Chpt 13: Understanding Research Results: Statistical Inference (PSY302) Flashcards
Researchers rarely, if ever, study entire ___________; their findings are based on sample data.
populations
Def: Inferential stats
are used to determine whether the results match what would happen if the experiments were repeatedly conducted with multiple samples.
They can also help answer whether it can be inferred that the diff in the sample means reflects a true diff in the population means
Equivalence of groups is achieved by experimentally _________ all other variables or by randomization. The assumption is that if the groups are ________, any diffs in the DV must be due to the effect of the IV.
controlling, equivalent
The diff between any 2 groups will always never be _____.
zero
Inferential stats allow researchers to make ___________ abt the true diff in population on the basis of the sample data.
They give the __________ that the diff between means reflects random error rather than a real diff.
inferences, probability
Def: Null hypothesis
that the population means are equal and that the observed diff is due to random error. That the IV has no effect, this is always going to be the counter/opposite to the research hypothesis.
Def: The research hypothesis
the population means are no equal; that the IV had an effect.
Def: Statistical significance
when there is a low probability that the diff between the obtained sample means was due to random error.
Def: Probability
the likelihood of the occurrence of some event or outcome.
You want to specify the probability that an event will happen if there’s no diff in the population
Def: Sampling distributions
are based on the assumption that the null is true.
Def: Sample size
the total number of observations, which has an impact on determinations of stat significance. Greater size= more confidence.
To use a statistical test, you gotta:
State the null
State the research hypothesis
Find the df and alpha level
Compute the means
Compute the SD and variance
Compute z-scores and t-scores
Compute the amount of error.
Do the Spearman or Pearson r
Def: T test
to examine whether or not the 2 groups are significantly diff or not.
The t uses 2 types of data:
The mean of both the sample and population
The sampling error
Def: Degrees of freedom (df)
represents the number of scores free to vary once the means are known.
Somewhat diff critical values of t are used depending on whether the test is ___________ or has __________.
one tailed, 2 tails
Def: Single tailed tests
research hypothesis specifies the direction of diff right off the bat here.
Def: 2 tailed tests
we don’t know which direction we’re going in our data.
Def: Analysis of variance/F test
more general statistical procedure than the t test.
When a study has only a single IV with 2 groups, F and t are identical.
Analysis of variance is also used when there are more than 2 levels of an IV and when a factorial design with 2 or more IV has been used
The F stat is a ratio of 2 types of variance:
Systematic variance
Error variance
Def: Systematic variance
The deviation of the group means front eh grand mean for mean score of everyone in the population/all groups.
Def: Error variance
the deviation of the individual scores in each group form their respective group means.
The ________ the F ratio is, the more likely it is that the results are _________.
larger, significant
Another effect size estimate used when comparing 2 means is called the ________, which expresses effect size in terms of SD units.
Cohen’s d
Def: Confidence interval
an interval of values within which there is a given level of confidence where the population value lies.
Represented in bar graphs as a vertical I-shaped line bound by upper and lower limits.
The goal of statistical significance is to help testers ________ if the obtained results are ________.
decide, reliable
The chosen significant level indicates how __________ you wish to be when making the decision.
confident
Significance results are most likely when a ________ sample size is used.
larger
The decision to reject the null hypothesis is based on the _________ rather than on certainties.
The decision might not be right, errors can result from the use of ________ stats.
probabilities, inferential
Using a decision matrix, there are ___ possible decision & 2 possible truths abt the population
2
Possible decisions: _______ the null or accept it.
reject
Def: Possible truths
the null is true or the null is false
Reject the null + null is true = _________
Type I error (a)
Null is false + null rejected = _________
Power (1-B)
Null is true + accept it = __________
Correct (1-a)
Null is false + accept null = __________
Type II error (B)
3 possible factors for a Type II error
Significance (alpha) level
Sample size
Effect size
Researchers have traditionally used either an alpha of __________ significance level in the decision to reject the null.
0.05 or 0.01
The chosen alpha level specifies the probability of ____________ error if the null is rejected.
Type I or II
The significance level chosen & the consequences of a type error are determined by the use of the ________.
results
The results of a single study could be non significant even when a _________ between the variables in the population does exist.
relationship
A meaningful result is more likely to be ____________ when the significance is very low.
overlooked
Sample sizes should be _____ enough to find a real effect.
big
Research should have a reasonably large sample to rule out the _________ that the sample was too small; evidence that the variables are not related should come from __________ studies.
possibility, multiple
The _________ of a test determines the optimal sample size based on the probability of correctly rejecting the null.
power
Effect sizes & desired power are what??
Smaller effect resizes require larger samples to be significant at a 0.05 level.
A higher desired power demands a greater sample size.
Researchers usually use a power between 0.70 and 0.90 to determine the sample size
Scientists attach little importance to the results of a _______ study.
single
A rich understanding of any phenomenon comes from the results of ___________ studies investigating the same variables
numerous
Instead of inferring population values on the basis for _______ investigation, a researcher can look at the results of several studies that _______ prev investigations.
single, replicate
Def: The Pearson r correlation coefficient (chpt 13
used to describe the strength of the relationship between 2 variables when both variables have interval or ratio scale properties
A statistical significance test helps to:
Decide whether to reject the null
Conclude that the true population correlation is action greater than 0
Most data analysis is carried out with statistical _________.
software
Major statistical programs are:
SPSS
SAS
SYSTAT
R (freely available)
Many ppl do most of the simple analysis using ______________
Microsoft Excel
Program analysis steps:
Input the data, entered in COLUMNS. Think of it as a matrix with rows & columns.
Provide instructions for the analysis
Interpret the output.
The variables that we study might have nominal, _______, interval, or ______ scale properties.
ordinal, ratio
Note that the nominal scale properties have _______ discrete values like male and female.
2(+)
Interval/ratio cale properties have many _______, like reaction time or rating scallions (or continuous variables)
values