Study Guide Test 3 PDF

Study guide test 3 1. Sampling: Definitions of terms o Population - all of the persons of interest for a particular study o Sample - smaller group of persons, from the population who actually participate in a study o Census - information gathered from an...

Study guide test 3 1. Sampling: Definitions of terms o Population - all of the persons of interest for a particular study o Sample - smaller group of persons, from the population who actually participate in a study o Census - information gathered from an entire population (ex: all persons with aphasia) o Inference - information from a sample applied to an entire population (ex: just 25 people with aphasia) o Parameters - numerical summary based on an entire population o Statistics - numerical summary based on a sample from a population Read about Sample Characteristics o Representative of the population ▪ sample is a good match to the population o Unbiased sample ▪ all members of a population have an equal opportunity of being selected o Biased sample ▪ some members of a population have an unequal opportunity, or perhaps no opportunity, of being selected o Read about Sources of bias in sampling: ▪ Failing to identify all members ▪ Using samples of convenience (only recruiting participants from a medical center nearby your university) ▪ Volunteerism (must be 100% voluntary) Biased sampling methods: 1. Convenience 2. Voluntary Unbiased sampling: 1. Stratified random 2. Multistage sampling 3. simple random sampling Read about Different types of Sampling Methods o Simple random sample – procedure in which every member of a population has equal chance of being selected as a participant ▪ Unbiased and representative o Systematic sampling – procedures in which every nth numbered person from a list is selected ▪ every 5th, 10th, or 20th person on the list will be selected o Stratified random sampling – researcher’s use population characteristics or “strata” in random sampling ▪ Strata examples: gender, age, SES, ethnicity, geographic region, urban/suburban/rural  o Cluster sampling – a random sample of predefined groups such as medical centers, classrooms, or communities ▪ Example: 10 participants with aphasia from 2 different medical settings o Multistage sampling – combines cluster sampling and simple random sampling o Purposive sampling – researchers actively recruit participants who have a predetermined characteristic ▪ looking for participants that correspond to the research question Sample size - the number of participants to include in the study Too small → could affect the validity of the conclusions obtained from group studies A sample that does not represent their population well Precision - how well a sample represents a population ○ larger sample yields greater precision Bias – source of bias in the selection process, reduces how well sample represents a population ○ larger sample does not reduce bias if your selection process is flawed Factors that determine Sample Size The size of the population of interest How variable the levels of performance are How frequent the trait is in the overall population Strategies for Sample size determination: ○ Percentage of population in sample Recruit higher percentage for small populations Population: 50 members; Sample: nearly all members Population: 100,000; Sample: 0.4% or 400 Recruit lower percentage from large populations Population: 100,000; Sample: 0.4% or 400 Increase sample size if: Behavior being measured is highly variable (ex: large group standard deviations) Behavior or trait occurs rarely in the population Difference between groups expected to be smaller  2. Data analysis – types of data ○ Levels of Measurement → many ways to measure an attribute or trait Measures range from written symbols to physiological data Example ways to “measure” a phoneme: ○ Transcribe with phonetic symbols ○ Generate an ultrasound image of articulation ○ Rate degree of distortion (mild to severe) ○ Record action potentials (ex: EMG – physiological data) ○ Measure oral/ nasal airflow during production 4 levels: Nominal – name or label an attribute or trait (no rank! just a number) 1. Assign instances to a category (the # differentiates) a. Mutually exclusive – for an attribute or trait, an instance or person fits only one category (Ex: Gender → Female: 1 and Male: 2) b. Exhaustive – every attribute or trait, an instance or person fits into a category (Ex: TD-2 and DLD- 1) Ordinal – rank ordered data; high to low relative to one another (#’s have meaningful data) 1. Compare ranks of different persons or instances 2. Info on relative position but not the amount of difference a. Coincidence – persons or instances can share ranks (ex: a tie) b. Precedence – ranks can be greater than or less than another rank (Ex: Class ranks → 1st rank: 100%, 2nd rank: 98%, 3rd rank: 92%) Interval – Info on which participants have higher or lower scores AND measure how much participants differ 1. No true zero 2. The numbers at the interval level can be added, subtracted, multiplied, or divided a. Ex: Scores on standardized tests, temperature (If the child scores 0 it doesn’t mean the child has an absence of language, it only means their skills weren’t being measured) Ratio – Info on which participants have higher or lower scores AND measure how much participants differ AND true zero 1. Manipulate numbers in all the ways of the interval level – addition, subtraction, multiplication, division 2. Compare values directly in ratios a. 40 is twice as much as 20 3. Examples: weight, intensity, duration, Electrophysiological measurements ○ Visual Representation of Data – know when to use each of these type Table – to provide participant info (demographic background, age, SES)  