Module 5 section 1 Flashcards
Levels of measurement
- Nominal
- Ordinal
- Interval
- Ratio
Nominal measurement
lowest in the hierarchy of measurements
- labelling or categorizing, classifying variables or events into categories (similar characteristics)
- no qualitative meaning
- true quantitative measurement
- no ranking occurs
Ordinal measurement
- ranks events or objects on some attribute
- assigning numbers to each category
- examples: shortest to tallest, ADL’s 1= completely dependent 2= needs another persons assistance 3= needs mechanical assistance 4= completely independent
- cannot be used for mathematical operations
Interval measurement
- involves ranking events or variables on a scale in which the intervals between the numbers are equal, the 0 value is arbitrarily set and does not have an absolute value.
- some addition and subtraction can occur
- can calculate the mean and standard deviation
- used with IQ testing
Ratio Measurement
- highest form of measurement
- equal intervals between numbers
- true 0 is on the scale, meaning there is a total absence of property at 0
- variables here are considered continuous
Classification of stats
- descriptive
- inferential
Descriptive stats
- describe and synthesize data
- includes:
- frequency distribution
- measures of central tendency
- measures of variability
Frequency distributions
- systematic listing of all the values of a variable from the lowest to the highest with the number of times each value is observed
- can be displayed in a frequency polygon
Measures of central tendency
- distribution has an average, or one number that represents the distribution of values
- this includes:
- mean
* median - mode
- mean
Mean
- sum of a set scores divided by the number of scores
- Example: 10 students scores are 55, 41, 46, 56, 45, 46, 58, 41, 50, 35. The sum of the scores is 473 and divided by 10 = 47.3
Median
- middle scores
- Examples: 1, 3, 5, 7 median would be between 3 and 5 there for it would be 4
- Examples: 2,4,6,8,10, median would
Mode
-the score that occurs the most frequently
Measures of variability
- are used to describe the dispersion or the spread of data.
- included:
- Range
- Percentile
- Standard deviation
Range
- the difference between the highest and lowest scores in the set.
Percentile
- assigns the score to specific place within the distribution .
- Example: 50th percentile means that there are 50% of cases are higher than you.
- 98th percentile means that 2% of cases are higher than you
Standard deviation
- most commonly used measure of variability
- is the average amount that each individual scare varies from the mean to the set of scores.
- the larger the standard deviation, the more variable the set of scores is.
Bivariate statistics
- allow a researcher to consider two variables together and describe the relationship between the variables
- used for correlational studies
Correlations
-tells the researcher to what extent the studies are related. Measured with a correlation coefficient
Correlation coefficient
- which is an index that describes the relationship between 2 variables
- the possible values range from -1.00 through .00 to +1.00.
- (- or +) indicate the positive or negative relationship between variables
Positive Correlation
-indicates that a high scores on one variable are paired with high scores on the other. and vise versa
Negative correlation
- indicates that low scores on one variable are paired with high scores on the other variable and vice versa
- (inverse)
Zero correlation
-when 2 variables are totally unrelated
Describing risk
- the purpose is to be able to identify the risk before and after the exposure to an intervention
- Includes:
- Absolute Risk (AR)
- Absolute risk reduction (ARR)
- Odds ratio (OR)
- number needed to treat (NNT)
Absolute risk (AR)
- simply the proportion of people who experienced an undesirable outcome in each group
Absolute risk reduction
-a comparison of the 2 risks is calculated by subtracting the AR for the exposed group from the AR of the unexposed group
Odds Ratio
-is the proportion of people with the adverse outcome relative to those without it
Number needed to be treated
-estimates how many people need to receive the intervention to prevent one undesirable outcome. Calculated by dividing 1 by ARR
Inferential Statistics
-based on the law of probability and are used to draw conclusions about the population on the bases of data obtained from the sample
-Purposes:
- to estimate the probability that the
sample accurately reflects the
population
- to test hypotheses about the population
-should be used when sample is randomly selected and the measurement scale is at the interval or ratio level
Sampling distributions
-selection of sample units by random selection
- are the most effective means of securing representative samples
-based on the assumption of random sampling from populations
-
Sampling errors
- refers to the variation in the statistical values that different samples of the population may present.
- the error will affect the statistical probability that the sample will accurately reflect the population
Parameter estimation
- estimating a population parameter
- such as a mean, a proportion or a difference in the mean of 2 groups
Confidence interval
-sample mean that establishes a range of values for the population value and the probabiliy that the population value falls within that range
Hypothesis testing
- commonly used for inferential statistics
- it enables the researchers to predict the outcome of their studies
- involves the null and research hypothesis
Null hypothesis
- states that there is no relationship between the independent and dependent variables
- example- there is no relationship between the independent and dependent variables
Research hypothesis
- aka scientific hypothesis
- is the prediction that the researcher makes about what will happening in the study
Type 1 error
- occurs when the researcher states that a relationship exists when there is none
- rejecting a null when you should accept
Type 2 error
- when the researcher states that a relationship does not exist when it does
- accepting a null when you should reject it
Level of significance
- usually expressed in terms of levels (p< 0.05) rather than an actual probability
- common levels are 0.01 and 0.05
Tests of statistical significance
-either done with parameter testing or nonparametric testing
Parameter
-refers to the occurrence of a variable in the total population
Statistic
- refers to the occurance in a smaller sample
Parametric testing
- use the sample stat to estimate the population parameter
-flexible and powerful, allow researcher to study effects on the other variable and there interaction
-Have 3 characteristics
1.focus on population parameters- require measurements at least on an interval scale
3.they involve other assumption, such as the
assumption that the variables are normally
distributed in the population
- require measurements at least on an interval scale
Nonparametric testing
- require fewer assumptions than the parametric method, because they are not based on population parameters and involve less restrictive asumptions about the shape of the distribution
- data must be measured on a nominal or ordinal scale
- most useful when data cannot be interpreted as interval level
Bivariant statistical testing
- used to analyze the relationship between the 2 variables
- Includes:
- t-tests
- analysis of variance
- chi-squared tests
- product moment correlation
Multivariate statistical analysis
- deals with 3 or more variables simultaneously