Week 13 Flashcards
What are the seven phases of research?
- idea-generating
- problem-definition phase
- procedure-design phase
- observation phase
- data analysis phase
- interpretation phase
- communication phase
What are the 5 types of research?
- Naturalistic observation
- case study
- correlational research
- differential research
- experimental research
What are the 6 basic principles of research?
- respect
- beneficence
- justice
- responsibility
- competence
- propriety
What was the 2007 ethical conduct statement made called?
National Statement on Ethical Conduct in Human Research
What is it called when going against the national ethical statement on conduct?
Scientific misconduct
What are the two types of data?
- Measurement data
2. Categorical data
What is the easiest way to describe reliability?
Consistency
What is the easiest way to describe validity?
Truthfulness, accuracy
Does validity require reliability or reliability require validity?
Validity requires reliability, but reliability does not necessarily require validity.
What are the three types of validity?
Test retest, parallel forms, split half
What are four types of validity?
Predictive, concurrent, content, discriminant
What are scores called after converting raw scores to standard scores?
Z-scores
What are z scores?
Scores are distance from mean, expressed in SDs
Define standard deviation
Summarises variability of sample scores
What is standard error?
Summarises variability of statistics
What is a confident interval?
Quantifies precision of estimate of population parameter
What is the p value?
The probability that an effect as large or larger as calculated in the sample would occur if the effect in the population was exactly 0.
What p value means the effect is statistically significant?
p < .05
Which p value is mean that it is not statistically significant?
p > .05
What happens when P > .05?
We fail to reject the null. However, this does not indicate that the null is true.
What have a direct relation to p-values?
Confidence intervals
The sample statistic is statistically significantly different from:
all values outside the CI.
What does it mean if two CIs do not overlap?
That their sample statistics are significantly different.
If samples are independent groups, when are sample statistics significantly different?
If CIs overlap <25% of the total length
When are samples are not independent, can overlapping CIs be used to infer statistical significance?
No.
What type of units is the size of an effect expressed as?
Standard units.
Describe the weak, moderate and strong effect of the correlation r
Weak = .1, Moderate = .3 and strong = .5
Describe the small, medium and large difference of means when using Cohen’s d:
small = 0.2, medium =0.5 and large = 0.8
When assessing proportions in a Chi-square, how is effect size measured?
an r-type measurement, the same as correlation
weak = .1, moderate = .3 and strong = .5
What is used to assess effect size in an ANOVA eta-squared?
weak/small = .01, moderate = /06 and strong/large = .14
Name three attributes of a Type 1 error?
- false positive
- incorrectly reject the null when it is true
- probability = p-value for statistical significance (a = alpha)
What are three attributes of a type 2 error?
- false negative
- failure to reject null when it is false
- probability = beta
What type of design can be used in a Chi-square?
experimental or non-experimental
What variables are used in a chi-square?
nominal/categorical
What are two types of chi-squares?
- Chi-square goodness of fit
2. chi square test of independence
What are the data requirements for a chi-square?
expected counts have to be greater or equal to 5
How is effect size measured in a chi-square?
Using Cramer’s V
What are four variable characteristics in correlation?
- interval/ratio
- univariate normality
- bivariate normality
- linearity
What are three things to watch out for in correlation?
- spurious correlation
- truncated range
- non-linearity
In correlation, design affects what?
Internal validity.
How does design effect internal validity in correlation?
- perhaps a third/lurking variable
2. directionality
What is more general than correlation?
Regression
What are two ways that regression enables prediction?
- predictor variable
2. outcome (criterion) variable
What are the three regression statistics?
- intercept (constant)
- B coefficient (standardised is Beta)
- R-squared
Experimental research is characterised by control. What are two ways this is done?
- Manipulate IV to cause a change in DV
2. Control extraneous and confounding variables to minimise threats to internal validity
What should be used when making inferences about means?
t-tests
What are the three types of t-tests?
- one sample t-test
- independent samples t-test
- paired samples t-test
What data is needed for the DV when using a t-test?
interval/ratio DV
Can a t-test be used in both an experimental and non-experimental design?
yes
What type of IV does both a paired and independent t-test require
categorical
What are two ways that t-tests test homogeneity of variance or independent samples?
Levene’s test and Hartley’s F max test
How is effect size measured in t-tests?
Using Cohen’s d
What type of omnibus test does a between subjects design use in ANOVA?
An independent groups
What type of ANOVA omnibus test use in a within subjects design?
repeated measures
With a one way ANOVA, how many independent variables must we have?
Only one.
What does a null hypothesis mean?
That all means are equal
What is the F statistic?
Ratio of between-groups and within-groups (error) variances
What are the data requirements/assumptions when using a one-way ANOVA?
The same for t-tests
What is a family wise error rate?
Inflated type 1 error rate
What are the three multiple comparison tests(idk if right)?
- Family wise error rate
- conservative vs liberal
- bonferroni;tukey, LSD (no correction)
How is effect size measured in a one-way ANOVA?
Eta-squared