Charts & Graphs – are ways to visually represent numerical information (Posters/presentation) Pie Chart – nominal, categorical measures 1. if you have percentages or proportions that total to 100% or 1.0 2. Ex: data about graduate students regarding where they plan to seek employment in the future Scatterplot – for illustrating the relationship between two, or three continuous measures 1. Depicting relationships between measures → in correlation research 2. Ex: Hypothetical test scores across age 3. Graph depicts → linear relationship 4. Sometimes they might be non linear too → U shaped graph 5. As x value (age) increases → Y value (test score) increases too Column and Bar graphs – Useful for illustrating the magnitude or frequency of one or more variables 1. Depict → group differences on measures → frequency counts, percentages of occurrence, and group means 2. Column and bar graphs → similar except for orientation of the display relative to the axes Line graph – common in research presentations and reports 1. All three types are useful → for frequencies, counts, percentages, and averages 2. Depicting several values in a series 3. For depicting a special kind of nonlinear relationship called an interaction (interaction effects between 2 variables) ○ Descriptive Statistics Frequencies and Percentages – measures that convey how often phenomena occurred in a data set 1. Data → Nominal level measurement 2. Proportions are similar to percentages but expressed as fractions of one Measures of Central Tendency- measures that convey information about typical and usual responses 1. Mode – category, response, or number that occurs the most frequently a. calculation of mode uses frequency information (how often it occurs in the data) b. Bimodal distributions are those with two “most frequent” outcomes 2. Median – number that occurs at the midpoint of a distribution a. Calculation either with a spreadsheet or from a list of ordered numbers b. Abbreviation is Mdn c. Ex: 50, 55, 55, 55, 60, 65, 70, 70, 75, 80, 85 Mode = 55  Median = 65 3. Mean – common measure of central tendency often called an “average” a. Calculation → summing all scores and dividing by the number of scores b. Abbreviation is M or X c. Ex: 55, 60, 65, 70, 75, 75, 75, 80, 85, 90, 95 Median = 75 Mean = 75 d. Outlier has greater effect on the mean (brings the mean value down) Measures of Variability – measures that convey how scores spread around the midpoint of a distribution 1. Minimum and maximum scores – simplest way to convey variability is to report the minimum and maximum scores a. Ex: 50, 50, 55, 60, 65, 70, 70, 75, 75, 80, 85, 90, 95 b. Minimum score is 50 c. Maximum score is 95 2. Range and interquartile range – a. Range → the difference between the minimum and maximum scores b. Interquartile range → is the difference between scores at the 75th quartile and 25th quartile 3. Standard deviation and variance – dispersion of scores around the mean a. Commonly used with interval and ratio level measures b. Abbreviation is SD i. SD is the square root of the variance ○ Means as Estimates 1. Margin of error – calculation of a margin of error a way to acknowledge possible errors in estimation  a. Hypothetical distribution of means that would emerge if you took many samples from a population b. Calculations from different samples would yield different estimates of the population mean c. Steps to calculate margin of error: i. Obtain the standard error ii. Decide on a level of confidence iii. Obtain z (population or t (sample) score corresponding to level of confidence iv. Enter values in the formula for margin of error (Margin of Error (MOE) = t.95 * SE d. Knowing margin of error allows you to generate a confidence interval ○ Different shapes of data Distributions – frequency polygon → a special type of graph to examine the shape of data distribution To construct a frequency polygon, ○ Scores → X axis ○ Frequency of each scores → Y axis ○ Use a graph such as a scatterplot with lines ○ Visually inspect the plot → decide whether or not the shape approximates that of a normal curve Distribution shapes: 1. Normal – the dispersion of the scores around the mean is symmetrical a. half of the scores fall above the mean and half fall below the mean b. The mean and median reflect the same value c. 68% of scores to fall within 1 SD on either side if the mean for the test d. 96% of scores will fall within 2 SDs of the mean e. Outliers – scores that separate quite a bit from the other scores f. When a set of scores has outliers → the shape of the frequency polygon changes – the distribution no longer approximates a normal curve 2. Positively skewed – has an abnormally long tail that stretches in the positive direction a. Scores tend to cluster on the left side of the distribution, with fewer observations on the right b. Mean is higher than the median 3. Negatively skewed – has an abnormally long tail that stretches in the negative direction a. Scores tend to cluster on the right side of the distribution, with fewer observations on the left b. Mean is lower than the median  4. Bimodal – is a probability distribution that has two peaks (or modes). This means that there are two distinct values that occur most frequently in the dataset. a. Exam scores: If a class has two distinct groups of students, one highly prepared and the other underprepared, the exam scores might show two peaks: one for the high-performing group and one for the low-performing group. 3. Data analysis – Inferential statistics – ○ Statistical significance, p-value and its importance: Probability that the null hypothesis is true: P-value 0.05 (5 in 100) 0.01 (1 in 100) 0.001 (1 in 1000) If your statistical analysis produces a p value less than 0.05 (p