Skip to main content
SAGE
Search form
  • 00:07

    SPEAKER 1: In this video, we lookat the most fundamental principleof research-- uncertainty.

  • 00:13

    SPEAKER 2: There is always uncertaintyin statistical decisions.The second purpose of statistics is to quantify and understandthis uncertainty.

  • 00:23

    SPEAKER 1: Remember, uncertainty is unavoidableand occurs when we use a sample to make an inferenceabout a population.Uncertainty is the result of randomness and chancein samples.Any sample can contain any set of participants,and each sample from the same population will be different.This chance difference between a sample and the populationis called sampling error.

  • 00:47

    SPEAKER 1 [continued]: We must never forget that uncertaintysets an absolute limit on what we can claim to know.

  • 00:52

    SPEAKER 2: We can still explore uncertainty.There are two core procedures for exploring uncertainty.

  • 00:60

    SPEAKER 1: First, the sampling distributionis the distribution of all possible samplesfrom any specific population and research design.It shows us the uncertainty about the samplesthat a population will produce.And the likelihood function showsus the uncertainty about which a populationa given sample could have come from.

  • 01:19

    SPEAKER 2: Next, we look at measuring uncertainty.The standard error is defined as the standard deviationof the sampling distribution.It is therefore the spread of all possible sample effectsizes.The standard error corresponds to the standard deviationof all possible sampling errors.So the standard error is a way of describingthe range of samples a population and design mightproduce, the typical sampling errors.

  • 01:48

    SPEAKER 1: The confidence interval from the likelihoodfunction is a way to quantify the uncertaintyabout a population given a sample.So the confidence interval is a wayof describing the range of possible populationsthat a sample came from.

  • 02:03

    SPEAKER 2: Finally, we look at working with uncertainty.If a confidence interval includes 0,e.g., the limits are minus 0.13 and plus 0.162, then thiswould mean that the true population effectsize could be 0, as well as any other effectsize within the interval.

  • 02:26

    SPEAKER 1: In the real world, the population effect sizeshould be somewhere in the middle of the sampleeffect sizes the different researchers find wheninvestigating the same topic.If we have access to an unbiased range of studiesof a particular effect, then theywill contain over estimates and under estimatesof the population effect.If we only have access to significant results,we will see a biased selection thatwill tend to contain only over estimates of the populationeffect.

  • 02:55

    SPEAKER 2: This all means something very practical.When you do your own piece of research,expect to see this sampling error occur.The sample effect size you get could be large, good luck,or it could be small, bad luck.

  • 03:12

    SPEAKER 1: For more information, see study.sagepub.com/statisticsforpsychology.

Video Info

Publisher: SAGE Publications Ltd.

Publication Year: 2019

Video Type:Tutorial

Methods: Sampling, Sampling error, Standard error, Confidence intervals, Sampling distribution

Keywords: uncertainty and error

Segment Info

Segment Num.: 1

Persons Discussed:

Events Discussed:

Keywords:

Abstract

Uncertainty occurs when a sample is used to make an inference about a population, as a result of randomness and chance in samples. Two core procedures for exploring uncertainty are described.

Looks like you do not have access to this content.

Uncertainty

Uncertainty occurs when a sample is used to make an inference about a population, as a result of randomness and chance in samples. Two core procedures for exploring uncertainty are described.

Copy and paste the following HTML into your website