Research

Hypothesis - predictions about behavior that can be tested (testing the theories/wonders)

Theory - using our observations to explain better (wonders)

Ex. - I wonder is sleep deprivation leads to bad driving (theory). I predict sleep deprivation will lead to worse driving (hypothesis). 

Causation - comes form us being able to randomize, we can generalize to the population, more random it is ~ more able we can generalize to the population

Sampling Bias - when you do not randomly select participants from the population ~ choose who you like to participate ~ don’t get diversity you need

Experimental Bias - when you sub-consciously put your opinions into the situation ~ focus on one piece of data rather than the other (best way to get rid of is double blind)

True Scientist - someone who doesn’t want a certain result, just wants to see what happens

Confirmation Bias - only look for the things that confirm what you already believe ~ looks at data that you want to believe rather than it as a whole, ignoring data that doesn’t confirm your beliefs

HIndsight Bias - say the following phrase “I knew that would happen” or “I knew they would win” 

Self-Report Bias - why people lie on the survey ~ over exaggerate your strengths and often minimize your weaknesses

False Consensus Effect - over estimate the extent to which people agree with you ~ “I thought everybody believed this,” “wait, you don’t believe this?”

Illusory Correlation - think theres a correlation but there really isn’t one ~ carrots can improve eyesight (superstition) 

Regression Toward The Mean - have an average, then one test or time period is severely better or worse, then goes back to the “average” 

Normal Distribution - normal bell curve, gives us a good idea of really nice data ~ mean, median, and mode are the same

Bimode Distribution - having two modes, indicates there should be two different studies

Social Desirability Bias - doing this for/because of the people who look/value it ~ exaggerates strengths for the thing, minimizes weakness for the thing

Positive correlation - arrows go up ~ the more junk food you eat, the more chance of heart disease

Negative correlation - arrows go down ~ less junk food you eat less chance of heart disease

Longitudinal Study - follow same person over months, years, or decades ~ what we want to do (pro - best data)(con - expensive, hard to find participants) 

Cross-Sectional Study - not over long periods of time, just look at different age groups and compare/contracts (con - not the same person, different age groups)

Meta-Analysis - when you look at a whole bunch of different studies on the same topic with the same variables 

What are the 5 types of research (“empirical investigation”), and what are the pros & cons for each?

5 Types of Research ("Empirical Investigation”)

  • Experiment

- Controlled lab (clinical)

- Pro(s)

   - Can say the word “cause” ~ were getting somewhere

- Con(s)

   - Expensive

   - Need participants

   - Can’t use it for Psych. theories because it is unethical

  • Case study

- 1 person or small group ~ not enough people who have this condition

- Pro(s)

   - Get a lot of data from them

- Con(s)

   - Unique to each person

  • Survey

   - Likert scale ~ strongly agree-> strongly disagree, 1->5 stars

              - Quantitative measures ~ numbers/average 

              - Qualitative measures ~ words/what we observe

- Pro(s)

   - Cheap

   - Easy to distribute

   - Patterns ~ data we can use

- Con(s)

   - Not always honest

   - Probably won’t get all/a lot of them back

  • Naturalistic observation

- Field study

- Can only be done ~ truely/ethicaly ~ on babies and animals

- Pro(s)

   -  Pure/raw honest data from observations

- Con(s)

   - If someone knows that your watching, its not their natural/real behaviors ~ hawthorne effect

  • Correlation

- Can’t manipulate the study

- No randomness 

- Predicts the other

- Relationship between 2 variables ~ say “strong relationship between ___ and ____”

- Pro(s) 

   - Relationship between 2 variables

- Con(s)

   - Cannot say the word “cause”



If my results to a study show a correlation of “-.86”, & my friend’s show a “+.52”, whose is a stronger relationship? [Me at -.86 (the correlational coefficient) means that I have a negative (“inverse”) correlation with the “r” number at -.86; my friend is showing a positive (“direct”) correlation with the correlational coefficient at +.52 (“r” number)] What would the “scatterplots” look like?  a. Which one of the numbers I mentioned would result from a study that says: “The more I vape/smoke, the greater my chances of developing lung cancer,” “The more t.v. you watch, the lower your grades go.”  b. What if a third person’s showed a “+1.13”, what would that mean?

