaneous regression: Adding all the predictors at once \- Hierarchical regression: Enter known predictors (such as from other research) into the model first in order of their importance in predicting the outcome \- Stepwise regression: Selects variables based on their individual contribution to the model's explanatory power -- Not recommended as variables might be considered bad predictors only because of what has already been put in the model \- Forward stepwise regression: Starts with an empty model and adds variables one at a time, selecting the variable that contributes the most to the model's predictive power at each step \- Backward stepwise regression: Starts with a model that includes all available predictors and removes variables one at a time, excluding the variable that contributes the least to the model's predictive power at each step Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen Assessing the improvement to the model at each stage: 2 \- *^F^~change~*=^(\ *N*−*k*^*~new~*−1) *R~change~* ^2^) *k~change~* ~(~1−*R~new~* \- Akaike information criterion (AIC): A measure of fit that penalises a model for having more variables -- If the AIC is bigger, the fit is worse; if the AIC is smaller, the fit is better Multicollinearity: When there is a strong correlation between two or more predictors -- If two predictors are perfectly correlated, then the values of *b* for each variable are interchangeable ↓ Problems with multicollinearity: 1\. As collinearity increases, so do the standard errors of the *b* coefficients -- Big standard errors for *b* coefficients mean more variability across samples, and a greater chance of (a) predictor equations that are unstable across samples too; and (b) *b* coefficients in the sample that are unrepresentative of those in the population 2\. It limits the size of *r* 3\. Multicollinearity between predictors makes it difficult to assess the individual importance of a predictor How to detect multicollinearity: \- Variance Inflation Factor (VIF): Indicates whether a predictor has a strong linear relationship with the other predictor(s) -- A high VIF suggests that a predictor variable is highly correlated with other predictors in the model \- If a VIF \> 10 then this indicates a serious problem with multicollinearity - If the average of all the VIF values \> 1 then the regression may be biased - Tolerance statistic: The reciprocal of the VIF (i.e. 1/VIF) \- A tolerance statistic close to 1 indicates low multicollinearity \- A tolerance statistic close to 0 indicates problematic multicollinearity Regression analysis with multiple predictors in SPSS: \- Look at scatterplots of the relationships between the outcome variable and the predictors \- 'Graphs' → 'Chart Builder' → 'Scatterplot Matrix' \- Drag all the variables of interest to the *x*-axis \- Check if the predictors have reasonably linear relationships with the outcome variable \- 'Analyze' → 'Regression' → 'Linear' \- Define the outcome variable in 'Dependent' \- Specify the predictor variable for the first block in 'Independent(s)' - Specify a second by clicking 'Next' and enter the new predictor(s) in 'Independent(s)' - Click on 'Statistics' and select the required statistics \- Under 'Casewise diagnostics', change the 'standard deviations' to 2 - Under 'Plots', some plots are useful for testing the assumptions → Click on 'Next' to define multiple plots Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen \- A plot of ZRESID (*y*-axis) against ZPRED (*x*-axis) is useful for testing the assumptions of independent errors, homoscedasticity and linearity \- A plot of SRESID (*y*-axis) against ZPRED (*x*-axis) will show up heteroscedasticity as well \- Under 'Save', you can calculate diagnostic variables for outliers and influential cases SPSS output: \- In the table 'Descriptives', the mean and standard deviation of each variable in the model is displayed \- The table 'Correlations' shows a correlation matrix containing the Pearson correlation coefficient between every pair of variables, the one-tailed significance of each correlation, and the number of cases contributing to each correlation \- In the table 'Model Summary', Model 1 refers to the first stage in the hierarchy with only one predictor; Model 2 refers to when all predictors are used \- The table 'ANOVA' provides the various sums of squares, the degrees of freedom associated with each and the resulting mean squares and the *F*-statistic → If the value under 'Sig.' \< 0,05, then the model results in a better prediction than the mean value \- The table 'Coefficients' shows the model parameters for all steps in the hierarchy - The first column under 'B' contains estimates for *b*-values \- The *P*-value associated with a *b*'s *t*-statistic (in the column 'Sig.') is the probability of getting a *t* at least as big as the one if the population value of *b* was zero → If 'Sig.' \< 0,05, then the predictor is a significant predictor \- The column 'Standardised Coefficients Beta' is the number of standard deviations that the outcome changes