aindreas/uni
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add second year

Browse Source
This commit is contained in:
Aindréas Ó hAodha
2023-12-07 01:19:12 +00:00
parent 3291e5c79e
commit 3d12031ab8
1168 changed files with 431409 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

100
second/semester1/logseq-stuff/logseq/bak/pages/Sampling/2022-09-30T10_01_42.381Z.Desktop.md Normal file
View File

@ -0,0 +1,100 @@
- #[[ST2001 - Statistics in Data Science I]]
- **Previous Topic:** [[Exploratory Data Analysis]]
- **Next Topic:** [[Probability]]
- **Relevant Slides:** ![Topic 3 - Sampling.pdf](../assets/Topic_3_-_Sampling_1663599787566_0.pdf)
-
- What is a **Parameter**? #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.5
card-next-schedule:: 2022-09-19T23:00:00.000Z
card-last-reviewed:: 2022-09-19T18:03:31.423Z
card-last-score:: 1
- A **parameter** is a single value summarising some feature or variable of interest in the population.
- It is usually unknown.
- What is **inference**? #card
card-last-interval:: 4.86
card-repeats:: 1
card-ease-factor:: 2.6
card-next-schedule:: 2022-09-24T13:58:35.788Z
card-last-reviewed:: 2022-09-19T17:58:35.789Z
card-last-score:: 5
- **Inference** is the process of making decisions about a population based on information in a sample.
- A consequence of **natural variation** is that two samples drawn form the same population will usually give different estimates of the population parameters.
-
- # Sampling
collapsed:: true
- What is **non-probabilistic sampling**? #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.5
card-next-schedule:: 2022-09-19T23:00:00.000Z
card-last-reviewed:: 2022-09-19T17:51:31.085Z
card-last-score:: 1
- **Non-probabilistic sampling** methods are techniques of obtaining a sample that is not chosen at random and may be subject to **sampling bias**.
- ## Simple Random Sample
- ### Difficulties:
- Obtaining a sampling frame (list of all experimental units).
- Possibly time consuming / expensive.
- Minority groups, by chance, may not be represented in the sample.
- ## Stratified Random Sampling #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.5
card-next-schedule:: 2022-09-26T23:00:00.000Z
card-last-reviewed:: 2022-09-26T12:16:41.996Z
card-last-score:: 1
- 1. Split entire population into **homogeneous groups**, called **strata**.
2. Take a Simple Random Sample from each stratum.
- ### Stratified VS Simple Random Sample
- Ensure representation from minority groups.
- Estimates of the population parameters per strata may be of interest.
- Possible reduction in cost per observation in the survey.
- Increased accuracy as reduced sampling error (less variation within a stratum).
-
- ### Difficulties
- Can you correctly allocate each individual to one & only one stratum?
- Should every group receive equal weight?
- What if some strata are more varied than others?
- Take into account mean, variance, and cost to get "optimal allocation".
- ## Cluster Sampling #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.5
card-next-schedule:: 2022-09-26T23:00:00.000Z
card-last-reviewed:: 2022-09-26T12:16:26.204Z
card-last-score:: 1
- Instead of randomly choosing individuals, a Simple Random Sample of collection or groups of individuals is taken.
- The population is broken up into regions or groups, usually a *natural partition*, called a **cluster**.
- Internally heterogeneous, homogeneous between the clusters.
- Clusters are assumed representation of the entire population.
- Small number of clusters are selected at random.
- Every individual within a cluster is observed.
-
- ### Cluster Over Stratified
- Sampling frame not necessarily needed.
- May be more practical and / or economical than Simple or Stratified Random Sampling.
- Will be biased if the entire cluster is not sampled.
- Careful if homogeneity within a cluster and heterogeneity between clusters as this can increase sample error.
- **Note:** In stratified sampling, all strata are sampled, while in cluster sampling only some clusters are sampled.
- # Studies & Experiments
- ## Observational Studies & Experiments #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.5
card-next-schedule:: 2022-09-26T23:00:00.000Z
card-last-reviewed:: 2022-09-26T12:12:25.456Z
card-last-score:: 1
- In an **Observational Study**, data is collected only be *observing* what occurs.
- E.g., surveys, historical records.
- When researchers want to investigate **causal relationships**, it's best to conduct an experiment.
- Usually there will be both an explanatory variable & a response variable.
- Be wary of confounding variables.
- ## Designed (Comparative) Study
- An experiment allows us to prove a cause-and-effect relationship.
- The experimenter must identify:
- at least one **explanatory variable**, called a **factor** to manipulate.
- at least one **response** variable to measure.
- The experimenter must also control any other **nuisance factors** that could influence the response.
- e.g., weather, day of the week.
-

