Stat 200 Exam #1

Simple random sampling

-Each member of the population has an equal likelihood of being selected -Helps ensure that our sample will represent the population of interest

Ex: An experiment is conducted at a high school where 16 students are asked to complete a two part exam: one part is written, while the other part is oral. When this two-part exam was given, 8 students were randomly assigned to do the written part first and the oral part second, while the other 8 students did the reverse order of the oral part first and the written part second. This is a... A.randomized comparative experiment B.observational comparative study C.matched-pairs (no randomization used) D.matched-pairs experiment (implies randomization)

D.matched-pairs experiment (implies randomization)

Ex: A company wants to know if placing green plants in workers' offices will help reduce employee absenteeism. Employees are randomly chosen to either be given or not be given a plant for their office. The number of sick days is determined for each participant in the study. Which components are a part of this study. This study is ___ and ___ A.single-blinded, uses a placebo B.singled-blinded, has a control group C.not blinded, uses a placebo D.not blinded, has a control group

D.not blinded, has a control group

Is associated implied? Low Iron Levels May Be Linked to Hearing Loss

YES

Is associated implied? Physician opioid prescribing decreases after learning of a patient's fatal overdose

YES

Is causation also implied? Physician opioid prescribing decreases after learning of a patient's fatal overdose

YES

What is a confounding variable also known as?

a lurking variable

Population

all individuals/observations of interest

Response bias

answer is not truthful

Randomized comparative experiment

assign treatment

double blinding

both participants and researches are uninformed

Observational studies can almost NEVER be used to establish...

causation

Randomized experiment can almost always be used to establish...

causation

Two variables are causally associated (shows causation) if...

changing the value of one variable directly influences the value of the other variable

Simple Random Sample (SRS)

each unit of the population has the same chance of being selected. This incorporates probability sampling

-How many cell phone have you had? -Number of apps you usually use? A. nominal B. ordinal C. continuous D. discrete

Discrete

Ex: Rate your sleep quality from 0 to 10, where 10 is "very well" A. nominal B. ordinal C. continuous D. discrete

Discrete

Ex: what type of variable is your Penn State student ID number? A. nominal B. ordinal C. continuous D. discrete

Discrete

Nonresponse bias

never receive survey or refuse to answer

internet surveys use...

non probability sampling

Sample

observations selected from the population, a subset

The best way to get a representative sample is to..

obtain a random sample

Stratified sampling

obtained a random sample of 3 schools from each division to obtain a simple random sample

Population or Sample? An instructor surveys all students in her class to determine if they prefer a take-home exam or an in-class exam. or Sample?

population

Phone surveys use...

probability sampling

What is the goal of matched pairs?

to get rid of some unwanted variation with the analysis

Two variables are associated when...

values for one are related to values of another

randomized double blind placebo control studies

"gold standard" in intervention-baed studies

Random assignment

-assigning subjects to different conditions in a way that they have equal chance of being placed in either condition -controls for confounding

Continuous

-assumes a range of values -often need an instrument -can come from a calculation -may be limited by instrument accuracy and research choice to round

Discrete

-countable list of distinct values -usually integers -typically used when a variable doesn't have too many choices

Categorical variable

-divides/places cases into groups or categories -records "word" responses or serves as an identifier

Ordinal

-do have logical ordering

Nominal

-don't have logical ordering -can also be identifiers

Quantitative variable

-measures or records a number -makes sense to do mathematical operations

Systematic sampling

-obtain a random number -Ex: select every other school

Randomized experiments

-randomly assign group 1 and group 2 -can include cause and effect statements -cancel out impact of confounding variables (bias)

Confounding variable

-related to both the explanatory and response variables -effect on response variable can't be separated from effect found within the explanatory variable

What is an experiment? Ex: Drinking water improves exam grades, research suggests

-researcher actively controls explanatory variables -determines which student gets/has a water bottle

Observational Studies / Experiment

-sample 1 and sample 2 -can include "association" statements -cant cancel out impact of confounding variables

Selection bias

-sample not truly random -sample not a simple random sample

Sample survey

-saves both time and money -needs to be representative

What is an observational study? Ex: Drinking water improves exam grades, research suggests

-simply observe whether or not student has a water bottle -this is what happened in the study

matched pairs experiment (implied randomization)

-with repeated measures: order of treatment -with pairs matched by similarity: assign treatment within each pair

Four pillars of inference...

1. Generalization 2. Causation 3. Estimation 4. Significance

When can we claim causation? (3 instances)

1. With an experiment where we also include random assignment 2. this randomization is not merely getting a simple random sample 3. with random assignment we can cancel out the impact of the confounding variable, because equally applied eliminate bias

Two caveats...

