SES - Statistical Tests of Differences Flashcards
What are interpreting results based on? Example?
Probability e.g. odds of winning a race.
What are interpreting results used for?
To interpret facts e.g. result of race.
Discussions form experiments can be what? Example? What must they be based on?
Black/white/shades of grey e.g. coaches discussion of the result, but must be based on evidence.
In relation to variables and stats tests of differences, there can be more than 1 what?
Independent variable.
What is a hypothesis?
A precise statement about the outcome of an experiment, based on the theory.
What 2 hypotheses does every theory have?
- ) Alternate hypothesis.
2. ) Null hypothesis.
HA? HO?
Alternate hypothesis.
Null hypothesis.
Alternate hypothesis? Null hypothesis?
Positive and according to theory.
Negative and contradicts the theory.
Experimental study?
A scientist actively manipulates/interferes - Manipulates an independent variable and measures the responses of the dependent variable.
Research process?
- ) Research question.
2. ) Hypotheses.
Statistical inference?
Process of drawing conclusions about the population based upon the sample data.
The larger the difference in results/standard deviation…
The more confident we are that they come from different populations.
P-value? What is it known as?
Describes the extent to which the observations were due to chance and systematic effects.
Known as the probability statistic.
When p = 0.10, chance %? systematic effects %?
10% chance of error.
90% systematic effects.
Critical p-value for significance and accepting the alternate hypothesis? Chance of error and systematic effect %?
Typically 0.05 in science.
5% chance of error.
95% due to systematic effect.
HO true and HO accepted?
Null hypothesis is accepted.
HO true and HA accepted?
Type I error.
HA true and HO accepted?
Type II error.
HA true and HA accepted?
Alternate hypothesis is accepted.
Type I error example?
Type II error example?
1 = Telling a man he's pregnant. 2 = Telling an obviously pregnant woman she's not pregnant.
What descriptive stats would you use for 2 samples of data in relation to stats tests of differences?
Means.
Standard deviations.
What is the objective for statistical tests of differences?
To determine whether the difference between the 2 means is large enough to reflect a real difference between the 2 populations from which the samples are drawn.
Why use a statistical test?
To help the researcher reach an objective decision about the data they have collected.
What is a problem with using statistical tests of differences?
2 different samples from the same population would give 2 slightly different means and SD’s because of the variance within that population.
Chance difference?
Differences in the means of 2 samples randomly selected from the same population.
What does it mean when T = 0 in a t-test?
There’s no difference in means.
The larger the T value, the larger the…
Difference between means.
What does a t-test tell you?
Tells you how significant the differences between groups are.
Tells you if those differences could have happened by chance.
What should you consider when constructing hypotheses?
- ) Difference or relationship?
- ) Dependent variable?
- ) Independent variable and the levels of the independent variable?
What does an independent sample t-test measure?
Difference between means of 2 samples made up of different people.
Assumptions of independent sample t-test?
- ) Interval/ratio data.
- ) Data must be normally distributed (i.e. skewness & kurtosis)
- ) Samples must be randomly selected from a population (i.e. statistical inference)
- ) Variance from each sample must be around equal.
How can you check if the variance from each sample is equal when carrying out an independent sample t-test?
If 1st SD is 2x > 2nd SD = violated.
Levene’s test?
Inferential statistic used to assess the equality of variances for a variable calculated for 2 or more groups.
Formula for a t-test?
T = X1 - X2 / SED
X1? (formula for t-test)
Mean of sample 1.
X2? (formula for t-test)
Mean of sample 2.
SED? (formula for t-test)
Standard error of the difference.
One-tailed test of difference? Example?
Investigator knows whether difference will be higher or lower than 2nd sample e.g. endurance runners’ VO2max vs sedentary subjects.
Two-tailed test of differences? Example?
Investigator does not know whether difference will be higher or lower than 2nd sample e.g. football players vs hockey players VO2max.
Before carrying out a t-test which assumptions should you check?
Interval or ratio data.
Random sampling.
Normality.
Writing result of “identifying if anticipation performance is significantly different between a group of club and a group of recreational tennis players” into a lab report example?
There was no significant difference between club players and recreational players’ anticipation scores, t(x) = x, p(>/