Chapter 4 - Sampling, Measurement, and Hypothesis Testing

Question 1

Sample

Answer 1

Subset of a larger group (population).

Question 2

Probability Sampling

Answer 2

Each member of the population has a specific probability of being selected.

Question 3

Convenience Sampling

Answer 3

Non-random samples, usually recruited from a pool of easily accessible individuals.

Question 4

Probability Sampling - Random Sampling

Answer 4

Each member of the population has an equal probability of being selected (ex. drawing names from a hat or assigning names to numbers and use a random number generator).

Question 5

Probability Sampling - Stratified Sampling

Answer 5

Uses random sampling but also ensures important subgroup proportions are maintained.

Example: Studying sex differences in a population comprised of 35% females and 65% males.
–> Randomly sample females until 35% of the sample is collected and then randomly males to complete the other 65% of the sample.

Question 6

Probability Sampling - Cluster Sampling

Answer 6

Uses predefined clusters of the population and randomly select from those clusters. Useful in you don’t have the contact info for every member of the population.

Example: Population of on-campus residents at SUNY Oswego.
–> Randomly select 3 of the 12 dorms (clusters).
–> Collect info from everyone in those clusters.

Question 7

Convenience Sampling - Purposive Sampling

Answer 7

Made up of specific types of people who are not a random sample (ex. college students, hospital inpatients, twins). Usually participants, “self-select”/volunteer for the study (can lead to biased sample).

Question 8

Convenience Sampling - Quota Sampling

Answer 8

Non-random samples that also ensure important subgroup proportions are maintained.

Example: Studying sex differences in a population comprised of 35% females and 65% males.
– > in a sample of 100, volunteers are accepted until 35 females and 65 males have been recruited.

Question 9

Convenience Sampling - Snowball Sampling

Answer 9

Non-random samples where participants are tasked with recruiting more participants. Often used in studies with small tight-knit populations (ex. college athletes or patients with a specific disorder).

Question 10

Reliability

Answer 10

When researchers are able to obtain the same results with CONSISTENCY. It’s related to MEASUREMENT ERROR.

Question 11

Test-Retest Reliability

Answer 11

Ability for the SAME experimenter to get the same results (in a test and retest).

Question 12

Interrater Reliability

Answer 12

Ability for DIFFERENT experimenters to get the same results.

Question 13

Validity

Answer 13

When researchers are ACCURATELY measuring something.

Question 14

Content Validity

Answer 14

Where test/measure appears to actually be measuring what it’s supposed to be. Usually based on wording of test questions.

Example: Anxiety
–> “Have you experienced a stressful event lately?” (not good)
–> “Does your heart race spontaneously?” (good)

Question 15

Construct & Criterion Validity

Answer 15

Measures/tests/constructs should correlate with things they’re related to and shouldn’t correlate with things they aren’t (ex. an intelligence test should predict college performance, but intelligence and depression should not be correlated).

Question 16

Effect Size

Answer 16

Informs the readers about the SIZE of a statistically significant difference.

Question 17

Descriptive Statistics

Answer 17

Information intended to describe or summarize that basic data that’s been collect.

Question 18

Measures of Central Tendency

Answer 18

Provide information about the TYPICAL SCORE in a sample (ex. mean, median, mode).

Question 19

Mean

Answer 19

The average score.

Question 20

Median

Answer 20

Middle score when numbers (data) are arranged lowest to highest.

*Useful when outliers (extreme scores) exist.

Question 21

Mode

Answer 21

The most frequent score.

Question 22

Measures of Variability

Answer 22

Provides information about the SPREAD of possible scores (ex. range, interquartile range/IQR, variance, standard deviation, histogram).

Question 23

Range

Answer 23

Difference between the highest score and the lowest score.

Question 24

Interquartile Range (IQR)

Answer 24

Differences between the 25th and 75th percentile of scores.

Question 25

Variance

Answer 25

Describes the distribution of scores relative to the mean. --> Sum each observation's squared difference from the mean and dividing the # of observations. --> Not usually reported but used in statistical analysis (ex. ANOVA, SD).

Question 26

Standard Deviation

Answer 26

Describes the average amount that scores deviate from the mean. --> Calculated by taking the square root of the variance.

Question 27

Histogram

Answer 27

Visually displays the frequency of all scores in the dataset. --> Individual scores plotted on x-axis. --> Frequency of those scores plotted on y-axis.

Question 28

Inferential Statistics

Answer 28

Allows you to take information from your sample and INFER CONCLUSIONS about the general population.

Question 29

Null Hypothesis Significance Testing (NHST)

Answer 29

Determines the probability that the results are due to chance.

Question 30

Null Hypothesis (H0)

Answer 30

Assumes there's NO difference in outcomes between two or more different conditions.

Question 31

Alternative Hypothesis

Answer 31

Assumes there IS a difference in outcomes between two or more different conditions.

Question 32

α (Alpha) Level or P-value

Answer 32

Probability of supporting H1 when H0 is actually true. --> Usually set to <.05, meaning we're only willing to call a result "significant" if there's less than a 5% chance of the H0 being true.

Question 33

Type I Error

Answer 33

When you reject the H0 (null) result, even though H0 is actually true. --> Means you find a significant result, even though it's not really there.

Question 34

Type II Error

Answer 34

Happens when you support H0 (fail to reject it) when H0 is actually false. --> You don't find a significant result, even though it's actually there.

Question 35

Reasons for Type I Error

Answer 35

Result just happens to be one of the "5% of times" that result would occur if H0 was true.

Question 36

Reasons for Type II Error

Answer 36

Not enough POWER to detect an effect. --> Power = 1- β (probability of not making a type II error). --> β = probability of type II error occurring.

Question 37

Ways to Increase Statistical Power

Answer 37

1. Larger sample size 2. Larger effect size 3. Less strict α level

Question 38

Publication Bias

Answer 38

Tendency for journals to only publish significant findings.

Question 39

The File Drawer Effect

Answer 39

Refers to the fact that non-significant findings get stored away and never seen.

Question 40

95% Confidence Intervals

Answer 40

Reflects range of values from a sample. We are 95% confident (statistically) that the actual value of the population is in this range. --> Non-overlapping CI's indicate meaningful differences between groups.