1. not all variables are associated 2. for variables that are associated, now all show causation

Two ways to obtain: 1. tactile 2. use technology

1. pick from a hat method 2. random generator

Matched pairs / paired design: data collection

1. repeated measures on the same unit / case 2. create pairs by matching two very similar participants then assign each unit to a different condition or treatment

A doctor's office wants to estimate the proportion of appointments where at least one medication was prescribed during 2019. The doctor had 520 appointments scheduled throughout the year. Which scheme fits the definition of a simple random sample of 52 appointments? A. Label all appointments from the year and study those numbered 1 through 520. Randomly select 52 appointments. B. Randomly choose one appointment scheduled for each of the 52 weeks. C. Choose the first appointment scheduled for each of the 52 weeks D. Randomly select 52 appointments from the 3:00 PM timeslot

A. Label all appointments from the year and study those numbered 1 through 520. Randomly select 52 appointments.

If study instead did this, would you have a randomized experiment? -Sample 1: random sample of n=32 cyclists from a population -Sample 2: random sample of n=30 controls from a population A.yes, study is now a randomized experiment B.no, still observational - but can now generalize to the population(s) C.unsure

B.no, still observational - but can now generalize to the population(s)

Ex: Does it matter whether blood pressure is measured over a sleeve or on a bare arm? A random sample of 15 patients from all patients at a doctor's practice was obtained. Each selected patient had their blood pressure (mm Hg) measured twice: first over a sleeve and then second on a bare arm. This is a... A.randomized comparative experiment B.observational comparative study C.matched-pairs (no randomization used) D.matched-pairs experiment E. I have no idea

C.matched-pairs (no randomization used)

-Time spent on phone -Dimensions of phone A. nominal B. ordinal C. continuous D. discrete

Continuous

Ex: How long did your sleep duration increase during the work week in minutes? A. nominal B. ordinal C. continuous D. discrete

Continuous

What are the two types of quantitative variables?

Continuous and discrete

Identify the explanatory and response variables: Cyclists who had peddled on an empty stomach incinerated about twice as much fat as those who had consumed a shake first.

Explanatory: if you had a shake or not Response: measurement fo change in fat

Statistical inference

Generalize conclusions from the sample to the population

Convenience sampling

I like the color "red" so select the 6 schools whose colors are red

Is associated implied? Feeling Guilty About Not Your Flossing Teeth. There's no need

NO

Is causation also implied? Feeling Guilty About Not Your Flossing Teeth. There's no need

NO

Is causation also implied? Low Iron Levels May Be Linked to Hearing Loss

NO

-Apple vs android -Do you have a phone? A. nominal B. ordinal C. continuous D. discrete

Nominal

Ex: Was your work time aligned with your circadian time? yes no A. nominal B. ordinal C. continuous D. discrete

Nominal

What are the 2 types of categorical variables?

Nominal and ordinal

Observational study or experiment? A study followed 98 puppies from birth to adulthood who enrolled in a training program. About 70% achieved success as guide dogs. More intense mothering early in life was associated with program failure. In addition, mothers whose nursing style required greater effort by puppies produced more successful offspring.

Observational study

-Model / year of your phone A. nominal B. ordinal C. continuous D. discrete

Ordinal

Population or Sample? All Presidents of Penn State University age started

Population

Ex: with 100 recruits, each had contaminated their hands and then instructed them to wash with soap and water: Subsequently randomly assigned: A. dry hands for 15 seconds with paper towels (n1=50) B. dry hands for 30 seconds with a hand dryer (n2=50) Measured: amount of reduction in bacterial count In the end, the scientists called it a draw: both methods dried the hands thoroughly and produced equivalent reductions in bacterial counts. Q1: Randomized experiment or observational study Q2: Unit / Case = ? Q3: Can the conclusion include a cause and effect statement? Yes or No Q4: What is the response and explanatory variables? Q5: Are there confounding variables?

Q1: Randomized experiment Q2: Recruit Q3: Yes Q4: Explanatory variable is the type of hand drying method and the response is the amount of reduction in bacterial count Q5: No confounding variables

Ex: The bone density of 32 male competitive cyclists, in their late 20s and early 30s, was compared to that of 30 controls: age-matched men who were active but not competitive athletes. -Sample 1, n=32 cyclists -Sample 2, n=30 controls Each recruited male was given a bone mineral density (BMD) test One result: Some of these cyclists had osteopenia in their spines, only one step below full-blown osteoporosis. Q1: Randomized experiment or observational study Q2: Individual unit: ? Q3: Can the conclusion include a cause and effect statement? Yes or No Q4: Can we suggest that an association has been found? Yes or No Q5: Explanatory and Response variable Q6: Is confounding variables a problem? Why?

Q1: observational study Q2: Male Q3: No Q4: Yes Q5: Explanatory variable is the cyclist or controls and the response variable is the BMD t-score Q6: Yes because there is no random assignment

Population or Sample? A highway safety researcher wants to determine the average distance at which all drivers can read a highway at night. She measures the distance for a sample of 50 drivers.

Sample

Population or Sample? The Daily Collegian wants to estimate the proportion of Penn State students with season football tickets who disapprove of the new mobile ticketing system when surveying 200 students

Sample

Results for this Gallup poll are based on telephone interviews conducted July 1-11, 2018, with a random sample of 1,033 national adults. Each sample of national adults includes a minimum quota of 70% cellphone respondents and 30% landline respondents, with additional minimum quotas by time zone within region. Landline and cellular telephone numbers are selected using random-digit-dial methods. Selection Bias? Y or N Nonresponse Bias? Y or N Response Bias? Y or N

Selection Bias? N Nonresponse Bias? Y Response Bias? Y

blinding is used to...

get rid of bias in studies

When is confounding a problem?

have an observational study -researcher does not actively control the variable values -observes the values as they naturally exist -cannot separate each variable's contribution to the change in the response variable (Bias exists)

single blinding

implement use of a placebo and participants are uninformed