when the predictor changes by one standard deviation -- Tells the importance of each predictor; the bigger absolute value, the more important \- The VIF-values and tolerance values can be read from the column 'Collinearity Statistics' \- The table 'Casewise Diagnostics' provides information about individual cases in the dataset and their influence on the regression model \- 'Standardised residuals' → No more than 5% of cases should have absolute values above 2, no more than about 1% should have absolute values above 2,5, and any case with a value above about 3 could be an outlier For a good overview of individual cases in the dataset, click on 'Analyze' → 'Reports' → 'Case Summaries...' and add the Mahalanobis distance, Cook's distance and Centered Leverage to the 'Variables' list **LECTURE W1.3** \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **W2.1** **ARTICLE BY MAREWSKI & OLSSON (2009) -- BEYOND THE NULL RITUAL: FORMAL MODELLING OF PSYCHOLOGICAL PROCESSES** Null ritual: NHST; which pits one's unspecified hypothesis against a null hypothesis postulating a 'zero correlation' or 'no differences between two population means' rather than against a competing hypothesis Ritual: A range of attributes, including (a) repetitions of the same action, (b) fixations on special features such as numbers, (c) anxieties about punishments for rule violations, and (d) wishful thinking -- Each of these characteristics is reflected in NHST Alternatives to NHST: \- Effect size measures \- Confidence intervals \- Graphical data displays \- Bayesian hypothesis testing Moving beyond the problems of null hypothesis significance testing → Addressing the vague nature of many psychological theories and making theories more precise by casting them as formal models Model: A simplified representation of the world that aims to explain observed data -- A formal instantiation of a theory that specifies the theory's predictions in mathematical equations or computer code Algorithmic level theories: Aim to describe and understand how cognitive processes work in terms of computational algorithms (i.e. step-by-step procedures for performing specific cognitive tasks) and how information is processed to produce behaviour -- 'How' Computational level theories: Focus on describing the overall goals, functions and purposes of cognitive processes without necessarily specifying the detailed algorithms or mechanisms that implement these processes -- 'What'/'why' Four interrelated benefits of increasing the precision of theories by casting them as models: 1\. Models allow the design of strong tests of theories → Models translate theoretical ideas into precise predictions by formalising the relationships between variables -- The precision of predictions enhances the strength of tests as it allows researchers to clearly specify the expected outcomes, making it easier to discern whether the data support or challenge the theory 2\. Models can sharpen research questions → If theories are underspecified, they can be used to 'explain' almost any observed empirical pattern -- One-word observations/equivocation: Labels that are broad in meaning and thus provide little specification of the underlying mechanisms -- Only when one starts modelling, one learns what a theory really predicts, and what it cannot account for 3\. Models can lead beyond theories built on the general linear model → NHST is only suited for the evaluation of simple hypotheses, but many relationships are more complex -- Advanced modelling techniques allow researchers to capture non-linear patterns in data, which is particularly relevant when studying complex cognitive processes or behaviours that may not adhere to linear relationships 4\. Modelling helps to address real-world problems → Modelling allows researchers to deal with natural confounds without destroying them; they can be built into the models Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen Two approaches to modelling: 1\. Task-specific modelling → The primary goal of this approach is to capture and understand the aspects of a specific, isolated task or phenomenon 2\. Broader perspective → Models are not only intended to explain a specific task or phenomenon but are integrated into a larger cognitive architecture that formally specifies the assumptions of a broader theory -- A cognitive architecture that serves as a general framework for modelling various cognitive phenomena \- Adaptive control of thought-rational (ACT-R): A cognitive architecture (i.e. a theoretical framework that models how human cognition works) that aims to explain and simulate how humans think, learn, and perform various cognitive tasks Model selection: The comparison of different models \- Psychological plausibility: Whether the computations postulated by a model are tractable in the world beyond the laboratory \- Falsifiability: Whether the model can be proven wrong or be shown to explain everything, and hence, nothing \- Parsimony: Which of competing models accounts for the data in the simplest way - Whether a model is consistent