Graphs of Correlations

  • Scatter plot

  • Dots everywhere ~ no slope, no correlation

  • Dots in a line ~ slope -> positive or negative correlation



Strength of a relationship

  • Closer to 0 - weaker 

  • Closer to 1 - stronger



Vape -> Lung Cancer

  • Positive correlation



T.V. -> Grades

  • Negative correlation



+1.13

  • A correlation of +1.13 means there is a mathematical error or a glitch in the software. Correlation coefficients (r) must strictly fall within the range of -1.00 to +1.00. A number outside of this range is statistically impossible and indicates that the data was calculated incorrectly.



From the “standard deviation” graph:  a. What % of people fall within 1 standard deviation of the mean (“z” score of 1)? Within 2 standard deviations from the median (“z” score of 2)?  b. On the Stanford-Binet test, what did Larry get if his score was 2 standard deviations above the mean?

Part A

  • Within 1 Standard Deviation Of The Mean ~ about 68% of people fall into this range

  • Within 2 Standard Deviations Of The Median ~ when it’s a perfectly symmetrical normal distribution, the mean and median are identical. So about 95% of people fall into this range



Part B

  • 130 or 132 making him classified as “gifted” and places him in the 98 percentile



What is the difference between “descriptive” & “inferential” statistics? (Don’t forget to explain all of the terms associated with “descriptive” stats, like “central tendency”, “variance”, “variability”, and “mode”) [For “inferential” stats, make sure to mention “statistical significance” (“p” value), “null hypothesis”, “alternative hypothesis”, and “Type I” & “Type II” errors.]

Descriptive statistics

  • Goal ~ to summarize, organize, and describe the features of a specific dataset

  • Data Scope ~ focuses solely on the collected dataset (no assumptions are made beyond the data at hand)

  • Common Tools ~ mean, median, mode, central tendency, variance, variability, and graphs



Core Concepts

- Central Tendency ~ Measures the center/typical value of the data set. Tels where the most of the data points cluster; three main metics (mean, median, mode). 

- Variability/Dispersion ~ How spread out/scattered the data points are around the center (if all numbers are nearly identical, variability is low; if they are widely spread, variability is high).

- Variance ~ Specific metric used to quantify variability. Calculates average of the squared differences from the mean (a higher variance means the data points are spread further from the average)



Inferential statistics

  • Goal ~ to make predictions, test hypotheses, and generalize findings to a broader population (generalization)

  • Data Scope ~ works with a representative sample to deduce properties of an unmeasured population

  • Common Tools ~ null/alternative hypotheses, p-values, t-tests, and Type I/II errors



Core Concepts

- Null Hypothesis ~ Baseline assumption that there is no effect, no difference, or no relationship between the variables being tested (if testing a new medicine, the null hypothesis states: "The medicine has zero effect on patient recovery.")

- Alternative Hypothesis ~ Claim you are actually trying to prove. States that there is a meaningful effect, difference, or relationship (The medicine significantly improves patient recovery)

- Statistical Significance and the “p” value ~ Framework used to decide wether to reject the null. The P-VALUE is the probability that the sample results happened purely at random. Low p-value ( _< .05) means the pattern in the data is highly unlikely to be an accident, giving the STATISTICAL SIGNIFICANT to reject the null and support the alternative hypothesis. 

- Type l Error (False Positive) ~ Happens when one incorrectly rejects a true null. When someone claims a pattern or effect exists when it does not (connected to p-value).

- Type ll Error (False Negative) ~ Happens when one fails to reject a false null. When someone misses a real pattern or effect because the data did not show enough evidence to detect it.