100
second/semester1/logseq-stuff/logseq/bak/pages/Sampling/2022-10-07T18_07_00.028Z.Desktop.md Normal file
View File

@ -0,0 +1,100 @@
- #[[ST2001 - Statistics in Data Science I]]
- **Previous Topic:** [[Exploratory Data Analysis]]
- **Next Topic:** [[Probability]]
- **Relevant Slides:** ![Topic 3 - Sampling.pdf](../assets/Topic_3_-_Sampling_1663599787566_0.pdf)
-
- What is a **Parameter**? #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.5
card-next-schedule:: 2022-10-07T23:00:00.000Z
card-last-reviewed:: 2022-10-07T10:42:17.374Z
card-last-score:: 1
- A **parameter** is a single value summarising some feature or variable of interest in the population.
- It is usually unknown.
- What is **inference**? #card
card-last-interval:: 4
card-repeats:: 2
card-ease-factor:: 2.7
card-next-schedule:: 2022-10-07T14:31:27.689Z
card-last-reviewed:: 2022-10-03T14:31:27.690Z
card-last-score:: 5
- **Inference** is the process of making decisions about a population based on information in a sample.
- A consequence of **natural variation** is that two samples drawn form the same population will usually give different estimates of the population parameters.
-
- # Sampling
collapsed:: true
- What is **non-probabilistic sampling**? #card
card-last-interval:: 9.76
card-repeats:: 3
card-ease-factor:: 2.46
card-next-schedule:: 2022-10-17T04:35:48.922Z
card-last-reviewed:: 2022-10-07T10:35:48.922Z
card-last-score:: 5
- **Non-probabilistic sampling** methods are techniques of obtaining a sample that is not chosen at random and may be subject to **sampling bias**.
- ## Simple Random Sample
- ### Difficulties:
- Obtaining a sampling frame (list of all experimental units).
- Possibly time consuming / expensive.
- Minority groups, by chance, may not be represented in the sample.
- ## Stratified Random Sampling #card
card-last-interval:: 4
card-repeats:: 2
card-ease-factor:: 2.36
card-next-schedule:: 2022-10-10T09:40:03.399Z
card-last-reviewed:: 2022-10-06T09:40:03.400Z
card-last-score:: 3
- 1. Split entire population into **homogeneous groups**, called **strata**.
2. Take a Simple Random Sample from each stratum.
- ### Stratified VS Simple Random Sample
- Ensure representation from minority groups.
- Estimates of the population parameters per strata may be of interest.
- Possible reduction in cost per observation in the survey.
- Increased accuracy as reduced sampling error (less variation within a stratum).
-
- ### Difficulties
- Can you correctly allocate each individual to one & only one stratum?
- Should every group receive equal weight?
- What if some strata are more varied than others?
- Take into account mean, variance, and cost to get "optimal allocation".
- ## Cluster Sampling #card
card-last-interval:: 1.47
card-repeats:: 2
card-ease-factor:: 2.36
card-next-schedule:: 2022-10-08T21:38:07.258Z
card-last-reviewed:: 2022-10-07T10:38:07.258Z
card-last-score:: 3
- Instead of randomly choosing individuals, a Simple Random Sample of collection or groups of individuals is taken.
- The population is broken up into regions or groups, usually a *natural partition*, called a **cluster**.
- Internally heterogeneous, homogeneous between the clusters.
- Clusters are assumed representation of the entire population.
- Small number of clusters are selected at random.
- Every individual within a cluster is observed.
-
- ### Cluster Over Stratified
- Sampling frame not necessarily needed.
- May be more practical and / or economical than Simple or Stratified Random Sampling.
- Will be biased if the entire cluster is not sampled.
- Careful if homogeneity within a cluster and heterogeneity between clusters as this can increase sample error.
- **Note:** In stratified sampling, all strata are sampled, while in cluster sampling only some clusters are sampled.
- # Studies & Experiments
- ## Observational Studies & Experiments #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.6
card-next-schedule:: 2022-10-07T23:00:00.000Z
card-last-reviewed:: 2022-10-07T10:38:47.387Z
card-last-score:: 1
- In an **Observational Study**, data is collected only be *observing* what occurs.
- E.g., surveys, historical records.
- When researchers want to investigate **causal relationships**, it's best to conduct an experiment.
- Usually there will be both an explanatory variable & a response variable.
- Be wary of confounding variables.
- ## Designed (Comparative) Study
- An experiment allows us to prove a cause-and-effect relationship.
- The experimenter must identify:
- at least one **explanatory variable**, called a **factor** to manipulate.
- at least one **response** variable to measure.
- The experimenter must also control any other **nuisance factors** that could influence the response.
- e.g., weather, day of the week.
-