with overarching theories of cognition - Descriptive adequacy: Which of the models provides the smallest deviation from existing data -- How well a model aligns with empirical data and real-world observations \- Generalisability: The ability of a model to predict new data \- Overfitting: When a model learns the data too well, capturing noise or random fluctuations in the data instead of the underlying patterns ↓ A model's generalisability can increase positively with the model's complexity, but only to the point at which the model is complex enough to capture *systematic* variations in the data -- Beyond that point, additional complexity can result in decreases in generalisability, because then the model may also start to absorb random variations in the data Different approaches to model selection: 1\. Practical approaches: Rely on the intuition that in a comparison of models, the one that can predict unseen data better than other models should be preferred \- Cross-validation: A statistical technique used to assess the performance and generalisation ability of a predictive model by diving the dataset into two main subsets: a training set (calibration) used to educate the model and a testing set (validation) used to evaluate the model's performance \- *K* -fold cross-validation: The data is partitioned into *K* subsets, and one of these *K* subsets is successively used as a calibration set, while the remaining (*K* - 1) subsets are used as the validation set 2\. Simulation approaches: A virtual representation or model of a system is created to observe and analyse its behaviour in a controlled, artificial environment 3. Theoretical approaches: Involve the development and formulation of models based on theoretical principles, domain knowledge or conceptual frameworks \- Information criteria (such as AIC, BIC and SIC) are used to evaluate the goodness of fit of different models to the observed data -- Incorporate a penalty term for the number of parameters in a model, promoting parsimony Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **CHAPTER BY SHADISH (2008) -- EVALUATING THEORIES** Theory: A concise statement about how we believe the world to be -- Organise observations and allow researchers to make predictions ↓ Science is about testing theories Two kinds of theories: 1\. Formal theories: Very precise, deductive and mathematical -- High degree of precision and unambiguous representation -- Expressed using mathematical symbols and syntax, which provide clear and concise expressions 2\. Verbal theories: Less rigid than formal theories, as they often deal with complex phenomena that may not be easily measurable -- Expressed using natural language -- Provide a flexible and accessible means of communication, allowing for nuanced descriptions and interpretations Characteristics of successful theories: 1\. Descriptive adequacy: Refers to the extent to which a theory accords with data, or: how well a theory describes a particular phenomenon -- Which of the models provides the smallest deviation from existing data (E.g.: a basic model that predicts student grades solely based on the number of hours spent studying has less descriptive adequacy than a model that incorporates multiple factors, such as study hours, previous grades, attendance and engagement in class discussions) \- NHST allows for comparison of a theory with data \- Drawbacks of NHST: \- Relies on dichotomous outcomes (i.e. significant or not significant) \- It is not possible to conclude that there is no difference \- Focuses on isolated relationships between variables, neglecting the broader theoretical context -- Does not provide insights into the underlying mechanisms or processes \- Advantages of formal models: \- Formal models allow for a more fine-grained analysis instead of a binary outcome \- Formal models can represent the mechanisms and processes proposed by a theory \- Formal models can be used for simulation and prediction -- Researchers can simulate the model under different conditions and compare the simulated outcomes to observed data, providing a more comprehensive evaluation of the theory 2\. Precision and interpretability: Refers to the level of detail, accuracy and specificity in a theory -- A precise theory is provided in specific and clear language -- Would all researchers interpret the theory in exactly the same way? 3\. Coherence and consistency: Refers to the consistency of the theory with other models, as well as internal coherence \- Circularity: Constantly returning to the same point or situation 4\. Falsifiability: A theory should be testable and it must be possible to prove the theory false through empirical observation or experimentation (**Popper**) Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen 5\. Post-diction and explanation: Refers to whether the theory actually provides a genuine explanation of the data 6\. Parsimony/Occam's razor: Refers to the idea that the simplest theory that explains a phenomenon is preferred -- Encourages minimising unnecessary complexity -- Not at the cost of descriptive adequacy! 