What are positive and negative skewed distributions? In a positively skewed scenario, what is the relationship between mean and median? Which is a better indicator of the results?

Skewed Distributions

  • Median is a better indicator than the mean, won’t be effected by a random high or low number

  • Positive ~ Right, the “tail” on the right side/high end, is longer/fatter. Most data points cluster on the left

  • Negative ~ Left, the “tail” on the left side/low end, is longer/fatter. Most data points cluster on the right



Positive Between Mean and Median

  • Values rank form lowest to highest, mode->median->mean

  • Mean is pulled up/to the right, by extreme high values

  • Median stays closer to the center/peak, largely unaffected by the extreme high values

  • Mean is greater than the median



Better Indicator

  • Median

  • The mean is sensitive to outliers and extreme values, which warps its representation of a normal data point (making the average look artificially high in a positive skew). The median splits the dataset cleanly in half (50% above, 50% below) regardless of how extreme the tail values are, making it a more robust measure.



What are “operational definitions”?

Operational Definitions

  • When you clearly define the variables in the experiment

  • Be as detailed as you can when you explain the variables

  • Very clear detailed instructions on the experiment on how you will use the variables

  • Wrong Ex. ~ We gave them a pill

  • Somewhat Good Ex. ~ I gave them a red pill weighing ____, twice a day after breakfast and dinner with _____ (drink)

  • Helps other scientists replicate study ~ to add/prove that this experiment was correct/information gathered is accurate




Describe a study (“experimental design”) where you want to see if the new diet pill you created can actually cause people to lose weight. In your explanation, make sure to use the “scientific method” terms: hypothesis, experimental & control groups, independent & dependent variables, single & double blind, confounding variables, and random sample & random assignment.

  1. Hypothesis

  2. Population

  3. Random Sample Assignment

  4. Control Group (Placebo) 4.   Experimental Group

              4.1 (i.v.) Manipulate     4.1 (d.v.) Measure

  1. Collect Data

  2. Analyze Data

  3. Publish



Hypothesis ~ If you _____ then _______

Population ~ People who would participate/a group that is most affected/related to the study, budget (large budget -> US or Virginia, Medium -> Loudoun County, Small -> Briar Woods)

Remove Bias ~ Random sample, random assignment, single blind, double blind

Random Sample ~ Pick people at random to join study (out of hat, computer program, etc.) 

Random Assignment ~ Pick people at random for each group (control or experimental)

Control Group ~ Placebo/fake pill/something that does jack shit

Experimental Group ~ Gets the i.v.

i.v. Manipulation ~ Time you sleep, how many pills you take

d.v. Measure ~ IQ

Single Blind ~ Don’t know what group your in

Double Blind ~ People who are running the experiment, excluding top researchers, don’t know which group is which

Confounding Variable ~ Anything that mess’ with the d.v. thats not the i.v. (Ex. ~ music increases IQ (test), starts to eat more salmon (considered “brain food”). Maybe salmon increased IQ instead of music)

Analyze data; 1 types -> two variations 

Descriptive Statistics ~ Numerical data used to measure and describe characteristics of groups. Includes measures of central tendency (the mean ~ average, median ~ middle number, and mode ~ most common number) and measures of variation (range ~ difference between highest and lowest value and standard deviation ~ how are the numbers distributed around the mean)



In research, we might need to use human participants. In order to be ethical, we must do what?

Ethics (for humans)

  • No harming them psychologically or physically

  • Informed consent ~ permission/willingness/freewill for adults

  • Informed assent ~ permission/willingness from kid but adult consent 

  • Limited deception

  • Debrief ~ Telling them what you are/plan to do after

  • Confidentiality ~ can’t publish things they didn’t approve of/don’t post personal info

  • Right to withdraw ~ can leave the experiment at any time

  • Right to refuse ~ still in the experiment but refuses/isn’t comfortable doing something