100
second/semester1/logseq-stuff/logseq/bak/pages/Sampling/2022-10-10T11_59_52.472Z.Desktop.md Normal file
View File

@ -0,0 +1,100 @@
- #[[ST2001 - Statistics in Data Science I]]
- **Previous Topic:** [[Exploratory Data Analysis]]
- **Next Topic:** [[Probability]]
- **Relevant Slides:** ![Topic 3 - Sampling.pdf](../assets/Topic_3_-_Sampling_1663599787566_0.pdf)
-
- What is a **Parameter**? #card
card-last-interval:: 2.6
card-repeats:: 2
card-ease-factor:: 2.6
card-next-schedule:: 2022-10-11T05:09:09.705Z
card-last-reviewed:: 2022-10-08T15:09:09.705Z
card-last-score:: 5
- A **parameter** is a single value summarising some feature or variable of interest in the population.
- It is usually unknown.
- What is **inference**? #card
card-last-interval:: 11.2
card-repeats:: 3
card-ease-factor:: 2.8
card-next-schedule:: 2022-10-18T19:20:23.932Z
card-last-reviewed:: 2022-10-07T15:20:23.932Z
card-last-score:: 5
- **Inference** is the process of making decisions about a population based on information in a sample.
- A consequence of **natural variation** is that two samples drawn form the same population will usually give different estimates of the population parameters.
-
- # Sampling
collapsed:: true
- What is **non-probabilistic sampling**? #card
card-last-interval:: 9.76
card-repeats:: 3
card-ease-factor:: 2.46
card-next-schedule:: 2022-10-17T04:35:48.922Z
card-last-reviewed:: 2022-10-07T10:35:48.922Z
card-last-score:: 5
- **Non-probabilistic sampling** methods are techniques of obtaining a sample that is not chosen at random and may be subject to **sampling bias**.
- ## Simple Random Sample
- ### Difficulties:
- Obtaining a sampling frame (list of all experimental units).
- Possibly time consuming / expensive.
- Minority groups, by chance, may not be represented in the sample.
- ## Stratified Random Sampling #card
card-last-interval:: 4
card-repeats:: 2
card-ease-factor:: 2.36
card-next-schedule:: 2022-10-10T09:40:03.399Z
card-last-reviewed:: 2022-10-06T09:40:03.400Z
card-last-score:: 3
- 1. Split entire population into **homogeneous groups**, called **strata**.
2. Take a Simple Random Sample from each stratum.
- ### Stratified VS Simple Random Sample
- Ensure representation from minority groups.
- Estimates of the population parameters per strata may be of interest.
- Possible reduction in cost per observation in the survey.
- Increased accuracy as reduced sampling error (less variation within a stratum).
-
- ### Difficulties
- Can you correctly allocate each individual to one & only one stratum?
- Should every group receive equal weight?
- What if some strata are more varied than others?
- Take into account mean, variance, and cost to get "optimal allocation".
- ## Cluster Sampling #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.36
card-next-schedule:: 2022-10-09T23:00:00.000Z
card-last-reviewed:: 2022-10-09T08:52:46.030Z
card-last-score:: 1
- Instead of randomly choosing individuals, a Simple Random Sample of collection or groups of individuals is taken.
- The population is broken up into regions or groups, usually a *natural partition*, called a **cluster**.
- Internally heterogeneous, homogeneous between the clusters.
- Clusters are assumed representation of the entire population.
- Small number of clusters are selected at random.
- Every individual within a cluster is observed.
-
- ### Cluster Over Stratified
- Sampling frame not necessarily needed.
- May be more practical and / or economical than Simple or Stratified Random Sampling.
- Will be biased if the entire cluster is not sampled.
- Careful if homogeneity within a cluster and heterogeneity between clusters as this can increase sample error.
- **Note:** In stratified sampling, all strata are sampled, while in cluster sampling only some clusters are sampled.
- # Studies & Experiments
- ## Observational Studies & Experiments #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.6
card-next-schedule:: 2022-10-08T23:00:00.000Z
card-last-reviewed:: 2022-10-08T15:24:45.845Z
card-last-score:: 1
- In an **Observational Study**, data is collected only be *observing* what occurs.
- E.g., surveys, historical records.
- When researchers want to investigate **causal relationships**, it's best to conduct an experiment.
- Usually there will be both an explanatory variable & a response variable.
- Be wary of confounding variables.
- ## Designed (Comparative) Study
- An experiment allows us to prove a cause-and-effect relationship.
- The experimenter must identify:
- at least one **explanatory variable**, called a **factor** to manipulate.
- at least one **response** variable to measure.
- The experimenter must also control any other **nuisance factors** that could influence the response.
- e.g., weather, day of the week.
-