7\. Originality: Refers to whether the theory is new, instead of it being a restatement of another theory 8\. Breadth: Refers to whether the theory has a restricted reach, or is applicable to a broad range of phenomena -- Theories should be as broad as possible, while still providing a genuine explanations of phenomena 9\. Usability: Refers to whether the theory can be used in practice 10\. Rationality: Refers to the idea that claims about the mind should be reasonable in the light of evolutionary history Rescorla-Wagner model: A formal model of the circumstances under which Pavlovian conditioning occurs → *ΔV* =*α* ( *λ*−*V* ) \- *ΔV*: The change in associative strength, which indicates the strength of the association between the CS and US \- α: The learning rate of the CS -- Signifies how quickly an organism forms an association between the CS and the US \- λ: The maximum associative strength that the CS can achieve -- A theoretical upper limit on the strength of the association Difference: 1\. Prediction: Making a forecast or estimation about a future event or outcome before it occurs 2\. Post-diction: Making a judgment or assessment about an event or outcome after it has already occurred ↓ Explanations of behaviour are often not predictive, but postdictive, because one simply cannot possess all the information needed to make a precise prediction of future acts Underfitting Overfitting Theory-ladenness: The idea that observations and interpretations are influenced by the theoretical frameworks or preexisting beliefs that individuals hold Types of hypotheses: 1\. Substantive hypothesis: A hypothesis that pertains to the substantive or real-world content of a study (E.g.: 'increased levels of exercise are associated with greater weight loss') Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen 2\. Research hypothesis: A specific statement that the researcher intends to test or evaluate through empirical research -- Theoretical variables (E.g.: 'participants who engage in a high-intensity exercise program will experience a greater reduction in weight compared to those in a low-intensity program') 3\. Statistical hypothesis: A formal statement regarding a population parameter that is being tested in a statistical analysis -- Concrete, operationalised variables (E.g.: 'there is a significant difference in mean weight loss between the high-intensity exercise group and the low-intensity exercise group') **ARTICLE BY DIENES (2008) -- KARL POPPER AND DEMARCATION** Potential falsifier: Any potential observation statement that could contradict a theory (E.g.: the statement 'this swan is black' is a falsifier of the hypothesis 'all swans are white') **Popper**: One theory is more falsifiable than another if the class of potential falsifiers is larger, and theories that are more falsifiable are preferred Theories gain falsification by being: \- Simpler → Simple theories typically involve fewer assumptions, which means that there are fewer elements in the overall web of beliefs that could be responsible for a failure to match predictions with observations \- Specific (E.g.: 'A is positively correlated with B' has more falsifiers than 'A is correlated with B'; the former would constitute a better form of theory than the latter) - Universal → Broader theories make predictions that cover a wide range of phenomena, but the key is whether these predictions are specific and testable A theory that allows everything explains nothing -- The more a theory forbids, the more it says about the world -- The 'empirical content' of a theory increases with its degree of falsifiability Corroboration: The degree of empirical support or confirmation that a scientific theory receives through successful predictions or passing empirical tests Ad hoc: A revision or addition of a theory that decreases falsifiability (E.g.: an ad hoc addition of the theory 'all swans are white' would be 'all swans are white, except this one') -- Revisions and amendments should always increase falsifiability Post hoc: Attempts to save theories in ways that do not suggest new tests Two critiques of Popper's view: 1\. No theory is falsifiable after all → Duhem-Quine problem: It is impossible to experimentally test a scientific hypothesis in isolation, because an empirical test of the hypothesis requires one or more background assumptions -- The theory is like a complex web, including the main idea, the supporting assumptions and the background knowledge, and testing the theory is like untangling a knot (i.e. all the threads are connected, and it is not clear which one is causing the issue) -- Challenges the idea of easily isolating and falsifying individual components of a scientific theory due to the interconnectedness of various elements within the broader system of beliefs 2\. All theories are falsified anyway → Acknowledgment of the ongoing and self correcting nature of the scientific process, where theories are subject to continual refinement in response to empirical investigation and changing understanding **LECTURE