Flashcards
Statistical Analysis and Design
Flashcards do conjunto (68)
- 1A correlational design is useful in... or in research where...real-world research where the variables are measures as they exist naturally
it is impossible or unethical to manipulate variables - 2Define Correlationa measure of strength and direction of the relationship (association) between two variables
- 3Define Covariancea measure of the average relationship between two variables
- 4A positive correlational relationship is one in which...as scores on one variable increase, scores on the other variable also in increase
- 5A negative correlational relationship is one in which...as scores on one variable increase, scores on the other variable decrease
- 6Define Correlational CoefficientA standard statistic of the covariance between two variables
- 7Correlational Coefficient ranges between-1 and +1
- 8A perfect correlation coefficient relationship of -1 or +1 occurs when...changes in score in one variable are related to perfectly predictable changes in scores in another vairable
- 9List some of the issues with correlationsThe third-variable problem
Direction of causality - 10Describe the third-variable problemin any correlation, causality between two variables often cannot be assumed because there may be other measured or unmeasured variables affecting the results
Whilst there may appear a relationship between variables X and Y, it may actually be the case that both are controlled by the unmeasured variable Z - 11Describe the Direction of causalityCorrelation coefficients say nothing about which variable causes the other to change
In some tests, the direction of the relationship may seem fairly obvious but the correlation does not confirm that - 12r is the symbol of...one of correlation coefficient when applied to a sample
- 13r can be any value between...-1 or +1
- 14The closer r is to +- 1 the ___ the relationship between the variablesstronger
- 15The polarity of the r-value tells you the ___ of the relationshipdirection
- 16Positive r-value =positive relationship
- 17Negative r-value =negative relationship
- 18When there is no relationship that means no...covariance
- 19Assumptions of a Pearson CorrelationData is continuous (interval or ratio)
Data should be in pairs
Absence of outliers
Linear relationship between points
Variables are normally distributed
Homoscedasticity in the data - 20Describe problems with covarianceCovariance depends upon the units on the measurement. The covariance of two variables measured in Miles might be 4.25, but if the same scores are converted to km, the covariance is 11.
- 21How to solve issues with covarainceDivide the covariance by the product of the standard deviations of both variables
- 22If your r-value is larger than the critical values at the appropriate alpha and df then your test issignificant at that alpha
- 23Define an outlierA datapoint/value that is very different from the data value that form the majority of the data set
- 24List the ways to spot outliersVisual inspection of the data
Z-scores and pre-determined cut offs - 25Describe Z-scoresZ-scores are based on the normal distribution and give your data a mean of 0 and a standard deviation of 1
Converts your data to Standard Deviation units. People use somewhere between +-2.5SD and +-3SD as cutoffs - 26If you set alpha =.05 what the means is that probability of a false positive is calculated asalpha = (1- .95)*100 = .05*100 = 5%
- 27Describe the FamilyWise Error Rate (FWER)The error rate increases for every analysis you run on your data
Increasing the number of tests you run on the same data increases the chances of a false positive - 28Describe the Bonferroni CorrectionAlpha value divided by the number of comparisons
- 29Define Linear Regressionis an analytical method that looks to model the relationship between an outcome variable and one or more explanatory predictor variables
- 30Whereas correlation was about the relationship between two variables, regression is about...predicting scores on one variable (the DV) based on other variables (the predictors)
- 31The outcome variable of a linear regression must be ____continuous
- 32The predictor variables of a linear regression can be...continuous, discrete or categorical, or a combination of these
- 33Describe a Simple Linear Regressionwhere one variable is predicted from one single predictor variable
- 34Describe a Multiple Linear Regressionwhere one variable is predicted from a combination of more than one predictor variables
- 35Define the interceptwhere the line of best fit crosses the y-axis
- 36Define the slope/regression coefficientthe change in Y associated with a one-unit increase in X
- 37Define the errora degree of error that is introduced by individual differences
- 38Regression is trying to find the model that...minimises the residuals
- 39Define the residualsthe difference between an actual and a predicted value
- 40Multiple Linear Regression extends the idea of...linearly predicting a DV from a single IV to the case where we linearly predict the DV from a combination of several unrelated IVs
- 41The slope coefficients in the multiple linear regression equation reflect...each predictors' indepedent contributions to the model based on partial correlation
- 42Variance Inflation Factors (VIF) should be all around between... Values of 5, 6 and higher....1 and 3
you may have issues - 43Define Type 1 ErrorFalse Positive
Accept something that isn't true
Publish an effect that nobody can replicate - 44Define Type 2 ErrorFalse Negative
Reject something that is there
You never publish an effect but you should have - 45Define the Grand MeanThe mean of the means of the conditions
It is the common element in all scores/measurements
Everything is measured against this - 46Define Within-groups varianceThe variability of each score from its group mean; attributed to error in measurement and individual differences
- 47Define Between-groups varianceThe variability of each group mean from the grand mean; attributed to effects of the IV, plus measurement error and individual differences
- 48The frequency distribution for F-ratio depends on...2 individual degrees of freedom
- 49In the F distribution, df1 is related to... Also known as...the number of conditions
between-groups df (or the numerator) - 50In the F distribution, df2 is related to... Also known as...the number of participants in each condition
the within-groups df or the error df (or the denominator) - 51The F distribution is not symmetrical and all the values are...positive
- 52If there is no effect of our manipulation the F-value will approach/be equal to...1
- 53In a F-distribution, as the number of participants increases...the critical values decrease
- 54Eta squared ranges between...0 and 1
- 55The closer the Eta squared value is to 1...the more your IV caused the effect
- 56The closer the Eta squared value is to 0...the more your effect is just error or individual differences
- 57Assumptions for the one-way between subjects ANOVAWe are trying to make an inference about the population form which our data comes
Parametric Data - all samples drawn from a normal distribution
Data is interval/ration
Homogeneity of variance - samples from different conditions have equal variance
Data is independent from each other - 58GES stands for...Generalised Eta Squared
- 59Generalised Eta Squared in one-way between-subjects ANOVA is...identical to partial eta squared
- 60Sphericity will not be returned unless...you have more than 2 conditions
- 61Sphericity is the comparison of...variance between pairs of conditions
- 62Problems with too many factorsincredibly complex
requires many participants
difficult to interpret - 63SS TTotal variability
- 64SS MVariance explained by the experimental manipulation
- 65SS RUnexplained variance
- 66SS AVariance explained by variable A
- 67SS BVariance explained by variable B
- 68SS AxBVariance explained by the interaction of A and B
Informações
- Descrição
- Psychology (2B) FlashCards sobre Statistical Analysis and Design, criado por Hannah Luke em 24-03-2021.
- Criado por
- Hannah Luke
- Visualizações
- 0
- Avaliação