100
second/semester1/logseq-stuff/logseq/bak/pages/Sampling/2022-10-20T09_05_58.001Z.Desktop.md Normal file
View File

@ -0,0 +1,100 @@
- #[[ST2001 - Statistics in Data Science I]]
- **Previous Topic:** [[Exploratory Data Analysis]]
- **Next Topic:** [[Probability]]
- **Relevant Slides:** ![Topic 3 - Sampling.pdf](../assets/Topic_3_-_Sampling_1663599787566_0.pdf)
-
- What is a **Parameter**? #card
card-last-interval:: 2.6
card-repeats:: 2
card-ease-factor:: 2.6
card-next-schedule:: 2022-10-11T05:09:09.705Z
card-last-reviewed:: 2022-10-08T15:09:09.705Z
card-last-score:: 5
- A **parameter** is a single value summarising some feature or variable of interest in the population.
- It is usually unknown.
- What is **inference**? #card
card-last-interval:: 11.2
card-repeats:: 3
card-ease-factor:: 2.8
card-next-schedule:: 2022-10-18T19:20:23.932Z
card-last-reviewed:: 2022-10-07T15:20:23.932Z
card-last-score:: 5
- **Inference** is the process of making decisions about a population based on information in a sample.
- A consequence of **natural variation** is that two samples drawn form the same population will usually give different estimates of the population parameters.
-
- # Sampling
collapsed:: true
- What is **non-probabilistic sampling**? #card
card-last-interval:: 9.76
card-repeats:: 3
card-ease-factor:: 2.46
card-next-schedule:: 2022-10-17T04:35:48.922Z
card-last-reviewed:: 2022-10-07T10:35:48.922Z
card-last-score:: 5
- **Non-probabilistic sampling** methods are techniques of obtaining a sample that is not chosen at random and may be subject to **sampling bias**.
- ## Simple Random Sample
- ### Difficulties:
- Obtaining a sampling frame (list of all experimental units).
- Possibly time consuming / expensive.
- Minority groups, by chance, may not be represented in the sample.
- ## Stratified Random Sampling #card
card-last-interval:: 4
card-repeats:: 2
card-ease-factor:: 2.36
card-next-schedule:: 2022-10-10T09:40:03.399Z
card-last-reviewed:: 2022-10-06T09:40:03.400Z
card-last-score:: 3
- 1. Split entire population into **homogeneous groups**, called **strata**.
2. Take a Simple Random Sample from each stratum.
- ### Stratified VS Simple Random Sample
- Ensure representation from minority groups.
- Estimates of the population parameters per strata may be of interest.
- Possible reduction in cost per observation in the survey.
- Increased accuracy as reduced sampling error (less variation within a stratum).
-