W1.3** \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen **W2.2** **ARTICLE BY CRONBACH (1957) -- THE TWO DISCIPLES OF SCIENTIFIC PSYCHOLOGY** Discipline: A method of asking questions and of testing answers to determine whether they are sound Scientific psychology operates as two distinct disciplines 1\. Experimental discipline: Focuses on controlled experimentation to establish cause and-effect relationships -- Involves laboratory studies with high internal validity - Sacrifices external validity for internal validity, focusing on tightly controlled experiments that may not reflect the complexity of real-world situations 2. Correlational discipline: Examines relationships among variables in natural settings -- Is concerned with external validity and generalisability but may lack the experimental control of the laboratory \- Lacks the ability to make strong causal claims due to the absence of experimental control Methodological pluralism: When researchers draw on a range of methods to address different research questions → Both disciplines are necessary for a comprehensive understanding of psychological phenomena -- Psychologists should combine the strengths of both approaches to develop a more comprehensive and realistic understanding of human behaviour \- *Example*: Correlational research, particularly when applied to the study of aptitude and personality, provides valuable insights into the complexity of human behaviour -- By understanding the relationships between variables in natural settings, researchers can gain insights into how individual differences (such as aptitude and personality) may influence responses to different treatments **ARTICLE BY KIEVIT ET AL. (2013) -- SIMPSON'S PARADOX IN PSYCHOLOGICAL SCIENCE: A PRACTICAL GUIDE** Simpson's paradox: A statistical phenomenon that describes that a relationship observed in a population could be reversed within all of the subgroups that make up that population (E.g.: a higher dosage of medicine may be associated with higher recovery rates at the population level, but within subgroups, a higher dosage may actually result in lower recovery rates) -- Might have important implications in real-life! Ergodicity: When the expected value of an activity performed by a group is the same as for an individual carrying out the same action over time Preventing Simpson's paradox: \- More longitudinal data collection \- More experimental research \- Data visualisation \- Cluster analysis: A method used in data analysis to group similar things together -- Helps to identify natural groupings, or 'clusters', within a dataset, making it easier to understand patterns and relationships **LECTURE W2.2** \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **W2.3** **CHAPTER 11.1-11.2 + 11.4 (FIELD) -- MODERATION, MEDIATION AND MULTICATEGORY PREDICTORS** Mediation: A situation when the relationship between a predictor variable and an outcome variable can be explained by their relationship to a third variable (i.e. the mediator) -- Occurs if the strength of the relationship between the predictor and outcome is reduced by including the mediator -- Suggests a consistent pathway or process by which the independent variable influences the dependent variable ↓ Three different models: 1\. A linear model predicting the outcome from the predictor variable (*c*) 2. A linear model predicting the mediator from the predictor variable (*a*) 3. A linear model predicting the outcome from both the predictor variable and the mediator (*c'* and *b*) Four conditions of mediation: 1\. The predictor variable must significantly predict the outcome variable in model 1 2. The predictor variable must significantly predict the mediator in model 2 3. The mediator must significantly predict the outcome variable in model 3 4. The predictor variable must predict the outcome variable less strongly in model 3 than in model 1 \- How much 'reduction' is sufficient to infer mediation: \- Mediation would occur if the relationship between the predictor and outcome was significant (*p* \< 0.05) when looked at in isolation but not significant (*p* \> 0.05) when the mediator is included too → Criticism: 'all-or-nothing' mindset \- Sobel test: Estimates the indirect effect and its significance -- If the Sobel test is significant it means that the predictor significantly affects the outcome variable via the mediator Effect sizes for mediation: \- Indirect effect = *ab* \- Indirect effect (partially standardised) = *^ab^* *s~outcome~* \- Indirect effect (standardised) = *^ab^* *s~outcome~×s~predictor~* \- *P~M~*~=~*^ab^~c~* → Indirect effect relative to the total effect Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen \- *^R^~M~*~=~*^ab^c^\'^* → Indirect effect relative to the direct effect \- *R ~M~*^2^ =*R~Y\ ,\ M~* ^2^ −(*R~Y\ ,MX~* ^2^ −*R~Y\ ,\ X~* ^2^), where *R~Y\ ,\ M~* ^2^ is variance explained by the mediator, *R~Y\ ,\ X~* ^2^ is ^2^ is variance explained by both variance explained by the predictor, and *R~Y\ ,\ MX~* Mediation in SPSS: \- 'Analyze' → 'Regression' → 