Descriptive Statistics Flashcards
Measures of central tendency
Denote the average, most frequent score (mean, median, mode)
Mode
The value occurring most often in a data set. Nominal data.
Distribution: mode sits as highest point in the curve
Median
Middle value of the data set; divides the distribution in half.
To calculate, order all responses, response in the exact middle of the data set is median.
Ordinal data (or other quantitative data)
Distribution: center value in set, regardless of curve
Mean
Mathematical middle of the data set
Sum of all responses divided by n- number of responses
Interval and ratio data
Distribution depends on range and skew of data- it is the most sensitive to change as it mathematically considers every score
Not appropriate to use when there are outliers because it is such a sensitive measure
What does it mean if mean, median, and mode are all the same?
symmetrical distribution
What does it mean if mean, median, and mode are different?
There will be skew.
When would median be a better indicator of central tendency than mean?
If skew is severe enough. And you may want to consider avoiding parametric stats
How many decimal places do you round to when calculating mean, median and mode?
Two
What type of data is best described by mode?
nominal
What type of data is best described by median?
ordinal, interval skewed and ratio skewed
What type of data is best described by mean?
interval normal and ratio normal
What happens to variance when you add more participants to your sample?
It mimics the population more
What are sources of variance?
- Participants/people
- Environment
- Measurement
What type of variance is error?
- Measurement
- Researcher
What are some measures of variability?
- range
- standard deviation
- variance
Range
- The simplest measure of variability
- defined by the lowest and highest values in the data set (subtract lowest from highest and the different is range)
- more common to actually report low and high values
Standard deviation
We figure out how much individual scores vary from the mean, then calculate what was the standard amount of variance from the mean. Take individual variance (each person’s deviation from the mean) and combine it to form a group variance.
What additional information do measures of variance give us beyond what measures of central tendency do?
Range and the standard amount of variation within the sample
Random error
occurs through random selection/assignment:
- characteristics don’t mimic population
- one group different than other before you start
- this is why some researchers use other types of sampling/assignment than true random
Systematic error
Mistake made on a repeated basis that affects accuracy (ex: clock consistently 5 min fast)
Errors of Omission
Individual collecting data does something wrong with or without realizing they’ve done something wrong (ie misread directions)
Errors of Commission
Individual knows better and still deviates from what they should:
- only use partial results
- uses a subtest that is not appropriate for a subgroup included in sample
Sources of Error
- Researcher
- Participants
- Measurement
- Environment
What effect does error have on power?
It reduces it.
What are skew and representational error often signs of?
Variance not found in population. One way to decrease skew and representational error is to increase your sample.
How is sample size related to power?
Power is a function of sample size:
more participants=less error; less error=more power; more participants=more power
What are ways to plan a powerful study?
- select an adequate sample size
- control for threats to research
Type I error
a false rejection of the null hypothesis (false positive)
related to our inferences: null hypothesis v. research hypothesis
Type II error
accepting the null hypothesis of no difference when it is not true (false negative)
related to our inferences: null hypothesis v. research hypothesis
What factors determine power?
effect size, the sample size, the alpha level, and the chosen power that you want (typically 80%)