- ### Difficulties
- Can you correctly allocate each individual to one & only one stratum?
- Should every group receive equal weight?
- What if some strata are more varied than others?
- Take into account mean, variance, and cost to get "optimal allocation".
- ## Cluster Sampling #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.36
card-next-schedule:: 2022-10-09T23:00:00.000Z
card-last-reviewed:: 2022-10-09T08:52:46.030Z
card-last-score:: 1
- Instead of randomly choosing individuals, a Simple Random Sample of collection or groups of individuals is taken.
- The population is broken up into regions or groups, usually a *natural partition*, called a **cluster**.
- Internally heterogeneous, homogeneous between the clusters.
- Clusters are assumed representation of the entire population.
- Small number of clusters are selected at random.
- Every individual within a cluster is observed.
-
- ### Cluster Over Stratified
- Sampling frame not necessarily needed.
- May be more practical and / or economical than Simple or Stratified Random Sampling.
- Will be biased if the entire cluster is not sampled.
- Careful if homogeneity within a cluster and heterogeneity between clusters as this can increase sample error.
- **Note:** In stratified sampling, all strata are sampled, while in cluster sampling only some clusters are sampled.
- # Studies & Experiments
- ## Observational Studies & Experiments #card
card-last-interval:: 3.45
card-repeats:: 2
card-ease-factor:: 2.46
card-next-schedule:: 2022-10-13T21:37:27.708Z
card-last-reviewed:: 2022-10-10T11:37:27.708Z
card-last-score:: 3
- In an **Observational Study**, data is collected only be *observing* what occurs.
- E.g., surveys, historical records.
- When researchers want to investigate **causal relationships**, it's best to conduct an experiment.
- Usually there will be both an explanatory variable & a response variable.
- Be wary of confounding variables.
- ## Designed (Comparative) Study
- An experiment allows us to prove a cause-and-effect relationship.
- The experimenter must identify:
- at least one **explanatory variable**, called a **factor** to manipulate.
- at least one **response** variable to measure.
- The experimenter must also control any other **nuisance factors** that could influence the response.
- e.g., weather, day of the week.
-

100
second/semester1/logseq-stuff/logseq/bak/pages/Sampling/2022-11-23T09_22_44.508Z.Desktop.md Normal file
View File