'PROCESS' \- Drag the outcome variable to the box labelled 'Y variable', the predictor variable to the box labelled 'X variable' and the mediator variable to the box labelled 'Mediator(s) M' \- Change 'Model number' to 4 \- Under 'Options', click on the desired statistics (E.g.: the Sobel test or effect sizes) SPSS output: **-** The first part of the output shows the results of the linear model between the predictor and the mediator -- If *p* \< 0,05, then the predictor significantly predicts the mediator (i.e. path *a*) **-** The second part of the output shows the results of the linear model between both the predictor and the mediator, and the outcome -- If *p* \< 0,05, then the predictor and the mediator significantly predict the outcome (i.e. paths *c'* and *b*) **-** The third part of the output shows the results of the linear model between the predictor and the outcome -- If *p* \< 0,05, then the predictor significantly predicts the outcome (i.e. path *c*) **-** The fourth part of the output displays the results for the total effect (= direct effect + indirect effect) of the predictor on the outcome, as well as the bootstrapped upper and lower limit of the confidence intervals → If the confidence interval of the indirect effect contains zero, then the mediator does not mediate the relationship between the predictor and the outcome **LECTURE W2.3** \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **W3.1** **CHAPTER 11.3 (FIELD) -- MODERATION, MEDIATION AND MULTICATEGORY PREDICTORS** Moderation: The combined effect of two or more predictor variables on an outcome -- Focuses on the conditions under which the relationship between two variables changes -- Suggests that the strength or direction of the relationship between the independent and dependent variables is contingent on the values of the moderator \- The interaction effect tells us whether moderation has occurred but the predictor and moderator have to be included for the interaction term to be valid Centring: The process of transforming a variable into deviations around a fixed point If the moderation effect is significant → Simple slopes analysis \- Zone of significance: Between these two values of the moderator the predictor does not significantly predict the outcome, whereas below the lower value and above the upper value of the moderator the predictor significantly predicts the outcome Moderation in SPSS: \- 'Analyze' → 'Regression' → 'PROCESS' \- Drag the outcome variable to the box labelled 'Y Variable', the predictor variable to the box labelled 'X Variable', and the moderator to the box labelled 'Mediator(s) M' - Change 'Model number' to 1 \- Select the 'Johnson-Neyman' test for the zone of significance SPSS output: \- The second part of the output shows the results of three models: 1\. When the value of the moderator is low 2\. At the mean value of the moderator 3\. When the value of the moderator is high \- The third part of the output will show the Johnson-Neyman test and its zone of significance **LECTURE W3.1** \- \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen **W3.2** **CHAPTER 2 (FURR & BACHARACH) -- SCALING** Concatenation operation: The operation of combining two or more strings, sequences, or sets end-to-end to create a new string, sequence or set (**Campbell**) Psychological measurement: The assignment of numerals to objects or events according to rules -- The measurement process succeeds if the numbers assigned to an attribute reflect the actual amounts of that attribute \- *Rules*: Scales of measurement → Proposed by **Stevens** Scaling: The way numerical values are assigned to psychological attributes Three numerical properties: 1\. Property of identity: The ability to reflect differences -- Categories of people - The categories must be mutually exclusive \- The categories must be exhaustive 2\. Property of order: The ability to reflect ranking 3\. Property of quantity: The ability to provide information about magnitude -- Conventional standardised units Absolute zero: Zero is the lowest value → Relative zero: Zero is the lowest value on a specific measurement Types of arbitrariness in units of measurement: 1\. Unit size 2\. Some units of measurement are not tied to any one type of object 3\. Some units of measurement can be used to measure different features of objects Four scales of measurement: 1\. Nominal (E.g.: hair colour) 2\. Ordinal (E.g.: education level) 3\. Interval (E.g.: temperature) 4\. Ratio (E.g.: weight) **ARTICLE BY LORD (1953) -- ON THE STATISTICAL TREATMENT OF FOOTBALL NUMBERS** Theory of admissible statistics: The idea that certain statistical measures are appropriate or meaningful for certain types of data (**Stevens**) -- Levels of measurement → One cannot perform a *t*-test on non-interval data! ↓ **Lord**: Introduces a thought experiment in which the use of an 'inadmissible' parametric test on nominal numbers leads to a legitimate conclusion -- Supports the view that levels of measurement should not influence choice of analysis → Measurement-statistics debate ↓ Lord shows that clever people who know what they are doing can still obtain sensible result Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **W3.3** **ARTICLE BY BLANTON & JACCARD (2006) -- ARBITRARY METRICS IN PSYCHOLOGY** Metric: The numbers that the observed measures take on when describing individuals' standings on the construct of interest \- Arbitrary in psychology: 1\. It is not known where a given score locates an individual on the underlying psychological dimension 2\. It is not known how a one-unit change on the observed score reflects the magnitude of change on the underlying dimension An individual's observed score on a response metric provides only an indirect assessment Reliability: Consistency of a measurement Validity: Accuracy of a measurement Matters of metric arbitrariness are of minor consequence for theory testing and theory development → They can be important for applied work (such as when one is trying to diagnose an individual's absolute standing on a dimension or when one wishes to gain a sense of the magnitude and importance of change) To reduce arbitrariness, test developers should build a strong empirical base that links specific test scores to meaningful events and that defines cutoff or threshold values that imply significantly heightened risks or benefits **ARTICLE BY LEBEL & PETERS (2011) -- FEARING THE FUTURE OF EMPIRICAL PSYCHOLOGY: BEM'S (2011) EVIDENCE OF PSI AS A CASE STUDY OF DEFICIENCIES IN MODAL RESEARCH PRACTICE** In article, the authors use Bem's article presenting experimental evidence for psi as a case study to highlight significant shortcomings in common research practices within empirical psychology **-** Addresses concerns related to an overemphasis on conceptual rather than close replication, insufficient scrutiny of measurement and experimental procedures, and flawed implementation of null hypothesis significance testing Deficiencies contribute to a biased interpretation of data -- Interpretation bias Advocates for a stronger emphasis on close replication, verification of measurement instruments and experimental procedures, and the use of more robust and diagnostic forms of NHST **LECTURE W3.3** \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Stuvia - Koop en Verkoop de Beste Samenvattingen **W4.1** **CHAPTER 2 (COYLE) -- INTRODUCTION TO QUALITATIVE PSYCHOLOGICAL RESEARCH** Quantitative research: Seeks to understand the causal or correlational relationship between variables through testing hypotheses \- Hypothetico-deductivism: Allows scientists to freely invent hypothetical theories to explain observed data, but requires that such hypotheses are indirectly tested by their empirical consequences → Top-down Qualitative research: Seeks to understand a phenomenon within a real-world context through the use of interviews and observation \- Epistemology: Branch of philosophy that is concerned with the theory of knowledge - Ontology: The assumptions we make about the nature of being, existence or reality - Positivism: Holds that there is a direct correspondence between our perception and the things in our world → Empiricism: Holds that our knowledge of the world arises through our sense of observation and perception \- Reflexivity: The acknowledgement by the researcher of the role played by their interpretative framework or speaking position Small q qualitative research: Uses qualitative tools and techniques but withing a hypothetico deductive framework Big Q qualitative research: The use of qualitative techniques within a qualitative paradigm which rejects notions of universal truth and emphasises contextualised understandings Nomothetic research approaches: Seek generalisable findings that uncover laws to explain objective phenomena Idiographic research approaches: Seek to examine individual cases in detail to understand an outcome Phenomenology: Methods that focus on obtaining detailed descriptions of experience as understood by those who have that experience in order to discern its essence Three ways in which we can view reality: 1\. Realism: Reality exists independent of the observer, and we can observe reality through research 2\. Critical realism: There exists a reality, but we cannot know it 3\. Relativism: Reality is dependent on the observer's observation → Social constructivism: Knowledge is not objective or absolute but is actively constructed by individuals within a social context Methodolatry: An excessive or uncritical adherence to a particular research methodology, treating it as if it were the only valid or superior approach to studying phenomena → Mixed methods approach: Using both qualitative and quantitative research methods in the same project → Pluralistic analysis **LECTURE W4.1** \- Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen?  Stuvia - Koop en Verkoop de Beste Samenvattingen **GOOD LUCK ON THE EXAM!** Gedownload door: jessescheele01 \| jessescheele01\@gmail.com Dit document is auteursrechtelijk beschermd, het verspreiden van dit document is strafbaar. € 912 per jaar extra verdienen? Powered by TCPDF (www.tcpdf.org)