@ -0,0 +1,100 @@
- #[[ST2001 - Statistics in Data Science I]]
- **Previous Topic:** [[Exploratory Data Analysis]]
- **Next Topic:** [[Probability]]
- **Relevant Slides:** ![Topic 3 - Sampling.pdf](../assets/Topic_3_-_Sampling_1663599787566_0.pdf)
-
- What is a **Parameter**? #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.6
card-next-schedule:: 2022-11-15T00:00:00.000Z
card-last-reviewed:: 2022-11-14T16:30:17.122Z
card-last-score:: 1
- A **parameter** is a single value summarising some feature or variable of interest in the population.
- It is usually unknown.
- What is **inference**? #card
card-last-interval:: 33.64
card-repeats:: 4
card-ease-factor:: 2.9
card-next-schedule:: 2022-11-22T23:24:34.466Z
card-last-reviewed:: 2022-10-20T08:24:34.467Z
card-last-score:: 5
- **Inference** is the process of making decisions about a population based on information in a sample.
- A consequence of **natural variation** is that two samples drawn form the same population will usually give different estimates of the population parameters.
-
- # Sampling
collapsed:: true
- What is **non-probabilistic sampling**? #card
card-last-interval:: 29.04
card-repeats:: 4
card-ease-factor:: 2.56
card-next-schedule:: 2022-12-13T20:02:44.059Z
card-last-reviewed:: 2022-11-14T20:02:44.059Z
card-last-score:: 5
- **Non-probabilistic sampling** methods are techniques of obtaining a sample that is not chosen at random and may be subject to **sampling bias**.
- ## Simple Random Sample
- ### Difficulties:
- Obtaining a sampling frame (list of all experimental units).
- Possibly time consuming / expensive.
- Minority groups, by chance, may not be represented in the sample.
- ## Stratified Random Sampling #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.36
card-next-schedule:: 2022-11-18T00:00:00.000Z
card-last-reviewed:: 2022-11-17T20:19:40.391Z
card-last-score:: 1
- 1. Split entire population into **homogeneous groups**, called **strata**.
2. Take a Simple Random Sample from each stratum.
- ### Stratified VS Simple Random Sample
- Ensure representation from minority groups.
- Estimates of the population parameters per strata may be of interest.
- Possible reduction in cost per observation in the survey.
- Increased accuracy as reduced sampling error (less variation within a stratum).
-
- ### Difficulties
- Can you correctly allocate each individual to one & only one stratum?
- Should every group receive equal weight?
- What if some strata are more varied than others?
- Take into account mean, variance, and cost to get "optimal allocation".
- ## Cluster Sampling #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.36
card-next-schedule:: 2022-11-18T00:00:00.000Z
card-last-reviewed:: 2022-11-17T20:18:48.138Z
card-last-score:: 1
- Instead of randomly choosing individuals, a Simple Random Sample of collection or groups of individuals is taken.
- The population is broken up into regions or groups, usually a *natural partition*, called a **cluster**.
- Internally heterogeneous, homogeneous between the clusters.
- Clusters are assumed representation of the entire population.
- Small number of clusters are selected at random.
- Every individual within a cluster is observed.
-
- ### Cluster Over Stratified
- Sampling frame not necessarily needed.
- May be more practical and / or economical than Simple or Stratified Random Sampling.
- Will be biased if the entire cluster is not sampled.
- Careful if homogeneity within a cluster and heterogeneity between clusters as this can increase sample error.
- **Note:** In stratified sampling, all strata are sampled, while in cluster sampling only some clusters are sampled.
- # Studies & Experiments
- ## Observational Studies & Experiments #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.46
card-next-schedule:: 2022-11-15T00:00:00.000Z
card-last-reviewed:: 2022-11-14T16:49:06.426Z
card-last-score:: 1
- In an **Observational Study**, data is collected only be *observing* what occurs.
- E.g., surveys, historical records.
- When researchers want to investigate **causal relationships**, it's best to conduct an experiment.
- Usually there will be both an explanatory variable & a response variable.
- Be wary of confounding variables.
- ## Designed (Comparative) Study
- An experiment allows us to prove a cause-and-effect relationship.
- The experimenter must identify:
- at least one **explanatory variable**, called a **factor** to manipulate.
- at least one **response** variable to measure.
- The experimenter must also control any other **nuisance factors** that could influence the response.
- e.g., weather, day of the week.
-

100
second/semester1/logseq-stuff/logseq/bak/pages/Sampling/2022-11-23T12_15_36.873Z.Desktop.md Normal file
View File

@ -0,0 +1,100 @@
- #[[ST2001 - Statistics in Data Science I]]
- **Previous Topic:** [[Exploratory Data Analysis]]
- **Next Topic:** [[Probability]]
- **Relevant Slides:** ![Topic 3 - Sampling.pdf](../assets/Topic_3_-_Sampling_1663599787566_0.pdf)
-
- What is a **Parameter**? #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.6
card-next-schedule:: 2022-11-15T00:00:00.000Z
card-last-reviewed:: 2022-11-14T16:30:17.122Z
card-last-score:: 1
- A **parameter** is a single value summarising some feature or variable of interest in the population.
- It is usually unknown.
- What is **inference**? #card
card-last-interval:: 33.64
card-repeats:: 4
card-ease-factor:: 2.9
card-next-schedule:: 2022-11-22T23:24:34.466Z
card-last-reviewed:: 2022-10-20T08:24:34.467Z
card-last-score:: 5
- **Inference** is the process of making decisions about a population based on information in a sample.
- A consequence of **natural variation** is that two samples drawn form the same population will usually give different estimates of the population parameters.
-
- # Sampling
collapsed:: true
- What is **non-probabilistic sampling**? #card
card-last-interval:: 29.04
card-repeats:: 4
card-ease-factor:: 2.56
card-next-schedule:: 2022-12-13T20:02:44.059Z
card-last-reviewed:: 2022-11-14T20:02:44.059Z
card-last-score:: 5
- **Non-probabilistic sampling** methods are techniques of obtaining a sample that is not chosen at random and may be subject to **sampling bias**.
- ## Simple Random Sample
- ### Difficulties:
- Obtaining a sampling frame (list of all experimental units).
- Possibly time consuming / expensive.
- Minority groups, by chance, may not be represented in the sample.
- ## Stratified Random Sampling #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.36
card-next-schedule:: 2022-11-18T00:00:00.000Z
card-last-reviewed:: 2022-11-17T20:19:40.391Z
card-last-score:: 1
- 1. Split entire population into **homogeneous groups**, called **strata**.
2. Take a Simple Random Sample from each stratum.
- ### Stratified VS Simple Random Sample
- Ensure representation from minority groups.
- Estimates of the population parameters per strata may be of interest.
- Possible reduction in cost per observation in the survey.
- Increased accuracy as reduced sampling error (less variation within a stratum).
-
- ### Difficulties
- Can you correctly allocate each individual to one & only one stratum?
- Should every group receive equal weight?
- What if some strata are more varied than others?
- Take into account mean, variance, and cost to get "optimal allocation".
- ## Cluster Sampling #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.36
card-next-schedule:: 2022-11-18T00:00:00.000Z
card-last-reviewed:: 2022-11-17T20:18:48.138Z
card-last-score:: 1
- Instead of randomly choosing individuals, a Simple Random Sample of collection or groups of individuals is taken.
- The population is broken up into regions or groups, usually a *natural partition*, called a **cluster**.
- Internally heterogeneous, homogeneous between the clusters.
- Clusters are assumed representation of the entire population.
- Small number of clusters are selected at random.
- Every individual within a cluster is observed.
-
- ### Cluster Over Stratified
- Sampling frame not necessarily needed.
- May be more practical and / or economical than Simple or Stratified Random Sampling.
- Will be biased if the entire cluster is not sampled.
- Careful if homogeneity within a cluster and heterogeneity between clusters as this can increase sample error.
- **Note:** In stratified sampling, all strata are sampled, while in cluster sampling only some clusters are sampled.
- # Studies & Experiments
- ## Observational Studies & Experiments #card
card-last-interval:: -1
card-repeats:: 1
card-ease-factor:: 2.46
card-next-schedule:: 2022-11-15T00:00:00.000Z
card-last-reviewed:: 2022-11-14T16:49:06.426Z
card-last-score:: 1
- In an **Observational Study**, data is collected only be *observing* what occurs.
- E.g., surveys, historical records.
- When researchers want to investigate **causal relationships**, it's best to conduct an experiment.
- Usually there will be both an explanatory variable & a response variable.
- Be wary of confounding variables.
- ## Designed (Comparative) Study
- An experiment allows us to prove a cause-and-effect relationship.
- The experimenter must identify:
- at least one **explanatory variable**, called a **factor** to manipulate.
- at least one **response** variable to measure.
- The experimenter must also control any other **nuisance factors** that could influence the response.
- e.g., weather, day of the week.
-